- Email: [email protected]
A novel multiplex fluorescent competitive PCR for copy number variation detection
Genomics xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Genomics journal homepage: www.elsevier.com/locate/ygeno
Original Article
A novel multiplex fluorescent competitive PCR for copy number variation detection Ke Chena, Shuang-shuang Dongb,c,d, Nan Wue,f,g, Zhi-hong Wue,f,g, Yu-xun Zhouh, Kai Lih, ⁎⁎ ⁎ Feng Zhangb,c,d, , Jun-hua Xiaoh, a
College of Environmental Science and Engineering, Donghua University, Shanghai, China Obstetrics and Gynecology Hospital, State Key Laboratory of Genetic Engineering at School of Life Sciences, Institute of Reproduction and Development, Fudan University, Shanghai, China c Key Laboratory of Reproduction Regulation of NHFPC, Collaborative Innovation Center of Genetics and Development, Fudan University, Shanghai, China d Shanghai Key Laboratory of Female Reproductive Endocrine Related Diseases, Shanghai, China e Department of Orthopedic Surgery, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, China f Beijing Key Laboratory for Genetic Research of Skeletal Deformity, Beijing, China g Medical Research Center of Orthopedics, Chinese Academy of Medical Sciences, Beijing, China h Institute of Biological Sciences and Biotechnology, Donghua University, Shanghai, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: CNV NMFC-PCR Blunt hairpin primer
The copy number variation (CNV) is an important genetic marker in cancer and other diseases. To detect CNVs of specific genetic loci, the multiplex ligation-dependent probe amplification (MLPA) is an appropriate approach, but the experimental optimization and probe synthesis are still great challenges. The multiplex competitive PCR is an alternative method for CNV detection. However, the construction of internal competitive template and establishment of a stable multiplex PCR system are the main limiting factors for this method. Here, we introduce a novel multiplex fluorescent competitive PCR (NMFC-PCR) for detecting CNVs. In this method, the blunt hairpin primers are used to rapidly establish a stable multiplex PCR system due to the reduction of non-specific amplification, and limited cycles' amplification is used to obtain the internal competitive template instead of artificial synthesis. With this method, we tested 21 clinical samples with potential LIM homeobox 1 (LHX1) or Tbox 6 (TBX6) deletion. Every three segments located on the LHX1 and TBX6 were selected as the target regions, while two segments located on X-chromosome and five segments located on autosome were selected as the reference regions for detecting CNVs. The results showed that the gender information of 21 samples can be accurately inferred by the copy number ratio (CNR) of X-chromosomal reference region to autosomal reference region (X/A), and 2 samples had one copy of LHX1 and 9 samples had one copy of TBX6. To evaluate the accuracy of NMFC-PCR, 5 random samples with CNV were also detected by array-based comparative genomic hybridization (aCGH), and the results of aCGH were consistent with the NMFC-PCR results. To further assess the performance of NMFC-PCR, 60 normal samples were simultaneously tested. The results showed that the gender results were exactly the same as known information, and CNVs of LHX1 or TBX6 were not found. In conclusion, the method is a cheap, efficient, accurate, and convenient competitive PCR method for CNV detection.
1. Introduction
the CNVs are as important as SNPs and STRs in studies of genetic structure and gene function [1,2]. In order to confirm the gene copy numbers, many methods have been proposed, such as fluorescent in situ hybridization, multiplex amplifiable probe hybridization (MAPH), MLPA, multiplex competitive PCR, droplet digital PCR (ddPCR), aCGH and CNV-seq [3–8]. The aCGH and CNV-seq are the best choices for high throughput CNV detection. However, they are not suitable for CNV
Copy number variation (CNV), which is 1 kb or larger segment addition or deletion in chromosome compared with a reference genome, has been recognized as an underlying factor for specific gene function and human disease [1]. The CNV occurs in > 1% of the population, and 12% of the human genome is variable in copy number [2]. Therefore, ⁎
Corresponding author. Correspondence to: F. Zhang, Obstetrics and Gynecology Hospital, State Key Laboratory of Genetic Engineering at School of Life Sciences, Institute of Reproduction and Development, Fudan University, Shanghai, China. E-mail addresses: [email protected] (F. Zhang), [email protected] (J.-h. Xiao). ⁎⁎
https://doi.org/10.1016/j.ygeno.2018.11.029 Received 7 May 2018; Received in revised form 9 November 2018; Accepted 28 November 2018 0888-7543/ © 2018 Elsevier Inc. All rights reserved.
Please cite this article as: Chen, K., Genomics, https://doi.org/10.1016/j.ygeno.2018.11.029
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
University, Shanghai, China) (Supplementary Data 5). Similarly, one female standard sample was chosen to be the internal competitive sample for each normal sample.
detection of specific genetic loci with many samples because of the high cost [5]. The MAPH and MLPA can economically detect target regions CNVs, but they are limited by the oligonucleotide synthesis, experimental optimization, and non-specific hybridization [4]. The ddPCR is a precise and accurate measurement of copy number and is suitable for detecting the sample with a wide range of copy numbers [9,10]. For each reaction, ddPCR can only detect the CNV of several genes according to the number of fluorescence, and the sample amount of ddPCR is relatively large, usually 10–100 ng per reaction [6,10]. Owing to specificity, flexibility, reproducibility and cost-efficient of the PCR amplification, the multiplex competitive PCR is suitable for CNV detection of target regions with large-scale samples [11–13]. The principle of multiplex competitive PCR is the co-amplification of a test template and an internal competitive template [13,14]. After co-amplification in one reaction using the universal fluorescent primers, the mixed product is detected by capillary electrophoresis, and then the CNVs of the test template are inferred by the fluorescence peak ratio of the test template to internal competitive template [3,5,14]. However, the internal competitive template is often constructed by artificial synthesis or plasmid construction [15,16], and the multiplex competitive PCR, which amplifies multiplex reference regions and target regions simultaneously, needs a stable multiplex PCR system to avoid non-specific products [3,14,17–19]. In this report, we describe the NMFC-PCR for CNV detection (Fig. 1). This method includes two rounds PCR. The first round PCR is a multiplex PCR to construct the test template and the internal competitive template individually. Specifically, two kinds of blunt hairpin primers with 4 bases length difference are used to amplify test sample and internal competitive sample by only five cycles, respectively [18]. The products of the test sample and the internal competitive sample are defined as the test template and the internal competitive template. Then the test template and the internal competitive template are mixed in equal amount and purified by AMPureXP beads. The second round PCR is a competitive PCR for CNV detection. Specifically, the mixed and purified products of the first round PCR are co-amplified using universal fluorescent primers, and the products of universal fluorescent primers are directly detected by capillary electrophoresis. Then the fluorescence peak ratios (FPRs) of the test template to the internal competitive template are calculated. For detecting CNV of the target region, the FPR of target region is compared with the FPR of reference region with known copy number. Based on the principle of multiplex competitive PCR, the FPR of target region to reference region represents their CNR, and therefore the copy number of target region of the test sample can be inferred [3,11,13]. To assess the performance of NMFCPCR, 21 clinical samples which have potential LHX1 or TBX6 deletion were tested, and the CNV detection results of 5 random samples with CNV were confirmed by aCGH. Further, additional 60 normal samples were also tested through the NMFC-PCR, and the results showed that gender information of 60 normal samples can be accurately inferred and CNVs of LHX1 or TBX6 were not found (Supplementary Table 5–6).
