Ensemble of nucleic acid absolute quantitation modules for copy number variation detection and RNA profiling

Ethical Statement

The research complies with all relevant ethical regulations. All procedures performed in studies involving human participants were approved by Institutional Review Board at MD Anderson (protocols PA16-0507 and PA19-0375), and were in accordance with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all participants. Participants received no compensation, and the data was not used for any treatment decisions.

Oligonucleotides and Reagents

All oligonucleotides were purchased from Integrated DNA Technologies (100uM in IDTE, pH 8.0). Oligonucleotide sequences are provided in Supplementary Data 1 QASeq primer sequences. Primers in the 2-plex QASeq panel are dual-HPLC purified; primers in other panels are standard-desalted to reduce cost. Conversion yield can be slightly reduced using standard-desalted primer but the median conversion yield is still over 60%. Phusion High-Fidelity DNA polymerase and deoxynucleoside triphosphates (dNTPs) were purchased from New England Biolabs. PowerUp SYBR Green Master Mix was purchased from Thermo Fisher Scientific. iTaq Universal SYBR Green Supermix was purchased from Bio-Rad Laboratories. AMPure XP was purchased from Beckman Coulter. NGS index primers (NEBNext Multiplex Oligos for Illumina) were purchased from New England Biolabs.

QASeq protocol

Library preparation consists of three PCR reactions: UMI PCR, nested PCR and index PCR, all performed on a T100 Thermal Cycler (Bio-Rad). In UMI PCR, the DNA sample was mixed with 1U Phusion High-Fidelity DNA polymerase, Phusion HF buffer, forward and outer reverse primers (15 nM each), and dNTPs (0.2 mM each) to reach a total volume of 50 μL.

Thermal cycling started with 30 s at 98 °C, followed by two cycles of 10 s at 98 °C, 30 min at 63 °C and 15 s at 72 °C, and then two cycles of 10 s at 98 °C, 15 s at 63 °C and 15 s at 72 °C, finally five cycles of 10 s at 98 °C and 30 s at 71 °C. During the last 5 min of the second 30 min at 63 °C, 1.5 μM of each universal primer was added while keeping the reactions inside the thermal cycler. After UMI PCR, 1.6X AMPure XP beads purification was performed.

In nested PCR, the eluate from the previous step was mixed with PowerUp SYBR Green Master Mix (1X final concentration) and 15 nM each inner reverse primer. Thermal cycling started with 3 min at 95 °C, followed by 2 cycles of 10 s at 95 °C and 30 min at 60 °C. The PCR product was purified by 1.6X AMPure XP beads.

Next, index PCR was performed; the eluate from the previous step was mixed with iTaq Universal SYBR Green Supermix (1X final concentration) and 250 nM each NEBNext index primers. Thermal cycling started with a 3 min incubation step at 95 °C, followed by 25 cycles of 10 s at 95 °C and 30 s at 65 °C, and finally 2 min at 65 °C. After index PCR, double-side size selection (0.4X, 0.4X ratio) was performed. Libraries were normalized and loaded onto an Illumina sequencer.

DNA extracted from FF or blood samples was sheared to 150 bp peak length using Covaris LE220 Focused Ultrasonicator before library preparation.

In RNA QASeq, RNA sample was firstly reverse transcribed to cDNA as input for QASeq protocol. RNA was mixed with dNTP (0.5 mM), Murine RNase Inhibitor (8 U), M-MuLv buffer (1X), M-MuLV Reverse Transcriptase (8 U), and random hexamer (6 µM). The mixture was incubated at 25 °C for 5 min, at 42 °C for 60 min, and then inactivated at 65 °C for 20 min. The reaction mixture was directly used as input for UMI PCR without purification.

Sequencing was performed on HiSeq or NextSeq (Illumina) with 2 × 150 bp paired-end reads and dual 8 bp index. The recommended sequencing depth is 90,000X for 8.3 ng human DNA input (see Supplementary Information Note 6 for more details). QASeq replicates were performed by the same operator, using the same DNA input, with library preparation performed in the same day and sequenced in the same run.

NGS data processing

NGS adapter sequences were first removed from FASTQ data using custom Python code; alignment was performed using Bowtie2 software³⁹. UMI grouping and CNV analysis were performed using custom Matlab code; a detailed description of the algorithm can be found in Supplementary Note 2 and Note 3. Mutation analysis was performed using custom Python and Matlab code; a detailed description can be found in Supplementary Note 4.

