The Spatiotemporal Proteome Landscape of Aging: Structural determinants of age-sensitive proteome remodeling

By employing a robotic pipeline to analyze 90 million single-cell images across thousands of yeast strains, this study constructs an age-resolved proteome atlas that reveals widespread spatial remodeling during aging and identifies biophysical structural determinants that dictate why specific proteins are susceptible to age-related breakdown.

The Gist

Imagine your body as a bustling, high-tech city. Inside every cell of this city, there are millions of tiny workers (proteins) doing specific jobs. Some work in the power plant (mitochondria), some in the city hall (nucleus), and some in the recycling center (vacuole). For the city to run smoothly, these workers need to be in the right place, at the right concentration, and working together in organized teams.

This paper is like a massive, high-definition surveillance project that tracked every single worker in a yeast cell (a tiny, single-celled organism) as it got old. The researchers wanted to answer a big question: What actually goes wrong inside a cell as it ages?

Here is the story of their discovery, broken down into simple concepts:

1. The Problem: The City Gets Chaotic

We know that as we age, our cells get tired and make mistakes. But scientists didn't have a complete map of how the workers get confused. Do they just stop working? Or do they wander off to the wrong neighborhoods?

The researchers built a robotic pipeline (like a super-fast assembly line) to isolate thousands of "old" yeast cells. They took over 90 million 3D photos of these cells, creating a time-lapse movie of the entire workforce as they aged.

2. The Discovery: The "Lost and Found" Phenomenon

Using advanced AI (like a super-smart security camera system), they mapped where every protein was located. They found that aging doesn't just mean "less work"; it means total chaos in the office.

• The Drifting Workers: In young cells, workers stay in their assigned departments. In old cells, workers start wandering. A protein that usually works in the power plant might drift into the cytoplasm (the open office space).
• The Wrong Neighborhoods: Some proteins completely moved houses. For example, proteins that usually stay in the "nucleus" (the city hall) started leaking out, while others from the "mitochondria" (power plant) ended up in the "cytosol" (the hallway).
• The Broken Teams: Proteins that usually work together in tight-knit teams (complexes) started falling apart. It's like a construction crew where the bricklayers and the cement mixers stop talking to each other, leading to a crumbling building.

3. The "Why": It's Written in Their DNA (Literally)

The most surprising part of the study is why some proteins wander off while others stay put.

The researchers realized that a protein's physical shape and chemical surface determine its fate. Think of it like this:

• The Sticky Worker: Some proteins have "sticky" patches on their surface (specifically, certain chemical groups like cysteine and lysine). As the cell ages, the environment becomes more "oxidizing" (like rust forming). These sticky patches react with the environment, causing the protein to clump up or detach from its team.
• The Smooth Worker: Other proteins have smooth, stable surfaces that don't react as easily. They stay in their office and keep working, even as the cell gets old.

The study found that if you look at a protein's 3D blueprint (its structure), you can predict whether it will be a "wanderer" or a "stayer" as the cell ages. It's as if the protein's own design contains a hidden expiration date.

4. The Big Picture: A City in Decline

As the yeast cells aged, the researchers saw a pattern:

• Organelles got bloated but empty: The "rooms" in the cell got bigger (like a house expanding), but the furniture (proteins) didn't fill the extra space, making the rooms feel empty and inefficient.
• Communication broke down: The walls between departments (organelles) started to leak. The power plant stopped talking to the recycling center.
• The "Flow" Map: They created a map showing how proteins flowed from one room to another over time. It showed that the cell's internal logistics system completely rewired itself, often in the wrong direction.

The Takeaway

This paper is a massive step forward because it moves beyond just counting how many proteins exist. Instead, it looks at where they are and how they interact.

The Analogy:
Imagine a library. In a young library, books are perfectly sorted on the shelves, and librarians know exactly where everything is.
In an old library (aging cell), the books start falling off the shelves, ending up in the wrong sections, and the librarians (protein complexes) stop working together.
This study didn't just count the books; it figured out that the books with rough, sticky covers are the ones most likely to get lost, while the smooth, hardcover books stay on the shelf.

