Targeted sequencing of mutations via RNA-templated gap filling of oligonucleotides for single-cell RNA-seq

This paper presents a novel method that utilizes BstFL DNA polymerase to perform RNA-templated gap filling and ligation, enabling simultaneous whole-transcriptome and targeted mutation profiling within single cells.

The Gist

Imagine you are trying to solve a massive jigsaw puzzle inside a tiny, bustling city (a single cell). You want to find specific, rare pieces that are slightly different from the rest (mutations) to understand what makes that city unique. However, the city is so crowded and the pieces are so small that finding these specific "odd" pieces is incredibly hard. Usually, you can only see the edges of the puzzle, missing the middle where the differences hide.

This paper introduces a clever new tool called GoT-Multi-Gap that acts like a magical "fill-in-the-blanks" machine to solve this problem.

Here is how it works, broken down into simple steps:

1. The Problem: The "Missing Middle"

In standard single-cell RNA sequencing (scRNA-seq), scientists look at the genetic instructions (mRNA) inside a cell. It's like reading a book, but the pages are torn and scattered. You can read the beginning and the end of a sentence, but often the middle part—where a mutation might be hiding—is missing or too faint to see. Previous methods were like trying to guess the middle of the sentence by only looking at the first and last words; if the mutation was in the middle, you missed it.

2. The Solution: The "Gap-Filling" Trick

The researchers invented a method to fill in those missing middle parts using a special molecular machine called Bst Polymerase. Think of this enzyme as a construction crew with a dual personality:

• Personality A (The Builder): It can read an RNA template and build a DNA bridge to fill a gap between two floating pieces of a puzzle.
• Personality B (The Sculptor): Once the bridge is built, it can chip away a tiny, useless piece at the end of the bridge to prepare it for the next step.

3. The Process: A Three-Act Play

Act 1: Setting the Stage (The Trap)
Scientists place two tiny "probes" (like magnetic clips) on either side of the mutation they want to find on the mRNA strand.

• The Left Clip: Is ready to connect.
• The Right Clip: Is deliberately broken. It has a "locked" end (a 5'-OH group) that prevents it from snapping onto the Left Clip immediately. This is a safety measure to stop random, false connections.

Act 2: The Construction (Filling the Gap)
The "Builder" enzyme (Bst Polymerase) arrives. It sees the gap between the two clips. It grabs the Left Clip and starts building a DNA bridge across the gap, copying the instructions from the RNA right up to the broken Right Clip.

• Crucial Point: If there is a mutation in the RNA, the bridge built will match that mutation. If it's a normal gene, the bridge matches the normal code.

Act 3: The Sculpting and Locking (The "Aha!" Moment)
Now, the "Sculptor" side of the enzyme kicks in. It chisels off that "locked" piece on the Right Clip, revealing a fresh, sticky end (a 5'-phosphate).

• Suddenly, the Right Clip is unlocked!
• A "glue" enzyme (SplintR Ligase) comes in and permanently snaps the Left and Right clips together.
• The Result: The two clips are now one single piece, but only if the bridge was successfully built across the gap. If the RNA wasn't there, or the bridge failed, they never snap together.

4. The Big Win: Seeing the Whole Picture

Once the clips are snapped together, the scientists can read the whole sequence, including the mutation that was previously hidden in the "gap."

Because this process is integrated into a standard workflow (10x Genomics), they can do two things at once for the same single cell:

1. Read the whole library: See which genes are active (the "phenotype" or what the cell is doing).
2. Find the typos: Detect specific mutations (the "genotype" or the cell's genetic identity).

Why This Matters

Think of it like a detective who can now interview a suspect (the cell) and ask two questions simultaneously:

• "What job are you doing right now?" (Gene expression)
• "Do you have a specific secret code in your past?" (Mutation)

Previously, you had to choose one question or the other, or you had to look at thousands of cells and guess which ones had the secret code. Now, you can look at one single cell and know exactly what it is doing and what its genetic secrets are. This is a huge leap forward for understanding cancer and how different cells within a tumor evolve and behave differently.

In short: They built a molecular "fill-in-the-blank" machine that turns invisible mutations into visible, readable signals, allowing scientists to read the full story of a single cell without missing a word.

Technical Summary

1. The Problem

Detecting somatic mutations in single-cell RNA sequencing (scRNA-seq) is critical for linking genotype to phenotype in cancer and other diseases. However, current methods face significant limitations:

• Sparse Coverage: scRNA-seq data is inherently sparse, making it difficult to capture specific mutation sites.
• Distance Constraints: Methods relying on capturing the 5' or 3' ends of transcripts often miss mutations located far from these termini.
• Throughput vs. Sensitivity Trade-off: Plate-based full-transcript sequencing offers sensitivity but is limited to hundreds of cells, while high-throughput probe-based methods (e.g., 10x Flex, RASL-seq) often struggle with sensitivity or require prior knowledge of specific alleles.
• Probe Design Limitations: Previous ligation-based approaches (e.g., GoT-Multi) require allele-specific probe designs for known variants. This limits their utility in complex mutational landscapes where the specific variant is unknown or where probe competition reduces performance.

