Detection of Candidate Circular RNAs to Monitor Anti-Hormonal Response in the Mammary Gland

This study utilized RNA sequencing data from genetically engineered mouse models to identify a specific set of circular RNAs that are reproducibly differentially regulated in the mammary gland by both tamoxifen and letrozole, suggesting their potential as non-invasive biomarkers for monitoring anti-hormonal therapy response in breast cancer.

The Gist

Imagine your body is a vast, bustling city. In this city, there are specific neighborhoods (tissues) like the "Mammary District," which can sometimes get overcrowded with troublemakers (cancer cells) if the city's power grid (hormones) gets out of whack.

Doctors have two main ways to fix this power grid: they can either cut the power supply entirely (using drugs like Letrozole) or install a blocker that stops the power from reaching the trouble spots (using drugs like Tamoxifen). These are called "anti-hormonal therapies."

The problem? Right now, doctors have to wait a long time or use expensive, invasive scans to see if these drugs are actually working. They are essentially waiting to see if the city lights go out before they know the plan is working.

This paper is about finding a "smoke signal" that tells us the drugs are working almost immediately.

The "Smoke Signal": Circular RNAs

Inside every cell in our body, there are tiny messengers called RNA. Usually, these messengers are straight lines (like a piece of string) that carry instructions. But sometimes, the cell ties the ends of the string together to make a loop. These are called Circular RNAs (circRNAs).

Think of these loops like indestructible rubber bands. Because they are closed loops, they don't unravel or get eaten by the cell's cleanup crew as easily as the straight strings do. They are tough, stable, and they can escape the cell, float into the bloodstream, and be caught in a simple blood test (a "liquid biopsy").

The Experiment: A Mouse City

The researchers didn't test this on humans first; they used a sophisticated "simulation" using mice.

• They created two different types of "mouse cities" that were prone to developing breast cancer (Genetically Engineered Mouse Models).
• They treated one group with the "power blocker" (Tamoxifen) and another with the "power cut" (Letrozole).
• They then looked at the "rubber bands" (circRNAs) floating in the mice's mammary tissue to see which ones changed when the drugs were introduced.

The Discovery: The "Universal Alarm"

The researchers were looking for a specific pattern: Which rubber bands changed in the exact same way, regardless of which drug was used?

If a rubber band changed only with Tamoxifen but not Letrozole, it was a "one-trick pony." But they found 35 special rubber bands that reacted to both drugs in the same way.

• 4 of them got louder (increased in number).
• 31 of them got quieter (decreased in number).

It's as if the city had a universal alarm system. Whether you cut the power or blocked the signal, the same 35 sirens went off. This suggests that these specific circular RNAs are a direct result of the body successfully responding to the treatment.

Why This Matters for Humans

The researchers then checked a map to see if these "mouse rubber bands" had human cousins.

• They found that 25 of these 35 had direct human equivalents.
• Even better, the "parent genes" that make these rubber bands are known to exist in human breast tissue.

This is a huge deal because it means we might soon be able to take a simple blood test from a woman taking breast cancer prevention drugs. If the levels of these specific 25 rubber bands shift in the right direction, the doctor knows immediately: "The drug is working!"

The Bottom Line

Currently, checking if breast cancer prevention drugs work is like waiting for a building to burn down to see if the sprinklers worked. This paper suggests we can instead look for a specific type of smoke (circular RNAs) that tells us the sprinklers are working while the fire is still being put out.

It's a "proof of concept" that we can use these tough, loop-shaped RNA molecules as a universal, non-invasive dashboard to monitor how well anti-hormonal therapies are performing, potentially saving lives by catching treatment failures much earlier.

Technical Summary

1. Problem Statement

Anti-hormonal therapies, such as Selective Estrogen Receptor Modulators (SERMs like tamoxifen) and Aromatase Inhibitors (AIs like letrozole), are the cornerstone of treatment and prevention for estrogen receptor-positive (ER+) breast cancer. However, current methods for monitoring therapeutic response rely on invasive imaging (e.g., mammography density changes) or blood tests measuring estrogen levels, which do not directly reflect the molecular response within the tissue. There is a critical need for non-invasive, genetically based biomarkers that can detect a positive molecular response to anti-hormonal therapy across different drug classes and mechanisms of action. Circular RNAs (circRNAs), a class of stable, non-coding RNAs detectable in plasma via liquid biopsy, are proposed as potential biomarkers, but a specific panel responsive to anti-hormonal treatment in breast tissue has not yet been established.

2. Methodology

The study employed a rigorous bioinformatic pipeline to identify circRNAs differentially regulated by anti-hormonal agents in preclinical models.