2.2. Primer design To detect CNVs of target regions, the reference regions are necessary for competitive PCR [3,11]. Six genes were selected as reference regions, which were ATP11C (Chromosome X), BRCA1 (Chromosome 17), CCDC132 (Chromosome 7), HSD17B12 (Chromosome 11), MTHFR (Chromosome 1) and MTRR (Chromosome 5), and each gene had one or two segments to design primers (Supplementary Data 1). Furthermore, every three segments located on the LHX1 and TBX6 were selected as the target regions for accurate CNV detection (Supplementary Data 1). The sequences of all regions were downloaded from the UCSC Genome Browser (hg38; http://genome.ucsc.edu/). The specific blunt hairpin primers of multiplex PCR were designed according to previously reported research [18]. The blunt hairpin primer consists of three parts from the 3′-end to the 5′-end (Fig. 1). The first 3′-end part has about 20 bp constituting the target primer region. The second part is the universal sequence (Supplementary Table 3). The third part has about 13 bp complementary to the first part to form a blunt hairpin structure. The reverse primers of the test sample and internal competitive sample were the same (Supplementary Table 2). Nevertheless, the forward primers of the test sample had four inserted bases (ACTG) to separate the PCR products of the test sample and internal competitive sample because of different lengths (Fig. 1). The product sizes of all the regions were between 220 and 374 bp (Supplementary Table 1). The universal fluorescent primers were used to amplify the products of the specific blunt hairpin primers in the next step. All the primers were synthesized by Bioligo Technologies (Shanghai, China). The universal fluorescent primers were purified by HPLC, and the other primers were supplied as a standard desalting grade. 2.3. Two rounds PCR The first round PCR is a multiplex PCR for constructing the test template and the internal competitive template individually. The test sample and internal competitive sample were respectively amplified five cycles with two kinds of blunt hairpin primers (4 bases length difference) (Fig. 1). The 10 μl PCR included 1× reaction buffer (2 mM MgCl2, 200 μM each dNTP and stabilizers, Nuhighbio, Suzhou, China), 1 U of EzAmp™ MPX Taq DNA Polymerase (Nuhighbio, Suzhou, China), 0.02 μM each specific primer, 40 ng human genome DNA. The following cycling program was used for the PCR: 95 °C for 15 min, 5 cycles of [94 °C for 30 s, 60 °C for 4 min]. The products of the test sample and internal competitive sample were mixed in equal amount and purified with AMPureXP beads according to the manufacturer's instructions (Beckman Coulter Inc., Brea, CA USA). The second round PCR is a competitive PCR for CNV detection. The products of the test sample and internal competitive sample were coamplified using universal fluorescent primers. The 20 μl PCR included 1× reaction buffer (2 mM MgCl2, 200 μM each dNTP and stabilizers, Nuhighbio, Suzhou, China), 2 U of EzAmp™ MPX Taq DNA Polymerase (Nuhighbio, Suzhou, China), 1 μM each universal fluorescent primer, 4 μl purified PCR products. The following cycling program was used for the PCR: 95 °C for 15 min, 30 cycles of [94 °C for 30 s, 65 °C for 1 min, 72 °C for 30 s]. The products of universal fluorescent primers were directly detected by ABI 3730XL genetic analyzer (Carlsbad, CA, USA). Raw data were analyzed by GeneMapper v4.0 (ABI). The products of the test sample and internal competitive sample were identified by their fragment sizes. In general, the fluorescence peak represented fluorescence value; therefore the fluorescence peaks of all regions were exported into a Microsoft Excel file to calculate the fluorescent peak ratio
2. Materials and methods 2.1. DNA samples The 21 clinical human genome samples (13 males and 8 females) and 3 standard samples (1 male and 2 males) were from the State Key Laboratory of Genetic Engineering at School of Life Sciences (Fudan University, Shanghai, China). The 3 standard samples were utilized as a self-comparison both as the test sample and its own internal competitive sample. Meanwhile, the 21 test samples have potential CNV with LHX1 or TBX6 deletion, and one female standard sample was chosen to be the internal competitive sample for each test sample (Supplementary Data 2). The 60 normal samples (17 males and 43 females) were from the Molecular Genetics Research Group of Donghua University (Donghua 2
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Fig. 1. Schematic overview of NMFC-PCR. In order to simplify the schematic, only two reference regions and one target region are illustrated. X-chromosomal reference region, autosomal reference region and target region have 2 segments, 5 segments and 3 segments, respectively. (a) Primer design. The blunt hairpin primers are designed for the targeted region, reference region 1 located on X-chromosome, and reference sequence 2 located on autosome. Each blunt hairpin primer consists of loop sequences about 13 bp, universal 20 bp flanking sequences (red), and unique 20–21 bp targeted arms (green/black/yellow). To separate the PCR products of the internal competitive sample (C) and test sample (S), the four key bases (ACTG, orange) are added to the forward specific primers of the test sample. (b) Two-round PCR amplification. The female internal competitive sample (C) and male test sample (S) were amplified. In the first round PCR, the internal competitive sample and test sample are amplified only five cycles, respectively. The products of the test sample and the internal competitive sample are defined as the test template and the internal competitive template, and are mixed in equal amount and purified. In the second round PCR, the mixed and purified products are amplified using universal fluorescent primers, and the products of universal fluorescent primers are directly detected by capillary electrophoresis. (c) Data Analysis. The products of the test sample and internal competitive sample are identified by their fragment sizes, and the fluorescence peak ratio (FPR) of each region is calculated. Based on known gender information, the female internal competitive sample has two copies reference region 1 and 2 respectively, and the male test sample just has one copy reference region 1 and two copies reference region 2. Therefore, the value of FPR of reference region 1 to reference region 2 is 0.5 in theory (the actual value is around 0.5) and indicates the reference region 1 in the test sample is 1 copy. Then the value of FPR of targeted region to reference region 2 is calculated and displayed as 0.5 (the actual value is also around 0.5), meaning the targeted region is 1 copy. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
CNR of the test sample to the internal competitive sample (Supplementary Data 3).
(Supplementary Data 3). 2.4. Data analysis
2.4.2. Correction coefficients for correcting amplification bias Although there is only 4 bases difference between the test template and the internal competitive template, it still leads to an amplification bias in the second round PCR even using universal fluorescent primers. Therefore, the FPRs of the test template to the internal competitive
2.4.1. Calculation of fluorescence peak ratio (FPR) The FPRs of the test template to the internal competitive template were calculated for subsequent copy number calculations, and the FPR of the test template to the internal competitive template represents the 3
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Table 1 Correction coefficients of three standard samples. Region
ATP11C-1 ATP11C-2 BRCA1 CCDC132 HSD17B12 MTHFR-2 MTRR LHX1–1 LHX1–2 LHX1–3 TBX6–1 TBX6–2 TBX6–3 Average
Chromosome
X X 17 7 11 1 5 17 17 17 16 16 16
Sample 1
Sample 2
Sample 3
Average
Peak ratio
Peak ratio
Peak ratio
Peak ratio
0.9759 1.1947 0.8564 0.8628 0.9682 0.8099 0.7011 0.9465 0.7375 1.3962 0.8089 1.0308 1.1852
0.9423 0.8983 0.8304 0.8221 0.8831 0.7442 0.7338 0.8039 0.7128 1.1838 0.8486 0.9184 0.9807
0.9134 0.9107 0.8645 0.8115 0.8631 0.8038 0.7781 0.8929 0.7605 1.2646 0.7754 0.9282 1.0024
0.9439 1.0012 0.8504 0.8321 0.9048 0.7860 0.7377 0.8811 0.7369 1.2816 0.8110 0.9591 1.0561
SD
CV
0.0313 0.1677 0.0178 0.0271 0.0558 0.0363 0.0387 0.0720 0.0239 0.1072 0.0367 0.0623 0.1123 0.0607
3.315% 16.74% 2.095% 3.255% 6.165% 4.614% 5.241% 8.174% 3.238% 8.365% 4.524% 6.491% 10.63% 6.373%
chromosomal reference region; “dca” = data correction of autosomal reference region (Supplementary Data 4). The ranges of CNRs of X/A in males and females represents 1 copy and 2 copies, respectively (Supplementary Datas 2 and 4).
template need to be corrected by the correction coefficients before they are used for copy number calculation. The correction coefficients are measured using the standard samples as the test samples and its own internal competitive samples, and therefore the internal competitive sample and the test sample are completely identical (equal mixing) and the theoretical FPR of each segment should be 1. The fluorescence peak ratios of the three standard samples were calculated, and the average values were defined as the correction coefficients for correcting amplification bias (Table 1).
