📄 medicine

An integrated single-cell transcriptomic pipeline identifies ZNF740/BRD3 and Cathepsin S as novel therapeutic targets in chronic active rim smoldering multiple sclerosis

This study presents an integrated single-cell transcriptomic pipeline that identifies ZNF740/BRD3, Cathepsin S, and DNMT1 as novel, blood-accessible therapeutic targets for treating chronic active rim smoldering multiple sclerosis through BET bromodomain inhibition, protease targeting, and epigenetic reprogramming, respectively.

Original authors: John Goodman

Published 2026-07-08

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: John Goodman

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine Multiple Sclerosis (MS) as a house that is slowly falling apart. While doctors have good tools to stop the house from catching fire suddenly (the acute attacks), they haven't figured out how to stop the slow, invisible rot that happens in the walls over time. This slow rot is called "smoldering MS."

The paper describes a new, high-tech detective pipeline designed to find the specific "rotting agents" causing this damage in a specific type of lesion called a Chronic Active Rim (CA-RIM). Think of this rim as the active edge of the damage, where the destruction is slowly expanding.

Here is how the researchers solved the mystery, using simple analogies:

The Detective Pipeline: Four Steps to the Culprit

The researchers built a four-step computer program to sift through mountains of genetic data, much like a detective using different tools to solve a crime.

The Bulk Scan (Phase 1): First, they looked at a "smoothie" of brain tissue from the damaged areas. They compared the ingredients in the damaged tissue against healthy tissue. This told them which genes were acting up (like finding that there is too much sugar in the smoothie), but it didn't tell them who was putting the sugar there.
The High-Res Zoom (Phase 2): This is the most important step. They took a massive collection of single-cell data (like looking at every individual ingredient in the smoothie separately) and used a smart AI (called scVI) to organize them. This allowed them to see exactly which type of cell was responsible for the trouble.
- The Correction: In an earlier draft, they thought a specific suspect was hiding in the "T-cell" crowd. However, they realized they accidentally mixed in data from COVID-19 patients (like mixing up two different crime scenes). Once they cleaned that up, the suspect's location changed.
The Background Check (Phase 3): They took their top suspects and ran them through a massive database of known drugs and biology. They asked: "Is this target easy to hit with a drug? Is it accessible in the blood?"
The Network Map (Phase 4): Finally, they drew a map of how these suspects connect to other known MS genes. If a suspect is friends with the "big bosses" of MS, they are a prime target.

The Three New Suspects (Therapeutic Targets)

The pipeline identified three specific "rotting agents" that the paper suggests could be stopped with new treatments.

1. The "Switch Master" (ZNF740 and BRD3)

The Analogy: Imagine a light switch (ZNF740) that is stuck in the "ON" position, keeping a loud alarm (inflammation) blaring. This switch is connected to a team of "readers" (BRD3 and other BET proteins) that keep the alarm going.
The Discovery: The researchers found that ZNF740 is a specific switch found mostly in CD8+ T-cells (a type of immune soldier) at the edge of the lesion. It is tightly linked to BRD3, a protein that helps read genetic instructions.
The Potential: Since we already have drugs that can "jam" these readers (called BET inhibitors), the paper suggests we could use them to flip the switch off and stop the T-cells from causing damage.

2. The "Scissors" (Cathepsin S / CTSS)

The Analogy: Imagine a pair of scissors (Cathepsin S) that is supposed to cut up trash, but in MS, it's cutting up the house's insulation (myelin) instead.
The Discovery: This "scissor" gene was found to be very active in the damaged tissue. Interestingly, the paper corrected a previous mistake: it is not primarily found in the T-cells as originally thought, but rather in a specific type of brain neuron (VIP GABAergic interneurons).
The Twist: Even though we aren't sure exactly which cell is holding the scissors in the lesion, the paper notes that these scissors are also floating freely in the blood. This makes them a perfect "liquid biopsy" marker—you could check a blood test to see how active the "scissors" are without needing a brain biopsy. There are also existing drugs that can blunt these scissors.

3. The "Memory Keeper" (DNMT1)

The Analogy: Imagine a librarian (DNMT1) who writes notes in the margins of a book to tell the cells what to be. In smoldering MS, this librarian is rewriting the notes so that the immune cells remember to be "angry" forever, even when they shouldn't be.
The Discovery: DNMT1 was the top-ranked target. It is an enzyme that changes how genes are read (epigenetics). The paper suggests it is reprogramming the T-cells at the lesion edge to stay in a permanent state of attack.
The Potential: There are existing drugs (used for blood cancers) that can stop this librarian from writing those angry notes, potentially "resetting" the immune cells.

What the Paper Does Not Claim

It is important to stick to what the paper actually says:

No Cures Yet: This is a computer-based discovery. The paper explicitly states these findings are "hypothesis-generating." They have not been tested in humans or animals yet.
No New Drugs: They did not invent new drugs. They found that existing types of drugs (BET inhibitors, DNMT inhibitors, Cathepsin S inhibitors) might work for this specific type of MS.
One Correction: The paper admits that an earlier version of their analysis had a data error (mixing in COVID data). They fixed it, which changed where they thought the "Scissors" (CTSS) were coming from, but the conclusion that the "Scissors" are a valid target remains.

The Bottom Line

The researchers built a sophisticated digital microscope to find three specific molecular "villains" (ZNF740/BRD3, Cathepsin S, and DNMT1) that are driving the slow, smoldering damage in MS. They argue that because these targets are accessible in the blood and have existing drugs that can hit them, they represent a promising new path to stop the disability caused by smoldering MS.

