Abstract

   Metabolic RNA labeling with high-throughput single-cell RNA sequencing
   (scRNA-seq) enables precise measurement of gene expression dynamics in
   complex biological processes, such as cell state transitions and
   embryogenesis. This technique, which tags newly synthesized RNA for
   detection through induced base conversions, relies on conversion
   efficiency, RNA integrity, and transcript recovery. These factors are
   influenced by the chosen chemical conversion method and platform
   compatibility. Despite its potential, a comprehensive comparison of
   chemical methods and platform compatibility has been lacking. Here, we
   benchmark ten chemical conversion methods using the Drop-seq platform,
   analyzing 52,529 cells. We find that on-beads methods, particularly the
   meta-chloroperoxy-benzoic acid/2,2,2-trifluoroethylamine combination,
   outperform in-situ approaches. To assess in vivo applications, we apply
   these optimized methods to 9883 zebrafish embryonic cells during the
   maternal-to-zygotic transition, identifying and experimentally
   validating zygotically activated transcripts, which enhanced zygotic
   gene detection capabilities. Additionally, we evaluate two commercial
   platforms with higher capture efficiency and find that on-beads
   iodoacetamide chemistry is the most effective. Our results provide
   critical guidance for selecting optimal chemical methods and scRNA-seq
   platforms, advancing the study of RNA dynamics in complex biological
   systems.

   Subject terms: Transcriptomics, Embryogenesis, Gene regulation
     __________________________________________________________________

   Zhang, Peng, and colleagues benchmark metabolic RNA labelling
   chemistries for time-resolved single-cell RNA sequencing, improving in
   vivo detection of newly synthesized RNA during zebrafish embryogenesis.

Introduction

   Single-cell RNA sequencing (scRNA-seq) has revolutionized our
   understanding of cellular heterogeneity and transcriptomic complexity.
   However, traditional scRNA-seq methods often fail to capture the
   temporal dynamics of RNA. Recent advances in time-resolved
   high-throughput scRNA-seq using metabolic labeling have provided deeper
   insights into RNA dynamics in complex biological processes^[52]1–[53]8.

   In metabolic labeling assays, nucleoside analogs, such as 4-Thiouridine
   (4sU)^[54]9–[55]15, 5-Ethynyluridine (5EU)^[56]16–[57]20, and
   6-Thioguanosine (6sG)^[58]1,[59]21–[60]25, are rapidly incorporated
   into newly synthesized RNA, creating a chemical tag that can be
   detected via sequencing by identifying chemical-induced conversions.
   This strategy is applicable to a wide range of model organisms,
   including humans^[61]5,[62]26, mice^[63]2,[64]6,
   zebrafish^[65]7,[66]8,[67]27, and fruit flies^[68]14. Metabolic RNA
   labeling combined with scRNA-seq has significantly enhanced our ability
   to quantitatively analyze RNA synthesis and degradation. This approach
   has enabled key discoveries, such as understanding cell-cycle dynamics
   in cultured cells and organoids^[69]3,[70]4,[71]16,[72]26, tracking RNA
   during embryogenesis^[73]7,[74]8,[75]27,[76]28, investigating
   transcriptional bursting^[77]29, and identifying rapid transcriptional
   responses during viral infection^[78]1.

   Metabolic labeling involves several crucial steps: incorporating 4sU
   into newly synthesized RNA, performing chemical conversion reactions,
   and ensuring compatibility with high-throughput scRNA-seq platforms.
   Key chemical conversion methods include SLAM-seq^[79]1,[80]9, which
   uses an iodoacetamide (IAA)-based reaction; TimeLapse-seq^[81]2,[82]10,
   which utilizes 2,2,2-trifluoroethylamine (TFEA) with
   meta-chloroperoxy-benzoic acid (mCPBA)/sodium periodate (NaIO[4]) and
   NH[4]Cl-based reactions; and TUC-seq^[83]30, which involves osmium
   tetroxide (OsO[4]) and ammonium chloride (NH[4]Cl). These steps
   determine the assay’s efficiency, including conversion efficiencies as
   indicated by T-to-C substitution rates, and RNA recovery rates,
   indicated by the number of genes and transcripts detected per cell.
   Although statistical methods have been developed to correct some of
   these limitations^[84]2,[85]10,[86]12, improving 4sU labeling and
   T-to-C conversion efficiency at the experimental level remains crucial
   for achieving more reliable and consistent outcomes, advancing our
   understanding of RNA dynamics in various biological processes.

