Systematic characterization of the ovarian landscape across mouse menopause models

This study systematically compares three mouse models of menopause—intact aging, VCD-induced follicle depletion, and Foxl2 haploinsufficiency—using multi-omics and machine learning to define their shared and distinct molecular and phenotypic features, thereby providing a framework for selecting context-appropriate preclinical models to study ovarian aging and its systemic health consequences.

The Gist

Imagine your body's ovaries as a highly specialized factory. This factory has two main jobs: producing eggs (for reproduction) and manufacturing hormones that keep your whole body running smoothly (like a thermostat for metabolism, bones, and mood).

As women age, this factory naturally runs out of raw materials (eggs) and eventually shuts down its main production line. This event is called menopause. But here's the problem: when the factory shuts down, the whole building (the body) starts to struggle. We know this happens, but scientists don't fully understand why the factory closing causes the rest of the building to crumble.

To figure this out, scientists need to test on animals. But for a long time, the "test factories" (mouse models) used in labs were like badly built replicas. Some stopped too suddenly, some kept running when they should have stopped, and none perfectly matched the slow, complex shutdown seen in humans.

This paper is like a comprehensive "User Manual" and "Quality Control Report" for three different types of mouse test factories. The researchers wanted to find out: Which mouse model actually behaves like a real human menopause?

Here is the breakdown of the three models they tested, explained with simple analogies:

1. The "Natural Aging" Model (The Slow Fade)

• What it is: Letting mice grow old naturally until their ovaries just... stop working.
• The Analogy: Imagine a factory that slowly runs out of parts over 20 years. The machines get rusty, the workers get tired, and production slows down gradually.
• The Result: This is the most "honest" model. It shows the slow decline, the buildup of "rust" (fibrosis), and the immune system getting confused (like security guards wandering around aimlessly). However, mice don't actually go through full menopause; they just enter a state of "low-power mode" called estropause. They never fully shut down the factory.

2. The "Chemical Attack" Model (The VCD Bomb)

• What it is: Injecting a chemical (VCD) that specifically targets and destroys the small, immature eggs in the ovaries.
• The Analogy: Imagine sending a specialized demolition crew into the factory to blow up the raw material storage rooms. The factory is forced to shut down quickly because it has no ingredients left.
• The Result: This model is great for seeing what happens when the factory is forced to close. It creates a rapid drop in hormones and a surge in the "emergency hormones" (FSH) that the body screams for. It mimics the crisis of menopause very well. However, because the shutdown is so sudden, it misses the slow, creeping changes that happen in natural aging. Also, the immune system in these mice actually shrinks away, which is the opposite of what happens in natural aging.

3. The "Genetic Glitch" Model (The Foxl2 Error)

• What it is: Using mice with a tiny genetic typo in a gene called Foxl2, which is essential for the factory workers (granulosa cells) to do their job.
• The Analogy: Imagine the factory workers are wearing the wrong uniforms or have forgotten their training manuals. The factory is still full of raw materials (eggs), and the machines are still running, but the workers are confused and inefficient.
• The Result: This was the surprise! Even though the mice still had plenty of eggs, their factory was in trouble. The walls were getting rusty (fibrosis), the security guards (immune cells) were panicking and gathering in huge numbers, and the workers were acting strangely.
• Why it matters: This model might be the best for studying the early warning signs of menopause. It shows that the body starts breaking down before the eggs are actually gone.

The New Tools: "Biological Clocks"

The researchers didn't just look at the ovaries; they built two new "clocks" to measure how old the ovaries really are, regardless of the mouse's actual birthday.

1. The Hormone Clock (OvAge): Think of this like checking the fuel gauge and oil pressure of a car. By measuring specific hormones in the blood, they can predict if the factory is "old" or "young."

   • Finding: The "Chemical Attack" mice looked very old (high fuel consumption). The "Genetic Glitch" mice looked surprisingly young on this clock, even though their factory was struggling.

