YeastSpotter: accurate and parameter-free web segmentation for microscopy images of yeast cells

Rational Tuning of the Concentration-independent Enrichment of Prion-like Domains in Stress Granules

Stress granules (SGs) are large ribonucleoprotein assemblies that form in response to acute stress in eukaryotes. SG formation is thought to be initiated by liquid-liquid phase separation (LLPS) of key proteins and RNA. These molecules serve as a scaffold for recruitment of client molecules. LLPS of scaffold proteins in vitro is highly concentration-dependent, yet biomolecular condensates in vivo contain hundreds of unique proteins, most of which are thought to be clients rather than scaffolds. Many proteins that localize to SGs contain low-complexity, prion-like domains (PrLDs) that have been implicated in LLPS and SG recruitment. The degree of enrichment of proteins in biomolecular condensates such as SGs can vary widely, but the underlying basis for these differences is not fully understood. Here, we develop a toolkit of model PrLDs to examine the factors that govern efficiency of PrLD recruitment to stress granules. Recruitment was highly sensitive to amino acid composition: enrichment in SGs could be tuned through subtle changes in hydrophobicity. By contrast, SG recruitment was largely insensitive to PrLD concentration at both a population level and single-cell level. These observations point to a model wherein PrLDs are enriched in SGs through either simple solvation effects or interactions that are effectively non-saturable even at high expression levels.

One-pot synthesis of cellobiose from sucrose using sucrose phosphorylase and cellobiose phosphorylase co-displaying Pichia pastoris as a reusable whole-cell biocatalyst

Cellobiose has received increasing attention in various industrial sectors, ranging from food and feed to cosmetics. The development of large-scale cellobiose applications requires a cost-effective production technology as currently used methods based on cellulose hydrolysis are costly. Here, a one-pot synthesis of cellobiose from sucrose was conducted using a recombinant Pichia pastoris strain as a reusable whole-cell biocatalyst. Thermophilic sucrose phosphorylase from Bifidobacterium longum (BlSP) and cellobiose phosphorylase from Clostridium stercorarium (CsCBP) were co-displayed on the cell surface of P. pastoris via a glycosylphosphatidylinositol-anchoring system. Cells of the BlSP and CsCBP co-displaying P. pastoris strain were used as whole-cell biocatalysts to convert sucrose to cellobiose with commercial thermophilic xylose isomerase. Cellobiose productivity significantly improved with yeast cells grown on glycerol compared to glucose-grown cells. In one-pot bioconversion using glycerol-grown yeast cells, approximately 81.2 g/L of cellobiose was produced from 100 g/L of sucrose, corresponding to 81.2% of the theoretical maximum yield, within 24 h at 60 °C. Moreover, recombinant yeast cells maintained a cellobiose titer > 80 g/L, even after three consecutive cell-recycling one-pot bioconversion cycles. These results indicated that one-pot bioconversion using yeast cells displaying two phosphorylases as whole-cell catalysts is a promising approach for cost-effective cellobiose production.

Automated, image-based quantification of peroxisome characteristics with perox-per-cell

SUMMARY

perox-per-cell automates cumbersome, image-based data collection tasks often encountered in peroxisome research. The software processes microscopy images to quantify peroxisome features in yeast cells. It uses off-the-shelf image processing tools to automatically segment cells and peroxisomes and then outputs quantitative metrics including peroxisome counts per cell and spatial areas. In validation tests, we found that perox-per-cell output agrees well with manually quantified peroxisomal counts and cell instances, thereby enabling high-throughput quantification of peroxisomal characteristics.

AVAILABILITY AND IMPLEMENTATION

The software is coded in Python. Compiled executables and source code are available at https://github.com/AitchisonLab/perox-per-cell.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

High-throughput classification of S. cerevisiae tetrads using deep learning

Meiotic crossovers play a vital role in proper chromosome segregation and evolution of most sexually reproducing organisms. Meiotic recombination can be visually observed in Saccharomyces cerevisiae tetrads using linked spore-autonomous fluorescent markers placed at defined intervals within the genome, which allows for analysis of meiotic segregation without the need for tetrad dissection. To automate the analysis, we developed a deep learning-based image recognition and classification pipeline for high-throughput tetrad detection and meiotic crossover classification. As a proof of concept, we analyzed a large image data set from wild-type and selected gene knock-out mutants to quantify crossover frequency, interference, chromosome missegregation, and gene conversion events. The deep learning-based method has the potential to accelerate the discovery of new genes involved in meiotic recombination in S. cerevisiae such as the underlying factors controlling crossover frequency and interference.

Ordered and disordered regions of the Origin Recognition Complex direct differential in vivo binding at distinct motif sequences

The Origin Recognition Complex (ORC) seeds replication-fork formation by binding to DNA replication origins, which in budding yeast contain a 17bp DNA motif. High resolution structure of the ORC-DNA complex revealed two base-interacting elements: a disordered basic patch (Orc1-BP4) and an insertion helix (Orc4-IH). To define the ORC elements guiding its DNA binding in vivo, we mapped genomic locations of 38 designed ORC mutants, revealing that different ORC elements guide binding at different sites. At silencing-associated sites lacking the motif, ORC binding and activity were fully explained by a BAH domain. Within replication origins, we reveal two dominating motif variants showing differential binding modes and symmetry: a non-repetitive motif whose binding requires Orc1-BP4 and Orc4-IH, and a repetitive one where another basic patch, Orc1-BP3, can replace Orc4-IH. Disordered basic patches are therefore key for ORC-motif binding in vivo, and we discuss how these conserved, minor-groove interacting elements can guide specific ORC-DNA recognition.

Toxicity of the model protein 3×GFP arises from degradation overload, not from aggregate formation

Although protein aggregation can cause cytotoxicity, such aggregates can also form to mitigate cytotoxicity from misfolded proteins, although the nature of these contrasting aggregates remains unclear. We previously found that overproduction (op) of a three green fluorescent protein-linked protein (3×GFP) induces giant aggregates and is detrimental to growth. Here, we investigated the mechanism of growth inhibition by 3×GFP-op using non-aggregative 3×MOX-op as a control in Saccharomyces cerevisiae. The 3×GFP aggregates were induced by misfolding, and 3×GFP-op had higher cytotoxicity than 3×MOX-op because it perturbed the ubiquitin-proteasome system. Static aggregates formed by 3×GFP-op dynamically trapped Hsp70 family proteins (Ssa1 and Ssa2 in yeast), causing the heat-shock response. Systematic analysis of mutants deficient in the protein quality control suggested that 3×GFP-op did not cause a critical Hsp70 depletion and aggregation functioned in the direction of mitigating toxicity. Artificial trapping of essential cell cycle regulators into 3×GFP aggregates caused abnormalities in the cell cycle. In conclusion, the formation of the giant 3×GFP aggregates itself is not cytotoxic, as it does not entrap and deplete essential proteins. Rather, it is productive, inducing the heat-shock response while preventing an overload to the degradation system.

