Intelligent sort-timing prediction for image-activated cell sorting.

Intelligent image-activated cell sorting (iIACS) has enabled high-throughput image-based sorting of single cells with artificial intelligence (AI) algorithms. This AI-on-a-chip technology combines fluorescence microscopy, AI-based image processing, sort-timing prediction, and cell sorting. Sort-timing prediction is particularly essential due to the latency on the order of milliseconds between image acquisition and sort actuation, during which image processing is performed. The long latency amplifies the effects of the fluctuations in the flow speed of cells, leading to fluctuation and uncertainty in the arrival time of cells at the sort point on the microfluidic chip. To compensate for this fluctuation, iIACS measures the flow speed of each cell upstream, predicts the arrival timing of the cell at the sort point, and activates the actuation of the cell sorter appropriately. Here, we propose and demonstrate a machine learning technique to increase the accuracy of the sort-timing prediction that would allow for the improvement of sort event rate, yield, and purity. Specifically, we trained an algorithm to predict the sort timing for morphologically heterogeneous budding yeast cells. The algorithm we developed used cell morphology, position, and flow speed as inputs for prediction and achieved 41.5% lower prediction error compared to the previously employed method based solely on flow speed. As a result, our technique would allow for an increase in the sort event rate of iIACS by a factor of ~2.

Assignment of unimodal probability distribution models for quantitative morphological phenotyping.

BACKGROUND: Cell morphology is a complex and integrative readout, and therefore, an attractive measurement for assessing the effects of genetic and chemical perturbations to cells. Microscopic images provide rich information on cell morphology; therefore, subjective morphological features are frequently extracted from digital images. However, measured datasets are fundamentally noisy; thus, estimation of the true values is an ultimate goal in quantitative morphological phenotyping. Ideal image analyses require precision, such as proper probability distribution analyses to detect subtle morphological changes, recall to minimize artifacts due to experimental error, and reproducibility to confirm the results. RESULTS: Here, we present UNIMO (UNImodal MOrphological data), a reliable pipeline for precise detection of subtle morphological changes by assigning unimodal probability distributions to morphological features of the budding yeast cells. By defining the data type, followed by validation using the model selection method, examination of 33 probability distributions revealed nine best-fitting probability distributions. The modality of the distribution was then clarified for each morphological feature using a probabilistic mixture model. Using a reliable and detailed set of experimental log data of wild-type morphological replicates, we considered the effects of confounding factors. As a result, most of the yeast morphological parameters exhibited unimodal distributions that can be used as basic tools for powerful downstream parametric analyses. The power of the proposed pipeline was confirmed by reanalyzing morphological changes in non-essential yeast mutants and detecting 1284 more mutants with morphological defects compared with a conventional approach (Box-Cox transformation). Furthermore, the combined use of canonical correlation analysis permitted global views on the cellular network as well as new insights into possible gene functions. CONCLUSIONS: Based on statistical principles, we showed that UNIMO offers better predictions of the true values of morphological measurements. We also demonstrated how these concepts can provide biologically important information. This study draws attention to the necessity of employing a proper approach to do more with less.

Poacic acid, a ß-1,3-glucan-binding antifungal agent, inhibits cell-wall remodeling and activates transcriptional responses regulated by the cell-wall integrity and high-osmolarity glycerol pathways in yeast.

As a result of the relatively few available antifungals and the increasing frequency of resistance to them, the development of novel antifungals is increasingly important. The plant natural product poacic acid (PA) inhibits ß-1,3-glucan synthesis in Saccharomyces cerevisiae and has antifungal activity against a wide range of plant pathogens. However, the mode of action of PA is unclear. Here, we reveal that PA specifically binds to ß-1,3-glucan, its affinity for which is ~30-fold that for chitin. Besides its effect on ß-1,3-glucan synthase activity, PA inhibited the yeast glucan-elongating activity of Gas1 and Gas2 and the chitin-glucan transglycosylase activity of Crh1. Regarding the cellular response to PA, transcriptional co-regulation was mediated by parallel activation of the cell-wall integrity (CWI) and high-osmolarity glycerol signaling pathways. Despite targeting ß-1,3-glucan remodeling, the transcriptional profiles and regulatory circuits activated by caspofungin, zymolyase, and PA differed, indicating that their effects on CWI have different mechanisms. The effects of PA on the growth of yeast strains indicated that it has a mode of action distinct from that of echinocandins, suggesting it is a unique antifungal agent.