2. The Genetic Clock: This is like reading the factory's internal logbook (DNA/RNA). It looks at the tiny instructions inside the cells to see if they are acting like a young factory or an old one.

   • Finding: This clock was more sensitive. It caught the "Genetic Glitch" mice acting old even though their hormone clock said they were young. This suggests that the genetic damage happens before the hormones change.

The Big Takeaway

This paper is a roadmap for scientists. It tells them:

• If you want to study the sudden crash of menopause, use the Chemical Attack model.
• If you want to study the slow, natural decline, use the Natural Aging model.
• If you want to catch the early warning signs and understand why the immune system goes haywire, use the Genetic Glitch model.

By choosing the right "test factory," scientists can finally figure out exactly how the loss of ovarian function leads to heart disease, bone loss, and memory issues in older women, and hopefully find ways to keep the "factory" running smoothly for longer.

Technical Summary

1. Problem Statement

Menopause is a critical transition linked to systemic health decline, including increased risks of cardiovascular disease, osteoporosis, and neurodegeneration. However, the mechanisms linking ovarian decline to systemic aging remain poorly understood. A major bottleneck is the lack of robust, age-relevant preclinical models that accurately recapitulate the complex physiological, endocrine, and molecular features of human menopause.

• Limitations of existing models:
  • Intact Aging: Mice do not undergo true menopause but enter "estropause," retaining low estrogen levels and lacking the post-menopausal endocrine milieu.
  • Ovariectomy (OVX): Causes acute, rapid hormone loss without the gradual follicular depletion or post-reproductive ovarian tissue effects seen in natural menopause.
  • VCD (4-vinylcyclohexene diepoxide): Induces follicle depletion but has historically been applied only to young adults, failing to model the intersection of ovarian failure with organismal aging.
  • Genetic Models: While Foxl2 haploinsufficiency is linked to human premature ovarian failure, its specific phenotypic trajectory compared to other models has not been systematically benchmarked.

2. Methodology

The authors established a comparative framework evaluating three distinct mouse models of ovarian aging using multi-omics approaches:

1. Models Studied:
   • Intact Aging: C57BL/6JNia females at 4 months (young) vs. 20 months (old/estropausal).
   • VCD Model: C57BL/6J females injected with VCD or vehicle at 3, 6, 8, or 10 months of age, analyzed 5 months post-injection.
   • Foxl2 Haploinsufficiency: Foxl2+/+ vs. Foxl2+/- littermates at 3–4, 8–10, and 12–13 months.
2. Experimental Techniques:
   • Histology: H&E staining for follicle counts; Picrosirius Red (fibrosis) and Sudan Black B (lipofuscin) for tissue remodeling.
   • Endocrine Profiling: Serum measurement of Anti-Müllerian Hormone (AMH), Follicle-Stimulating Hormone (FSH), and Inhibin A (INHBA).
   • Single-Cell RNA Sequencing (scRNA-seq): Profiling of ~102,000 high-quality cells across all models to define cell type composition and transcriptional states.
   • Validation: RNAscope in situ hybridization and flow cytometry for immune cell quantification.
   • Computational Modeling:
     • OvAge: A hormone-based machine learning clock (Random Forest) predicting ovarian age.
     • Transcriptome-based Clocks: Lasso regression models trained on granulosa and theca cell gene expression to predict molecular age.
     • Advanced Analyses: Augur (cell-type separability), WGCNA (co-expression networks), trajectory analysis (LRT), and Transposable Element (TE) expression analysis.

3. Key Contributions

• Unified Benchmarking: First systematic comparison of intact aging, chemical (VCD), and genetic (Foxl2) models across histological, hormonal, and single-cell transcriptomic scales.
• Novel Aging Clocks: Development of "OvAge" (hormone-based) and transcriptome-based clocks to quantify ovarian aging and age acceleration non-invasively.
• Cell-Type Resolution: Comprehensive annotation of 19 ovarian cell types (8 non-immune, 11 immune), revealing model-specific shifts in cellular composition and immune microenvironments.
• Interactive Resources: Creation of public R Shiny applications for predicting ovarian age and exploring the annotated scRNA-seq datasets.