Automated, image-based quantification of peroxisome characteristics with perox-per-cell

perox-per-cell automates cumbersome, image-based data collection tasks often encountered in peroxisome research. The software processes microscopy images to quantify peroxisome features in yeast cells. It uses off-the-shelf image processing tools to automatically segment cells and peroxisomes and then outputs quantitative metrics including peroxisome counts per cell and spatial areas. In validation tests, we found that perox-per-cell output agrees well with manually-quantified peroxisomal counts and cell instances, thereby enabling high-throughput quantification of peroxisomal characteristics. The software is available at https://github.com/AitchisonLab/perox-per-cell.

PyF2F: a robust and simplified fluorophore-to-fluorophore distance measurement tool for Protein interactions from Imaging Complexes after Translocation experiments

Structural knowledge of protein assemblies in their physiological environment is paramount to understand cellular functions at the molecular level. Protein interactions from Imaging Complexes after Translocation (PICT) is a live-cell imaging technique for the structural characterization of macromolecular assemblies in living cells. PICT relies on the measurement of the separation between labelled molecules using fluorescence microscopy and cell engineering. Unfortunately, the required computational tools to extract molecular distances involve a variety of sophisticated software programs that challenge reproducibility and limit their implementation to highly specialized researchers. Here we introduce PyF2F, a Python-based software that provides a workflow for measuring molecular distances from PICT data, with minimal user programming expertise. We used a published dataset to validate PyF2F's performance.

A system for inducible mitochondria-specific protein degradation in vivo

Targeted protein degradation systems developed for eukaryotes employ cytoplasmic machineries to perform proteolysis. This has prevented mitochondria-specific analysis of proteins that localize to multiple locations, for example, the mitochondria and the nucleus. Here, we present an inducible mitochondria-specific protein degradation system in Saccharomyces cerevisiae based on the Mesoplasma florum Lon (mf-Lon) protease and its corresponding ssrA tag (called PDT). We show that mitochondrially targeted mf-Lon protease efficiently and selectively degrades a PDT-tagged reporter protein localized to the mitochondrial matrix. The degradation can be induced by depleting adenine from the medium, and tuned by altering the promoter strength of the MF-LON gene. We furthermore demonstrate that mf-Lon specifically degrades endogenous, PDT-tagged mitochondrial proteins. Finally, we show that mf-Lon-dependent PDT degradation can also be achieved in human mitochondria. In summary, this system provides an efficient tool to selectively analyze the mitochondrial function of dually localized proteins.

From pixels to insights: Machine learning and deep learning for bioimage analysis

Bioimage analysis plays a critical role in extracting information from biological images, enabling deeper insights into cellular structures and processes. The integration of machine learning and deep learning techniques has revolutionized the field, enabling the automated, reproducible, and accurate analysis of biological images. Here, we provide an overview of the history and principles of machine learning and deep learning in the context of bioimage analysis. We discuss the essential steps of the bioimage analysis workflow, emphasizing how machine learning and deep learning have improved preprocessing, segmentation, feature extraction, object tracking, and classification. We provide examples that showcase the application of machine learning and deep learning in bioimage analysis. We examine user-friendly software and tools that enable biologists to leverage these techniques without extensive computational expertise. This review is a resource for researchers seeking to incorporate machine learning and deep learning in their bioimage analysis workflows and enhance their research in this rapidly evolving field.

A computer vision and residual neural network (ResNet) combined method for automated and accurate yeast replicative aging analysis of high-throughput microfluidic single-cell images

With the rapid development of microfluidic platforms in high-throughput single-cell culturing, laborious operation to manipulate massive budding yeast cells (Saccharomyces cerevisiae) in replicative aging studies has been greatly simplified and automated. As a result, large datasets of microscopy images bring challenges to fast and accurately determine yeast replicative lifespan (RLS), which is the most important parameter to study cell aging. Based on our microfluidic diploid yeast long-term culturing (DYLC) chip that features 1100 traps to immobilize single cells and record their proliferation and aging via time-lapse imaging, herein, a dedicated algorithm combined with computer vision and residual neural network (ResNet) was presented to efficiently process tremendous micrographs in a high-throughput and automated manner. The image-processing algorithm includes following pivotal steps: (i) segmenting multi-trap micrographs into time-lapse single-trap sub-images, (ii) labeling 8 yeast budding features and training the 18-layer ResNet, (iii) converting the ResNet predictions in analog values into digital signals, (iv) recognizing cell dynamic events, and (v) determining yeast RLS and budding time interval (BTI) ultimately. The ResNet algorithm achieved high F1 scores (over 92%) demonstrating the effectiveness and accuracy in the recognition of yeast budding events, such as bud appearance, daughter dissection and cell death. Therefore, the results conduct that similar deep learning algorithms could be tailored to analyze high-throughput microscopy images and extract multiple cell behaviors in microfluidic single-cell analysis.

Optogenetic spatial patterning of cooperation in yeast populations

Microbial communities are shaped by complex metabolic interactions such as cooperation and competition for resources. Methods to control such interactions could lead to major advances in our ability to better engineer microbial consortia for synthetic biology applications. Here, we use optogenetics to control SUC2 invertase production in yeast, thereby shaping spatial assortment of cooperator and cheater cells. Yeast cells behave as cooperators (i.e., transform sucrose into hexose, a public good) upon blue light illumination or cheaters (i.e., consume hexose produced by cooperators to grow) in the dark. We show that cooperators benefit best from the hexoses they produce when their domain size is constrained between two cut-off length-scales. From an engineering point of view, the system behaves as a bandpass filter. The lower limit is the trace of cheaters' competition for hexoses, while the upper limit is defined by cooperators' competition for sucrose. Cooperation mostly occurs at the frontiers with cheater cells, which not only compete for hexoses but also cooperate passively by letting sucrose reach cooperators. We anticipate that this optogenetic method could be applied to shape metabolic interactions in a variety of microbial ecosystems.