(5) CNR of T/A = PRdct/PRdca with “dct” = data correction of target region (Supplementary Data 4). (6) According to the CNV detection range, the copy number of target region is determined. The CNR of T/A falls within the range of CNRs of X/A in males, and represents 1 copy. The CNR of T/A falls within the range of CNRs of X/A in females, and represents 2 copies.
2.4.3. Data correction The correction coefficient represents the amplification bias between the test template and the internal competitive template. The process of data correction is to divide the original FPR with the correction coefficient, which is called data correction (Supplementary Data 4).
2.5. High-resolution assay of comparative genomic hybridization microarray The aCGH is the gold standard for detection of CNV [17]. To evaluate the CNV detection accuracy of multiplex fluorescent competitive PCR, the clinical samples 7, 13, 19, 20 and 21, which were suggested to have LHX1 or TBX6 deletion across all three target regions (Supplementary Table 4), were further analyzed by Agilent SurePrint G3 human 1 × 1 M microarray. DNA processing, microarray handling, and data analysis were conducted following the Agilent oligonucleotide CGH protocol (version 6.0) [20].
2.4.4. CNV detection range Even after data correction for correcting the amplification bias, the copy number ratio (CNR) of the target region to autosomal reference region (T/A) is still a value within a given range rather than an exact value of 0.5 or 1 due to systematic errors. The copy number of X/A is gender-related, and the male and female have 1 copy and 2 copies of Xchromosomal reference region, respectively. Therefore, we used the range of CNRs of X-chromosomal reference region to autosomal reference region (X/A) to determine the copy number of other regions. The range of CNRs of X/A in males represents 1 copy, and the CNR of T/ A falls within the males' range of CNRs of X/A meaning target region is 1 copy. The range of CNRs of X/A in females represents 2 copies, and the CNR of T/A falls within the females' range of CNRs of X/A meaning the target region is 2 copies (Supplementary Data 4). Copy number calculation of target region: The FPR of target region (the average value of three segments) to autosomal reference region (the average value of five segments located on autosome) represents the CNR of T/A, and then the copy number of target region is determined by CNV detection range. In short:
3. Results 3.1. Amplification bias and the correction coefficients Although there is 4 bases difference between the test template and the internal competitive template, it still caused small amplification bias. For data analysis, the average FPRs of three standard samples, which represented this amplification bias, were used as coefficient calculations for correcting the original FPR for each region (Supplementary Data 3–4). Each average peak ratio of correction coefficient was close to 1 (theoretical value) with a range of 0.7377 to 1.2816. The standard deviations (SDs) and coefficient of variations (CVs) of all regions were shown in Table 1.
(1) PR = Peakts/Peakic with “PR” = peak ratio; “ts” = test sample; “ic” = internal competitive sample (Supplementary Data 3). (2) PRc = average values of peak ratios of three standard samples with “c” = correction coefficients (Table 1). (3) PRdc = PR/PRc with “dc” = data correction (Supplementary Data 4). (4) CNR of X/A = (PRdcx/PRdca)males with “dcx” = data correction of X-
3.2. CNV detection range The FPRs of the 21 clinical samples were corrected by the correction coefficients, and then were used to calculate the CNR of X/A detailed in Materials and Methods (Supplementary Data 4). According to gender difference, the CNR of X/A had a clear-cut distinction between male and female corresponding to one and two X-chromosomes, respectively [9]. The results showed 13 clinical male test samples with a value of 0.4896 ± 0.0637 (range, 0.3889 to 0.5840) and 8 clinical female test 4
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
3.4. aCGH data To further evaluate the CNV detection accuracy of NMFC-PCR, the clinical samples 7, 13, 19, 20 and 21, which were detected LHX1 or TBX6 deletion, were confirmed by aCGH. The results of aCGH showed that clinical sample 13 had LHX1 deletion, and the clinical sample 7, 19, 20 and 21 had TBX6 deletion (Fig. 4, Supplementary Fig. 1). Therefore the results of aCGH were consistent with the NMFC-PCR results. 3.5. The CNV detection results of the 60 normal samples With the NMFC-PCR, the 60 normal samples were also tested (Supplementary Data 5–6). The results showed the gender information of 60 normal samples can be accurately inferred (Table 2), while the CNR of LHX1 or TBX6 to autosomal reference region in the 60 normal samples were not significantly different (Supplementary Table 5–6).
Fig. 2. The copy number ratios of X-chromosomal reference region to autosomal reference region for the 21 clinical test samples (13 males and 8 females). The square represents a clinical male sample, and the circle represents a clinical female sample. ***: p = 0.0002 < 0.001.
4. Discussion
samples with a value of 0.8404 ± 0.1000 (range, 0.7766 to 1.0594) (Fig. 2). Then the range of CNRs of X/A in the clinical males was used as the CNV detection range of 1 copy, and the same range in the clinical females was used as the CNV detection range of 2 copies.
We have presented the NMFC-PCR for CNVs detection of specific genetic loci. In this method, the blunt hairpin primers, which can effectively reduce non-specific products, are used to establish a stable and efficient multiplex PCR system [18]. The NMFC-PCR showed high stability and accuracy. For the three standard samples, the average SD of all regions was < 0.0607, and the average coefficient of variations (CVs) of all regions was < 6.373% (Table 1). For the 21 clinical samples, the average SD of all groups which had differences in CNRs compared with autosomal reference region was < 0.0696 (Figs. 2–3). For the 60 normal samples, the average SD of all groups which had differences in CNRs compared with autosomal reference region was < 0.0822 (Table 2). In this method, the test sample and internal competitive sample are respectively amplified by five cycles for constructing the test template and the internal competitive template, which is an important improvement for multiplex competitive PCR. Because the internal competitive template in the traditional multiplex competitive PCR is often constructed by artificial synthesis (long oligonucleotide), and needs to be accurately quantified and diluted to the concentration of the same order of magnitude as the test template (genomic DNA) [3,4]. However, the internal competitive template in the NMFC-PCR is the same order of magnitude as the test template through only five cycles of PCR to avoid complicated dilution and mixing processes, and the internal competitive template and the test template are all long oligonucleotides (the PCR products of the first round PCR) which can reduce the amplification bias. Meanwhile, the five cycles' amplification ensures that the CNR of the test template to the internal competitive template represents the CNR of the test sample
3.3. CNV detection of LHX1 and TBX6 for 21 clinical samples The CNR of T/A was calculated (Supplementary Table 4). With the CNV detection range, the copy numbers of LHX1 or TBX6 for 21 clinical test samples were determined (Fig. 3). For LHX1, the CNRs compared with autosomal reference region were divided into two groups with significant differences (Fig. 3a). The group 1 included clinical sample 8 and 13 with an average CNR of 0.4935 ± 0.0245 (range, 0.4762 to 0.5108), and the group 2 includes the 19 other clinical samples with an average CNR of 0.9792 ± 0.0900 (range, 0.8313 to 1.2159). According to the CNV detection range, the clinical sample 8 and 13 in the group 1 had LHX1 deletion, and the other clinical samples in group 2 had 2 copies of LHX1 (Fig. 3a). Similarly, the CNRs of TBX6 were also divided into two groups (Fig. 3b). The group 1 included clinical sample 7, 10, 11, 12, 16, 18, 19, 20 and 21 with an average CNR of 0.4689 ± 0.0585 (range, 0.4075 to 0.5823), and the group 2 includes the 12 other clinical samples with an average CNR of 1.0491 ± 0.1205 (range, 0.8299 to 1.2207). For copy number calculation, the clinical sample 7, 10, 11, 12, 16, 18, 19, 20 and 21 in group 1 had TBX6 deletion, and the other clinical samples in group 2 had 2 copies of TBX6 (Fig. 3b).
