scGlyco-seq: High-Fidelity Multiplexed Targeted Single-Cell RNA-Seq for Glycogene Profiling

Introduction: : Glycosylation is an abundant molecular modification of proteins and lipids that regulates all aspects of normal human physiology and disease. The biosynthesis of these complex carbohydrates or glycans is regulated by the synergistic action of ~200 enzymes called glycosyltransferases (GTs) [1]. The quantification of the GT transcripts is, however, not easily possible using traditional single-cell RNA sequencing (scRNA-seq) because the GTs only constitute ~1% of the human transcriptome and are expressed at low abundance [2]. While currently available target enrichment strategies such as CaptureSeq [3] and TEQUILA-seq [4] improve transcript detection, they involve complex workflows, high cost, and poor compatibility with droplet-based scRNA-seq platforms. To address these limitations, the current paper describes scGlyco-seq, an integrated computational and experimental workflow for multiplex, targeted quantification of glycogenes at single cell level. The method extends RACE (Rapid Amplification of cDNA Ends) principles for targeted sequencing of GT transcripts in a variety of cancer and blood cell types.

Materials and

Methods: : Computational algorithm: A python script was written to design multiplex PCR primers for amplifying human GT transcripts (Fig. 1A). The program integrates Primer3 software for targeted amplification of either the 3’ or 5’ transcript ends, with all primers having similar melting temperature and yielding similarly sized products. Transcript ID for each gene was selected based on either abundance in selected cell types using data from the CCLE (Cancer Cell Line Encyclopedia) or using MANE (Matched Annotation from NCBI and EML-EBI)-select transcript IDs from NCBI. Absence of sequence overlaps with other genes in the human transcriptome was confirmed. In this manner, 10 unique primers were designed for each target transcript. These transcripts were designed both for ‘outer round PCR’ that amplifies a large section of the gene and a subset ‘inner round PCR’ which results in smaller product. Genetic algorithm was used to select one primer per transcript, while minimizing primer homodimer and heterodimer formation based on GC content, 3’-overlap and binding free energy. Experimental

Methods: The designed primers were applied to various cancer cell lines and human blood cells. Studies were conducted to confirm design principles in both bulk and single-cell sequencing contexts. Primer performance was assessed using standard PCR, qPCR, fragment analysis, and next-generation sequencing with a custom data analysis pipeline. The platform was then applied to primary blood cells using a droplet-based single-cell system to enable targeted amplification of GT transcripts from bead-captured mRNA.

Results, Conclusions, and Discussions:: The computational algorithm reliably generated high-quality primer sets targeting both 3′ and 5′ transcript ends, enabling specific, high-fidelity amplification across a range of plex sizes without evidence of primer dimerization or non-specific product formation. Multiplex primer pools of increasing complexity (5-, 10-, 24-, 50, 89-, and 170-plex) were experimentally tested using nested PCR under optimized temperature, primer and template concentration conditions. As expected, real-time qPCR showed progressive increase in RFU values and corresponding reduction in Ct values as primer complexity increased. Electropherograms from fragment analysis demonstrated specific amplification exclusively at the expected product sizes, with signal intensity and product size diversity increasing proportionally with plex level (Fig. 1B). Next generation sequencing (NGS) quantified transcript count for individual genes using unique molecular identifiers (UMI) attached to primer handles. In one example (Fig. 1C), five glycosyltransferase transcripts (C1GALT1, OGA, OGT, ST3GAL1, and ST3GAL3) were amplified in plexes of increasing complexity. As seen, the relative transcript counts did not vary dramatically even upon increasing PCR complexity from 5-genes to 170-genes. Such consistent amplification efficiency without detectable dropout or bias was verified for 100+ primer sets. Glyco-seq analysis of six human cancer cell lines (HEK293, HL60, Colo205, MDA-MB-231, LS174T, and Calu3) demonstrated GT transcript variations across cell types (Fig. 1D). Principal component analysis (PCA) also showed that each cancer cell type yields distinct glycosylation profiles. There was also broad agreement between transcript values obtained using the new glyco-seq workflow and conventional whole transcriptome measurements which were undertaken in parallel (Pearson correlation coefficient > 0.75-0.8). The method could be extended to single-cell workflows by performing targeted amplification on bead-captured transcripts derived from PBMCs and neutrophils. When combined with CiTE-Seq workflows, cell identifications could be gleaned along with targeted transcript measurements. Results show that the expression of selected sialyltransferases is augmented in neutrophilic and monocytic populations. Distinct evolution of glycosylation patterns was also noted as naïve T-cell subtypes differentiate to terminally differentiated effectors. Overall, this study presents a scalable, high-fidelity computational framework for targeted single-cell glycogene profiling, adaptable to diverse biological contexts including immune monitoring, disease studies, and expanded applications beyond glycosylation.

Acknowledgements and/or References (Optional):: 1. Groth, T., et al., GlycoEnzOnto: a GlycoEnzyme pathway and molecular function ontology. Bioinformatics, 2022. 38(24): p. 5413-5420.
2. Chrysinas, P., et al., Cell- and tissue-specific glycosylation pathways informed by single-cell transcriptomics. NAR Genom Bioinform, 2024. 6(4): p. lqae169.
3. Mercer, T.R., et al., Targeted RNA sequencing reveals the deep complexity of the human transcriptome. Nature biotechnology, 2012. 30(1): p. 99-104.
4. Wang, F., et al., TEQUILA-seq: a versatile and low-cost method for targeted long-read RNA sequencing. Nature Communications, 2023. 14(1): p. 4760.