🧬 biology

Ribozyme-optimized CVS-N2cΔG Rabies Virus with Hyper-diverse Barcodes for High-fidelity Single‑Neuron Projection Connectivity

This study reports a ribozyme-optimized CVS-N2cΔG rabies virus platform that utilizes Twister-based configurations to dramatically enhance barcode diversity and rescue efficiency, thereby enabling high-fidelity, collision-free mapping of single-neuron projection connectivity in the mouse visual pathway.

Original authors: Lei Jin, He Zhang, Daoyin Liu, Xueping Gao, Xiangyu Ge, Yahui Li, Xia Zhang

Published 2026-06-26

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Lei Jin, He Zhang, Daoyin Liu, Xueping Gao, Xiangyu Ge, Yahui Li, Xia Zhang

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Picture: Mapping the Brain's "Wiring Diagram"

Imagine the human brain is a massive, bustling city with billions of people (neurons) and trillions of phone lines (synapses) connecting them. Scientists want to draw a perfect map of who is calling whom. To do this, they use a special "spy virus" (a modified Rabies virus) that jumps from one neuron to the one it's connected to, leaving a unique digital "tag" or barcode behind.

If the virus tags a neuron in the visual cortex (V1), scientists can look at the tags to see which neurons in the visual relay station (LGN) sent the signal.

The Problem:
The city is so big that the scientists ran out of unique phone numbers. If they only have 1,000 unique tags but need to label 10,000 neurons, many neurons get the same tag by accident. This creates a "glitch" in the map. It looks like one neuron is talking to a huge crowd, when really, it's just a mistake where two different people were given the same ID card. This is called a barcode collision.

The Solution: A Better "Printer" for the Tags

The researchers (Lei Jin, He Zhang, and team) realized the problem wasn't the tags themselves, but the machine printing them. They were using an old, slow printer (standard ribozymes) that often jammed or printed messy copies, leading to fewer unique tags than they hoped for.

They decided to upgrade the printer by swapping out its internal parts for a brand-new, high-speed model called the Twister ribozyme.

Analogy 1: The Factory Assembly Line

Think of making the virus like running a factory assembly line.

The Old Machine: The factory used a slow, clunky tool to cut the raw material (RNA) at the end of the line. Sometimes it cut too early, sometimes too late, or the material got crumpled. This meant many products were defective, and the factory could only produce a small batch of unique items before running out of good ones.
The New Machine: The team installed a "Twister" tool that cuts the material instantly and perfectly every time.
The Result: The factory didn't just get faster; it became 100 times more efficient. They could now produce a massive library of unique tags without the "jamming" errors.

The Experiments: Testing the New Machine

1. The "Mini-Factory" Test (Screening)
First, they built a tiny, simplified version of the virus factory (called a minigenome) just to test different cutting tools. They tried many combinations.

Finding: The combination of a standard tool at the start and the new "Twister" tool at the end worked best. It made the virus rescue efficiency jump by two orders of magnitude (100x).

2. The "Real World" Test (In Vivo)
Next, they used the best tools to make the actual virus libraries and injected them into mice brains to trace the connection between the LGN and the visual cortex (V1).

The Old Way: The standard virus library was like a library with only 700 unique books. When they tried to map the brain, the "collision" rate was high. The map looked messy, showing neurons connected to dozens of inputs that didn't actually exist (false positives).
The New Way: The "Twister" virus library had 75 times more unique tags that survived the trip.
- The Result: The map became incredibly clear. Instead of seeing a neuron connected to 57 inputs (a glitch), they saw it connected to only 3–6 inputs. This matched what we know biology should look like: a sparse, clean network.

The Key Takeaway: "Collision-Free" Mapping

The paper makes a crucial point about how to read the data. Even with the amazing new "Twister" virus, some collisions still happen. However, the researchers developed a new way to look at the data:

Old Method: Look at all the data and try to guess what's real.
New Method: Only look at the neurons that have a 100% unique tag (a "collision-free" cell).

By filtering out the messy data and only trusting the clean, unique tags, they could reconstruct the brain's wiring diagram with high fidelity. The new virus didn't change the brain's biology; it just removed the "static" on the radio so they could hear the true signal.

Summary in One Sentence

The researchers upgraded the "scissors" used to cut the genetic code of a tracing virus, creating a super-efficient machine that produces millions of unique ID tags, allowing scientists to draw a clear, error-free map of how individual neurons in the brain are connected.

Technical Summary: Ribozyme-optimized CVS-N2cΔG Rabies Virus with Hyper-diverse Barcodes

Problem Statement

High-resolution neural circuit mapping using barcoded rabies virus (RV) is fundamentally limited by a "numbers problem." To map billions of neurons without "barcode collision" (where distinct neurons acquire identical barcodes, creating artificial "spurious convergence"), library diversity must exceed the starter cell population by orders of magnitude. Current CVS-N2cΔG systems, while superior in neurotoxicity and trans-synaptic efficiency compared to SAD-B19 strains, suffer from low rescue efficiency and packaging bottlenecks. These limitations lead to severe "founder effects" and "jackpot clones" during viral production, drastically reducing effective library diversity. Furthermore, existing methods often lack systematic, cell-centric benchmarking to verify collision reduction in vivo, leaving topological distortions in reconstructed networks unaddressed.

Methodology

1. Minigenome-Based Screening Platform (CVS-N2cmini)

To isolate terminal RNA processing from the complexities of viral replication, the authors engineered CVS-N2cmini, a replication-incompetent minigenome. This system deletes all five viral structural genes (N, P, M, G, L) and supplies the Matrix (M) protein in trans. This allows for the precise screening of ribozyme configurations at the 5' and 3' termini without the confounding variables of autonomous replication.