Using Dimensionality Reduction to Visualize Phenotypic Changes in High-Throughput Microscopy

High-throughput microscopy has enabled screening of cell phenotypes at unprecedented scale. Systematic identification of cell phenotype changes (such as cell morphology and protein localization changes) is a major analysis goal. Because cell phenotypes are high-dimensional, unbiased approaches to detect and visualize the changes in phenotypes are still needed. Here, we suggest that changes in cellular phenotype can be visualized in reduced dimensionality representations of the image feature space. We describe a freely available analysis pipeline to visualize changes in protein localization in feature spaces obtained from deep learning. As an example, we use the pipeline to identify changes in subcellular localization after the yeast GFP collection was treated with hydroxyurea.

Mesoscale molecular assembly is favored by the active, crowded cytoplasm

The mesoscale organization of molecules into membraneless biomolecular condensates is emerging as a key mechanism of rapid spatiotemporal control in cells1. Principles of biomolecular condensation have been revealed through in vitro reconstitution2. However, intracellular environments are much more complex than test-tube environments: They are viscoelastic, highly crowded at the mesoscale, and are far from thermodynamic equilibrium due to the constant action of energy-consuming processes3. We developed synDrops, a synthetic phase separation system, to study how the cellular environment affects condensate formation. Three key features enable physical analysis: synDrops are inducible, bioorthogonal, and have well-defined geometry. This design allows kinetic analysis of synDrop assembly and facilitates computational simulation of the process. We compared experiments and simulations to determine that macromolecular crowding promotes condensate nucleation but inhibits droplet growth through coalescence. ATP-dependent cellular activities help overcome the frustration of growth. In particular, actomyosin dynamics potentiate droplet growth by reducing confinement and elasticity in the mammalian cytoplasm, thereby enabling synDrop coarsening. Our results demonstrate that mesoscale molecular assembly is favored by the combined effects of crowding and active matter in the cytoplasm. These results move toward a better predictive understanding of condensate formation in vivo.

Peroxisomal compartmentalization of amino acid biosynthesis reactions imposes an upper limit on compartment size

Cellular metabolism relies on just a few redox cofactors. Selective compartmentalization may prevent competition between metabolic reactions requiring the same cofactor. Is such compartmentalization necessary for optimal cell function? Is there an optimal compartment size? Here we probe these fundamental questions using peroxisomal compartmentalization of the last steps of lysine and histidine biosynthesis in the fission yeast Schizosaccharomyces japonicus. We show that compartmentalization of these NAD+ dependent reactions together with a dedicated NADH/NAD+ recycling enzyme supports optimal growth when an increased demand for anabolic reactions taxes cellular redox balance. In turn, compartmentalization constrains the size of individual organelles, with larger peroxisomes accumulating all the required enzymes but unable to support both biosynthetic reactions at the same time. Our reengineering and physiological experiments indicate that compartmentalized biosynthetic reactions are sensitive to the size of the compartment, likely due to scaling-dependent changes within the system, such as enzyme packing density.

Evaluating single-cell variability in proteasomal decay

Gene expression is a stochastic process that leads to variability in mRNA and protein abundances even within an isogenic population of cells grown in the same environment. This variation, often called gene-expression noise, has typically been attributed to transcriptional and translational processes while ignoring the contributions of protein decay variability across cells. Here we estimate the single-cell protein decay rates of two degron GFPs in Saccharomyces cerevisiae using time-lapse microscopy. We find substantial cell-to-cell variability in the decay rates of the degron GFPs. We evaluate cellular features that explain the variability in the proteasomal decay and find that the amount of 20s catalytic beta subunit of the proteasome marginally explains the observed variability in the degron GFP half-lives. We propose alternate hypotheses that might explain the observed variability in the decay of the two degron GFPs. Overall, our study highlights the importance of studying the kinetics of the decay process at single-cell resolution and that decay rates vary at the single-cell level, and that the decay process is stochastic. A complex model of decay dynamics must be included when modeling stochastic gene expression to estimate gene expression noise.

The molecular grammar of protein disorder guiding genome-binding locations

Intrinsically disordered regions (IDRs) direct transcription factors (TFs) towards selected genomic occurrences of their binding motif, as exemplified by budding yeast's Msn2. However, the sequence basis of IDR-directed TF binding selectivity remains unknown. To reveal this sequence grammar, we analyze the genomic localizations of >100 designed IDR mutants, each carrying up to 122 mutations within this 567-AA region. Our data points at multivalent interactions, carried by hydrophobic-mostly aliphatic-residues dispersed within a disordered environment and independent of linear sequence motifs, as the key determinants of Msn2 genomic localization. The implications of our results for the mechanistic basis of IDR-based TF binding preferences are discussed.

The dynamin Vps1 mediates Atg9 transport to the sites of autophagosome formation

Autophagy is a key process in eukaryotes to maintain cellular homeostasis by delivering cellular components to lysosomes/vacuoles for degradation and reuse of the resulting metabolites. Membrane rearrangements and trafficking events are mediated by the core machinery of autophagy-related (Atg) proteins, which carry out a variety of functions. How Atg9, a lipid scramblase and the only conserved transmembrane protein within this core Atg machinery, is trafficked during autophagy remained largely unclear. Here, we addressed this question in yeast Saccharomyces cerevisiae and found that retromer complex and dynamin Vps1 mutants alter Atg9 subcellular distribution and severely impair the autophagic flux by affecting two separate autophagy steps. We provide evidence that Vps1 interacts with Atg9 at Atg9 reservoirs. In the absence of Vps1, Atg9 fails to reach the sites of autophagosome formation, and this results in an autophagy defect. The function of Vps1 in autophagy requires its GTPase activity. Moreover, Vps1 point mutants associated with human diseases such as microcytic anemia and Charcot-Marie-Tooth are unable to sustain autophagy and affect Atg9 trafficking. Together, our data provide novel insights on the role of dynamins in Atg9 trafficking and suggest that a defect in this autophagy step could contribute to severe human pathologies.

The monitoring of oil production process by deep learning based on morphology in oleaginous yeasts

BACKGROUND

Monitoring jar fermenter-cultured microorganisms in real time is important for controlling productivity of bioproducts in large-scale cultivation settings. Morphological data is used to understand the growth and fermentation states of these microorganisms during monitoring. Oleaginous yeasts are used for their high productivity of single-cell oils but the relationship between lipid productivity and morphology has not been elucidated in these organisms.