Fig. 3. The distribution of copy number ratios of LXH1 and TBX6 to autosomal reference region for 21 clinical samples. (a) Shown is the copy number ratio distribution of LHX1. **: p = 0.0095 < 0.01. (b) Shown is the copy number distribution of TBX6. The square represents a one copy sample, and the circle represents a two copies sample. ***: p = 0.0001 < 0.001. 5
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Fig. 4. CNV results of aCGH. For aCGH, the green (deletion), black (no obvious change), and red (addition) dots show the relative intensities of tested genomic [3]. The LHX1 or TBX6 is indicated by red arrow. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the CNV detection results were consistent with the aCGH results, which confirmed the accuracy of NMFC-PCR (Figs. 2–4). For the 60 normal samples, the average SD of all groups with differences in CNRs was < 0.0822, and the gender results of the 60 normal samples were exactly the same as known information and there are not significantly different in the CNR of LHX1 or TBX6 to autosomal reference region (Table 2). The results further confirmed the accuracy and stability of NMFC-PCR. The NMFC-PCR is a simple, cheap, and convenient method to detect CNV. However, the technical replication of testing sample or using more detection region located on the target gene can further improve the precision and accuracy of NMFC-PCR. In this report, the NMFC-PCR can accurately detect one copy deletion (Table 2). But if we need to detect the duplication of target region, we need a positive sample which is copy number gain and used as a control to determine CNV detection range. In the future, using more than one fluorescein can increase the detection throughput by capillary electrophoresis [4]. Meanwhile, the NMFC-PCR could be coupled with the next generation sequencing. The two kinds of the blunt hairpin primers would only have sequence difference instead of length difference in the first round PCR, and the adapter primers of the next generation sequencing could be used in the second round PCR replacing the universal fluorescent primers in the second round PCR [5,11]. Then the products of adapter primers could be sequenced on the next generation sequencing platform, and the read
Table 2 Groups of copy number ratios of X-chromosomal reference region, LHX1, and TBX6 to autosomal reference region in 60 normal samples. Gene ID
Group
Average (theoretical value)
SD
Range
X-chromosome
Male Female 1 1
0.5212 1.0375 1.0662 1.0224
0.0332 0.0858 0.1323 0.0775 0.0822
0.4655–0.5839 0.8774–1.2037 0.8131–1.3804 0.8787–1.1929
LHX1 TBX6 Average
(0.5000) (1.0000) (1.0000) (1.0000)
to the internal competitive sample, and the CNV detection results confirm this inference. During the competitive PCR (the second round PCR), there is amplification bias between the test sample template and the internal competitive template, even if they had only four bases (ACTG) differences (Supplementary Table 1). Therefore the FPR of the test template to the internal competitive template must be corrected by the correction coefficients to reduce the amplification bias (Supplementary Data 3–4). After PCR bias correction, we used the range of the CNR of X/A for CNV detection and correcting system errors. This copy number calculation was stable and accurate. For the 21 clinical samples, the average SD of all groups with differences in CNRs was < 0.0696, and
6
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
count ratio would replace the FPR for calculating CNV of target region [18].
[6]
5. Conclusions [7]
We have presented the NMFC-PCR method for CNV detection. The method is a cost-efficient and convenient competitive PCR method, which is suitable for CNV detection of multiplex loci with large-scale samples. Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ygeno.2018.11.029.
[10]
Acknowledgments
[11]
This work was supported by grants from the National Natural Science Foundation of China (Grant no.31772550).
[12]
Conflict of interest
[13]
[8] [9]
[14]
The authors declare that they have no conflict of interest. [15]
References
[16] [1] L. Feuk, A.R. Carson, S.W. Scherer, Structural variation in the human genome [J], Nat. Rev. Genet. 7 (2) (2006) 85–97, https://doi.org/10.1038/nrg1767. [2] P.J. Hastings, J.R. Lupski, S.M. Rosenberg, et al., Mechanisms of change in gene copy number [J], Nat. Rev. Genet. 10 (8) (2009) 551–564, https://doi.org/10. 1038/nrg2593. [3] R. Du, C. Lu, Z. Jiang, et al., Efficient typing of copy number variations in a segmental duplication-mediated rearrangement hotspot using multiplex competitive amplification [J], J. Hum. Genet. 57 (8) (2012) 545–551, https://doi.org/10.1038/ jhg.2012.66. [4] L.N. Sellner, G.R. Taylor, MLPA and MAPH: new techniques for detection of gene deletions [J], Hum. Mutat. 23 (5) (2004) 413–419, https://doi.org/10.1002/humu. 20035. [5] J. Duan, J.G. Zhang, H.W. Deng, et al., Comparative studies of copy number
[17]
[18] [19]
[20]
7
variation detection methods for next-generation sequencing technologies [J], PLoS ONE 8 (3) (2013) e59128, https://doi.org/10.1371/journal.pone.0059128. B.J. Hindson, K.D. Ness, D.A. Masquelier, et al., High-throughput droplet digital PCR system for absolute quantitation of DNA copy number [J], Anal. Chem. 83 (22) (2011) 8604–8610, https://doi.org/10.1021/ac202028g. C. Xie, M.T. Tammi, CNV-seq, a new method to detect copy number variation using high-throughput sequencing [J], BMC Bioinforma. 10 (80) (2009), https://doi.org/ 10.1186/1471-2105-10-80. S.A. Hardwick, I.W. Deveson, T.R. Mercer, Reference standards for next-generation sequencing [J], Nat. Rev. Genet. 18 (8) (2017) 473. E. Mazaika, J. Homsy, Digital droplet PCR: CNV analysis and other applications [J], Curr. Protoc. Human Genet. 82 (7) (2014), https://doi.org/10.1002/0471142905. hg0724s82 241–13. A.D. Bell, C.L. Usher, S.A. Mccarroll, Analyzing copy number variation with droplet digital PCR [J], Methods Mol. Biol. 1768 (2018) 143–160, https://doi.org/10. 1007/978-1-4939-7778-9_9. D. Goossens, L.N. Moens, E. Nelis, et al., Simultaneous mutation and copy number variation (CNV) detection by multiplex PCR-based GS-FLX sequencing [J], Hum. Mutat. 30 (3) (2009) 472–476, https://doi.org/10.1002/humu.20873. L. Mamanova, A.J. Coffey, C.E. Scott, et al., Target-enrichment strategies for nextgeneration sequencing [J], Nat. Methods 7 (2) (2010) 111–118, https://doi.org/10. 1038/nmeth.1419. L. Zentilin, M. Giacca, Competitive PCR for precise nucleic acid quantification [J], Nat. Protoc. 2 (9) (2007) 2092–2104, https://doi.org/10.1038/nprot.2007.299. G.L. You, Q.L. Ding, Y.L. Lu, et al., Characterization of large deletions in the F8 gene using multiple competitive amplification and the genome walking technique [J], J. Thromb. Haemost. 11 (6) (2013) 1103–1110. Y. Lu, Y. Xin, J. Dai, et al., Spectrum and origin of mutations in sporadic cases of haemophilia A in China [J], Haemophilia 24 (2) (2018) 291–298. N. Ding, X. Yang, L. Zhang, et al., Association of β-defensin gene copy number variations with ankylosing spondylitis in Chinese population: a case-control study [J], Jpn. J. Rheumatol. 26 (1) (2016) 146–150. Li J, Lin L-H, Wang J, Et Al., Quantitative analysis of multiple genes' expressions based on a novel competitive RT-PCR assay [J], Anal. Bioanal. Chem. 405 (4) (2013) 1353–1360. K. Chen, Y. Zhou, Li K, Et Al., Multiplex PCR with the blunt hairpin primers for next generation sequencing [J], Biotechnol. Bioprocess Eng. 22 (3) (2017) 347–351. D. Liang, Y. Peng, Lv W, Et Al., Copy number variation sequencing for comprehensive diagnosis of chromosome disease syndromes [J], J. Mol. Diagn. 16 (5) (2014) 519–526, https://doi.org/10.1016/j.jmoldx.2014.05.002. Y. Chen, L. Guo, J. Chen, et al., Genome-wide CNV analysis in mouse induced pluripotent stem cells reveals dosage effect of pluripotent factors on genome integrity [J], BMC Genomics 15 (1) (2014) 79.