• Data Source: Total RNA sequencing (RNA-seq) data from mammary gland tissues of two distinct Genetically Engineered Mouse Models (GEMMs) of estrogen pathway-induced breast cancer:
  1. MMTV-rtTA/tet-op-Esr1: Models estrogen receptor alpha (ER$\alpha$) overexpression.
  2. MMTV-rtTA/tet-op-CYP19A1: Models aromatase overexpression (increasing local estrogen).
  • Conditions: Mice were treated with either tamoxifen or letrozole (2-month exposure) or left untreated. Each condition included $n=3$ independent samples per genotype, combined for analysis ($n=6$).
• CircRNA Detection Pipeline:
  • Utilized the nf-core/circrna pipeline.
  • Back-Splice Junction (BSJ) Detection: Employed multiple algorithms (CIRIquant, CircExplorer2, CircRNA finder, DCC, Find circ, MapSplice, Segemehl) to ensure robust detection.
  • Annotation: Annotated against murine databases (CircAtlas, CircBase, CircBank, CIRCpedia, TSCD) using the GRCm39 reference genome.
• Differential Expression Analysis:
  • Used DESeq2 with a design formula controlling for age, transgene status, and induction.
  • Filters: Retained circRNAs with $\ge$ 10 counts in at least 3 samples.
  • Significance Criteria: Adjusted p-value ($p_{adj}$) < 0.05 and absolute log2 fold change ($|log2FC|$) > 1.
• Cross-Species Validation:
  • Identified orthologous regions in humans using UCSC LiftOver tools.
  • Filtered for host genes known to be expressed in human breast epithelial cells (using The Human Protein Atlas and primary high-risk breast cell data).
  • Cross-referenced with human circRNA databases (circAtlas, circBase, etc.) to confirm the existence of human orthologs.

3. Key Contributions

• Identification of a Common Response Signature: The study successfully identified a set of circRNAs that are commonly differentially regulated by two distinct classes of anti-hormonal drugs (tamoxifen and letrozole) and across two distinct molecular pathways of breast cancer generation (ER$\alpha$ overexpression vs. Aromatase overexpression).
• Proof of Principle for Liquid Biopsy: It establishes a rational foundation for using circRNAs as mechanism-agnostic molecular readouts for estrogen pathway suppression, suggesting their utility as non-invasive biomarkers.
• Human Translation: The study bridges the gap between mouse models and human application by identifying 25 candidate orthologous circRNAs derived from host genes expressed in human breast tissue, with 5 already documented in human databases.

4. Key Results

• Differential Expression:
  • Tamoxifen: Identified 61 differentially expressed (DE) circRNAs.
  • Letrozole: Identified 51 DE circRNAs.
  • Common Overlap: 35 circRNAs were commonly regulated by both drugs in the same direction.
    • 4 Up-regulated: Fmn1, Gm34744/Bcam, Hmg20a, Scn4a.
    • 31 Down-regulated: Including Csn1s2a, Rn18s-rs5, Bud23, Alg13, Rpn1, Zfp148, etc.
• Validation and Filtering:
  • Database Support: 94% (33/35) of the identified mouse circRNAs were supported by at least one established murine circRNA database, confirming they are genuine circRNAs.
  • Fold Change: All identified circRNAs showed >2-fold change; 20 of the 31 down-regulated circRNAs showed >4-fold change, indicating high reliability.
  • Host Gene Independence: In most cases, the differential expression of circRNAs was not paralleled by significant changes in their linear host gene expression, suggesting circRNA regulation is independent of host gene transcription levels.
• Human Relevance:
  • 88% of the host genes for the 35 candidates are expressed in human breast primary epithelial cells.
  • 25 orthologous human circRNAs were identified.
    • 5 were previously documented in human breast tissue databases (e.g., hsa-FMN1_0002, hsa-CSNK1G3_0001, hsa-ZNF148_0018, hsa-MBNL1, circRpn1).
    • 20 represent newly identified potential structures (host genes expressed in human breast but not yet listed in human circRNA databases).
• Circular-to-Linear Ratios (CLR): Calculated CLRs ranged from 0.002 to 0.60, indicating varying abundance relative to linear transcripts.

5. Significance and Future Directions

• Clinical Utility: This study provides a candidate panel of circRNAs that could be developed into a liquid biopsy test to monitor early therapeutic response to anti-hormonal therapy. This could help identify treatment failures early and validate new drug candidates.
• Mechanism-Agnostic: Because the circRNAs respond to both SERMs and AIs, and across different genetic drivers of breast cancer, they represent a robust biomarker class for estrogen pathway suppression regardless of the specific drug or tumor subtype.
• Limitations & Next Steps:
  • The study is bioinformatic; experimental validation (qPCR, in situ hybridization) in mouse and human tissues is required.
  • The study did not assess exosome secretion patterns, which is crucial for circRNAs to be detectable in peripheral blood.
  • Functional mechanisms (e.g., miRNA sponging) were not tested.
  • Future work will focus on validating these candidates in human peripheral blood to establish a clinical biomarker panel.

In conclusion, this paper demonstrates that specific circRNAs in the mammary gland undergo reproducible, significant changes in response to anti-hormonal therapy, offering a promising, non-invasive avenue for monitoring breast cancer treatment efficacy.