RESULTS

In this study, we investigated the relationship between the morphology of oleaginous yeasts (Lipomyces starkeyi and Rhodosporidium toruloides were used) and their cultivation state in a large-scale cultivation setting using a real-time monitoring system. We combined this with deep learning by feeding a large amount of high-definition cell images obtained from the monitoring system to a deep learning algorithm. Our results showed that the cell images could be grouped into 7 distinct groups and that a strong correlation existed between each group and its biochemical activity (growth and oil-productivity).

CONCLUSIONS

This is the first report describing the morphological variations of oleaginous yeasts in a large-scale cultivation, and describes a promising new avenue for improving productivity of microorganisms in large-scale cultivation through the use of a real-time monitoring system combined with deep learning.

KEY POINTS

• A real-time monitoring system followed the morphological change of oleaginous yeasts. • Deep learning grouped them into 7 distinct groups based on their morphology. • A correlation between the cultivation state and the shape of the yeast was observed.

microbeSEG: A deep learning software tool with OMERO data management for efficient and accurate cell segmentation

In biotechnology, cell growth is one of the most important properties for the characterization and optimization of microbial cultures. Novel live-cell imaging methods are leading to an ever better understanding of cell cultures and their development. The key to analyzing acquired data is accurate and automated cell segmentation at the single-cell level. Therefore, we present microbeSEG, a user-friendly Python-based cell segmentation tool with a graphical user interface and OMERO data management. microbeSEG utilizes a state-of-the-art deep learning-based segmentation method and can be used for instance segmentation of a wide range of cell morphologies and imaging techniques, e.g., phase contrast or fluorescence microscopy. The main focus of microbeSEG is a comprehensible, easy, efficient, and complete workflow from the creation of training data to the final application of the trained segmentation model. We demonstrate that accurate cell segmentation results can be obtained within 45 minutes of user time. Utilizing public segmentation datasets or pre-labeling further accelerates the microbeSEG workflow. This opens the door for accurate and efficient data analysis of microbial cultures.

Segmentation, tracking and cell cycle analysis of live-cell imaging data with Cell-ACDC

BACKGROUND

High-throughput live-cell imaging is a powerful tool to study dynamic cellular processes in single cells but creates a bottleneck at the stage of data analysis, due to the large amount of data generated and limitations of analytical pipelines. Recent progress on deep learning dramatically improved cell segmentation and tracking. Nevertheless, manual data validation and correction is typically still required and tools spanning the complete range of image analysis are still needed.

RESULTS

We present Cell-ACDC, an open-source user-friendly GUI-based framework written in Python, for segmentation, tracking and cell cycle annotations. We included state-of-the-art deep learning models for single-cell segmentation of mammalian and yeast cells alongside cell tracking methods and an intuitive, semi-automated workflow for cell cycle annotation of single cells. Using Cell-ACDC, we found that mTOR activity in hematopoietic stem cells is largely independent of cell volume. By contrast, smaller cells exhibit higher p38 activity, consistent with a role of p38 in regulation of cell size. Additionally, we show that, in S. cerevisiae, histone Htb1 concentrations decrease with replicative age.

CONCLUSIONS

Cell-ACDC provides a framework for the application of state-of-the-art deep learning models to the analysis of live cell imaging data without programming knowledge. Furthermore, it allows for visualization and correction of segmentation and tracking errors as well as annotation of cell cycle stages. We embedded several smart algorithms that make the correction and annotation process fast and intuitive. Finally, the open-source and modularized nature of Cell-ACDC will enable simple and fast integration of new deep learning-based and traditional methods for cell segmentation, tracking, and downstream image analysis. Source code: https://github.com/SchmollerLab/Cell_ACDC.

Transcription feedback dynamics in the wake of cytoplasmic mRNA degradation shutdown

In the last decade, multiple studies demonstrated that cells maintain a balance of mRNA production and degradation, but the mechanisms by which cells implement this balance remain unknown. Here, we monitored cells' total and recently-transcribed mRNA profiles immediately following an acute depletion of Xrn1-the main 5'-3' mRNA exonuclease-which was previously implicated in balancing mRNA levels. We captured the detailed dynamics of the adaptation to rapid degradation of Xrn1 and observed a significant accumulation of mRNA, followed by a delayed global reduction in transcription and a gradual return to baseline mRNA levels. We found that this transcriptional response is not unique to Xrn1 depletion; rather, it is induced earlier when upstream factors in the 5'-3' degradation pathway are perturbed. Our data suggest that the mRNA feedback mechanism monitors the accumulation of inputs to the 5'-3' exonucleolytic pathway rather than its outputs.

Massive expression of cysteine-containing proteins causes abnormal elongation of yeast cells by perturbing the proteasome

The enhanced green fluorescent protein (EGFP) is considered to be a harmless protein because the critical expression level that causes growth defects is higher than that of other proteins. Here, we found that overexpression of EGFP, but not a glycolytic protein Gpm1, triggered the cell elongation phenotype in the budding yeast Saccharomyces cerevisiae. By the morphological analysis of the cell overexpressing fluorescent protein and glycolytic enzyme variants, we revealed that cysteine content was associated with the cell elongation phenotype. The abnormal cell morphology triggered by overexpression of EGFP was also observed in the fission yeast Schizosaccharomyces pombe. Overexpression of cysteine-containing protein was toxic, especially at high-temperature, while the toxicity could be modulated by additional protein characteristics. Investigation of protein aggregate formation, morphological abnormalities in mutants, and transcriptomic changes that occur upon overexpression of EGFP variants suggested that perturbation of the proteasome by the exposed cysteine of the overexpressed protein causes cell elongation. Overexpression of proteins with relatively low folding properties, such as EGFP, was also found to promote the formation of SHOTA (Seventy kDa Heat shock protein-containing, Overexpression-Triggered Aggregates), an intracellular aggregate that incorporates Hsp70/Ssa1, which induces a heat shock response, while it was unrelated to cell elongation. Evolutionary analysis of duplicated genes showed that cysteine toxicity may be an evolutionary bias to exclude cysteine from highly expressed proteins. The overexpression of cysteine-less moxGFP, the least toxic protein revealed in this study, would be a good model system to understand the physiological state of protein burden triggered by ultimate overexpression of harmless proteins.