High-dimensional single-cell phenotyping reveals extensive haploinsufficiency.

Haploinsufficiency, a dominant phenotype caused by a heterozygous loss-of-function mutation, has been rarely observed. However, high-dimensional single-cell phenotyping of yeast morphological characteristics revealed haploinsufficiency phenotypes for more than half of 1,112 essential genes under optimal growth conditions. Additionally, 40% of the essential genes with no obvious phenotype under optimal growth conditions displayed haploinsufficiency under severe growth conditions. Haploinsufficiency was detected more frequently in essential genes than in nonessential genes. Similar haploinsufficiency phenotypes were observed mostly in mutants with heterozygous deletion of functionally related genes, suggesting that haploinsufficiency phenotypes were caused by functional defects of the genes. A global view of the gene network was presented based on the similarities of the haploinsufficiency phenotypes. Our dataset contains rich information regarding essential gene functions, providing evidence that single-cell phenotyping is a powerful approach, even in the heterozygous condition, for analyzing complex biological systems.

AI-based forecasting of ethanol fermentation using yeast morphological data.

Several industries require getting information of products as soon as possible during fermentation. However, the trade-off between sensing speed and data quantity presents challenges for forecasting fermentation product yields. In this study, we tried to develop AI models to forecast ethanol yields in yeast fermentation cultures, using cell morphological data. Our platform involves the quick acquisition of yeast morphological images using a nonstaining protocol, extraction of high-dimensional morphological data using image processing software, and forecasting of ethanol yields via supervised machine learning. We found that the neural network algorithm produced the best performance, which had a coefficient of determination of >0.9 even at 30 and 60 min in the future. The model was validated using test data collected using the CalMorph-PC(10) system, which enables rapid image acquisition within 10 min. AI-based forecasting of product yields based on cell morphology will facilitate the management and stable production of desired biocommodities.

Implications of maintenance of mother-bud neck size in diverse vital processes of Saccharomyces cerevisiae.

The mother-bud neck is defined as the boundary between the mother cell and bud in budding microorganisms, wherein sequential morphological events occur throughout the cell cycle. This study was designed to quantitatively investigate the morphology of the mother-bud neck in budding yeast Saccharomyces cerevisiae. Observation of yeast cells with time-lapse microscopy revealed an increase of mother-bud neck size through the cell cycle. After screening of yeast non-essential gene-deletion mutants with the image processing software CalMorph, we comprehensively identified 274 mutants with broader necks during S/G2 phase. Among these yeasts, we extensively analyzed 19 representative deletion mutants with defects in genes annotated to six gene ontology terms (polarisome, actin reorganization, endosomal tethering complex, carboxy-terminal domain protein kinase complex, DNA replication, and maintenance of DNA trinucleotide repeats). The representative broad-necked mutants exhibited calcofluor white sensitivity, suggesting defects in their cell walls. Correlation analysis indicated that maintenance of mother-bud neck size is important for cellular processes such as cell growth, system robustness, and replicative lifespan. We conclude that neck-size maintenance in budding yeast is regulated by numerous genes and has several aspects that are physiologically significant.

Simulated microgravity triggers characteristic morphology and stress response in Saccharomyces cerevisiae.