Contents lists available at ScienceDirect
Genomics journal homepage: www.elsevier.com/locate/ygeno
Original Article
A novel multiplex fluorescent competitive PCR for copy number variation detection Ke Chena, Shuang-shuang Dongb,c,d, Nan Wue,f,g, Zhi-hong Wue,f,g, Yu-xun Zhouh, Kai Lih, ⁎⁎ ⁎ Feng Zhangb,c,d, , Jun-hua Xiaoh, a
College of Environmental Science and Engineering, Donghua University, Shanghai, China Obstetrics and Gynecology Hospital, State Key Laboratory of Genetic Engineering at School of Life Sciences, Institute of Reproduction and Development, Fudan University, Shanghai, China c Key Laboratory of Reproduction Regulation of NHFPC, Collaborative Innovation Center of Genetics and Development, Fudan University, Shanghai, China d Shanghai Key Laboratory of Female Reproductive Endocrine Related Diseases, Shanghai, China e Department of Orthopedic Surgery, Peking Union Medical College Hospital, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing, China f Beijing Key Laboratory for Genetic Research of Skeletal Deformity, Beijing, China g Medical Research Center of Orthopedics, Chinese Academy of Medical Sciences, Beijing, China h Institute of Biological Sciences and Biotechnology, Donghua University, Shanghai, China b
A R T I C LE I N FO
A B S T R A C T
Keywords: CNV NMFC-PCR Blunt hairpin primer
The copy number variation (CNV) is an important genetic marker in cancer and other diseases. To detect CNVs of specific genetic loci, the multiplex ligation-dependent probe amplification (MLPA) is an appropriate approach, but the experimental optimization and probe synthesis are still great challenges. The multiplex competitive PCR is an alternative method for CNV detection. However, the construction of internal competitive template and establishment of a stable multiplex PCR system are the main limiting factors for this method. Here, we introduce a novel multiplex fluorescent competitive PCR (NMFC-PCR) for detecting CNVs. In this method, the blunt hairpin primers are used to rapidly establish a stable multiplex PCR system due to the reduction of non-specific amplification, and limited cycles' amplification is used to obtain the internal competitive template instead of artificial synthesis. With this method, we tested 21 clinical samples with potential LIM homeobox 1 (LHX1) or Tbox 6 (TBX6) deletion. Every three segments located on the LHX1 and TBX6 were selected as the target regions, while two segments located on X-chromosome and five segments located on autosome were selected as the reference regions for detecting CNVs. The results showed that the gender information of 21 samples can be accurately inferred by the copy number ratio (CNR) of X-chromosomal reference region to autosomal reference region (X/A), and 2 samples had one copy of LHX1 and 9 samples had one copy of TBX6. To evaluate the accuracy of NMFC-PCR, 5 random samples with CNV were also detected by array-based comparative genomic hybridization (aCGH), and the results of aCGH were consistent with the NMFC-PCR results. To further assess the performance of NMFC-PCR, 60 normal samples were simultaneously tested. The results showed that the gender results were exactly the same as known information, and CNVs of LHX1 or TBX6 were not found. In conclusion, the method is a cheap, efficient, accurate, and convenient competitive PCR method for CNV detection.
1. Introduction
the CNVs are as important as SNPs and STRs in studies of genetic structure and gene function [1,2]. In order to confirm the gene copy numbers, many methods have been proposed, such as fluorescent in situ hybridization, multiplex amplifiable probe hybridization (MAPH), MLPA, multiplex competitive PCR, droplet digital PCR (ddPCR), aCGH and CNV-seq [3–8]. The aCGH and CNV-seq are the best choices for high throughput CNV detection. However, they are not suitable for CNV
Copy number variation (CNV), which is 1 kb or larger segment addition or deletion in chromosome compared with a reference genome, has been recognized as an underlying factor for specific gene function and human disease [1]. The CNV occurs in > 1% of the population, and 12% of the human genome is variable in copy number [2]. Therefore, ⁎
Corresponding author. Correspondence to: F. Zhang, Obstetrics and Gynecology Hospital, State Key Laboratory of Genetic Engineering at School of Life Sciences, Institute of Reproduction and Development, Fudan University, Shanghai, China. E-mail addresses: [email protected] (F. Zhang), [email protected] (J.-h. Xiao). ⁎⁎
https://doi.org/10.1016/j.ygeno.2018.11.029 Received 7 May 2018; Received in revised form 9 November 2018; Accepted 28 November 2018 0888-7543/ © 2018 Elsevier Inc. All rights reserved.
Please cite this article as: Chen, K., Genomics, https://doi.org/10.1016/j.ygeno.2018.11.029
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
University, Shanghai, China) (Supplementary Data 5). Similarly, one female standard sample was chosen to be the internal competitive sample for each normal sample.
detection of specific genetic loci with many samples because of the high cost [5]. The MAPH and MLPA can economically detect target regions CNVs, but they are limited by the oligonucleotide synthesis, experimental optimization, and non-specific hybridization [4]. The ddPCR is a precise and accurate measurement of copy number and is suitable for detecting the sample with a wide range of copy numbers [9,10]. For each reaction, ddPCR can only detect the CNV of several genes according to the number of fluorescence, and the sample amount of ddPCR is relatively large, usually 10–100 ng per reaction [6,10]. Owing to specificity, flexibility, reproducibility and cost-efficient of the PCR amplification, the multiplex competitive PCR is suitable for CNV detection of target regions with large-scale samples [11–13]. The principle of multiplex competitive PCR is the co-amplification of a test template and an internal competitive template [13,14]. After co-amplification in one reaction using the universal fluorescent primers, the mixed product is detected by capillary electrophoresis, and then the CNVs of the test template are inferred by the fluorescence peak ratio of the test template to internal competitive template [3,5,14]. However, the internal competitive template is often constructed by artificial synthesis or plasmid construction [15,16], and the multiplex competitive PCR, which amplifies multiplex reference regions and target regions simultaneously, needs a stable multiplex PCR system to avoid non-specific products [3,14,17–19]. In this report, we describe the NMFC-PCR for CNV detection (Fig. 1). This method includes two rounds PCR. The first round PCR is a multiplex PCR to construct the test template and the internal competitive template individually. Specifically, two kinds of blunt hairpin primers with 4 bases length difference are used to amplify test sample and internal competitive sample by only five cycles, respectively [18]. The products of the test sample and the internal competitive sample are defined as the test template and the internal competitive template. Then the test template and the internal competitive template are mixed in equal amount and purified by AMPureXP beads. The second round PCR is a competitive PCR for CNV detection. Specifically, the mixed and purified products of the first round PCR are co-amplified using universal fluorescent primers, and the products of universal fluorescent primers are directly detected by capillary electrophoresis. Then the fluorescence peak ratios (FPRs) of the test template to the internal competitive template are calculated. For detecting CNV of the target region, the FPR of target region is compared with the FPR of reference region with known copy number. Based on the principle of multiplex competitive PCR, the FPR of target region to reference region represents their CNR, and therefore the copy number of target region of the test sample can be inferred [3,11,13]. To assess the performance of NMFCPCR, 21 clinical samples which have potential LHX1 or TBX6 deletion were tested, and the CNV detection results of 5 random samples with CNV were confirmed by aCGH. Further, additional 60 normal samples were also tested through the NMFC-PCR, and the results showed that gender information of 60 normal samples can be accurately inferred and CNVs of LHX1 or TBX6 were not found (Supplementary Table 5–6).