The Effect of Lithium on the Budding Yeast Saccharomyces cerevisiae upon Stress Adaptation

Lithium salts are used in the treatment of mood disorders, cancer, and Alzheimer's disease. It has been shown to prolong life span in several phyla; however, not yet in budding yeast. In our study, we investigate the influence of lithium on yeast cells' viability by characterizing protein aggregate formation, cell volume, and molecular crowding in the context of stress adaptation. While our data suggest a concentration-dependent growth inhibition caused by LiCl, we show an extended long-term survival rate as an effect of lithium addition upon glucose deprivation. We show that caloric restriction mitigates the negative impact of LiCl on cellular survival. Therefore, we suggest that lithium could affect glucose metabolism upon caloric restriction, which could explain the extended long-term survival observed in our study. We find furthermore that lithium chloride did not affect an immediate salt-induced Hsp104-dependent aggregate formation but cellular adaptation to H2O2 and acute glucose starvation. We presume that different salt types and concentrations interfere with effective Hsp104 recruitment or its ATP-dependent disaggregase activity as a response to salt stress. This work provides novel details of Li+ effect on live eukaryotic cells which may also be applicable in further research on the treatment of cancer, Alzheimer's, or other age-related diseases in humans.

YeastMate: Neural network-assisted segmentation of mating and budding events in S. cerevisiae

SUMMARY

Here, we introduce YeastMate, a user-friendly deep learning-based application for automated detection and segmentation of Saccharomyces cerevisiae cells and their mating and budding events in microscopy images. We build upon Mask R-CNN with a custom segmentation head for the subclassification of mother and daughter cells during lifecycle transitions. YeastMate can be used directly as a Python library or through a stand-alone GUI application and a Fiji plugin as easy to use frontends.

AVAILABILITY AND IMPLEMENTATION

The source code for YeastMate is freely available at https://github.com/hoerlteam/YeastMate under the MIT license. We offer installers for our software stack for Windows, macOS and Linux. A detailed user guide is available at https://yeastmate.readthedocs.io.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

A convolutional neural network for segmentation of yeast cells without manual training annotations

MOTIVATION

Single-cell time-lapse microscopy is a ubiquitous tool for studying the dynamics of complex cellular processes. While imaging can be automated to generate very large volumes of data, the processing of the resulting movies to extract high-quality single-cell information remains a challenging task. The development of software tools that automatically identify and track cells is essential for realizing the full potential of time-lapse microscopy data. Convolutional neural networks (CNNs) are ideally suited for such applications, but require great amounts of manually annotated data for training, a time-consuming and tedious process.

RESULTS

We developed a new approach to CNN training for yeast cell segmentation based on synthetic data and present (i) a software tool for the generation of synthetic images mimicking brightfield images of budding yeast cells and (ii) a convolutional neural network (Mask R-CNN) for yeast segmentation that was trained on a fully synthetic dataset. The Mask R-CNN performed excellently on segmenting actual microscopy images of budding yeast cells, and a density-based spatial clustering algorithm (DBSCAN) was able to track the detected cells across the frames of microscopy movies. Our synthetic data creation tool completely bypassed the laborious generation of manually annotated training datasets, and can be easily adjusted to produce images with many different features. The incorporation of synthetic data creation into the development pipeline of CNN-based tools for budding yeast microscopy is a critical step toward the generation of more powerful, widely applicable and user-friendly image processing tools for this microorganism.

AVAILABILITY AND IMPLEMENTATION

The synthetic data generation code can be found at https://github.com/prhbrt/synthetic-yeast-cells. The Mask R-CNN as well as the tuning and benchmarking scripts can be found at https://github.com/ymzayek/yeastcells-detection-maskrcnn. We also provide Google Colab scripts that reproduce all the results of this work.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

The zinc cluster transcription factor Rha1 is a positive filamentation regulator in Candida albicans

Zinc cluster transcription factors (TFs) are essential fungal regulators of gene expression. In the pathogen Candida albicans, the gene orf19.1604 encodes a zinc cluster TF regulating filament development. Hyperactivation of orf19.1604, which we have named RHA1 for Regulator of Hyphal Activity, generates wrinkled colony morphology under nonhyphal growth conditions, triggers filament formation, invasiveness, and enhanced biofilm formation and causes reduced virulence in the mouse model of systemic infection. The strain expressing activated Rha1 shows up-regulation of genes required for filamentation and cell-wall-adhesion-related proteins. Increased expression is also seen for the hyphal-inducing TFs Brg1 and Ume6, while the hyphal repressor Nrg1 is downregulated. Inactivation of RHA1 reduces filamentation under a variety of filament-inducing conditions. In contrast to the partial effect of either single mutant, the double rha1 ume6 mutant strain is highly defective in both serum- and Spider-medium-stimulated hyphal development. While the loss of Brg1 function blocks serum-stimulated hyphal development, this block can be significantly bypassed by Rha1 hyperactivity, and the combination of Rha1 hyperactivity and serum addition can generate significant polarization even in brg1 ume6 double mutants. Thus, in response to external signals, Rha1 functions with other morphogenesis regulators including Brg1 and Ume6, to mediate filamentation.

Investigating molecular crowding during cell division and hyperosmotic stress in budding yeast with FRET

Cell division, aging, and stress recovery triggers spatial reorganization of cellular components in the cytoplasm, including membrane bound organelles, with molecular changes in their compositions and structures. However, it is not clear how these events are coordinated and how they integrate with regulation of molecular crowding. We use the budding yeast Saccharomyces cerevisiae as a model system to study these questions using recent progress in optical fluorescence microscopy and crowding sensing probe technology. We used a Förster Resonance Energy Transfer (FRET) based sensor, illuminated by confocal microscopy for high throughput analyses and Slimfield microscopy for single-molecule resolution, to quantify molecular crowding. We determine crowding in response to cellular growth of both mother and daughter cells, in addition to osmotic stress, and reveal hot spots of crowding across the bud neck in the burgeoning daughter cell. This crowding might be rationalized by the packing of inherited material, like the vacuole, from mother cells. We discuss recent advances in understanding the role of crowding in cellular regulation and key current challenges and conclude by presenting our recent advances in optimizing FRET-based measurements of crowding while simultaneously imaging a third color, which can be used as a marker that labels organelle membranes. Our approaches can be combined with synchronized cell populations to increase experimental throughput and correlate molecular crowding information with different stages in the cell cycle.

