Modelling the polygenicity and clinical heterogeneity of human depression in mice to identify biomarkers of antidepressant response

This study introduces a new stable polygenic mouse model of depression (H-FST) that, unlike its predecessor (H-TST), exhibits distinct behavioral and molecular profiles with opposite antidepressant responses, thereby enabling the identification of human biomarkers specific to clinical subgroups of treatment responders.

The Gist

Imagine that Depression is a massive, chaotic storm. For decades, scientists have tried to predict the weather and build better umbrellas (medications), but they've been struggling because they've been treating every storm as if it were the same. They assumed that if a person is sad, they must have the same "weather pattern" inside their brain.

This paper is like a team of meteorologists realizing, "Wait a minute. Some storms are dry and hot (anxiety without sadness), while others are freezing and wet (sadness without anxiety). We need different umbrellas for different storms."

Here is the story of how they built two new "weather stations" (mouse models) to figure out why some people respond to antidepressants and others don't.

1. The Problem: One Size Doesn't Fit All

Currently, about 30% of people with depression don't get better with standard antidepressants. Why? Because depression is polygenic (caused by hundreds of tiny genetic factors) and heterogeneous (everyone experiences it differently).

Most previous mouse models were like building a house with only one brick. They tested mice with just one bad gene or one bad experience. But real human depression is more like a house built from thousands of different, slightly broken bricks. The researchers wanted to build a mouse model that felt like that messy, complex reality.

2. The Experiment: Breeding Two Different "Personalities"

The scientists started with a huge mix of 8 different mouse breeds (a genetic "smoothie") to ensure variety. Then, they put them through two different "stress tests":

• The Tail Suspension Test (TST): Imagine hanging a mouse by its tail. If it stops struggling and just hangs there, it's "giving up."
• The Forced Swim Test (FST): Imagine dropping a mouse in a bucket of water. If it stops swimming and just floats, it's "giving up."

They bred two distinct groups of mice based on how they reacted:

1. The H-TST Group: These mice were bred to be the ones who gave up quickly in the Tail Suspension test.
2. The H-FST Group: These mice were bred to be the ones who gave up quickly in the Swimming test.

3. The Big Surprise: They Are Totally Different

You might think these two groups of "giving up" mice would be identical twins. They weren't. They were like two different types of storms.

• The H-TST Mice (The "Anxious & Sad" Storm): These mice didn't just give up; they were also anxious (afraid of open spaces), anhedonic (they stopped enjoying sweet treats), and had sleep problems. They were a complete package of misery.
• The H-FST Mice (The "Quiet Sadness" Storm): These mice gave up in the water, but they were normal in every other way. They weren't anxious, they still loved their sweets, and they slept like babies.

The Analogy: Think of H-TST as a person who is depressed, anxious, and can't sleep. Think of H-FST as a person who is just deeply sad and unmotivated, but not anxious.

4. The Medicine Test: Different Keys for Different Locks

The researchers then tried to "cure" these mice with different drugs.

• The H-FST Mice (Quiet Sadness): When given standard antidepressants (SSRIs like Prozac), they bounced back! Their swimming improved. They were like a lock that opens easily with a standard key.
• The H-TST Mice (Anxious Sadness): The standard drugs barely worked on them. However, when given a drug that targets a different system (glutamate receptors), they improved. They were a high-security lock that needed a special, complex key.

The Lesson: This explains why some humans don't respond to Prozac. They might have the "H-TST" type of depression (anxious, sleepless), which needs a different treatment approach.

5. The Brain Clues: The Wiring is Different

The scientists looked inside the mice's brains (specifically the prefrontal cortex, the CEO of the brain).

• H-TST Brains: Were flooded with signs of inflammation (like a body fighting a cold) and had too much "excitatory" wiring (too much noise).
• H-FST Brains: Were quiet but had a different problem: too much "inhibitory" wiring (too much silence/braking).

It's like the H-TST brain is a car engine revving too high and overheating, while the H-FST brain is a car with the parking brake stuck on. You can't fix both with the same mechanic.

6. The Human Connection: Predicting the Future

Finally, the researchers looked at human data. They found that the genes that were "broken" in the anxious H-TST mice matched the genes found in women with depression who also have anxiety.

They then tested a specific list of genes in a group of women taking a drug called Duloxetine.

• The Result: They found 4 specific genes (AKAP13, CLCN7, P4HB, FBLN1) that could predict who would get better.
  • If a woman had high levels of these genes, she was likely to respond well to the drug.
  • If she had low levels, she likely wouldn't.

The Takeaway

This paper is a giant step toward Precision Psychiatry.

Instead of saying, "Here is a pill for depression," doctors might soon be able to say, "Let's look at your genetic profile. You have the 'Anxious-Sleepless' type of depression, so we need a drug that targets inflammation and glutamate, not just serotonin."

By creating these two distinct mouse models, the researchers built a map that helps us understand that depression isn't one disease—it's many different storms, and we finally have the tools to build the right umbrella for each one.

Technical Summary

1. Problem Statement

Major Depressive Disorder (MDD) affects hundreds of millions globally, yet approximately 30% of patients do not respond to standard antidepressant treatments. Current rodent models of MDD typically rely on a single genetic mutation or a single environmental stressor, failing to recapitulate the polygenic nature and clinical heterogeneity (e.g., varying symptom profiles like anxiety, anhedonia, and sleep disturbances) observed in human patients. Consequently, there is a lack of biological markers to predict treatment response or stratify patients into clinically relevant subgroups.

2. Methodology

The study employed a multi-faceted approach combining behavioral genetics, transcriptomics, proteomics, and translational bioinformatics.