Reduction of gravity results in changes in gene expression and morphology in the budding yeast Saccharomyces cerevisiae. We studied the genes responsible for the morphological changes induced by simulated microgravity (SMG) using the yeast morphology data. We comprehensively captured the features of the morphological changes in yeast cells cultured in SMG with CalMorph, a high-throughput image-processing system. Statistical analysis revealed that 95 of 501 morphological traits were significantly affected, which included changes in bud direction, the ratio of daughter to mother cell size, the random daughter cell shape, the large mother cell size, bright nuclei in the M phase, and the decrease in angle between two nuclei. We identified downregulated genes that impacted the morphological changes in conditions of SMG by focusing on each of the morphological features individually. Gene Ontology (GO)-enrichment analysis indicated that morphological changes under conditions of SMG were caused by cooperative downregulation of 103 genes annotated to six GO terms, which included cytoplasmic ribonucleoprotein granule, RNA elongation, mitotic cell cycle phase transition, nucleocytoplasmic transport, protein-DNA complex subunit organization, and RNA localization. P-body formation was also promoted under conditions of SMG. These results suggest that cooperative downregulation of multiple genes occurs in conditions of SMG.

Genome editing to generate nonfoam-forming sake yeast strains.

Mutations frequently occur during breeding of sake yeasts and result in unexpected phenotypes. Here, genome editing tools were applied to develop an ideal nonfoam-forming sake yeast strain, K7GE01, which had homozygous awa1∆/awa1∆ deletion alleles that were responsible for nonfoam formation and few off-target mutations. High-dimensional morphological phenotyping revealed no detectable morphological differences between the genome-edited strain and its parent, while the canonical nonfoam-forming strain, K701, showed obvious morphological changes. Small-scale fermentation tests also showed differences between components of sake produced by K7GE01 and K701. The K7GE01 strain produced sake with significant differences in the concentrations of ethyl acetate, malic acid, lactic acid, and acetic acid, while K701 produced sake with more differences. Our results indicated genuine phenotypes of awa1∆/awa1∆ in sake yeast isolates and showed the usefulness of genome editing tools for sake yeast breeding.

Global study of holistic morphological effectors in the budding yeast Saccharomyces cerevisiae.

BACKGROUND: The size of the phenotypic effect of a gene has been thoroughly investigated in terms of fitness and specific morphological traits in the budding yeast Saccharomyces cerevisiae, but little is known about gross morphological abnormalities. RESULTS: We identified 1126 holistic morphological effectors that cause severe gross morphological abnormality when deleted, and 2241 specific morphological effectors with weak holistic effects but distinctive effects on yeast morphology. Holistic effectors fell into many gene function categories and acted as network hubs, affecting a large number of morphological traits, interacting with a large number of genes, and facilitating high protein expression. Holistic morphological abnormality was useful for estimating the importance of a gene to morphology. The contribution of gene importance to fitness and morphology could be used to efficiently classify genes into functional groups. CONCLUSION: Holistic morphological abnormality can be used as a reproducible and reliable gene feature for high-dimensional morphological phenotyping. It can be used in many functional genomic applications.

Profiling of the effects of antifungal agents on yeast cells based on morphometric analysis.

The incidence of fungal infection and evolution of multidrug resistance have increased the need for new antifungal agents. To gain further insight into the development of antifungal drugs, the phenotypic profiles of currently available antifungal agents of three classes-ergosterol, cell wall and nucleic acid biosynthesis inhibitors-were investigated using yeast morphology as a chemogenomic signature. The comparison of drug-induced morphological changes with the deletion of 4718 non-essential genes not only confirmed the mode of action of the drugs but also revealed an unexpected connection among ergosterol, vacuolar proton-transporting V-type ATPase and cell-wall-targeting drugs. To improve, simplify and accelerate drug development, we developed a systematic classifier that sorts a newly discovered compound into a class with a similar mode of action without any mutant information. Using well-characterized agents as target unknown compounds, this method successfully categorized these compounds into their respective classes. Based on our data, we suggest that morphological profiling can be used to develop novel antifungal drugs.

Unveiling nonessential gene deletions that confer significant morphological phenotypes beyond natural yeast strains.

