Transcriptomics and proteomics of projection neurons in a circuit linking hippocampus with dorsolateral prefrontal cortex in the human brain

This study utilizes single-cell transcriptomics and proteomics of projection neurons within the hippocampal-prefrontal circuit to reveal schizophrenia-associated molecular alterations, including region-specific phosphorylation changes, co-expression networks, and disrupted directional connectivity driven by glial interactions and inhibitory neuropeptide downregulation.

The Gist

Imagine the human brain as a massive, bustling city. In this city, different neighborhoods (brain regions) need to talk to each other to keep everything running smoothly. One of the most important conversations happens between the Hippocampus (the city's library and memory center) and the Prefrontal Cortex (the CEO's office responsible for planning and decision-making).

In people with Schizophrenia (SCZ), this conversation often breaks down. But for a long time, scientists trying to understand why were looking at the city from a helicopter, seeing only the general traffic flow (bulk tissue). They couldn't see the specific mechanics of the individual cars or the drivers.

This paper is like sending a team of microscopic inspectors down to the street level to look at the actual drivers (neurons) and their engines (proteins) to see what's going wrong.

Here is a simple breakdown of what they found, using some everyday analogies:

1. The New Microscope: Laser Capture Microdissection (LCM)

The Old Way: Imagine trying to understand a specific type of car by scooping up a bucket of water from a river that contains mud, leaves, fish, and cars. You can't tell which car is which. This is what "bulk tissue" studies did—they mixed all brain cells together.
The New Way: The researchers used a high-tech laser (like a super-precise pair of scissors) to snip out only the specific drivers (excitatory projection neurons) that travel between the library and the CEO's office. This gave them a crystal-clear view of the actual machinery, free from the "mud" of other cell types.

2. The Translation Problem: Reading the Blueprint vs. Building the House

The Analogy: Think of RNA (transcriptomics) as the architectural blueprints, and Proteins (proteomics) as the actual bricks and mortar built from those blueprints.
The Finding: Usually, just because a blueprint says "build a wall" doesn't mean the wall gets built perfectly. Sometimes the blueprints are stable, but the construction crew gets confused.
The Discovery: The team found that the blueprints and the bricks matched up very well only when the blueprint was "stable" (didn't change much over time). In Schizophrenia patients, the blueprints for certain "glue" proteins (cell junctions) were surprisingly stable, but the blueprints for the "synapse" (the connection points between neurons) were wobbly and unstable. This suggests that in Schizophrenia, the brain is struggling to keep its connection points steady.

3. The One-Way Street of Risk

The Analogy: Imagine a relay race. The Hippocampus (CA1) passes a baton to the Subiculum (SUB), which then passes it to the CEO's office.
The Finding: The researchers discovered that the "risk" for Schizophrenia flows in one specific direction. It's like a broken baton being passed from the CA1 (the start of the race) that messes up the runner in the SUB (the next leg).
The Metaphor: It's not that the runner in the SUB is naturally slow; it's that the runner in the CA1 is handing them a baton that is already on fire. The study shows that genes related to Schizophrenia risk in the CA1 directly predict problems in the synaptic connections of the SUB, but not the other way around.

4. The Glue and the Guard

The Analogy: Neurons need to stick together (glue) and talk to each other (guards).
The Finding:

• Too much Glue: In Schizophrenia, the neurons are sticking too tightly to the "construction workers" (glial cells). It's like the construction crew is hugging the drivers too hard, distracting them.
• Missing Guards: The "security guards" (inhibitory neurons) who usually calm down the drivers are missing in action. Specifically, the guards that use chemical messengers like Somatostatin and NPY are gone. Without these guards, the drivers (excitatory neurons) get too excited and chaotic.

5. The "Solute Carrier" Problem

The researchers found a specific group of genes called "Solute Carriers" (SLC) that were acting up.
The Analogy: Imagine these genes are the porters in a hotel who carry luggage (ions and nutrients) in and out of rooms.
The Finding: In Schizophrenia, the porters in the library (CA1) and the CEO's office (DLPFC) are dropping the luggage. They aren't moving the right chemicals in and out of the cells. This disrupts the balance of electricity in the brain, making it hard for the library and the CEO's office to understand each other.

