📄 molecular biology

Rank Normalization Enables Cross Pathway Comparison in Degradation Resolved Proteome Turnover Analysis

This study introduces a rank normalization strategy that enables the cross-pathway comparison of protein turnover rates between the ubiquitin-proteasome and autophagy-lysosomal systems by converting absolute stabilization values into percentile scores, thereby revealing proteome-wide degradation preferences and providing quantitative evidence for proteaphagy.

Original authors: Gao, Z., Xiao, Q., Kelly, J. W., Powers, E. T.

Published 2026-10-02

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Gao, Z., Xiao, Q., Kelly, J. W., Powers, E. T.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Inside every living cell, a constant battle for balance rages. Proteins, the molecular machines that build and run our bodies, are constantly being made and broken down. This delicate equilibrium, known as proteostasis, is essential for health. When it fails, cells accumulate damaged or unnecessary proteins, a process linked to aging, cancer, and neurodegenerative diseases. To maintain this balance, cells rely on two main garbage disposal systems. One is the ubiquitin-proteasome system, a highly selective machine that chews up short-lived or misfolded proteins. The other is the autophagy-lysosome pathway, a broader system that can digest large chunks of cellular material, including entire organelles. Understanding which proteins go to which trash bin is crucial for understanding how cells stay healthy, but figuring out the specific route for thousands of different proteins at once has been a major analytical hurdle.

The problem is that these two disposal systems do not react to experimental interference in the same way. When scientists try to block one system to see what happens, the other often responds differently, creating a mismatch in the data. Imagine trying to compare the height of a mountain to the depth of a canyon by looking at a map where the vertical scale for mountains is stretched out and the scale for canyons is squashed; the comparison becomes misleading. In this study, researchers at The Scripps Research Institute faced exactly this problem. They wanted to map out which proteins rely on the proteasome and which rely on the lysosome, but the drugs they used to block these systems produced vastly different amounts of change. One drug caused massive shifts in protein lifespans, while the other caused only tiny shifts. If they simply compared the raw numbers, the results would be skewed, making it look like almost everything was being handled by the proteasome simply because that drug caused bigger changes.

To solve this, the team developed a new way of looking at the data that focuses on relative ranking rather than absolute size. They treated cells with a drug to block the proteasome and another to block the lysosome, then tracked how long thousands of proteins lasted before being broken down. Instead of comparing the raw time differences, they asked a simpler question: for each drug, how did a specific protein's behavior compare to all the other proteins in the same experiment? They converted these behaviors into a score from zero to one, representing where a protein stood in line relative to its peers. A protein that was among the most sensitive to the lysosome-blocker got a high score on that scale, even if the actual time it survived was shorter than a protein that was only moderately sensitive to the proteasome-blocker. This method, which they call rank normalization, allowed them to place both drugs on a common scale, revealing the true preferences of the proteins without being fooled by the different magnitudes of the drugs' effects.

Using this new approach, the researchers analyzed nearly four thousand proteins in human cells. They found that the two disposal systems have distinct preferences. Proteins that are part of the cell's internal membrane structures and those involved in making energy were most likely to be cleared by the lysosome. In contrast, proteins involved in building the cell's structural framework and those that help regulate the cell's internal environment were more likely to be handled by the proteasome. Most surprisingly, the study revealed that the proteasome itself is not immune to the lysosome. The researchers found that when the lysosome pathway was active, the components of the proteasome machine were preferentially sensitive to lysosomal inhibition, suggesting that the cell recycles its own garbage disposal units through the lysosome. This phenomenon, known as proteaphagy, had been seen in other organisms but was not well understood in human cells.

To confirm this unexpected finding, the team activated the cell's autophagy system using a drug that mimics starvation. They observed that the levels of proteasome components dropped, and they used a specialized imaging technique to see the proteasomes physically moving toward the lysosomes, where they would be broken down. This provided direct evidence that the cell can choose to degrade its own proteasome machinery when needed. The study does not claim to have solved every mystery of protein turnover, nor does it provide a final list of every protein's fate. Instead, it offers a robust method for comparing different cellular processes that behave differently. By shifting the focus from how much a protein changes to how it ranks among its peers, the researchers have provided a clearer map of cellular maintenance. This framework suggests that the cell's cleanup crews are not just random scavengers but are organized with specific preferences, ensuring that the right proteins are removed by the right system to keep the cell functioning smoothly.

Technical Summary: Rank Normalization Enables Cross Pathway Comparison in Degradation Resolved Proteome Turnover Analysis

Problem Statement
Protein homeostasis (proteostasis) relies on two primary intracellular degradation systems: the ubiquitin-proteasome system (UPS) and the autophagy-lysosome pathway (ALP). While pulse stable isotope labeling by amino acids in cell culture (pSILAC) allows for the quantification of protein turnover rates and the identification of degradation pathways, a significant analytical challenge arises when comparing responses to different pathway perturbations. Specifically, pharmacological inhibitors of the UPS (e.g., marizomib/MZB) and ALP (e.g., bafilomycin A1/BafA1) generate responses with markedly different dynamic ranges and effect size distributions.