Altered Protein Abundance and Localization Inferred from Sites of Alternative Modification by Ubiquitin and SUMO

Protein modification by ubiquitin or SUMO can alter the function, stability or activity of target proteins. Previous studies have identified thousands of substrates that were modified by ubiquitin or SUMO on the same lysine residue. However, it remains unclear whether such overlap could result from a mere higher solvent accessibility, whether proteins containing those sites are associated with specific functional traits, and whether selectively perturbing their modification by ubiquitin or SUMO could result in different phenotypic outcomes. Here, we mapped reported lysine modification sites across the human proteome and found an enrichment of sites reported to be modified by both ubiquitin and SUMO. Our analysis uncovered thousands of proteins containing such sites, which we term Sites of Alternative Modification (SAMs). Among more than 36,000 sites reported to be modified by SUMO, 51.8% have also been reported to be modified by ubiquitin. SAM-containing proteins are associated with diverse biological functions including cell cycle, DNA damage, and transcriptional regulation. As such, our analysis highlights numerous proteins and pathways as putative targets for further elucidating the crosstalk between ubiquitin and SUMO. Comparing the biological and biochemical properties of SAMs versus other non-overlapping modification sites revealed that these sites were associated with altered cellular localization or abundance of their host proteins. Lastly, using S. cerevisiae as model, we show that mutating the SAM motif in a protein can influence its ubiquitination as well as its localization and abundance.

Yeast cell segmentation in microstructured environments with deep learning

Cell segmentation is a major bottleneck in extracting quantitative single-cell information from microscopy data. The challenge is exasperated in the setting of microstructured environments. While deep learning approaches have proven useful for general cell segmentation tasks, previously available segmentation tools for the yeast-microstructure setting rely on traditional machine learning approaches. Here we present convolutional neural networks trained for multiclass segmenting of individual yeast cells and discerning these from cell-similar microstructures. An U-Net based semantic segmentaiton approach, as well as a direct instance segmentation approach with a Mask R-CNN are demonstrated. We give an overview of the datasets recorded for training, validating and testing the networks, as well as a typical use-case. We showcase the methods' contribution to segmenting yeast in microstructured environments with a typical systems or synthetic biology application. The models achieve robust segmentation results, outperforming the previous state-of-the-art in both accuracy and speed. The combination of fast and accurate segmentation is not only beneficial for a posteriori data processing, it also makes online monitoring of thousands of trapped cells or closed-loop optimal experimental design feasible from an image processing perspective. Code is and data samples are available at https://git.rwth-aachen.de/bcs/projects/tp/multiclass-yeast-seg.

A functionally divergent intrinsically disordered region underlying the conservation of stochastic signaling

Stochastic signaling dynamics expand living cells' information processing capabilities. An increasing number of studies report that regulators encode information in their pulsatile dynamics. The evolutionary mechanisms that lead to complex signaling dynamics remain uncharacterized, perhaps because key interactions of signaling proteins are encoded in intrinsically disordered regions (IDRs), whose evolution is difficult to analyze. Here we focused on the IDR that controls the stochastic pulsing dynamics of Crz1, a transcription factor in fungi downstream of the widely conserved calcium signaling pathway. We find that Crz1 IDRs from anciently diverged fungi can all respond transiently to calcium stress; however, only Crz1 IDRs from the Saccharomyces clade support pulsatility, encode extra information, and rescue fitness in competition assays, while the Crz1 IDRs from distantly related fungi do none of the three. On the other hand, we find that Crz1 pulsing is conserved in the distantly related fungi, consistent with the evolutionary model of stabilizing selection on the signaling phenotype. Further, we show that a calcineurin docking site in a specific part of the IDRs appears to be sufficient for pulsing and show evidence for a beneficial increase in the relative calcineurin affinity of this docking site. We propose that evolutionary flexibility of functionally divergent IDRs underlies the conservation of stochastic signaling by stabilizing selection.

Sequence-based features that are determinant for tail-anchored membrane protein sorting in eukaryotes

The correct targeting and insertion of tail-anchored (TA) integral membrane proteins is critical for cellular homeostasis. TA proteins are defined by a hydrophobic transmembrane domain (TMD) at their C-terminus and are targeted to either the ER or mitochondria. Derived from experimental measurements of a few TA proteins, there has been little examination of the TMD features that determine localization. As a result, the localization of many TA proteins are misclassified by the simple heuristic of overall hydrophobicity. Because ER-directed TMDs favor arrangement of hydrophobic residues to one side, we sought to explore the role of geometric hydrophobic properties. By curating TA proteins with experimentally determined localizations and assessing hypotheses for recognition, we bioinformatically and experimentally verify that a hydrophobic face is the most accurate singular metric for separating ER and mitochondria-destined yeast TA proteins. A metric focusing on an 11 residue segment of the TMD performs well when classifying human TA proteins. The most inclusive predictor uses both hydrophobicity and C-terminal charge in tandem. This work provides context for previous observations and opens the door for more detailed mechanistic experiments to determine the molecular factors driving this recognition.

Signal-mediated localization of Candida albicans pheromone response pathway components

 

A MAPK cascade consists of three kinases, (MEKK, MEK and MAPK), that are sequentially activated in response to a stimulus and serve to transmit signals. In C. albicans and in yeast, an MAPK cascade is linked to the pheromone pathway through a scaffold protein (Cst5 and Ste5, respectively). Cst5 is much shorter and lacks key domains compared to Ste5, so in C. albicans, other elements, in particular the MEKK Ste11, play key roles in controlling the associations and localizations of network components.

Abstract

Candida albicans opaque cells release pheromones to stimulate cells of opposite mating type to activate their pheromone response pathway. Although this fungal pathogen shares orthologous proteins involved in the process with Saccharomyces cerevisiae, the pathway in each organism has unique characteristics. We have used GFP-tagged fusion proteins to investigate the localization of the scaffold protein Cst5, as well as the MAP kinases Cek1 and Cek2, during pheromone response in C. albicans. In wild-type cells, pheromone treatment directed Cst5-GFP to surface puncta concentrated at the tips of mating projections. These puncta failed to form in cells defective in either the Gα or β subunits. However, they still formed in response to pheromone in cells missing Ste11, but with the puncta distributed around the cell periphery in the absence of mating projections. These puncta were absent from hst7Δ/Δ cells, but could be detected in the ste11Δ/Δ hst7Δ/Δ double mutant. Cek2-GFP showed a strong nuclear localization late in the response, consistent with a role in adaptation, while Cek1-GFP showed a weaker, but early increase in nuclear localization after pheromone treatment. Activation loop phosphorylation of both Cek1 and Cek2 required the presence of Ste11. In contrast to Cek2-GFP, which showed no localization signal in ste11Δ/Δ cells, Cek1-GFP showed enhanced nuclear localization that was pheromone independent in the ste11Δ/Δ mutant. The results are consistent with CaSte11 facilitating Hst7-mediated MAP kinase phosphorylation and also playing a potentially critical role in both MAP kinase and Cst5 scaffold localization.

