Periodic Reporting for period 1 - CancerFLCloneSeq (Clone-based full-length RNA-seq for early diagnosis of cancer)
Berichtszeitraum: 2022-12-01 bis 2024-05-31
The project aim is to overcome the challenge of intra-tumor genetic heterogeneity in cancer treatment, where drug resistance often arises due to the growth of pre-existing sub-clones with mutations or altered states. Early detection of these subclonal cells can improve cancer therapy, but current single-cell RNA sequencing (RNA-seq) has limitations, such as high dropout rates, which hinder accurate profiling of mutations and gene expression. Bulk sequencing, while commonly used, cannot detect mutations below a 5% frequency. We propose a novel 3D clone-based full-length RNA sequencing technology, which has shown improved sensitivity compared to standard single-cell RNA-seq. Applied to lung adenocarcinoma cells, this method identified previously undetectable cancer stem-like subpopulations, suggesting it could be a powerful tool for detecting rare mutations and improving personalized cancer treatment.
In order to capture all types of RNA present in cells while excluding ribosomal RNA, we employed reverse transcription (RT) with a combination of two types of primers. We computationally selected a set of 220 hexamers designed to selectively enrich for all non-rRNA targets. In silico alignment of these primers on the transcriptome revealed that, for every 100 nt of total RNA, 4.4 sequences align, capturing 99.9% of total RNA, thus providing ample priming coverage. (Supplementary Figure 1.A) To enhance priming efficiency, polyTN8 primer was added to the primer mixtures. The polyTN8 primer is composed of three parts: a polyT of 12 nucleotides, followed by CCC, and ending with a random octamer. The polyT anneals to the polyA tail of mRNA, with CCC serving as a bridge so that the random octamer can anneal with upstream part of the mRNA. Since mRNA in cells naturally forms secondary structures (Georgakopoulos-Soares, Parada, and Hemberg 2022), the folded RNA would be spatially closer to the polyA tail for N8 annealing. We integrated the rDS and polyTN8 primers to single cell barcode beads with handles for library preparation as well as Unique Molecule Identifier (UMI)s for PCR duplication removal. With droplet-based microfluidics platform, we introduced sc-rDSeq as a high-throughput single-cell, full-length total RNA sequencing method. (Figure 1 A) We implemented a ramping-cycling-short RT strategy to improve the reaction efficiency. The temperature is gradually ramped from 4°C to 50°C, with a corresponding hold time determined by the melting temperature of the assigned rDS sequences, and the ramping program is repeated for 20 cycles to enhance the annealing of primers that were not initially annealed. Furthermore, to prevent strand displacement that is known to occur during RT reactions (Martín-Alonso et al. 2020), the total reaction time for the RT (50°C) is limited to 5 minutes. With all these modifications, the RT efficiency was significantly improved
To capture all RNA types while excluding ribosomal RNA (rRNA), reverse transcription (RT) was performed using two primer types, including 220 hexamers selected to enrich non-rRNA targets. The primers provided 99.9% RNA coverage. Additionally, a polyTN8 primer enhanced priming efficiency by targeting the polyA tails of mRNA. These primers were integrated into barcode beads for single-cell RNA sequencing (sc-rDSeq) using a droplet-based microfluidics platform. A ramping-cycling RT strategy improved annealing and reaction efficiency. The sc-rDSeq method demonstrated high sensitivity for full-length RNA, outperforming inDrops in gene detection, RNA biotype coverage, and capturing unique molecules per cell.
Figure 1 capture. sc-rDSeq showed high sensitivity for full-length total RNA in single cell with high-throughput. (A) Single cell and barcode beads are co-encapsulated by microfluidics platform. The primers on the barcode beads are equipped with handles for library preparation and single cell barcode and ended with rDS (red) or polyN8 (blue), which target both polyA (+) and polyA (-) RNA transcripts from random places. (B) Gene-body coverage across all genes based on merged single cells data showing strong 3’ bias for inDrops, mostly uniform with a small 3’ bias for sc-rDSeq, and highly uniform with a small 5’ bias for reads came from rDS primers (excluding reads from polyTN8 primers). Each gene is divided into 40 segments and counts failing into each segment are calculated for coverage. Ribbon area shows the SEM around mean. (C) The boxplot displays nUMI/cell for sc-rDSeq and inDrop. On average, sc-rDSeq captures 90k nUMI/cell, inDrop captures 7.4k nUMI/cell. (D) Distribution of uniquely mapped reads in each type of genomic locations with merged single cells data. (E) Percentage of genes detected by sc-rDSeq compared to inDrops. 80% of the detected genes were detected by both methods, 17.3% were uniquely detected by sc-rDSeq and 2.8% were detected only by inDrops. (F) Number of genes by each type of the RNAs uniquely detected by sc-rDSeq and inDrops. (G) Distribution of single cells by the fraction of transcripts per biotype comparing sc-rDSeq to inDrops. Selected biotypes are mRNA, lncRNA, miRNA and sncRNA.
Figure 1 capture. sc-rDSeq showed high sensitivity for full-length total RNA in single cell with high-throughput. (A) Single cell and barcode beads are co-encapsulated by microfluidics platform. The primers on the barcode beads are equipped with handles for library preparation and single cell barcode and ended with rDS (red) or polyN8 (blue), which target both polyA (+) and polyA (-) RNA transcripts from random places. (B) Gene-body coverage across all genes based on merged single cells data showing strong 3’ bias for inDrops, mostly uniform with a small 3’ bias for sc-rDSeq, and highly uniform with a small 5’ bias for reads came from rDS primers (excluding reads from polyTN8 primers). Each gene is divided into 40 segments and counts failing into each segment are calculated for coverage. Ribbon area shows the SEM around mean. (C) The boxplot displays nUMI/cell for sc-rDSeq and inDrop. On average, sc-rDSeq captures 90k nUMI/cell, inDrop captures 7.4k nUMI/cell. (D) Distribution of uniquely mapped reads in each type of genomic locations with merged single cells data. (E) Percentage of genes detected by sc-rDSeq compared to inDrops. 80% of the detected genes were detected by both methods, 17.3% were uniquely detected by sc-rDSeq and 2.8% were detected only by inDrops. (F) Number of genes by each type of the RNAs uniquely detected by sc-rDSeq and inDrops. (G) Distribution of single cells by the fraction of transcripts per biotype comparing sc-rDSeq to inDrops. Selected biotypes are mRNA, lncRNA, miRNA and sncRNA.