2.2. Primer design To detect CNVs of target regions, the reference regions are necessary for competitive PCR [3,11]. Six genes were selected as reference regions, which were ATP11C (Chromosome X), BRCA1 (Chromosome 17), CCDC132 (Chromosome 7), HSD17B12 (Chromosome 11), MTHFR (Chromosome 1) and MTRR (Chromosome 5), and each gene had one or two segments to design primers (Supplementary Data 1). Furthermore, every three segments located on the LHX1 and TBX6 were selected as the target regions for accurate CNV detection (Supplementary Data 1). The sequences of all regions were downloaded from the UCSC Genome Browser (hg38; http://genome.ucsc.edu/). The specific blunt hairpin primers of multiplex PCR were designed according to previously reported research [18]. The blunt hairpin primer consists of three parts from the 3′-end to the 5′-end (Fig. 1). The first 3′-end part has about 20 bp constituting the target primer region. The second part is the universal sequence (Supplementary Table 3). The third part has about 13 bp complementary to the first part to form a blunt hairpin structure. The reverse primers of the test sample and internal competitive sample were the same (Supplementary Table 2). Nevertheless, the forward primers of the test sample had four inserted bases (ACTG) to separate the PCR products of the test sample and internal competitive sample because of different lengths (Fig. 1). The product sizes of all the regions were between 220 and 374 bp (Supplementary Table 1). The universal fluorescent primers were used to amplify the products of the specific blunt hairpin primers in the next step. All the primers were synthesized by Bioligo Technologies (Shanghai, China). The universal fluorescent primers were purified by HPLC, and the other primers were supplied as a standard desalting grade. 2.3. Two rounds PCR The first round PCR is a multiplex PCR for constructing the test template and the internal competitive template individually. The test sample and internal competitive sample were respectively amplified five cycles with two kinds of blunt hairpin primers (4 bases length difference) (Fig. 1). The 10 μl PCR included 1× reaction buffer (2 mM MgCl2, 200 μM each dNTP and stabilizers, Nuhighbio, Suzhou, China), 1 U of EzAmp™ MPX Taq DNA Polymerase (Nuhighbio, Suzhou, China), 0.02 μM each specific primer, 40 ng human genome DNA. The following cycling program was used for the PCR: 95 °C for 15 min, 5 cycles of [94 °C for 30 s, 60 °C for 4 min]. The products of the test sample and internal competitive sample were mixed in equal amount and purified with AMPureXP beads according to the manufacturer's instructions (Beckman Coulter Inc., Brea, CA USA). The second round PCR is a competitive PCR for CNV detection. The products of the test sample and internal competitive sample were coamplified using universal fluorescent primers. The 20 μl PCR included 1× reaction buffer (2 mM MgCl2, 200 μM each dNTP and stabilizers, Nuhighbio, Suzhou, China), 2 U of EzAmp™ MPX Taq DNA Polymerase (Nuhighbio, Suzhou, China), 1 μM each universal fluorescent primer, 4 μl purified PCR products. The following cycling program was used for the PCR: 95 °C for 15 min, 30 cycles of [94 °C for 30 s, 65 °C for 1 min, 72 °C for 30 s]. The products of universal fluorescent primers were directly detected by ABI 3730XL genetic analyzer (Carlsbad, CA, USA). Raw data were analyzed by GeneMapper v4.0 (ABI). The products of the test sample and internal competitive sample were identified by their fragment sizes. In general, the fluorescence peak represented fluorescence value; therefore the fluorescence peaks of all regions were exported into a Microsoft Excel file to calculate the fluorescent peak ratio
2. Materials and methods 2.1. DNA samples The 21 clinical human genome samples (13 males and 8 females) and 3 standard samples (1 male and 2 males) were from the State Key Laboratory of Genetic Engineering at School of Life Sciences (Fudan University, Shanghai, China). The 3 standard samples were utilized as a self-comparison both as the test sample and its own internal competitive sample. Meanwhile, the 21 test samples have potential CNV with LHX1 or TBX6 deletion, and one female standard sample was chosen to be the internal competitive sample for each test sample (Supplementary Data 2). The 60 normal samples (17 males and 43 females) were from the Molecular Genetics Research Group of Donghua University (Donghua 2
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Fig. 1. Schematic overview of NMFC-PCR. In order to simplify the schematic, only two reference regions and one target region are illustrated. X-chromosomal reference region, autosomal reference region and target region have 2 segments, 5 segments and 3 segments, respectively. (a) Primer design. The blunt hairpin primers are designed for the targeted region, reference region 1 located on X-chromosome, and reference sequence 2 located on autosome. Each blunt hairpin primer consists of loop sequences about 13 bp, universal 20 bp flanking sequences (red), and unique 20–21 bp targeted arms (green/black/yellow). To separate the PCR products of the internal competitive sample (C) and test sample (S), the four key bases (ACTG, orange) are added to the forward specific primers of the test sample. (b) Two-round PCR amplification. The female internal competitive sample (C) and male test sample (S) were amplified. In the first round PCR, the internal competitive sample and test sample are amplified only five cycles, respectively. The products of the test sample and the internal competitive sample are defined as the test template and the internal competitive template, and are mixed in equal amount and purified. In the second round PCR, the mixed and purified products are amplified using universal fluorescent primers, and the products of universal fluorescent primers are directly detected by capillary electrophoresis. (c) Data Analysis. The products of the test sample and internal competitive sample are identified by their fragment sizes, and the fluorescence peak ratio (FPR) of each region is calculated. Based on known gender information, the female internal competitive sample has two copies reference region 1 and 2 respectively, and the male test sample just has one copy reference region 1 and two copies reference region 2. Therefore, the value of FPR of reference region 1 to reference region 2 is 0.5 in theory (the actual value is around 0.5) and indicates the reference region 1 in the test sample is 1 copy. Then the value of FPR of targeted region to reference region 2 is calculated and displayed as 0.5 (the actual value is also around 0.5), meaning the targeted region is 1 copy. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
CNR of the test sample to the internal competitive sample (Supplementary Data 3).
(Supplementary Data 3). 2.4. Data analysis
2.4.2. Correction coefficients for correcting amplification bias Although there is only 4 bases difference between the test template and the internal competitive template, it still leads to an amplification bias in the second round PCR even using universal fluorescent primers. Therefore, the FPRs of the test template to the internal competitive
2.4.1. Calculation of fluorescence peak ratio (FPR) The FPRs of the test template to the internal competitive template were calculated for subsequent copy number calculations, and the FPR of the test template to the internal competitive template represents the 3
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Table 1 Correction coefficients of three standard samples. Region
ATP11C-1 ATP11C-2 BRCA1 CCDC132 HSD17B12 MTHFR-2 MTRR LHX1–1 LHX1–2 LHX1–3 TBX6–1 TBX6–2 TBX6–3 Average
Chromosome
X X 17 7 11 1 5 17 17 17 16 16 16
Sample 1
Sample 2
Sample 3
Average
Peak ratio
Peak ratio
Peak ratio
Peak ratio
0.9759 1.1947 0.8564 0.8628 0.9682 0.8099 0.7011 0.9465 0.7375 1.3962 0.8089 1.0308 1.1852
0.9423 0.8983 0.8304 0.8221 0.8831 0.7442 0.7338 0.8039 0.7128 1.1838 0.8486 0.9184 0.9807
0.9134 0.9107 0.8645 0.8115 0.8631 0.8038 0.7781 0.8929 0.7605 1.2646 0.7754 0.9282 1.0024
0.9439 1.0012 0.8504 0.8321 0.9048 0.7860 0.7377 0.8811 0.7369 1.2816 0.8110 0.9591 1.0561
SD
CV
0.0313 0.1677 0.0178 0.0271 0.0558 0.0363 0.0387 0.0720 0.0239 0.1072 0.0367 0.0623 0.1123 0.0607
3.315% 16.74% 2.095% 3.255% 6.165% 4.614% 5.241% 8.174% 3.238% 8.365% 4.524% 6.491% 10.63% 6.373%
chromosomal reference region; “dca” = data correction of autosomal reference region (Supplementary Data 4). The ranges of CNRs of X/A in males and females represents 1 copy and 2 copies, respectively (Supplementary Datas 2 and 4).
template need to be corrected by the correction coefficients before they are used for copy number calculation. The correction coefficients are measured using the standard samples as the test samples and its own internal competitive samples, and therefore the internal competitive sample and the test sample are completely identical (equal mixing) and the theoretical FPR of each segment should be 1. The fluorescence peak ratios of the three standard samples were calculated, and the average values were defined as the correction coefficients for correcting amplification bias (Table 1).