Digital droplet PCR

ddPCR CNV Assays from Bio-Rad were used in this study. Specifically, ddPCR Copy Number Assay: ERBB2, Human (Fluorophore: FAM, UniqueAssayID: dHsaCP1000116) and ddPCR Copy Number Assay: AGO1 (EIF2C1), Human (Fluorophore: HEX, UniqueAssayID: dHsaCP2500349) were purchased. Reaction setup, thermal cycling conditions and data acquisition were performed according to Bio-Rad protocol for ddPCR Copy Number Variation Assays. 10 ng of input DNA were used for each reaction. The ddPCR replicates were performed by the same operator, using the same DNA input, with droplet generation and PCR reaction performed in the same day and analyzed by Droplet Reader in the same run.

Samples

Fresh frozen (FF) breast tissue samples from breast cancer patients were purchased from OriGene Technologies. ERBB2 status of the tumor tissue measured by immunohistochemistry (IHC) was obtained from the vendor. Genomic DNA from FF samples and buffy coat of blood samples was extracted using QIAamp DNA Mini (Qiagen) following the manufacturer’s protocol.

56 plasma samples from 15 ERBB2 + metastatic breast cancer patients in de-identified format were collected from MD Anderson Cancer Center. Patient Characteristics are summarized in Supplementary Note 7.

Cell-free DNA was extracted from plasma using QIAamp MinElute ccfDNA Mini Kit (Qiagen) following the manufacturer’s protocol. Samples were quantified by qPCR with Human cell-line gDNA NA18537 as reference. The concentration calculated from qPCR reflects the amplifiable DNA.

Normal human placenta FFPE was purchased from BioChain. Total RNA from FFPE was extracted using RNeasy FFPE Kit (Qiagen). Human liver total RNA was purchased from Takara Bio. Human whole blood samples from healthy people were purchased from Zen-Bio Inc. RNA from fresh total blood was extracted using Monarch Total RNA Miniprep Kit (New England BioLabs).

ERBB2-positive cell-line (SK-BR-3) DNA was in the National Institute of Standards and Technology (NIST) Standard Reference Material 2373, and was purchased from ATCC (product name: SRM NIST-2373).

Infer tumor ERBB2 ploidy from plasma

Here ({{{{{{rm{Fraction}}}}}}}left({{{{{{rm{tumor}}}}}}; {{{{{rm{clone}}}}}}; i}right)) is the fraction of circulation DNA derived from tumor subclone i in cfDNA, ({{{{{{rm{Ploidy}}}}}}}left({{{{{{rm{tumor}}}}}}; {{{{{rm{clone}}}}}}; i}right)) is the ERBB2 ploidy in pure tumor clone i, and(,{{{{{{rm{VAF}}}}}}}({{{{{{rm{tumor}}}}}}; {{{{{rm{clone}}}}}}; i})) is the mutation VAF in pure tumor clone i.

Considering the complexity of tumor heterogeneity, here we proof-of-concept demonstrate the feasibility of inferring tumor ERBB2 ploidy from plasma, with two assumptions being made: (1) there is only one subclone in the tumor or we take the average of the tumor tissue to treat tumor as a whole, and (2) the VAF in that pure tumor is 50% (monoallelic).

When there is only one clone or we take the average of the tumor, equations are converted to:

We took pathogenic mutation observed in baseline cfDNA with VAF between 1% and 30% for tumor fraction calculation, to avoid the influence of SNP. Baseline mutation in 2 patients (1834 and 3669) were identified. We hypothesized mutation VAF in pure tumor is 50% (monoallelic), so that:

If the VAF in pure tumor is 100% (biallelic), tumor fraction calculated from Eq. (5) is 2-fold of the true value. Overall, we envision the inferred mean tumor ERBB2 ploidy based on Eqs. (3) and (5) should be within two-fold of the true value to help estimate tumor tissue information from plasma cfDNA.

RNAseq

Library preparation was performed using NEBNext Ultra II RNA Library Prep Kit for Illumina. Ribosomal RNA depletion was performed using NEBNext rRNA Depletion Kit v2. Raw fastq reads were initially quality filtered using Trimmomatic v0.39. Specifically, individual reads were trimmed to the longest continuous segment for which phred quality score (Q) was ≥20 (Q  ≥  20 represents ∼99% accuracy per nucleotide position). Reads shorter than 50 bp after trimming were discarded. Next, libraries were aligned to the human reference genome (GRCh38) using bowtie2 v2.4.4. After alignment, sam files were sorted and converted to bam files using samtools v1.12. HTseqv0.13.5 in mode ‘intersection-strict’ and with additional parameter ‘–minaqual 1’ was used to estimate the number of reads that mapped to each gene of interest. Finally, StringTie v2.1.7 was used to calculate TPM-normalized gene abundance.

Nanostring and microarray

Extracted RNA samples were sent to Amsbio LLC for Nanostring test using nCounter Breast Cancer 360 V2 Panel, and were sent to UT Southwestern Medical Center Microarray Core Facility for GeneChip Human Transcriptome Array 2.0 (HTA 2.0) test.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.