Modular biosynthesis of plant hemicellulose and its impact on yeast cells

BACKGROUND

The carbohydrate polymers that encapsulate plants cells have benefited humans for centuries and have valuable biotechnological uses. In the past 5 years, exciting possibilities have emerged in the engineering of polysaccharide-based biomaterials. Despite impressive advances on bacterial cellulose-based hydrogels, comparatively little is known about how plant hemicelluloses can be reconstituted and modulated in cells suitable for biotechnological purposes.

RESULTS

Here, we assembled cellulose synthase-like A (CSLA) enzymes using an optimized Pichia pastoris platform to produce tunable heteromannan (HM) polysaccharides in yeast. By swapping the domains of plant mannan and glucomannan synthases, we engineered chimeric CSLA proteins that made β-1,4-linked mannan in quantities surpassing those of the native enzymes while minimizing the burden on yeast growth. Prolonged expression of a glucomannan synthase from Amorphophallus konjac was toxic to yeast cells: reducing biomass accumulation and ultimately leading to compromised cell viability. However, an engineered glucomannan synthase as well as CSLA pure mannan synthases and a CSLC glucan synthase did not inhibit growth. Interestingly, Pichia cell size could be increased or decreased depending on the composition of the CSLA protein sequence. HM yield and glucose incorporation could be further increased by co-expressing chimeric CSLA proteins with a MANNAN-SYNTHESIS-RELATED (MSR) co-factor from Arabidopsis thaliana.

CONCLUSION

The results provide novel routes for the engineering of polysaccharide-based biomaterials that are needed for a sustainable bioeconomy. The characterization of chimeric cellulose synthase-like enzymes in yeast offers an exciting avenue to produce plant polysaccharides in a tunable manner. Furthermore, cells modified with non-toxic plant polysaccharides such as β-mannan offer a modular chassis to produce and encapsulate sensitive cargo such as therapeutic proteins.

YeastNet: Deep-Learning-Enabled Accurate Segmentation of Budding Yeast Cells in Bright-Field Microscopy

Accurate and efficient segmentation of live-cell images is critical in maximizing data extraction and knowledge generation from high-throughput biology experiments. Despite recent development of deep-learning tools for biomedical imaging applications, great demand for automated segmentation tools for high-resolution live-cell microscopy images remains in order to accelerate the analysis. YeastNet dramatically improves the performance of the non-trainable classic algorithm, and performs considerably better than the current state-of-the-art yeast-cell segmentation tools. We have designed and trained a U-Net convolutional network (named YeastNet) to conduct semantic segmentation on bright-field microscopy images and generate segmentation masks for cell labeling and tracking. YeastNet enables accurate automatic segmentation and tracking of yeast cells in biomedical applications. YeastNet is freely provided with model weights as a Python package on GitHub.

A novel allele of SIR2 reveals a heritable intermediate state of gene silencing

Genetic information acquires additional meaning through epigenetic regulation, the process by which genetically identical cells can exhibit heritable differences in gene expression and phenotype. Inheritance of epigenetic information is a critical step in maintaining cellular identity and organismal health. In Saccharomyces cerevisiae, one form of epigenetic regulation is the transcriptional silencing of two mating-type loci, HML and HMR, by the SIR-protein complex. To focus on the epigenetic dimension of this gene regulation, we conducted a forward mutagenesis screen to identify mutants exhibiting an epigenetic or metastable silencing defect. We utilized fluorescent reporters at HML and HMR, and screened yeast colonies for epigenetic silencing defects. We uncovered numerous independent sir1 alleles, a gene known to be required for stable epigenetic inheritance. More interestingly, we recovered a missense mutation within SIR2, which encodes a highly conserved histone deacetylase. In contrast to sir1Δ, which exhibits states that are either fully silenced or fully expressed, this sir2 allele exhibited heritable states that were either fully silenced or expressed at an intermediate level. The heritable nature of this unique silencing defect was influenced by, but not completely dependent on, changes in rDNA copy number. Therefore, this study revealed a heritable state of intermediate silencing and linked this state to a central silencing factor, Sir2.

Protein context shapes the specificity of SH3 domain-mediated interactions in vivo

Protein–protein interactions (PPIs) between modular binding domains and their target peptide motifs are thought to largely depend on the intrinsic binding specificities of the domains. The large family of SRC Homology 3 (SH3) domains contribute to cellular processes via their ability to support such PPIs. While the intrinsic binding specificities of SH3 domains have been studied in vitro, whether each domain is necessary and sufficient to define PPI specificity in vivo is largely unknown. Here, by combining deletion, mutation, swapping and shuffling of SH3 domains and measurements of their impact on protein interactions in yeast, we find that most SH3s do not dictate PPI specificity independently from their host protein in vivo. We show that the identity of the host protein and the position of the SH3 domains within their host are critical for PPI specificity, for cellular functions and for key biophysical processes such as phase separation. Our work demonstrates the importance of the interplay between a modular PPI domain such as SH3 and its host protein in establishing specificity to wire PPI networks. These findings will aid understanding how protein networks are rewired during evolution and in the context of mutation-driven diseases such as cancer.

The SRC Homology 3 (SH3) domains mediate protein–protein interactions (PPIs). Here, the authors assess the SH3-mediated PPIs in yeast, and show that the identity of the protein itself and the position of the SH3 both affect the interaction specificity and thus the PPI-dependent cellular functions.

InstantDL: an easy-to-use deep learning pipeline for image segmentation and classification

BACKGROUND

Deep learning contributes to uncovering molecular and cellular processes with highly performant algorithms. Convolutional neural networks have become the state-of-the-art tool to provide accurate and fast image data processing. However, published algorithms mostly solve only one specific problem and they typically require a considerable coding effort and machine learning background for their application.

