Pervasive mislocalization of pathogenic coding variants underlying human disorders

Jessica Lacoste, Marzieh Haghighi, Shahan Haider, Zhen‐Yuan Lin, Dmitri Segal, Chloe Reno, Wesley Wei Qian, Xueting Xiong, Hamdah Shafqat Abbasi, Pearl V. Ryder, Rebecca A. Senft, Beth A. Cimini, Frederick P. Roth, Michael S. Calderwood, David E. Hill, Marc Vidal, S. Stephen Yi, Nidhi Sahni, Jian Peng, Anne‐Claude Gingras, Shantanu Singh, Anne E. Carpenter, Mikko TaipalebioRxiv 2023 · DOI

Abstract

Widespread sequencing has yielded thousands of missense variants predicted or confirmed as disease-causing. This creates a new bottleneck: determining the functional impact of each variant - largely a painstaking, customized process undertaken one or a few genes or variants at a time. Here, we established a high-throughput imaging platform to assay the impact of coding variation on protein localization, evaluating 3,547 missense variants of over 1,000 genes and phenotypes. We discovered that mislocalization is a common consequence of coding variation, affecting about one-sixth of all pathogenic missense variants, all cellular compartments, and recessive and dominant disorders alike. Mislocalization is primarily driven by effects on protein stability and membrane insertion rather than disruptions of trafficking signals or specific interactions. Furthermore, mislocalization patterns help explain pleiotropy and disease severity and provide insights on variants of unknown significance. Our publicly available resource will likely accelerate the understanding of coding variation in human diseases.

Introduction

We are currently in an era of "genomic glut." While we can sequence human genomes rapidly, we struggle to understand the functional impact of the thousands of missense variants (single-letter changes in DNA) we discover. These are often labeled as "Variants of Unknown Significance" (VUS), creating a massive bottleneck in clinical diagnosis.

While previous research has focused heavily on how mutations affect protein stability or specific protein-protein interactions, the impact on subcellular localization has been largely ignored. If a protein is supposed to be on the cell surface to act as a channel but gets stuck in the center of the cell, it cannot function. Computational tools currently predict that fewer than 2% of variants cause mislocalization, but this study suggests the true scale of the problem is much larger.

Methods and Experimental Design

To address this gap, researchers built a massive experimental pipeline to move beyond one-gene-at-a-time studies. The study utilized several key libraries to cover a broad range of human biology, including the hmORFeome 1.1 (3,046 variants), a kinase/chaperone library, and a cancer library, totaling 3,547 variants across 1,282 genes.

Overview of the experimental pipeline, dataset characteristics, and validation metrics.

The authors expressed these protein variants in human cells (HeLa, RPE1, and U2OS) and used automated confocal microscopy to capture millions of images. To ensure accuracy, every image was analyzed both by human experts (visual annotation) and a custom computational pipeline using CellProfiler to extract 1,293 morphological features per cell. They developed an "Impact Score" based on the Pearson correlation between the wild-type (normal) protein's distribution and the mutant's distribution. Any variant flagged as "mislocalized" underwent a second, independent round of testing to confirm the result.

Key Insights and Results

The study revealed that mislocalization is a pervasive mechanism of disease, affecting roughly 16% (one-sixth) of pathogenic variants. This makes mislocalization nearly as common a disease mechanism as the loss of protein-protein interactions.

Key findings include:

  • Secretory Pathway Vulnerability: Proteins that travel through the ER and Golgi are the most vulnerable; 23% of pathogenic variants in these proteins were mislocalized.
  • Stability Over Signals: Surprisingly, most proteins do not get lost because their "zip code" (signal peptide) is broken. Instead, mutations make the protein unstable, causing the cell's quality control machinery (like Hsp70 chaperones) to trap them in the ER for degradation.
  • Transmembrane Domain (TMD) Sensitivity: Mutations in TMDs were four times more likely to cause mislocalization than mutations elsewhere in the protein.
  • Predicting Severity: The degree of mislocalization often matched clinical severity. For example, variants causing mild Pelizaeus-Merzbacher disease showed partial membrane localization, while severe variants showed a total loss of membrane presence.
  • VUS Reclassification: The study provided functional evidence to suggest that several variants currently listed as "VUS" in clinical databases should be reclassified as pathogenic.

Discussion and Future Directions

While this high-throughput platform represents a significant advance, some limitations remain. Most of the screen was performed in HeLa cells; while localization is generally consistent across cell lines, some tissue-specific proteins might behave differently in their native environments. Additionally, the use of 3xFLAG-V5 tags, while small, could potentially alter the behavior of highly sensitive proteins.

Future work may involve moving from "arrayed" screens to "pooled" optical screens using DNA barcoding, which could allow for the profiling of all ~200,000 known human pathogenic variants. Furthermore, this platform could be used for phenotypic screening to find drugs that "rescue" mislocalized proteins, offering a path toward new treatments for genetic disorders. The authors have released the entire dataset of images and scores as an open-access resource for the scientific community.

Summary updated

Abstract

Widespread sequencing has yielded thousands of missense variants predicted or confirmed as disease-causing. This creates a new bottleneck: determining the functional impact of each variant - largely a painstaking, customized process undertaken one or a few genes or variants at a time. Here, we established a high-throughput imaging platform to assay the impact of coding variation on protein localization, evaluating 3,547 missense variants of over 1,000 genes and phenotypes. We discovered that mislocalization is a common consequence of coding variation, affecting about one-sixth of all pathogenic missense variants, all cellular compartments, and recessive and dominant disorders alike. Mislocalization is primarily driven by effects on protein stability and membrane insertion rather than disruptions of trafficking signals or specific interactions. Furthermore, mislocalization patterns help explain pleiotropy and disease severity and provide insights on variants of unknown significance. Our publicly available resource will likely accelerate the understanding of coding variation in human diseases.
BibTeX
@misc{lacoste2023pervasive,
  author = {Jessica Lacoste and Marzieh Haghighi and Shahan Haider and Zhen‐Yuan Lin and Dmitri Segal and Chloe Reno and Wesley Wei Qian and Xueting Xiong and Hamdah Shafqat Abbasi and Pearl V. Ryder and Rebecca A. Senft and Beth A. Cimini and Frederick P. Roth and Michael S. Calderwood and David E. Hill and Marc Vidal and S. Stephen Yi and Nidhi Sahni and Jian Peng and Anne‐Claude Gingras and Shantanu Singh and Anne E. Carpenter and Mikko Taipale},
  title = {{Pervasive mislocalization of pathogenic coding variants underlying human disorders}},
  year = {2023},
  doi = {10.1101/2023.09.05.556368},
  howpublished = {bioRxiv},
}
Figure