Investigation of Different Library Preparation and Tissue of Origin Deconvolution Methods for Urine and Plasma cfDNA Methylome Analysis.
Author(s): Kueng N, Sidler D, Banz V, Largiadèr CR, Ng CKY, Amstutz U
Publication: Diagnostics (Basel), 2023, Vol. 13, Page
PubMed ID: 37568867 PubMed Review Paper? No
Purpose of Paper
This paper compared sequencing metrics, methylation rates, and the tissue of origin of cell-free DNA (cfDNA) isolated from the plasma and urine of transplant patients and healthy volunteers; cfDNA was analyzed by single-strand and double-strand next generation sequencing (NGS) methods.
Conclusion of Paper
More uniquely mapped deduplicated reads, a higher mapping rate, and lower duplication rate were obtained when double-strand libraries were used rather than single-strand libraries, but single-strand libraries had a higher CpG methylation rate. Further, within-read methylation bias was only observed in double-strand libraries. The percentages of cfDNA derived from each of the eight possible tissues of origin were comparable between the two library preparation methods. However, double-strand libraries had a higher percentage of cfDNA that could not be assigned to one of the eight tissues of origin than single-stranded libraries. Fragment level deconvolution using a signature representing 36 cell types identified 3 cell types that were highly represented in plasma cfDNA and 10 cell types that were more highly represented in urine cfDNA when double-stranded sequencing was employed than single-stranded sequencing.
The mapping rate was lower for urine than plasma regardless of whether single- or double-strand libraries were constructed, but the duplication rate was only higher in urine than plasma when single-strand libraries were used. In double-strand libraries, within-read methylation bias was higher and CpG methylation rate was lower for urine than plasma. The Jagged ends (Jagged Index-Unmethylated) detected during the end-repair step in the double-strand sequencing protocol were higher in urine than plasma cfDNA and in the plasma of the two healthy volunteers than the seven transplant patients.
Studies
-
Study Purpose
This study compared sequencing metrics, methylation rates, and the tissue of origin of cfDNA isolated from the plasma and urine of transplant patients and healthy volunteers that were analyzed by single-strand and double-strand NGS methods. Blood was collected from three liver and four kidney transplant patients and two healthy volunteers into K3 EDTA tubes and urine was collected from the liver and kidney transplant patients in LoBind tubes and preserved with Streck Cell-Free DNA Urine Preserve. Within 2 h of collection, plasma was separated by centrifugation at 2,000 g for 15 min followed by 3,800 g for 10 min and stored at −20°C until DNA extraction. Plasma was thawed at 4°C before extraction using the QIAamp Circulating Nucleic Acid Kit. Urine specimens were centrifuged at 3000 g for 15 min and stored at −20°C or −80°C. Urine supernatant was thawed at room temperature and DNA was extracted using the Quick-DNA Urine Kit or using a quaternary ammonium anion exchange resin-based Q Sepharose method. cfDNA was quantified with the Qubit 1X dsDNA HS Assay Kit and stored at -80°C. Single-strand and double-strand cfDNA libraries were prepared using the SRSLY NanoPlus Kit and the NEBNext Enzymatic Methyl-seq Kit, respectively, and sequenced on a Novaseq 6000. Single-strand libraries were prepared from 7 of the urine and 8 of the plasma cfDNA extracts. Sequencing data was deduplicated and methylation status was extracted using Bismark. To allow comparison between methods, the NEBNext data was down sampled to the same average coverage as the SRSLY data using samtools. Individual CpG site-based deconvolution was performed using the deconvolution algorithm published by Erger et al. with 25 reference tissues. Fragment-level deconvolution was performed using non-negative least squares fragment-level deconvolution algorithm and a novel reference atlas.
Summary of Findings:
More uniquely mapped deduplicated reads, a higher mapping rate, and lower duplication rate were obtained using double-strand libraries than single-strand libraries (39 million versus 30 million; 83% versus 78%, P<0.001; and 9.4% versus 7.3%, P<0.0001, respectively). When broken down by specimen type, the mapping rate was lower for urine than plasma for both single-strand (74% versus 82%, P<0.001) and double-strand libraries (81% versus 85%, P<0.01), but the duplication rate was only higher in urine compared to plasma when single-strand libraries were used (9.8% vs. 9.0%, P<0.05). The enzymatic conversion was high regardless of library preparation method (99.4% and 99.5% for single-strand and double-strand libraries, respectively). No within-read methylation bias was observed in single-strand libraries from either plasma or urine cfDNA. In contrast, double-strand libraries showed bias as reduced methylation was observed at the beginning and end of the second read in plasma cfDNA, decreasing methylation over read 1, and low but increasing methylation at the start of read 2 in cfDNA from urine. The CpG methylation rate was higher for single-strand sequencing than double-strand sequencing of plasma cfDNA (80.2% versus 75.8%, P≤0.001) and urine cfDNA (79.4% versus 56.0%, P≤0.0001), and the double-strand sequencing CpG methylation rate was much lower for urine than plasma (56.0% versus 75.8%, P≤0.0001). The percentage of cfDNA that was derived from each of eight tissues of origin was comparable between the two library preparation methods, but double-strand libraries had a higher percentage of cfDNA that could not be assigned (to one of the eight tissues of origin) than single-stranded libraries for both plasma (5.9% versus 0.04%, P≤0.0001) and urine cfDNA (32.3% versus 1.9%, P≤0.0001). Fragment level deconvolution using a signature representing 36 cell types identified 3 cell type signatures that were higher in plasma and 10 cell type signatures that were higher in urine cfDNA when double-stranded sequencing was employed than single-stranded sequencing. The Jagged ends (Jagged Index-Unmethylated) detected during the end-repair in the double-strand sequencing protocol were higher in urine than plasma cfDNA (63.7 versus 18.6) and in the plasma of the two healthy volunteers than the seven transplant patients (22.3 versus 17.6).
Biospecimens
Preservative Types
- Frozen
Diagnoses:
- Other diagnoses
- Normal
Platform:
Analyte Technology Platform DNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Preaquisition Diagnosis/ patient condition Transplant recipient
Healthy volunteer
Next generation sequencing Specific Technology platform Single-strand library
Double-strand library
Biospecimen Acquisition Biospecimen location Urine
Plasma