BACKGROUND: Phenotypes are variable within species, with high phenotypic variation in the fitness and cell morphology of natural yeast strains due to genetic variation. A gene deletion collection of yeast laboratory strains also contains phenotypic variations, demonstrating the involvement of each gene and its specific function. However, to date, no study has compared the phenotypic variations between natural strains and gene deletion mutants in yeast. RESULTS: The morphological variance was compared between 110 most distinct gene deletion strains and 36 typical natural yeast strains using a generalized linear model. The gene deletion strains had higher morphological variance than the natural strains. Thirty-six gene deletion mutants conferred significant morphological changes beyond that of the natural strains, revealing the importance of the genes with high genetic interaction and specific cellular functions for species conservation. CONCLUSION: Based on the morphological analysis, we discovered gene deletion mutants whose morphologies were not seen in nature. Our multivariate approach to the morphological diversity provided a new insight into the evolution and species conservation of yeast.

Hyperspectral imaging techniques for the characterization of Haematococcus pluvialis (Chlorophyceae).

A hyperspectral imaging camera was combined with a bright-field microscope to investigate the intracellular distribution of pigments in cells of the green microalga Haematococcus pluvialis, a synonym for H. lacustris (Chlorophyceae). We applied multivariate curve resolution to the hyperspectral image data to estimate the pigment contents in culture and revealed that the predicted values were consistent with actual measurements obtained from extracted pigments. Because it was possible to estimate pigment contents in every pixel, the intracellular distribution of the pigments was investigated during various life-cycle stages. Astaxanthin was localized specifically at the eyespot of zoospores in early culture stages. Then, it became widely distributed in cells, but subsequently localized differently than the chl. Integrated with our recently developed image-processing program "HaematoCalMorph," the hyperspectral imaging system was useful for monitoring intracellular distributions of pigments during culture as well as for studying cellular responses under various conditions.

Image-based monitoring system for green algal Haematococcus pluvialis (Chlorophyceae) cells during culture.

The green microalga Haematococcus pluvialis accumulates the red pigment astaxanthin accompanied by morphological changes under stress conditions, including nutrient depletion, continuous light and high temperature. To investigate the physiological state of the algal cells, we developed the digital image-processing software called HaematoCalMorph. The software automatically outputs 25 single-cell measurements of cell morphology and pigments based on color, bright-field microscopic images. Compared with manual inspection, the output values of cell shape were reliable and reproducible. The estimated pigment content fits the values calculated by conventional methods. Using a random forests classifier, we were able to distinguish flagellated cells from immotile cells and detect their transient appearance in culture. By performing principal components analysis, we also successfully monitored time-dependent morphological and colorimetric changes in culture. Thus, combined with multivariate statistical techniques, the software proves useful for studying cellular responses to various conditions as well as for monitoring population dynamics in culture.

Profilin is required for Ca2+ homeostasis and Ca2+-modulated bud formation in yeast.

A cls5-1 mutant of Saccharomyces cerevisiae is specifically sensitive to high concentrations of Ca2+, with elevated intracellular calcium content and altered cell morphology in the presence of 100 mM Ca2+. To reveal the mechanisms of the Ca2+-sensitive phenotype, we investigated the gene responsible and its interacting network. We demonstrated that CLS5 is identical to PFY1, encoding profilin. Involvement of profilin in the maintenance of intracellular Ca2+ homeostasis was supported by the fact that both exchangeable and non-exchangeable intracellular Ca2+ pools in the cls5-1 mutant are higher than those of the wild-type strain. Several mutations of the genes whose proteins physically interact with profilin resulted in the Ca2+-sensitive phenotype. Examination of the intracellular Ca2+ pools indicated that Bni1p, Bem1p, Rho1p, and Cla4p are also required for the maintenance of Ca2+ homeostasis. Quantitative morphological analysis revealed that the Ca2+-induced morphological changes in cls5-1 cells are similar to bem1 and cls4-1 cells. Common Ca2+-induced morphological changes were an increase in cell size and a decrease of the ratio of budded cells in the population. Since a mutation allele of cls4-1 is located in the CDC24 gene, we suggest that profilin, Bem1p, and Cdc24p are required for Ca2+-modulated bud formation. Thus, profilin is involved in Ca2+ regulation in two ways: the first is Ca2+ homeostasis by coordination with Bni1p, Bem1p, Rho1p, and Cla4p, and the second is the requirement of Ca2+ for bud formation by coordination with Bem1p and Cdc24p.