2. Methodology: GoT-Multi-Gap

The authors developed a novel assay called GoT-Multi-Gap, which integrates an RNA-templated gap-filling mechanism into the 10x Genomics Flex workflow. The core innovation is the use of Bst DNA polymerase (Full Length) from Bacillus stearothermophilus to convert unknown mutations into ligation-competent substrates without needing allele-specific probes.

The Molecular Mechanism:

1. Probe Design: Two flanking oligonucleotide probes (Left-Hand Side [LHS] and Right-Hand Side [RHS]) hybridize adjacently to the target mRNA.
   • The LHS probe hybridizes upstream of the mutation.
   • The RHS probe hybridizes downstream. Crucially, its 5' end is modified with a 5'-OH group, rendering it initially ligation-incompetent.
2. Gap Filling (Reverse Transcription): Bst FL polymerase utilizes its reverse transcriptase activity to extend the 3' end of the LHS probe across the gap (containing the mutation) until it reaches the RHS probe.
3. Nick Translation (Probe Activation): Bst FL utilizes its nick translation activity to cleave the 5'-OH terminal nucleotide of the RHS probe. This exposes a 5'-phosphate on the next nucleotide, converting the RHS probe into a ligation-competent substrate.
4. Ligation: An RNA-templated DNA ligase (SplintR) seals the nick between the extended LHS and the activated RHS, creating a continuous sequencing-ready product.
5. Integration: The ligated product is captured by 10x Genomics gel beads (via the pCS1 sequence on the RHS), allowing simultaneous generation of libraries for whole-transcriptome gene expression and targeted mutation genotyping.

Optimization:
The authors incorporated exonuclease-resistant backbone modifications into the downstream probe to fine-tune the competition between Bst's polymerase and nick-translation activities, significantly improving gap-filling efficiency.

3. Key Contributions

• Novel Enzymatic Strategy: Demonstrated the use of Bst FL polymerase for RNA-templated gap filling and nick translation to activate probes, bypassing the need for pre-designed allele-specific probes.
• Scalable Workflow: Successfully integrated this chemistry into the high-throughput 10x Genomics Flex platform, enabling the profiling of thousands of cells simultaneously.
• Dual-Modal Profiling: Achieved simultaneous whole-transcriptomic profiling and multiplexed mutation detection from the same single cell without perturbing standard gene expression metrics.
• Systematic Characterization: Provided a comprehensive analysis of variables affecting mutation capture efficiency, including gap size, GC content, and mutation position.

4. Results

The method was validated using a mixture of three cell lines (MCF-7, SK-BR-3, and LnCAP) with 61 known cell-line-specific single nucleotide variants (SNVs).

• Performance & Specificity:
  • The assay successfully genotyped 38 targets (with sufficient cell counts).
  • Mutant calls showed high specificity (80–100%) to the expected cell lines.
  • Negative controls (loci with no variants) showed wildtype sequences with minimal background noise.
• Impact on Transcriptomics:
  • The gap-filling step did not negatively impact the 10x Flex workflow.
  • UMI counts, gene counts, and gene-wise correlations remained comparable to standard Flex experiments.
  • Cell populations remained clearly distinguishable based on canonical marker genes (e.g., ESR1, ERBB2, AR).
• Determinants of Efficiency:
  • Target Abundance: The primary determinant of detection sensitivity. A moderate positive correlation (Pearson's R = 0.37) was found between gene expression levels and genotyping rates.
  • GC Content: Probes with GC content between 44–72% performed optimally. Sensitivity dropped significantly outside this range.
  • Gap Length & Position: Unlike in situ assays where longer gaps reduced efficiency, single-cell data showed no correlation between gap length or mutation position within the gap and capture efficiency (likely masked by sequence heterogeneity).
• Robustness: The method successfully captured variants across the full length of the gap with comparable sensitivity.

5. Significance

This study presents a scalable, robust solution for detecting expressed mutations in single cells. By overcoming the limitations of allele-specific probe design and sparse coverage, GoT-Multi-Gap enables:

• Systematic Interrogation: The ability to screen for a wide range of mutations (including unknown variants) across the transcriptome in high-throughput settings.
• Genotype-Phenotype Linking: Direct correlation of somatic mutations with transcriptional states in individual cells, which is vital for understanding tumor heterogeneity, clonal evolution, and drug resistance mechanisms.
• Future Applications: The approach lays the groundwork for large-scale studies of genetic variation in complex tissues, potentially revolutionizing how researchers analyze cancer evolution and somatic diseases at single-cell resolution.