(5) CNR of T/A = PRdct/PRdca with “dct” = data correction of target region (Supplementary Data 4). (6) According to the CNV detection range, the copy number of target region is determined. The CNR of T/A falls within the range of CNRs of X/A in males, and represents 1 copy. The CNR of T/A falls within the range of CNRs of X/A in females, and represents 2 copies.
2.4.3. Data correction The correction coefficient represents the amplification bias between the test template and the internal competitive template. The process of data correction is to divide the original FPR with the correction coefficient, which is called data correction (Supplementary Data 4).
2.5. High-resolution assay of comparative genomic hybridization microarray The aCGH is the gold standard for detection of CNV [17]. To evaluate the CNV detection accuracy of multiplex fluorescent competitive PCR, the clinical samples 7, 13, 19, 20 and 21, which were suggested to have LHX1 or TBX6 deletion across all three target regions (Supplementary Table 4), were further analyzed by Agilent SurePrint G3 human 1 × 1 M microarray. DNA processing, microarray handling, and data analysis were conducted following the Agilent oligonucleotide CGH protocol (version 6.0) [20].
2.4.4. CNV detection range Even after data correction for correcting the amplification bias, the copy number ratio (CNR) of the target region to autosomal reference region (T/A) is still a value within a given range rather than an exact value of 0.5 or 1 due to systematic errors. The copy number of X/A is gender-related, and the male and female have 1 copy and 2 copies of Xchromosomal reference region, respectively. Therefore, we used the range of CNRs of X-chromosomal reference region to autosomal reference region (X/A) to determine the copy number of other regions. The range of CNRs of X/A in males represents 1 copy, and the CNR of T/ A falls within the males' range of CNRs of X/A meaning target region is 1 copy. The range of CNRs of X/A in females represents 2 copies, and the CNR of T/A falls within the females' range of CNRs of X/A meaning the target region is 2 copies (Supplementary Data 4). Copy number calculation of target region: The FPR of target region (the average value of three segments) to autosomal reference region (the average value of five segments located on autosome) represents the CNR of T/A, and then the copy number of target region is determined by CNV detection range. In short:
3. Results 3.1. Amplification bias and the correction coefficients Although there is 4 bases difference between the test template and the internal competitive template, it still caused small amplification bias. For data analysis, the average FPRs of three standard samples, which represented this amplification bias, were used as coefficient calculations for correcting the original FPR for each region (Supplementary Data 3–4). Each average peak ratio of correction coefficient was close to 1 (theoretical value) with a range of 0.7377 to 1.2816. The standard deviations (SDs) and coefficient of variations (CVs) of all regions were shown in Table 1.
(1) PR = Peakts/Peakic with “PR” = peak ratio; “ts” = test sample; “ic” = internal competitive sample (Supplementary Data 3). (2) PRc = average values of peak ratios of three standard samples with “c” = correction coefficients (Table 1). (3) PRdc = PR/PRc with “dc” = data correction (Supplementary Data 4). (4) CNR of X/A = (PRdcx/PRdca)males with “dcx” = data correction of X-
3.2. CNV detection range The FPRs of the 21 clinical samples were corrected by the correction coefficients, and then were used to calculate the CNR of X/A detailed in Materials and Methods (Supplementary Data 4). According to gender difference, the CNR of X/A had a clear-cut distinction between male and female corresponding to one and two X-chromosomes, respectively [9]. The results showed 13 clinical male test samples with a value of 0.4896 ± 0.0637 (range, 0.3889 to 0.5840) and 8 clinical female test 4
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
3.4. aCGH data To further evaluate the CNV detection accuracy of NMFC-PCR, the clinical samples 7, 13, 19, 20 and 21, which were detected LHX1 or TBX6 deletion, were confirmed by aCGH. The results of aCGH showed that clinical sample 13 had LHX1 deletion, and the clinical sample 7, 19, 20 and 21 had TBX6 deletion (Fig. 4, Supplementary Fig. 1). Therefore the results of aCGH were consistent with the NMFC-PCR results. 3.5. The CNV detection results of the 60 normal samples With the NMFC-PCR, the 60 normal samples were also tested (Supplementary Data 5–6). The results showed the gender information of 60 normal samples can be accurately inferred (Table 2), while the CNR of LHX1 or TBX6 to autosomal reference region in the 60 normal samples were not significantly different (Supplementary Table 5–6).
Fig. 2. The copy number ratios of X-chromosomal reference region to autosomal reference region for the 21 clinical test samples (13 males and 8 females). The square represents a clinical male sample, and the circle represents a clinical female sample. ***: p = 0.0002 < 0.001.
4. Discussion
samples with a value of 0.8404 ± 0.1000 (range, 0.7766 to 1.0594) (Fig. 2). Then the range of CNRs of X/A in the clinical males was used as the CNV detection range of 1 copy, and the same range in the clinical females was used as the CNV detection range of 2 copies.
We have presented the NMFC-PCR for CNVs detection of specific genetic loci. In this method, the blunt hairpin primers, which can effectively reduce non-specific products, are used to establish a stable and efficient multiplex PCR system [18]. The NMFC-PCR showed high stability and accuracy. For the three standard samples, the average SD of all regions was < 0.0607, and the average coefficient of variations (CVs) of all regions was < 6.373% (Table 1). For the 21 clinical samples, the average SD of all groups which had differences in CNRs compared with autosomal reference region was < 0.0696 (Figs. 2–3). For the 60 normal samples, the average SD of all groups which had differences in CNRs compared with autosomal reference region was < 0.0822 (Table 2). In this method, the test sample and internal competitive sample are respectively amplified by five cycles for constructing the test template and the internal competitive template, which is an important improvement for multiplex competitive PCR. Because the internal competitive template in the traditional multiplex competitive PCR is often constructed by artificial synthesis (long oligonucleotide), and needs to be accurately quantified and diluted to the concentration of the same order of magnitude as the test template (genomic DNA) [3,4]. However, the internal competitive template in the NMFC-PCR is the same order of magnitude as the test template through only five cycles of PCR to avoid complicated dilution and mixing processes, and the internal competitive template and the test template are all long oligonucleotides (the PCR products of the first round PCR) which can reduce the amplification bias. Meanwhile, the five cycles' amplification ensures that the CNR of the test template to the internal competitive template represents the CNR of the test sample
3.3. CNV detection of LHX1 and TBX6 for 21 clinical samples The CNR of T/A was calculated (Supplementary Table 4). With the CNV detection range, the copy numbers of LHX1 or TBX6 for 21 clinical test samples were determined (Fig. 3). For LHX1, the CNRs compared with autosomal reference region were divided into two groups with significant differences (Fig. 3a). The group 1 included clinical sample 8 and 13 with an average CNR of 0.4935 ± 0.0245 (range, 0.4762 to 0.5108), and the group 2 includes the 19 other clinical samples with an average CNR of 0.9792 ± 0.0900 (range, 0.8313 to 1.2159). According to the CNV detection range, the clinical sample 8 and 13 in the group 1 had LHX1 deletion, and the other clinical samples in group 2 had 2 copies of LHX1 (Fig. 3a). Similarly, the CNRs of TBX6 were also divided into two groups (Fig. 3b). The group 1 included clinical sample 7, 10, 11, 12, 16, 18, 19, 20 and 21 with an average CNR of 0.4689 ± 0.0585 (range, 0.4075 to 0.5823), and the group 2 includes the 12 other clinical samples with an average CNR of 1.0491 ± 0.1205 (range, 0.8299 to 1.2207). For copy number calculation, the clinical sample 7, 10, 11, 12, 16, 18, 19, 20 and 21 in group 1 had TBX6 deletion, and the other clinical samples in group 2 had 2 copies of TBX6 (Fig. 3b).