The Big Picture

This study is a huge step forward because it didn't just look at the "noise" of the whole brain. It zoomed in on the specific conversation between two critical areas.

In summary: Schizophrenia isn't just a general "glitch" in the brain. It looks like a specific breakdown in a relay race where:

1. The porters (SLC genes) are dropping the luggage.
2. The blueprints for the connections are shaking.
3. The security guards are missing, leaving the drivers too hyper.
4. And the "broken baton" is being passed from the hippocampus to the rest of the circuit, causing the whole system to stumble.

By understanding these specific mechanical failures, scientists hope to build better tools to fix the relay race, rather than just trying to calm down the whole city.

Technical Summary

1. Problem Statement

Schizophrenia (SCZ) is a complex psychiatric disorder with hundreds of identified genetic risk loci, yet the biological mechanisms by which these spatially distant genes converge to disrupt specific brain circuits remain unclear.

• Limitations of Current Approaches: Most postmortem studies rely on bulk tissue homogenates, which mix cell types and obscure cell-specific molecular signatures. While single-nucleus RNA sequencing (snRNA-seq) offers cell-type resolution, it captures only nuclear RNA, missing cytoplasmic mRNA transported to synapses, and suffers from 3' amplification bias and isoform resolution issues.
• The Gap: There is a lack of high-granularity, multi-omics data (transcriptome and proteome) specifically targeting excitatory projection neurons within the specific hippocampal-prefrontal circuit (CA1/Subiculum $\to$ DLPFC) implicated in SCZ pathogenesis. Furthermore, the relationship between transcriptomic and proteomic changes in this circuit, and the directional flow of risk across regions, remains poorly understood.

2. Methodology

The study employed a multi-modal approach using postmortem human brain tissue from 10 SCZ patients and 10 matched healthy controls (CTR).

• Sample Collection & Enrichment:
  • Laser Capture Microdissection (LCM): Used to isolate specific layers of projection neurons: Layer III pyramidal neurons in the Dorsolateral Prefrontal Cortex (DLPFC), and pyramidal neurons in the CA1 and Presubiculum (SUB) of the hippocampus. This ensured an excitatory neuron-enriched sample, avoiding the noise of bulk tissue.
  • Parallel snRNA-seq: Performed on CA1 tissue to assess cell-type interactions (neuron-glia-inhibitory) within the region.
• Omics Profiling:
  • Transcriptomics: RNA sequencing (RNA-seq) without pre-amplification to quantify gene and transcript levels.
  • Proteomics: Tandem Mass Tag (TMT) quantitative proteomics to measure protein abundance and phosphorylation states.
• Analytical Framework:
  • Comparative Analysis: Compared LCM data against bulk tissue data from the Lieber Institute for Brain Development (LIBD) repository.
  • Machine Learning: Random Forest classifiers were used to predict brain region identity based on gene, transcript, and protein data to assess information content.
  • Translatability & Stability: Calculated a "translatability score" (RNA-protein correlation) and correlated it with gene expression stability using longitudinal iPSC-derived neuron data (DIV56 vs. DIV70).
  • Network Analysis: Weighted Gene Co-expression Network Analysis (WGCNA) was used to identify SCZ-risk modules replicated across transcript and protein layers.
  • Interregional Connectivity: Novel techniques were used to compute interregional co-expression networks to test if SCZ risk genes in one region (e.g., CA1) predict synaptic gene expression in a downstream region (e.g., SUB).
  • Cell-Cell Communication: CellChat analysis was applied to snRNA-seq data to map ligand-receptor interactions between cell types in SCZ vs. CTR.

3. Key Contributions

• LCM Superiority: Demonstrated that LCM-enriched samples provide significantly higher resolution for regional identity discrimination (>90% accuracy for transcriptome, >97% for proteome) compared to bulk tissue (<70%).
• Multi-Omics Integration: Established a robust pipeline for integrating transcriptomic and proteomic data from the same neuronal lysates, identifying genes with high "translatability" (consistent RNA-protein expression).
• Directional Circuit Risk: Developed a method to map the directional flow of SCZ risk, showing that risk genes in CA1 predict downstream synaptic dysfunction in SUB, but not vice versa.
• Phosphoproteomics in SCZ: Identified specific alterations in protein phosphorylation, particularly in the hippocampus, suggesting cytoskeletal and synaptic trafficking disruptions.