RESULTS

We have thus developed InstantDL, a deep learning pipeline for four common image processing tasks: semantic segmentation, instance segmentation, pixel-wise regression and classification. InstantDL enables researchers with a basic computational background to apply debugged and benchmarked state-of-the-art deep learning algorithms to their own data with minimal effort. To make the pipeline robust, we have automated and standardized workflows and extensively tested it in different scenarios. Moreover, it allows assessing the uncertainty of predictions. We have benchmarked InstantDL on seven publicly available datasets achieving competitive performance without any parameter tuning. For customization of the pipeline to specific tasks, all code is easily accessible and well documented.

CONCLUSIONS

With InstantDL, we hope to empower biomedical researchers to conduct reproducible image processing with a convenient and easy-to-use pipeline.

A Selective Autophagy Pathway for Phase-Separated Endocytic Protein Deposits

Autophagy eliminates cytoplasmic content selected by autophagy receptors, which link cargo to the membrane-bound autophagosomal ubiquitin-like protein Atg8/LC3. Here, we report a selective autophagy pathway for protein condensates formed by endocytic proteins in yeast. In this pathway, the endocytic protein Ede1 functions as a selective autophagy receptor. Distinct domains within Ede1 bind Atg8 and mediate phase separation into condensates. Both properties are necessary for an Ede1-dependent autophagy pathway for endocytic proteins, which differs from regular endocytosis and does not involve other known selective autophagy receptors but requires the core autophagy machinery. Cryo-electron tomography of Ede1-containing condensates, at the plasma membrane and in autophagic bodies, shows a phase-separated compartment at the beginning and end of the Ede1-mediated selective autophagy route. Our data suggest a model for autophagic degradation of macromolecular protein complexes by the action of intrinsic autophagy receptors.

A convolutional neural network segments yeast microscopy images with high accuracy

The identification of cell borders ('segmentation') in microscopy images constitutes a bottleneck for large-scale experiments. For the model organism Saccharomyces cerevisiae, current segmentation methods face challenges when cells bud, crowd, or exhibit irregular features. We present a convolutional neural network (CNN) named YeaZ, the underlying training set of high-quality segmented yeast images (>10 000 cells) including mutants, stressed cells, and time courses, as well as a graphical user interface and a web application ( www.quantsysbio.com/data-and-software ) to efficiently employ, test, and expand the system. A key feature is a cell-cell boundary test which avoids the need for fluorescent markers. Our CNN is highly accurate, including for buds, and outperforms existing methods on benchmark images, indicating it transfers well to other conditions. To demonstrate how efficient large-scale image processing uncovers new biology, we analyze the geometries of ≈2200 wild-type and cyclin mutant cells and find that morphogenesis control occurs unexpectedly early and gradually.

AI-Assisted Forward Modeling of Biological Structures

The rise of machine learning and deep learning technologies have allowed researchers to automate image classification. We describe a method that incorporates automated image classification and principal component analysis to evaluate computational models of biological structures. We use a computational model of the kinetochore to demonstrate our artificial-intelligence (AI)-assisted modeling method. The kinetochore is a large protein complex that connects chromosomes to the mitotic spindle to facilitate proper cell division. The kinetochore can be divided into two regions: the inner kinetochore, including proteins that interact with DNA; and the outer kinetochore, comprised of microtubule-binding proteins. These two kinetochore regions have been shown to have different distributions during metaphase in live budding yeast and therefore act as a test case for our forward modeling technique. We find that a simple convolutional neural net (CNN) can correctly classify fluorescent images of inner and outer kinetochore proteins and show a CNN trained on simulated, fluorescent images can detect difference in experimental images. A polymer model of the ribosomal DNA locus serves as a second test for the method. The nucleolus surrounds the ribosomal DNA locus and appears amorphous in live-cell, fluorescent microscopy experiments in budding yeast, making detection of morphological changes challenging. We show a simple CNN can detect subtle differences in simulated images of the ribosomal DNA locus, demonstrating our CNN-based classification technique can be used on a variety of biological structures.

Learning unsupervised feature representations for single cell microscopy images with paired cell inpainting

Cellular microscopy images contain rich insights about biology. To extract this information, researchers use features, or measurements of the patterns of interest in the images. Here, we introduce a convolutional neural network (CNN) to automatically design features for fluorescence microscopy. We use a self-supervised method to learn feature representations of single cells in microscopy images without labelled training data. We train CNNs on a simple task that leverages the inherent structure of microscopy images and controls for variation in cell morphology and imaging: given one cell from an image, the CNN is asked to predict the fluorescence pattern in a second different cell from the same image. We show that our method learns high-quality features that describe protein expression patterns in single cells both yeast and human microscopy datasets. Moreover, we demonstrate that our features are useful for exploratory biological analysis, by capturing high-resolution cellular components in a proteome-wide cluster analysis of human proteins, and by quantifying multi-localized proteins and single-cell variability. We believe paired cell inpainting is a generalizable method to obtain feature representations of single cells in multichannel microscopy images.

To understand the cell biology captured by microscopy images, researchers use features, or measurements of relevant properties of cells, such as the shape or size of cells, or the intensity of fluorescent markers. Features are the starting point of most image analysis pipelines, so their quality in representing cells is fundamental to the success of an analysis. Classically, researchers have relied on features manually defined by imaging experts. In contrast, deep learning techniques based on convolutional neural networks (CNNs) automatically learn features, which can outperform manually-defined features at image analysis tasks. However, most CNN methods require large manually-annotated training datasets to learn useful features, limiting their practical application. Here, we developed a new CNN method that learns high-quality features for single cells in microscopy images, without the need for any labeled training data. We show that our features surpass other comparable features in identifying protein localization from images, and that our method can generalize to diverse datasets. By exploiting our method, researchers will be able to automatically obtain high-quality features customized to their own image datasets, facilitating many downstream analyses, as we highlight by demonstrating many possible use cases of our features in this study.

Proteome-wide signatures of function in highly diverged intrinsically disordered regions

Intrinsically disordered regions make up a large part of the proteome, but the sequence-to-function relationship in these regions is poorly understood, in part because the primary amino acid sequences of these regions are poorly conserved in alignments. Here we use an evolutionary approach to detect molecular features that are preserved in the amino acid sequences of orthologous intrinsically disordered regions. We find that most disordered regions contain multiple molecular features that are preserved, and we define these as 'evolutionary signatures' of disordered regions. We demonstrate that intrinsically disordered regions with similar evolutionary signatures can rescue function in vivo, and that groups of intrinsically disordered regions with similar evolutionary signatures are strongly enriched for functional annotations and phenotypes. We propose that evolutionary signatures can be used to predict function for many disordered regions from their amino acid sequences.