2. Ribozyme Engineering and Screening

The study systematically evaluated combinatorial libraries of self-cleaving ribozymes at the genomic termini of the CVS-N2c plasmid.

Candidates: Standard Hammerhead (HH), Hepatitis Delta Virus (HdVRz), optimized RzB, and the ultra-fast Twister ribozyme (including mirror variants).
Screening: The authors screened these combinations in the CVS-N2cmini system to optimize rescue efficiency and then validated the top candidates in the full infectious CVS-N2cΔG system.
Selection: Based on cleavage kinetics and rescue titers, HH + Twister and RzB + Twister were selected for library construction.

3. Library Construction and Viral Production

Three library configurations were constructed:

Canonical: HH + HdVRz (Control).
Optimized 1: HH + Twister.
Optimized 2: RzB + Twister.
Libraries were generated by assembling synthetic degenerate 20-bp barcode pools into the CVS-N2cΔG backbone. The libraries underwent a standard production pipeline: Plasmid pool $\rightarrow$ P0 rescue $\rightarrow$ P1 amplification (B19G envelope) $\rightarrow$ EnvA pseudotyping.

4. In Vivo Validation and Dual-Resolution Analysis

The libraries were injected into the primary visual cortex (V1) of mice to trace monosynaptic inputs from the lateral geniculate nucleus (LGN).

Sequencing: Single-cell droplet-based sequencing (DNBelab C-TaiM4) was used to capture viral barcodes from V1 (starter) and LGN (presynaptic) neurons.
Dual-Resolution Framework: The authors developed a three-tiered analysis to assess collision artifacts:
1. Barcode-centric: Quantifying the proportion of barcode species infecting multiple cells.
2. Cell-centric: Quantifying the fraction of host cells that are "collision-free" (infected only by unique barcodes).
3. Topological Reconstruction: Reconstructing network connectivity (V1 in-degree) using only collision-free cells to filter out false-positive hubs.

Key Results

1. Rescue Efficiency and Titer Improvement

Minigenome Screen: The RzB + Twister combination yielded the highest intracellular rescue efficiency (0.69% EGFP+), a 13.8-fold increase over the canonical HH + HdVRz control.
Full Virus Rescue: In the CVS-N2cΔG system, HH + Twister and RzB + Twister achieved peak viral titers approximately two orders of magnitude higher (94–99-fold improvement) than the canonical control. The optimized viruses also exhibited faster kinetics, reaching peak titers at Day 6 compared to Day 15 for the canonical strain.

2. Preservation of Barcode Diversity

Effective Library Capacity: The canonical library suffered a catastrophic 475-fold reduction in effective capacity (Shannon entropy-based $E = 2^H$ ) from plasmid to EnvA stage (330,981 $\rightarrow$ 694).
Optimized Performance: The HH + Twister library retained an effective capacity of 52,067 at the EnvA stage (from a plasmid input of 156,110), representing a 75-fold improvement over the canonical system.
Collision Rates: The collision rate (zero-distance edit distance) dropped from 0.377% (canonical) to 0.006% (HH + Twister), a 63-fold reduction.

3. In Vivo Mapping Performance

Diversity Recovery: In the V1-LGN circuit, the HH + Twister library yielded 4.2-fold more unique barcodes in V1 compared to the canonical library.
Collision-Free Cells: The fraction of collision-free cells in V1 increased from 1.9% (canonical) to 8.5% (HH + Twister) and 11.9% (RzB + Twister).
Topological Correction:
- Uncorrected: The canonical library suggested a biologically implausible V1 in-degree of 56.8 inputs per neuron (driven by collision artifacts).
- Corrected: When restricted to collision-free cells, all library types converged to a sparse, physiological architecture of 3–6 LGN inputs per V1 neuron.
- Distribution: The canonical library showed a right-skewed distribution (median in-degree = 27), whereas the HH + Twister library concentrated at biologically plausible values (median = 7).

Significance and Claims

The paper claims that ribozyme-mediated optimization of terminal RNA processing is a causal determinant of barcode diversity and downstream connectivity fidelity. By replacing the kinetically slow HdVRz ribozyme with the ultra-fast Twister ribozyme, the authors resolved the trade-off between the low toxicity of the CVS-N2c strain and the need for high library diversity.

Key contributions include:

A 75-fold expansion in effective barcode capacity, enabling high-fidelity single-neuron projection connectivity at scale.
Synergistic Strategy: The work demonstrates that maximizing library diversity (to increase the absolute yield of collision-free cells) must be paired with cell-centric computational filtering to resolve residual false-positive hubs.
Methodological Standard: The authors propose a "three-tiered" analysis framework (barcode-centric, cell-centric, and topological correction) as a necessary reporting standard for barcoded connectomics, analogous to doublet detection in transcriptomics, to ensure reported network architectures reflect genuine biology rather than library-specific errors.
Platform Generalizability: While focused on CVS-N2cΔG, the ribozyme optimization strategy is presented as potentially generalizable to other negative-strand RNA viruses (e.g., Sendai, Measles) that rely on terminal genome promoters.

The study concludes that the ribozyme-optimized CVS-N2cΔG platform provides a robust foundation for high-fidelity monosynaptic projection mapping, effectively unmasking the sparse, parallel organization of neural circuits (specifically LGN to V1) that was previously obscured by collision-induced artifacts.