Technical Summary: An Integrated Single-Cell Transcriptomic Pipeline Identifies ZNF740/BRD3 and Cathepsin S as Novel Therapeutic Targets in Chronic Active Rim Smoldering Multiple Sclerosis

Problem Statement

Smoldering multiple sclerosis (MS) is characterized by slowly expanding chronic active rim (CA-RIM) lesions, which drive a significant portion of irreversible disability independent of acute relapses. Despite their clinical relevance, the molecular drivers of the CA-RIM microenvironment remain poorly understood, and no approved disease-modifying therapies (DMTs) specifically target this pathology. Existing studies often fail to bridge the gap between bulk lesion-specific differential expression data and the single-cell resolution required to assign candidate targets to actionable cell types.

Methodology

The authors developed a four-phase computational pipeline to integrate bulk transcriptomics with single-cell resolution and network medicine:

Phase 1 (Bulk Differential Expression): Bulk RNA-seq data from CA-RIM lesions versus control white matter (GSE108000) were analyzed. After batch correction (ComBat) and statistical testing (Welch t-test, Benjamini–Hochberg FDR), 1,065 differentially expressed genes (DEGs) were identified, with 394 significantly upregulated (log2FC ≥ 0.5, FDR < 0.1).
Phase 2 (Single-Cell Atlas & Cross-Modal Integration): A 32,239-cell atlas was constructed by harmonizing 11,000 cells from the CELLxGENE Census MS dataset with 21,239 CD8+ T-cell profiles from GSE193770 (6 MS patients, 4 healthy controls). Technical artifacts were removed using scVI (a variational autoencoder) to generate a 19-cluster Leiden graph. Phase 1 DEGs were mapped onto this atlas to assign cell-type specificity and compute a priority score ( $S_{priority}$ ) based on effect size, significance, single-cell expression, blood accessibility (GTEx), and CA-RIM specificity.
Phase 3 (Target Validation): The top 25 candidates were validated against OpenTargets, ChEMBL, UniProt, and STRING-DB. A validation score ( $S_{val}$ ) was calculated by adding discrete bonuses for druggability, existing small molecules, and MS associations to the Phase 2 priority score.
Phase 4 (Network Medicine Proximity): STRING-DB interactors for the top candidates were compared against a curated 50-gene MS seed set (derived from IMSGC GWAS and neuroinflammation literature). Network proximity scores were computed to identify targets with strong connectivity to established MS disease mechanisms. An independent replication cohort (GSE138614) was used to assess directional concordance.

Key Results

The pipeline identified three primary therapeutic axes, supported by orthogonal computational evidence:

ZNF740/BRD3 Axis:
- Ranking: 2nd highest priority score ( $S_{priority}$ = 0.548).
- Cellular Context: Highly enriched in CD8+ T cells within the CA-RIM lesion rim.
- Mechanism: STRING analysis reveals ZNF740 is a C2H2-type zinc finger transcription factor directly connected to the BET bromodomain family (BRD2, BRD3, BRD4) and the cytokine IFNG. This suggests a BET-dependent transcriptional program driving interferon signaling in lesion-resident T cells.
- Status: A novel target with no known drugs in ChEMBL/OpenTargets, representing a new entry point for BET inhibitor repurposing.
Cathepsin S (CTSS):
- Ranking: 15th by priority score ( $S_{priority}$ = 0.379) but 2nd in combined repurposing score ( $S_{val}$ × proximity = 0.040).
- Cellular Context: In the corrected atlas (excluding contaminated COVID-19 data), CTSS is most highly expressed in VIP GABAergic cortical interneurons, not CD8+ T cells as previously hypothesized. However, it shows the highest blood expression (TPM = 229.8) of all candidates.
- Mechanism: Strongly connected to MS seed genes CSF1R and PTPRC. It is a potent protease involved in antigen presentation and myelin degradation.
- Status: Highly druggable (pChEMBL = 10.0, >100 compounds) and blood-accessible, serving as both a therapeutic target and a liquid biopsy marker.
DNMT1:
- Ranking: 1st priority score ( $S_{priority}$ = 0.715).
- Cellular Context: Ubiquitous expression but substantial presence in CD8+ T cells and VIP GABAergic interneurons.
- Mechanism: Connected to STAT3 and chromatin silencing hubs. Upregulation suggests active epigenetic reprogramming of T-cell effector identity.
- Status: Highly druggable (approved drugs: decitabine, azacitidine) and blood-accessible.
Replication & Validation:
- CTSS replicated directionally in the independent bulk cohort (GSE138614) with consistent effect size.
- ZNF740 and DNMT1 did not replicate in bulk tissue due to their cell-type-specific expression (dilution in bulk samples), confirming the necessity of the single-cell integration step.
- LINGO1 emerged as the top network proximity hit (0.120), serving as an internal positive control (known target, though absent from the single-cell atlas and below blood-accessibility thresholds).

Significance and Claims

The paper claims to have identified three previously unreported therapeutic axes for smoldering MS that are:

Novel: Specifically, the ZNF740/BRD3 axis and the specific application of DNMT1 inhibition in CA-RIM T cells are not currently represented in MS drug programs.
Blood-Accessible: All three targets meet GTEx criteria for blood accessibility, enabling non-invasive monitoring via liquid biopsy.
Orthogonally Supported: The targets are validated through a convergence of bulk differential expression, single-cell cell-type specificity, multi-database druggability, and network medicine proximity to MS seed genes.

The authors emphasize that these findings are hypothesis-generating and computationally derived. They explicitly note limitations, including the post-mortem nature of the tissue (precluding longitudinal analysis), the small sample size of the single-cell cohort, and the lack of functional validation in human tissue or animal models. The study concludes that this integrated pipeline offers a systematic path from discovery to translational target selection for a disease subtype currently lacking targeted therapies.