4. Key Results

A. Histological and Tissue Remodeling

• Aging & VCD Models: Both showed significant depletion of follicles (primordial to antral), increased fibrosis, and lipofuscin accumulation, mimicking natural ovarian decline.
• Foxl2 Model: Surprisingly, Foxl2+/- mice did not show significant follicle depletion compared to wild-type. However, they exhibited increased fibrosis and lipofuscin, indicating premature stromal aging despite preserved follicle numbers.

B. Endocrine Alterations

• Aging & VCD: Both models displayed the classic menopausal signature: decreased AMH and INHBA, and elevated FSH.
• Foxl2 Model: Showed a divergent endocrine profile. While FSH was elevated and INHBA reduced, AMH levels were paradoxically increased. This suggests Foxl2 deficiency alters follicle recruitment dynamics (potentially via AMH interaction) rather than simply depleting the reserve. Consequently, the "Ovarian Health Index" did not flag Foxl2+/- mice as impaired, unlike the other models.

C. Aging Clocks (OvAge & Transcriptomic)

• OvAge (Hormone-based): VCD treatment caused significant age acceleration (predicted age > chronological age). Foxl2+/- mice showed no significant age acceleration, consistent with their preserved AMH levels.
• Transcriptome-based Clocks: Both VCD and Foxl2+/- models showed molecular age acceleration in granulosa cells. Notably, the transcriptome-based clock detected aging in Foxl2+/- mice even when the hormone-based clock did not, suggesting molecular perturbations precede overt endocrine collapse.

D. Single-Cell Transcriptomics & Immune Landscape

• Immune Shifts:
  • Aging: Increased immune cell abundance (immune expansion).
  • VCD: Drastic reduction in immune cells, likely due to follicle loss and tissue remodeling.
  • Foxl2: Significant expansion of immune cells (NK, NKT, T cells), suggesting an immune-primed environment distinct from chemical ablation.
• Transcriptional Trajectories:
  • Granulosa Cells: Consistently the most sensitive cell type across all models.
  • Mitochondrial Dysregulation: A convergent theme across all models, though the direction of change varied (e.g., VCD downregulated respiration genes; Foxl2 upregulated them in some contexts).
  • Pathways: VCD recapitulated many aging signatures but lacked the immune expansion of natural aging. Foxl2 haploinsufficiency showed divergent stromal remodeling and unique immune activation patterns.

E. Regulatory Networks

• WGCNA & TF Activity: Identified conserved modules related to mitochondrial metabolism and ribosomal function.
• Transcription Factors: Reduced activity of cell cycle regulators (E2F, Myc) and increased activity of stress/metabolic regulators (Foxo1, Srebf1) were observed in granulosa cells across models.

5. Significance and Implications

• Model Selection Guidance: The study provides a decision framework for researchers:
  • Use Intact Aging for studying gradual physiological decline and immune remodeling.
  • Use VCD for studying acute follicle depletion and endocrine collapse, but note its failure to model immune expansion.
  • Use Foxl2 Haploinsufficiency for studying early, subclinical ovarian dysfunction, stromal remodeling, and immune dysregulation that precedes overt follicle loss.
• Mechanistic Insight: The findings suggest that ovarian aging is not a uniform process; different perturbations (chronological, chemical, genetic) trigger distinct molecular trajectories.
• Biomarker Development: The "OvAge" and transcriptome-based clocks offer tools to detect ovarian aging earlier than traditional histological methods, potentially identifying windows for therapeutic intervention.
• Systemic Health: By clarifying the specific molecular signatures of ovarian decline, this work lays the groundwork for understanding how ovarian dysfunction drives systemic pathologies like cardiovascular disease and neurodegeneration.

In conclusion, this paper transforms the field of reproductive aging by moving beyond single-model studies to a comparative, multi-omics framework, revealing that while different models share core features of ovarian decline, they diverge significantly in immune response, endocrine feedback, and molecular trajectory.