This study was organized into two parts with regard to the experiments. The first part involved an evaluation of karyotypically defined chromosomally abnormal single-cell DNA deriving from tissues with chromosomal abnormalities. The second part involved a retrospective blinded assessment of WGA products, selected from 157 consecutive clinical PGS cycles performed on single blastomeres that were biopsied from cleavage-stage embryos in the period of July 2014–November 2016. The karyotype of parents was shown in Additional file 1: Table S4, and therefore, we can analyze whether the CNV in embryo is inherited or de novo. Written informed consent was obtained from all participants in this study. This study was approved by the Ethics Committee of Women’s Hospital, School of Medicine, Zhejiang University.
Artificial single-cell-like DNA sample preparation
We extracted DNA from abortive villi samples from 21 cases with chromosomal abnormalities. The types of chromosomal abnormalities are listed in Additional file 1: Table S1. And a sample of lymphocytes of a normal male is also used. The DNA concentration was determined by the Qubit dsDNA High Sensitivity Assay kit (Life Technologies, Carlsbad, USA). Then, DNAs were diluted to 15 pg/μl. Next, 1 μl DNA was prepared for whole genome amplification using two commercial kits based on DOP-PCR or SurePlex strategies. The chimeric samples were mixed using two DNAs of different karyotypes. Three nanograms of one DNA sample and 6 ng of the other were mixed and then diluted to 60 pg/μl, with 1 μl used for whole genome amplification. The kits used in this study were the Genome Plex® Single Cell WGA Kit (DOP-PCR) and the Rubicon Genomics Pico PLEX® WGA Kit (SurePlex). All of the experimental operations followed the manufacturer’s instructions.
WGA of single-cell genomic DNA with different WGA kits
Single blastomere was isolated as described previously. Briefly, following sufficient dissociation and dilution of cells, single cell was randomly picked up using a mouth pipette under a microscope and washed three times in phosphate-buffered saline to avoid exogenous DNA contamination, after which they were transferred into a PCR tube. Single-cell isolation was confirmed by microscopy to ensure that only one cell was inside each tube. After the sample collection, biopsied blastomere was subjected to WGA amplification for both SSP-PGD/S and array-CGH analysis using SurePlex kits (BlueGnome).
WGA products were processed according to the BlueGnome 24sure plus protocol. Cell lysis and amplification was performed following the manufacturer’s instructions using the SurePlex Amplification system (Bluegnome, Cambridge, UK). As a positive control, 2.5 μl of female control DNA (G1521; Promega; 187 ng/μl) was used at a concentration of 25 pg/μl. The blank was equal to 2.5 μl of PBS. All of the samples were purified using Agencourt® AMPure® XP (Beckman Coulter, cat. no. A63881) according to the manufacturer’s protocol. The concentration was measured using the Qubit dsDNA High Sensitivity Assay kit (Life technologies, Carlsbad, USA).
Array-CGH experiment and analysis
All of the samples, amplified by SurePlex, were analyzed using the 24-Sure+ array (Bluegnome). The CNAs observed on this array should also be observed on the reference array profile. The array-CGH was performed according to the 24-Sure+ Protocol (Bluegnome), with a male genomic DNA sample as the reference. Copy number calls automatically generated by the Bluefuse Multi (Bluegnome) were assessed manually. The BlueFuse algorithm was based on calculating the median log2 ratio of all of the chromosomes for the detection of gains and losses. A median log2 ratio value of 0.3 or more indicated chromosome gains, whereas values of − 0.3 or less indicated chromosome losses.
Cell lysis and amplification were performed using the GenomePlex Single Cell WGA Kit (Sigma-Aldrich WGA4, Darmstadt, Germany) following the manufacturer’s instructions. As a positive control, 1 μl of male control DNA was used with a concentration of 15 pg/μl. The DNA samples used in this step were the same as the positive control. The blank was 1 μl of PBS. All of the samples were purified and measured following the manufacturer’s protocol, as described previously.
WGA amplification products were ligated to sequencing adapters using kits in accordance with the manufacturers’ protocols. In brief, 200 ng of the WGA products was fragmented to an average size distribution of 150 bp with the S2 Focused Ultrasonicator with Adaptive Focused Acoustics (AFA) technology (Covaris, Woburn, USA). Subsequently, libraries of the fragmented samples were created using the Ion Xpress™ Plus Fragment Library Kit, following the manufacturer’s protocol. Sequencing was performed on Ion Proton (Thermo Fisher Scientific, MA, USA). A total of 15 samples with each a different index were multiplexed on one chip. Samples were pooled at 20 nM each and diluted to a final concentration of 20 pM. Sequencing of 15 samples at 150 bp on an Ion Proton should lead to average genome coverage of 0.25× per sample (5 M reads/sample).
Reads were aligned to the human genomic reference sequences (hg19) using the BWA. Reads that were unmapped or had multiple primary alignment records were filtered in the alignment file, using an in-house Perl script. Duplicate reads were also removed. To eliminate the effect of WGA and sequencing bias, a two-step correction process was applied. In the first step, LOESS regression of the reads ratio of each 20 kbbin was used to smooth the GC bias. Then, in the second step, bins of outliers were masked to remove bias from the WGA. Further, chromosomal aneuploidies were detected by the Z-score method in comparison with control samples. Z-score > 3 was considered to be trisomy, while Z-score < − 3 was considered to be monosomy. Candidate CNVs (copy number variations) were detected by the CBS algorithm based on ratios of each 500 kbbin, which were accumulated by corrected values of 20 kbbins. The segments that had a reads ratio of< 1.4 were considered to be microdeletions, while the segments that had a reads ratio of > 2.6 were considered to be microduplications. The CNVs were selected by P value < 0.01 for a 10,000 times random permutation. Embryos were considered to be “implantable” if there were no aneuploidies or CNVs.