Prediction of ethanol fermentation under stressed conditions using yeast morphological data.

A high sugar concentration is used as a starting condition in alcoholic fermentation by budding yeast, which shows changes in intracellular state and cell morphology under conditions of high-sugar stress. In this study, we developed artificial intelligence (AI) models to predict ethanol yields in yeast fermentation cultures under conditions of high-sugar stress using cell morphological data. Our method involves the extraction of high-dimensional morphological data from phase contrast images using image processing software, and predicting ethanol yields by supervised machine learning. The neural network algorithm produced the best performance, with a coefficient of determination (R2) of 0.95, and could predict ethanol yields well even 60 min in the future. Morphological data from cells cultured in low-glucose medium could not be used for accurate prediction under conditions of high-glucose stress. Cells cultured in high-concentration glucose medium were similar in terms of morphology to cells cultured under high osmotic pressure. Feeding experiments revealed that morphological changes differed depending on the fermentation phase. By monitoring the morphology of yeast under stress, it was possible to understand the intracellular physiological conditions, suggesting that analysis of cell morphology can aid the management and stable production of desired biocommodities.

Is AI essential? Examining the need for deep learning in image-activated sorting of Saccharomyces cerevisiae.

Artificial intelligence (AI) has become a focal point across a multitude of societal sectors, with science not being an exception. Particularly in the life sciences, imaging flow cytometry has increasingly integrated AI for automated management and categorization of extensive cell image data. However, the necessity of AI over traditional classification methods when extending imaging flow cytometry to include cell sorting remains uncertain, primarily due to the time constraints between image acquisition and sorting actuation. AI-enabled image-activated cell sorting (IACS) methods remain substantially limited, even as recent advancements in IACS have found success while largely relying on traditional feature gating strategies. Here we assess the necessity of AI for image classification in IACS by contrasting the performance of feature gating, classical machine learning (ML), and deep learning (DL) with convolutional neural networks (CNNs) in the differentiation of Saccharomyces cerevisiae mutant images. We show that classical ML could only yield a 2.8-fold enhancement in target enrichment capability, albeit at the cost of a 13.7-fold increase in processing time. Conversely, a CNN could offer an 11.0-fold improvement in enrichment capability at an 11.5-fold increase in processing time. We further executed IACS on mixed mutant populations and quantified target strain enrichment via downstream DNA sequencing to substantiate the applicability of DL for the proposed study. Our findings validate the feasibility and value of employing DL in IACS for morphology-based genetic screening of S. cerevisiae, encouraging its incorporation in future advancements of similar technologies.

Targeted Mutations Produce Divergent Characteristics in Pedigreed Sake Yeast Strains.

Modification of the genetic background and, in some cases, the introduction of targeted mutations can play a critical role in producing trait characteristics during the breeding of crops, livestock, and microorganisms. However, the question of how similar trait characteristics emerge when the same target mutation is introduced into different genetic backgrounds is unclear. In a previous study, we performed genome editing of AWA1, CAR1, MDE1, and FAS2 on the standard sake yeast strain Kyokai No. 7 to breed a sake yeast with multiple excellent brewing characteristics. By introducing the same targeted mutations into other pedigreed sake yeast strains, such as Kyokai strains No. 6, No. 9, and No. 10, we were able to create sake yeasts with the same excellent brewing characteristics. However, we found that other components of sake made by the genome-edited yeast strains did not change in the exact same way. For example, amino acid and isobutanol contents differed among the strain backgrounds. We also showed that changes in yeast cell morphology induced by the targeted mutations also differed depending on the strain backgrounds. The number of commonly changed morphological parameters was limited. Thus, divergent characteristics were produced by the targeted mutations in pedigreed sake yeast strains, suggesting a breeding strategy to generate a variety of sake yeasts with excellent brewing characteristics.

Analysis of the biological activity of a novel 24-membered macrolide JBIR-19 in Saccharomyces cerevisiae by the morphological imaging program CalMorph.