• Generation of Polygenic Mouse Models:

  • Founders: An eight-way cross of inbred mouse strains (A/J, ABP/Le, BALB/cJ, C57BL/6J, C3H/HeJ, CBA/H, DBA/2J, SWXL-4) created a complex genetic background.
  • Selection: Researchers performed bidirectional selective breeding over 10–15 generations based on extreme phenotypes in the Forced Swim Test (FST).
  • Lines Generated:
    • H-FST: "Helpless" mice selected for high immobility in FST.
    • NH-FST: "Non-helpless" mice selected for low immobility in FST.
  • Comparison: These new lines were compared to previously established H-TST/NH-TST lines (selected via Tail Suspension Test).

• Behavioral Characterization:

  • Assessed depressive-like behaviors (FST, Tail Suspension Test), anhedonia (Sucrose Preference Test), anxiety (Elevated Plus Maze, Light/Dark Box), and sleep architecture (Polysomnography).
  • Pharmacological Validation: Tested acute and chronic responses to Selective Serotonin Reuptake Inhibitors (SSRIs: fluoxetine, escitalopram) and a Group II mGluR antagonist (LY341495).

• Molecular Analysis (Prefrontal Cortex - PFC):

  • Transcriptomics: RNA microarray analysis of the PFC in female mice to identify Differentially Expressed Genes (DEGs) between helpless and non-helpless lines.
  • Pathway Analysis: Gene Ontology (GO) enrichment and Rank-Rank Hypergeometric Overlap (RRHO) analysis to compare signatures with other rodent models (e.g., chronic mild stress, Brd1 mutants).
  • Proteomics: Western blot quantification of Vesicular Glutamate Transporter 1 (VGLUT1) and Vesicular GABA Transporter (VGAT) to assess excitatory/inhibitory (E/I) balance.

• Translational Human Analysis:

  • Data Integration: Compared mouse PFC DEGs with human whole-blood gene expression datasets (GSE98793: MDD with/without Generalized Anxiety Disorder [GAD]; GSE146446: MDD patients treated with duloxetine).
  • Biomarker Identification: Identified genes shared between mouse models and human subgroups to predict treatment response in a clinical trial cohort.

3. Key Results

A. Phenotypic Divergence (H-FST vs. H-TST)

Despite both models exhibiting "helplessness" (high immobility) in stress tests, they represent distinct clinical subtypes:

• H-TST Mice: Exhibit a "complex" depressive phenotype including anhedonia, anxiety, and sleep disturbances (increased paradoxical sleep, decreased slow-wave sleep). They show poor response to SSRIs but a good response to the glutamate antagonist LY341495.
• H-FST Mice: Exhibit pure depressive-like behavior (high immobility) without anhedonia, anxiety, or sleep disturbances. They show a robust response to SSRIs (fluoxetine, escitalopram) but a poor response to LY341495.

B. Molecular Mechanisms in the Prefrontal Cortex

• Transcriptomic Specificity: Only ~7.5% of DEGs were shared between H-TST and H-FST models.
  • H-TST Signature: Enriched in immunity and inflammation pathways (suggesting a neuroinflammatory mechanism).
  • H-FST Signature: Enriched in cell signaling and neurotransmission pathways, specifically glutamate receptor signaling.
• Opposite E/I Imbalance:
  • H-TST: Increased VGLUT1 (excitatory) with stable VGAT $\rightarrow$ Excitatory bias.
  • H-FST: Decreased VGLUT1 and increased VGAT $\rightarrow$ Inhibitory bias.
  • Conclusion: The two models represent opposite ends of the excitatory/inhibitory synaptic spectrum.

C. Translational Biomarkers

• Anxiety Subtyping: A signature of 155 downregulated genes was identified in both H-TST mice and human MDD patients with comorbid GAD (anxious depression).
• Treatment Prediction: Using a duloxetine clinical trial cohort (GSE146446), the study identified specific biomarkers predicting response:
  • Responders: Showed baseline differences in AKAP13, CLCN7, and P4HB (genes associated with the H-FST/non-anxious profile). These genes increased in expression after treatment in responders.
  • Non-Responders: Showed baseline overexpression of FBLN1 (associated with the H-TST/anxious profile).
  • Significance: These genes serve as potential biomarkers to stratify patients likely to respond to SSRIs (like duloxetine) versus those who may require alternative mechanisms (e.g., glutamatergic agents).

4. Significance and Impact

• Modeling Heterogeneity: The study successfully demonstrates that bidirectional selection based on different stress tests (FST vs. TST) generates distinct polygenic models that mimic specific human MDD subtypes (anxious/anhedonic vs. non-anxious).
• Mechanistic Insight: It links specific molecular pathways (neuroinflammation vs. neurotransmission) and synaptic imbalances (E/I ratio) to differential drug responses, supporting the "excitatory/inhibitory imbalance" hypothesis in depression.
• Precision Psychiatry: The identification of a gene-expression signature (AKAP13, CLCN7, P4HB, FBLN1) that predicts duloxetine response in women offers a potential translational tool for patient stratification. This moves the field toward precision medicine, where treatment selection is guided by biological markers rather than trial-and-error.
• Therapeutic Implications: The findings suggest that patients with an "excitatory bias" (similar to H-TST) may benefit from glutamatergic modulators (e.g., ketamine, mGluR antagonists), while those with an "inhibitory bias" (similar to H-FST) may respond better to SSRIs.

In summary, this paper provides a robust framework for understanding the biological basis of antidepressant resistance and response by leveraging polygenic mouse models to bridge the gap between animal behavior and human clinical heterogeneity.