4. Key Results

A. Methodological Validation & Granularity

• Cell Type Enrichment: LCM samples showed a significant increase in excitatory neuronal subtypes (e.g., CA1 Excit A, DLPFC Excit A) compared to bulk tissue.
• Regional Identity: Machine learning models predicted brain regions with 92.5–94.6% accuracy using LCM transcript data and 97.4–97.5% using proteomic data, significantly outperforming bulk gene-level data (69%).
• Co-expression Strength: Neuronal gene sets showed significantly stronger co-expression in LCM data than in bulk, indicating better preservation of neuronal pathways.

B. Transcriptome-Proteome Translatability

• Stability Correlation: A strong positive relationship was found between gene-protein translatability and the temporal stability of gene expression in iPSC-derived neurons. Highly stable genes showed strong RNA-protein correlation.
• SCZ Effects: In SCZ, genes unique to SCZ-derived stability were enriched for cell adhesion and nuclear processes, whereas common stable genes were enriched for synaptic terms.
• Phosphorylation: SCZ patients exhibited significantly increased phosphorylation of peptides in the hippocampus (CA1 and SUB) compared to controls, implicating cytoskeletal processes.

C. SCZ Risk Co-expression Networks

• Replicated Modules: Identified SCZ risk modules replicated across both transcript and protein networks.
• Region-Specific Ontologies:
  • DLPFC & CA1: Enriched for transmembrane transporter activity (specifically Solute Carrier genes like SLC12A5, SLC6A17).
  • SUB: Enriched for postsynaptic glutamatergic synapse processes (e.g., GRIA2, LRRTM2).
• Key Genes: Eight genes were replicated across regions, including SLC12A5 (KCC2, critical for GABA polarity), GRIA2 (AMPA receptor subunit), and LRRTM2 (AMPA receptor positioning).

D. Interregional Circuit Effects

• Directional Connectivity: SCZ risk genes in CA1 showed significant interregional connectivity to excitatory synaptic genes in the SUB (downstream), but not the reverse. This suggests a unidirectional mechanism where CA1 dysfunction drives SUB synaptic deficits.
• Diagnosis Effects: Neuronal gene sets showed reduced transcriptomic coupling (coordination) between CA1 and SUB in SCZ patients compared to controls, though this did not survive strict multiple comparison correction due to sample size.

E. Cell-Cell Interactions (snRNA-seq)

• Glia vs. Neurons: SCZ CA1 excitatory neurons showed increased interactions with glial cells (related to cell junctions/adhesion).
• Inhibitory Loss: There was a significant decrease in interactions between excitatory and inhibitory neurons in SCZ.
• Neuropeptide Deficit: Specific canonical inhibitory inputs (Somatostatin and Neuropeptide Y) to CA1 excitatory clusters were absent or reduced in SCZ, suggesting a disruption in excitatory-inhibitory balance.

5. Significance

• Mechanistic Insight: The study moves beyond bulk associations to propose a specific molecular mechanism: SCZ risk converges on transmembrane transporters in CA1 (e.g., SLC12A5), leading to disrupted excitatory-inhibitory balance and altered cytoskeletal dynamics (via phosphorylation). This dysfunction propagates directionally to impair postsynaptic glutamatergic function in the SUB.
• Circuit-Level Pathology: It provides molecular evidence for the "disrupted connectivity" phenotype observed in neuroimaging, linking it to specific gene networks and protein modifications in the hippocampal-prefrontal circuit.
• Methodological Advance: Validates LCM as a superior alternative to bulk tissue for studying circuit-specific neuropsychiatric disorders and demonstrates the necessity of integrating proteomics and phosphorylation data to capture post-transcriptional regulatory mechanisms missed by RNA-seq alone.
• Therapeutic Targets: Highlights specific targets such as AMPA receptor regulation (GRIA2, LRRTM2) and chloride transport (SLC12A5) as potential points of intervention for restoring circuit function in SCZ.