To investigate the biological activity of a novel 24-membered macrolide compound, JBIR-19, isolated from the culture broth of the entomopathogenic fungus Metarhizium sp. fE61, morphological changes in yeast cells were examined using the automated image-processing program CalMorph. Principal components analysis was used to elucidate dynamic changes in the phenotypes, revealing two independent effects of JBIR-19 in yeast cells: bud elongation and increased size of the actin region. Using a fitness assay, we identified the genes required for robust growth in the presence of JBIR-19. Among these were CCW12, YLR111W, and DHH1, which are also involved in abnormal bud morphology. Based on these results and others, we predict intracellular targets of JBIR-19 and its functional interactions.

Deep imaging flow cytometry.

Imaging flow cytometry (IFC) has become a powerful tool for diverse biomedical applications by virtue of its ability to image single cells in a high-throughput manner. However, there remains a challenge posed by the fundamental trade-off between throughput, sensitivity, and spatial resolution. Here we present deep-learning-enhanced imaging flow cytometry (dIFC) that circumvents this trade-off by implementing an image restoration algorithm on a virtual-freezing fluorescence imaging (VIFFI) flow cytometry platform, enabling higher throughput without sacrificing sensitivity and spatial resolution. A key component of dIFC is a high-resolution (HR) image generator that synthesizes "virtual" HR images from the corresponding low-resolution (LR) images acquired with a low-magnification lens (10×/0.4-NA). For IFC, a low-magnification lens is favorable because of reduced image blur of cells flowing at a higher speed, which allows higher throughput. We trained and developed the HR image generator with an architecture containing two generative adversarial networks (GANs). Furthermore, we developed dIFC as a method by combining the trained generator and IFC. We characterized dIFC using Chlamydomonas reinhardtii cell images, fluorescence in situ hybridization (FISH) images of Jurkat cells, and Saccharomyces cerevisiae (budding yeast) cell images, showing high similarities of dIFC images to images obtained with a high-magnification lens (40×/0.95-NA), at a high flow speed of 2 m s-1. We lastly employed dIFC to show enhancements in the accuracy of FISH-spot counting and neck-width measurement of budding yeast cells. These results pave the way for statistical analysis of cells with high-dimensional spatial information.

Jerveratrum-Type Steroidal Alkaloids Inhibit ß-1,6-Glucan Biosynthesis in Fungal Cell Walls.

The limited number of available effective agents necessitates the development of new antifungals. We report that jervine, a jerveratrum-type steroidal alkaloid isolated from Veratrum californicum, has antifungal activity. Phenotypic comparisons of cell wall mutants, K1 killer toxin susceptibility testing, and quantification of cell wall components revealed that ß-1,6-glucan biosynthesis was significantly inhibited by jervine. Temperature-sensitive mutants defective in essential genes involved in ß-1,6-glucan biosynthesis, including BIG1, KEG1, KRE5, KRE9, and ROT1, were hypersensitive to jervine. In contrast, point mutations in KRE6 or its paralog SKN1 produced jervine resistance, suggesting that jervine targets Kre6 and Skn1. Jervine exhibited broad-spectrum antifungal activity and was effective against human-pathogenic fungi, including Candida parapsilosis and Candida krusei. It was also effective against phytopathogenic fungi, including Botrytis cinerea and Puccinia recondita. Jervine exerted a synergistic effect with fluconazole. Therefore, jervine, a jerveratrum-type steroidal alkaloid used in pharmaceutical products, represents a new class of antifungals active against mycoses and plant-pathogenic fungi. IMPORTANCE Non-Candida albicans Candida species (NCAC) are on the rise as a cause of mycosis. Many antifungal drugs are less effective against NCAC, limiting the available therapeutic agents. Here, we report that jervine, a jerveratrum-type steroidal alkaloid, is effective against NCAC and phytopathogenic fungi. Jervine acts on Kre6 and Skn1, which are involved in ß-1,6-glucan biosynthesis. The skeleton of jerveratrum-type steroidal alkaloids has been well studied, and more recently, their anticancer properties have been investigated. Therefore, jerveratrum-type alkaloids could potentially be applied as treatments for fungal infections and cancer.