MZB directly and irreversibly inhibits proteasome catalytic subunits, often producing large absolute increases in protein half-lives. In contrast, BafA1 inhibits lysosomal acidification indirectly, typically resulting in smaller absolute half-life shifts. Consequently, direct comparison of absolute half-life changes or fold-changes is biased; proteins may appear preferentially dependent on the UPS simply because the UPS inhibitor produces a broader response distribution, rather than reflecting true biological pathway preference. Furthermore, these perturbations are mechanistically asymmetric and can trigger distinct secondary stress responses (e.g., the integrated stress response upon proteasome inhibition), complicating the interpretation of absolute stabilization magnitudes. A standardized framework to address this dynamic range mismatch in comparative turnover analysis was previously lacking.

Methodology
The authors employed a multi-step experimental and analytical strategy in HeLa cells:

Experimental Design: Two independent biological experiments were conducted using pSILAC. Cells were switched from light to heavy isotope-labeled media at the time of treatment with vehicle (DMSO), MZB, or BafA1. Samples were collected at offset time series (Experiment 1: 1, 2, 5, 8, 16, 24 h; Experiment 2: 1, 3, 6, 12, 18, 24 h) to broaden temporal coverage.
Kinetic Fitting: Using the KdeggeR framework, the authors estimated protein loss rates ( $k_{loss}$ ) by fitting the decay of relative isotope abundance (RIA) of pre-existing light-labeled proteins. Degradation rate constants ( $k_{deg}$ ) were derived by correcting for cell division dilution using a distribution-based inference method rather than measured division rates. Half-lives ( $t_{1/2}$ ) were calculated as $\ln(2)/k_{deg}$ .
Rank Normalization Strategy: To resolve the mismatch in dynamic ranges, the authors developed a within-condition rank normalization approach:
- Log2 fold changes in half-life (relative to DMSO) were calculated for each protein under MZB and BafA1 conditions.
- Proteins were ranked by the magnitude of their response within each condition.
- Ranks were converted to percentile scores using the Hazen plotting position: $p = (r - 0.5)/n$ , where $r$ is the rank and $n$ is the total number of proteins.
- This transformation places both perturbations on a common relative response scale (0 to 1), where values below 0.5 indicate stronger stabilization by MZB treatment, and values above 0.5 indicate stronger stabilization by BafA1 treatment.
Biological Validation: The normalized data were integrated with a curated Proteostasis Network (PN) annotation. Additionally, the authors performed orthogonal validation using mTOR inhibition (Torin1) to induce autophagy, followed by immunoblotting and proximity ligation assays (PLA) to assess proteasome-LC3 interactions.

Key Results

Dynamic Range Mismatch: Direct comparison of raw data confirmed a severe bias. The median half-life increase under MZB was 11.3 hours, compared to only 1.5 hours under BafA1. Proteins like GAPDH and EGFR, known to have lysosomal degradation routes, appeared strongly proteasome-preferring based on absolute stabilization magnitude alone.
Normalization Efficacy: The rank normalization successfully decoupled pathway preference from perturbation magnitude. The resulting 2D preference map (MZB percentile vs. BafA1 percentile) revealed four distinct response regimes.
Functional Organization: Gene ontology enrichment of the four regimes revealed coherent biological patterns. For instance, proteins preferentially stabilized by lysosomal inhibition were enriched for mitochondrial and organellar ribosome terms, while those stabilized by proteasome inhibition were enriched for sensory and eye development terms.
Proteostasis Network (PN) Insights: Integration with PN annotations revealed that many components of the degradation machinery itself are subject to coordinated regulation. Notably, both proteasome core particle and regulatory particle subunits showed significant relative sensitivity to lysosomal inhibition (ALP preference), supporting the concept of "proteaphagy" (selective autophagic turnover of proteasomes).
Orthogonal Validation: Treatment with Torin1 reduced proteasome regulatory subunit abundance and increased the proximity between proteasome subunits (PSMB5) and the autophagy marker LC3. This effect was further enhanced by co-treatment with BafA1, confirming that proteasomes are recruited to autophagic structures and degraded via the lysosomal pathway upon autophagy activation.

Significance and Claims
The paper establishes rank normalization as a practical analytical strategy for comparing degradation-resolved proteome turnover across perturbations with unequal response distributions. The authors claim that this method allows researchers to:

Report Relative Pathway Preference: The normalized scores indicate how strongly a protein responds to a specific perturbation relative to the global response of that condition, rather than estimating the absolute fraction of turnover mediated by a pathway.
Resolve Obscured Biology: By removing the bias of effect size magnitude, the method reveals coordinated regulation of proteostasis machinery (e.g., lysosomal turnover of proteasomes) that is difficult to detect via direct comparison of absolute half-life changes.
General Applicability: While demonstrated here with proteasome and lysosome inhibitors in HeLa cells, the framework is applicable to other turnover datasets involving genetic perturbations, alternative inhibitors, or targeted protein degradation strategies (e.g., PROTACs, AUTOTACs), provided the scores are interpreted as relative responses within each perturbation.

The authors maintain a modest scope, acknowledging that the method addresses the analytical consequence of unequal distributions but does not eliminate protein-specific secondary effects (e.g., stress responses) or distinguish between biological pathway usage and pharmacological mechanism. They emphasize that orthogonal perturbations remain necessary for validating individual pathway assignments.

More like this