Deep homology and design of proteasome chaperone proteins in Candida auris

Jackson R. Rapala^1,2, Mohammad Siddiq^3,4, Patricia J. Wittkopp^3,4, Matthew J. O’Meara^2,5*, Teresa R. O’Meara^{1* 1}Department of Microbiology and Immunology, University of Michigan Medical School; Ann Arbor MI 48019 USA. ²Gilbert S. Omenn Department of Computational Medicine and Bioinformatics, University of Michigan Medical School; Ann Arbor MI 48019 USA. ³Department of Ecology and Evolutionary Biology, University of Michigan; Ann Arbor MI 48019 USA. ⁴Department of Molecular, Cellular, and Developmental Biology, University of Michigan; Ann Arbor MI 48019 USA. ⁵Department of Medicinal Chemistry, University of Michigan, Ann Arbor, MI 48109, USA ^*Corresponding author. Email: tromeara@umich.edu, maom@umich.edu

Abstract

A central tenet of biology is that protein structure mediates the sequence-function relationship. Recently, there has been excitement about the promise of advances in protein structure modeling to generate hypotheses about sequence-structure-function relationships based on successes with controlled benchmarks. Here, we leverage structural similarity to identify rapidly evolving proteasome assembly chaperones and characterize their function in the emerging fungal pathogen Candida auris. Despite the large sequence divergence, we demonstrate conservation of structure and function across hundreds of millions of years of evolution, representing a case of rapid neutral evolution. Using the functional constraints on structure from these naturally evolved sequences, we prospectively designed de novo chaperones and demonstrate that these artificial proteins can rescue complex biological processes in the context of the whole cell.

Analysis Scripts

This repository contains the analysis scripts for the study of C. auris Poc4 sequence, structure, function:

1. Gather sequence and structure homology
2. Redesign sequences and predict their structure
3. Analyze sequence and structure

To run these scripts, check out this repository

git clone https://github.com/maomlab/Poc4
cd Poc4
# follow the instructions in install.sh to setup the environment

Installation should be feasible in 30 minutes, but if you run into trouble please reach out for assitance.

Then run the scripts in the scripts directory one at a time. This will use data in the data/ directory, and create files in the intermediate_data/ and product/directories. This will use a mix of different local, compute-cluster jobs using both CPU and GPU resources. Specific paths and parameters may need to adjusted to work with your system.

The computational workflow builds on the following tools:

• Frame2seq
• ESM
• AlphaFold
• IPSEA
• FoldSeek
• ParallelFold
• MPLearn
• BioPython
• Apache Arrow
• H2O AutoML
• SLURM

Overview of the workflow

1. Generate initial templates of C. auris Poc4/IRC25 wildtype complex using AlphaFold3 (web)
2. Search for structural orthologs using FoldSeek (web)
3. Generate diverse sequnece designs for Poc4 given predicted Poc4/IRC25 backbone using Frame2seq on GPU
4. Compute MSAs for designs using ParallelFold CPU SLURM version of AlphaFold2
5. Predict structures for designs using ParallelFold GPU SLURM version of AlphaFold2
6. Generate sequence embeddings for Poc4, designs and orthologs
7. Plot and anlyze embeddings using MPLearn wrapper for UMAP and R

License and Attribution

The computational workflow is released under the MIT License. If you find this workflow useful, please cite:

Rapala JR, Siddiq M, Wittkopp PJ, O'Meara MJ, O'Meara Deep TR,
Deep homology and design of proteasome chaperone proteins in Candida auris,
bioRxiv (2025), DOI: 10.1101/2025.05.14.654010, PMCID: PMC12132334