Why it matters:
By understanding that a protein's physical structure makes it vulnerable to aging, scientists might be able to design drugs to "smooth out" those sticky patches or protect the vulnerable proteins. This could help us keep our cellular cities running smoothly for much longer.

Technical Summary

1. Problem Statement

Aging is characterized by a decline in cellular function and widespread changes in mRNA and protein abundance. However, a critical gap exists in understanding how the spatiotemporal organization of the proteome (localization, concentration, and interaction states) changes at the single-cell level during aging.

• Limitations of Current Knowledge: Most omics studies focus on bulk abundance, failing to capture subcellular relocalization, aggregation, or local concentration changes. There is a known uncoupling between mRNA and protein levels with age.
• Technical Barrier: Replicative aged cells are rare and fragile, making systematic, proteome-wide enrichment and imaging technically challenging. Previous studies were limited to hypothesis-driven, low-throughput analyses of individual proteins.
• Core Question: Does aging broadly remodel subcellular protein localization and concentration? What biophysical or structural properties make certain proteins susceptible to age-associated disorganization while others remain stable?

2. Methodology

The authors developed a comprehensive, high-throughput pipeline combining robotics, advanced microscopy, and deep learning to map the yeast proteome across the replicative lifespan.

• Biological Model: Saccharomyces cerevisiae (budding yeast) replicative aging.
• Strain Library: Utilized a seamless C-terminal mNeonGreen (mNG) tagged library covering 5,661 strains (94% of the yeast proteome), preserving native 3'UTRs to minimize regulatory perturbations.
• High-Throughput Robotic Pipeline:
  • Developed an automated enrichment protocol using a Tecan Fluent liquid-handling robot to isolate old mother cells via biotin-streptavidin magnetic bead labeling.
  • Acquired 90+ million single-cell 3D fluorescence images (467,456 image stacks) spanning young to old (15+ divisions) cells.
  • Stained cells with Calcofluor White (cell wall) and Wheat Germ Agglutinin (WGA) for bud scar counting.
• Deep Learning Analysis:
  • DeepAge: A YOLO-based neural network trained to count bud scars in 3D images to assign precise replicative age bins (accuracy: 93%).
  • Ensemble Localization Model: Combined a 2D transfer-learned DeepLoc model (for signal-to-noise robustness) and a 3D ResNet-50 model (for volumetric spatial context). This ensemble achieved 95% mean accuracy in classifying 17 subcellular compartments without needing additional reference markers.
  • PIFiA (Protein Image-based Functional Annotation): Used high-dimensional feature embeddings (2,560 dimensions) from the imaging models to infer protein-protein associations and complex stoichiometry based on morphological similarity.
• Complementary Biochemistry:
  • LiP-MS (Limited Proteolysis-Mass Spectrometry): Used to profile proteome-wide changes in protein surface accessibility (interaction interfaces) in young vs. aged cells.
  • Structural Modeling: Leveraged AlphaFold predicted structures to extract 41 biophysical features (e.g., surface cysteines, radius of gyration) to train logistic regression models predicting protein stability.

3. Key Contributions

1. First Age-Resolved Single-Cell Proteome Atlas: Generated a dataset of 90 million single-cell images mapping localization, concentration, and aggregation for nearly the entire yeast proteome across the replicative lifespan.
2. Methodological Innovation: Established a scalable robotic pipeline for enriching rare aged cells and developed a 3D deep learning ensemble that outperforms 2D methods in resolving complex subcellular topologies.
3. Discovery of Structural Determinants: Identified that intrinsic protein structural features (specifically surface chemistry and compactness) predict susceptibility to aging, bridging molecular biophysics with systems-level aging phenotypes.
4. Integration of Modalities: Successfully integrated imaging-derived interaction networks (PIFiA) with biochemical surface accessibility data (LiP-MS) to validate widespread remodeling of protein complexes.