   Several methods have been developed to integrate metabolic labeling
   with high-throughput scRNA-seq platforms, such as scNT-seq^[87]2,
   scSLAM-seq^[88]1,[89]7, sci-fate^[90]3, sci-fate2^[91]4, 10×
   Genomics-based method^[92]8, and Well-TEMP-seq^[93]5,[94]6. These
   methods are built upon different scRNA-seq platforms, each with unique
   technical adaptations. For example, scNT-seq^[95]2 is based on the
   Drop-seq platform^[96]31 and leverages a microfluidic device, a
   strategy also implemented in the commercial 10× Genomics system^[97]32.
   Well-TEMP-seq^[98]5,[99]6 employs a microwell-based system, while
   sci-fate and sci-fate2 utilize the sci-RNA-seq
   approach^[100]3,[101]4,[102]33, which relies on multiple rounds of
   split-pool barcoding. The key distinction among these methods is the
   timing of chemical conversion, which occur either before or after
   single-cell encapsulation, potentially affecting conversion rates.
   scNT-seq^[103]2 relies on the home-brew Drop-seq platform^[104]31,
   while Well-TEMP-seq^[105]5,[106]6 relies on the Well-Paired-seq
   platform^[107]34, both of which use the same barcoded beads from
   Drop-seq and enable chemical conversion on naked capture RNA attached
   to barcoded beads after cell lysis. In contrast, sci-fate, sci-fate2
   and 10× Genomics-based methods employ in-situ IAA-based chemical
   conversion within cells before mRNA releasing from the intact cell
   (Supplementary Fig. [108]1). Compared to the relatively low cell
   capture rate of the home-brew Drop-seq platform (~5%)^[109]31, in-situ
   IAA chemical conversion coupled with commercial platforms, such as 10×
   Genomics^[110]32 and MGI C4^[111]35, with higher capture rates
   (~50%)^[112]32 can be more effective for studying unique biological
   systems, such as early-stage embryos, where only a limited number of
   cells are available^[113]8. Although these methods have been
   demonstrated in different cell lines and biological systems, they vary
   in conversion efficiency, RNA recovery rate, and compatibility with
   scRNA-seq platforms^[114]2,[115]3. Given these differences and their
   potential impact on RNA dynamics analysis, a systematic and unbiased
   comparison of chemical conversion methods and their compatibility with
   different single-cell platforms is needed.

   To address this gap, we tested ten chemical conversion methods with
   varying reagent components and buffer conditions (Fig. [116]1a, b),
   including comparisons of in-situ and on-beads conditions using the same
   cell line. Our work provides direct comparisons and recommendations for
   time-resolved scRNA-seq methods using metabolic labeling. We further
   demonstrated that the recommended method effectively identifies
   zygotically activated transcripts in zebrafish embryogenesis.
   Additionally, we compared the 10× Genomics and MGI C4, two commercial
   single-cell platforms with high cell capture efficiency to the
   home-brew Drop-seq platform. The results highlight the strengths and
   weaknesses of each system, offering guidance for selecting the most
   appropriate chemical reaction and single-cell platform.

Fig. 1. Experimental design for benchmarking chemical conversion methods.

   [117]Fig. 1
   [118]Open in a new tab

   a Workflow for high-throughput scRNA-seq using metabolic labeling in
   ZF4 cells. ZF4 cells were labeled with 4-thiouridine (4sU, 100 μM),
   followed by cell dissociation and fixation. Chemical conversion was
   performed either before or after single-cell encapsulation on the
   Drop-seq platform. Newly synthesized transcripts were detected via
   sequencing by identifying chemical-induced T-to-C substitutions. b
   Summary of the ten chemical conversion methods evaluated in this study,
   including key parameters such as the main reagent, buffer pH,
   temperature, reaction time, and relevant references. “In-situ” refers