Fig. 3. The distribution of copy number ratios of LXH1 and TBX6 to autosomal reference region for 21 clinical samples. (a) Shown is the copy number ratio distribution of LHX1. **: p = 0.0095 < 0.01. (b) Shown is the copy number distribution of TBX6. The square represents a one copy sample, and the circle represents a two copies sample. ***: p = 0.0001 < 0.001. 5
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
Fig. 4. CNV results of aCGH. For aCGH, the green (deletion), black (no obvious change), and red (addition) dots show the relative intensities of tested genomic [3]. The LHX1 or TBX6 is indicated by red arrow. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the CNV detection results were consistent with the aCGH results, which confirmed the accuracy of NMFC-PCR (Figs. 2–4). For the 60 normal samples, the average SD of all groups with differences in CNRs was < 0.0822, and the gender results of the 60 normal samples were exactly the same as known information and there are not significantly different in the CNR of LHX1 or TBX6 to autosomal reference region (Table 2). The results further confirmed the accuracy and stability of NMFC-PCR. The NMFC-PCR is a simple, cheap, and convenient method to detect CNV. However, the technical replication of testing sample or using more detection region located on the target gene can further improve the precision and accuracy of NMFC-PCR. In this report, the NMFC-PCR can accurately detect one copy deletion (Table 2). But if we need to detect the duplication of target region, we need a positive sample which is copy number gain and used as a control to determine CNV detection range. In the future, using more than one fluorescein can increase the detection throughput by capillary electrophoresis [4]. Meanwhile, the NMFC-PCR could be coupled with the next generation sequencing. The two kinds of the blunt hairpin primers would only have sequence difference instead of length difference in the first round PCR, and the adapter primers of the next generation sequencing could be used in the second round PCR replacing the universal fluorescent primers in the second round PCR [5,11]. Then the products of adapter primers could be sequenced on the next generation sequencing platform, and the read
Table 2 Groups of copy number ratios of X-chromosomal reference region, LHX1, and TBX6 to autosomal reference region in 60 normal samples. Gene ID
Group
Average (theoretical value)
SD
Range
X-chromosome
Male Female 1 1
0.5212 1.0375 1.0662 1.0224
0.0332 0.0858 0.1323 0.0775 0.0822
0.4655–0.5839 0.8774–1.2037 0.8131–1.3804 0.8787–1.1929
LHX1 TBX6 Average
(0.5000) (1.0000) (1.0000) (1.0000)
to the internal competitive sample, and the CNV detection results confirm this inference. During the competitive PCR (the second round PCR), there is amplification bias between the test sample template and the internal competitive template, even if they had only four bases (ACTG) differences (Supplementary Table 1). Therefore the FPR of the test template to the internal competitive template must be corrected by the correction coefficients to reduce the amplification bias (Supplementary Data 3–4). After PCR bias correction, we used the range of the CNR of X/A for CNV detection and correcting system errors. This copy number calculation was stable and accurate. For the 21 clinical samples, the average SD of all groups with differences in CNRs was < 0.0696, and
6
Genomics xxx (xxxx) xxx–xxx
K. Chen et al.
count ratio would replace the FPR for calculating CNV of target region [18].
[6]
5. Conclusions [7]
We have presented the NMFC-PCR method for CNV detection. The method is a cost-efficient and convenient competitive PCR method, which is suitable for CNV detection of multiplex loci with large-scale samples. Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ygeno.2018.11.029.
[10]
Acknowledgments
[11]
This work was supported by grants from the National Natural Science Foundation of China (Grant no.31772550).
[12]
Conflict of interest
[13]
[8] [9]
[14]
The authors declare that they have no conflict of interest. [15]
References
[16] [1] L. Feuk, A.R. Carson, S.W. Scherer, Structural variation in the human genome [J], Nat. Rev. Genet. 7 (2) (2006) 85–97, https://doi.org/10.1038/nrg1767. [2] P.J. Hastings, J.R. Lupski, S.M. Rosenberg, et al., Mechanisms of change in gene copy number [J], Nat. Rev. Genet. 10 (8) (2009) 551–564, https://doi.org/10. 1038/nrg2593. [3] R. Du, C. Lu, Z. Jiang, et al., Efficient typing of copy number variations in a segmental duplication-mediated rearrangement hotspot using multiplex competitive amplification [J], J. Hum. Genet. 57 (8) (2012) 545–551, https://doi.org/10.1038/ jhg.2012.66. [4] L.N. Sellner, G.R. Taylor, MLPA and MAPH: new techniques for detection of gene deletions [J], Hum. Mutat. 23 (5) (2004) 413–419, https://doi.org/10.1002/humu. 20035. [5] J. Duan, J.G. Zhang, H.W. Deng, et al., Comparative studies of copy number
[17]
[18] [19]
[20]
7
variation detection methods for next-generation sequencing technologies [J], PLoS ONE 8 (3) (2013) e59128, https://doi.org/10.1371/journal.pone.0059128. B.J. Hindson, K.D. Ness, D.A. Masquelier, et al., High-throughput droplet digital PCR system for absolute quantitation of DNA copy number [J], Anal. Chem. 83 (22) (2011) 8604–8610, https://doi.org/10.1021/ac202028g. C. Xie, M.T. Tammi, CNV-seq, a new method to detect copy number variation using high-throughput sequencing [J], BMC Bioinforma. 10 (80) (2009), https://doi.org/ 10.1186/1471-2105-10-80. S.A. Hardwick, I.W. Deveson, T.R. Mercer, Reference standards for next-generation sequencing [J], Nat. Rev. Genet. 18 (8) (2017) 473. E. Mazaika, J. Homsy, Digital droplet PCR: CNV analysis and other applications [J], Curr. Protoc. Human Genet. 82 (7) (2014), https://doi.org/10.1002/0471142905. hg0724s82 241–13. A.D. Bell, C.L. Usher, S.A. Mccarroll, Analyzing copy number variation with droplet digital PCR [J], Methods Mol. Biol. 1768 (2018) 143–160, https://doi.org/10. 1007/978-1-4939-7778-9_9. D. Goossens, L.N. Moens, E. Nelis, et al., Simultaneous mutation and copy number variation (CNV) detection by multiplex PCR-based GS-FLX sequencing [J], Hum. Mutat. 30 (3) (2009) 472–476, https://doi.org/10.1002/humu.20873. L. Mamanova, A.J. Coffey, C.E. Scott, et al., Target-enrichment strategies for nextgeneration sequencing [J], Nat. Methods 7 (2) (2010) 111–118, https://doi.org/10. 1038/nmeth.1419. L. Zentilin, M. Giacca, Competitive PCR for precise nucleic acid quantification [J], Nat. Protoc. 2 (9) (2007) 2092–2104, https://doi.org/10.1038/nprot.2007.299. G.L. You, Q.L. Ding, Y.L. Lu, et al., Characterization of large deletions in the F8 gene using multiple competitive amplification and the genome walking technique [J], J. Thromb. Haemost. 11 (6) (2013) 1103–1110. Y. Lu, Y. Xin, J. Dai, et al., Spectrum and origin of mutations in sporadic cases of haemophilia A in China [J], Haemophilia 24 (2) (2018) 291–298. N. Ding, X. Yang, L. Zhang, et al., Association of β-defensin gene copy number variations with ankylosing spondylitis in Chinese population: a case-control study [J], Jpn. J. Rheumatol. 26 (1) (2016) 146–150. Li J, Lin L-H, Wang J, Et Al., Quantitative analysis of multiple genes' expressions based on a novel competitive RT-PCR assay [J], Anal. Bioanal. Chem. 405 (4) (2013) 1353–1360. K. Chen, Y. Zhou, Li K, Et Al., Multiplex PCR with the blunt hairpin primers for next generation sequencing [J], Biotechnol. Bioprocess Eng. 22 (3) (2017) 347–351. D. Liang, Y. Peng, Lv W, Et Al., Copy number variation sequencing for comprehensive diagnosis of chromosome disease syndromes [J], J. Mol. Diagn. 16 (5) (2014) 519–526, https://doi.org/10.1016/j.jmoldx.2014.05.002. Y. Chen, L. Guo, J. Chen, et al., Genome-wide CNV analysis in mouse induced pluripotent stem cells reveals dosage effect of pluripotent factors on genome integrity [J], BMC Genomics 15 (1) (2014) 79.