4. Key Results

A. Proteome-Wide Spatial Remodeling

• Widespread Disorganization: Aging causes a "blurring" of compartmental boundaries. While most proteins shift within their original compartments (morphological changes), a distinct set of 215 proteins undergoes cross-compartment relocalization.
• Key Relocalization Events:
  • Nucleolus to Nucleoplasm: Ribosome biogenesis factors (e.g., Utp14) disperse, occurring earlier than Extrachromosomal rDNA circle (ERC) accumulation.
  • Nuclear to Cytosol: DNA repair factors (RPA complex) form cytosolic foci, suggesting mtDNA leakage and activation of innate immune-like signaling.
  • Mitochondrial to Cytosol/Vacuole/Peroxisome: Proteins like Leu4 (leucine biosynthesis) and Mnp1 (mitoribosomal) relocalize, potentially decoupling metabolic pathways.
• Inter-Organelle Connectivity: Aging weakens membrane contact sites (e.g., Nucleus-Vacuole Junctions, ER-Mitochondria, PM-ER), leading to a loss of coordinated lipid and calcium signaling.

B. Organelle Size and Concentration Dynamics

• Organelle Dilution: While absolute organelle sizes increase with cell growth, cell-normalized sizes decrease for most organelles (nucleus, mitochondria, peroxisomes), except the vacuole.
• Concentration vs. Abundance: Total protein abundance often increases with age, but local concentrations frequently decrease due to organelle expansion outpacing protein synthesis. This leads to "dilution" of functional machinery.
• Stoichiometric Imbalance: Multi-protein complexes (e.g., mitoribosomes, proteasomes) exhibit heterogeneous subunit behavior, with some subunits aggregating or decreasing while others remain stable, disrupting complex assembly.

C. Protein Interaction Network Rewiring

• Weakening of Complexes: PIFiA analysis revealed a broad weakening of feature-profile co-clustering among complex subunits in aged cells.
• Surface Accessibility Changes: LiP-MS confirmed that ~800 proteins exhibit significant changes in surface accessibility, indicating altered interaction interfaces.
• Correlation: ~62% of protein pairs showing altered co-clustering in imaging also showed altered surface accessibility in MS, validating the imaging-based inference of interaction remodeling.

D. Structural Determinants of Aging Susceptibility

• Predictive Modeling: A logistic regression model trained on 28 structural features (derived from AlphaFold) could distinguish "aging-stable" from "aging-unstable" proteins with high accuracy (AUC 0.86).
• Key Predictors: The strongest predictors of instability were:
  • Surface-exposed Cysteine and Lysine residues: High chemical reactivity makes them vulnerable to redox stress and non-enzymatic PTMs.
  • Isoelectric Point (pI): Surface charge properties.
  • pLDDT and Radius of Gyration: Lower confidence scores and larger radii (less compact/ordered structures) correlate with instability.
• Cross-Species Conservation: The yeast-trained model successfully predicted age-associated instability in human proteins (UUPAs/UDPAs) with an AUC of 0.71, suggesting these biophysical principles are evolutionarily conserved.

5. Significance

• Mechanistic Insight: The study moves beyond "what changes" to "why it changes," proposing that intrinsic structural fragility (encoded by surface chemistry and compactness) predisposes specific proteins to age-related dysfunction.
• Unifying Theory: It links stochastic molecular damage (e.g., oxidation of cysteines) to reproducible population-level aging hallmarks, suggesting that the proteome's architecture encodes its own "aging trajectory."
• Resource: Provides a foundational, single-cell, single-age atlas for the aging community to dissect mechanistic links between proteome breakdown and cellular senescence.
• Future Directions: The framework offers a scalable approach to test longevity interventions, environmental exposures, and disease states by monitoring spatiotemporal proteome integrity.

In summary, this work demonstrates that aging is not merely a loss of protein abundance but a structural and spatial breakdown of the proteome, driven by the intrinsic biophysical vulnerability of specific proteins to the aging cellular environment.