Humanized yeast to model human biology, disease and evolution

Abstract

For decades, budding yeast, a single-cellular eukaryote, has provided remarkable insights into human biology. Yeast and humans share several thousand genes despite morphological and cellular differences and over a billion years of separate evolution. These genes encode critical cellular processes, the failure of which in humans results in disease. Although recent developments in genome engineering of mammalian cells permit genetic assays in human cell lines, there is still a need to develop biological reagents to study human disease variants in a high-throughput manner. Many protein-coding human genes can successfully substitute for their yeast equivalents and sustain yeast growth, thus opening up doors for developing direct assays of human gene function in a tractable system referred to as 'humanized yeast'. Humanized yeast permits the discovery of new human biology by measuring human protein activity in a simplified organismal context. This Review summarizes recent developments showing how humanized yeast can directly assay human gene function and explore variant effects at scale. Thus, by extending the 'awesome power of yeast genetics' to study human biology, humanizing yeast reinforces the high relevance of evolutionarily distant model organisms to explore human gene evolution, function and disease.

Keywords: Functional complementation; Functional replaceability; Humanized yeast; Orthology.

Fig. 1.

Swapping conserved human genes in yeast. (A) General outline of yeast humanization assays. Shared human protein-coding sequences are cloned in yeast expression vectors. The wild-type yeast that express functional essential genes are viable, as indicated by colonies growing on Petri dishes. The deletion or conditional knockout of the essential yeast gene causes lethality, resulting in an absence of colonies on Petri dishes. However, yeast are viable if the expression of the orthologous human gene can compensate for the function of the yeast counterpart, despite nearly a billion years of divergent evolution, as indicated by growth on Petri dishes similar to that of wild-type yeast. (B) There is extensive genetic polymorphism in critical human genes, and some of these mutations often lead to disease. Yeast shares 2146 orthologs with humans, of which 702 are essential in yeast. Comparatively, owing to gene amplification, humans share 3942 genes with yeast, of which 961 human orthologs relate to corresponding essential counterparts in yeast [data obtained from Inparanoid (Sonnhammer and Östlund, 2015)]. By functionally replacing the conserved human genes in yeast, the resulting humanized yeast become a tractable system for testing human genetic variation in the context of a simplified cell. These yeast–human gene swaps allow researchers to characterize genetic or protein–protein interactions relevant to disease, build entire human pathways in yeast, generate personalized yeast strains for each unique human variant, identify genetic suppressors of human disease, and provide a platform to identify novel therapeutics. Image concept credit: Andrew Horton.

Fig. 2.

Humanized yeast is an advantageous system to model human disease. (A) The past two decades of research have revealed millions of gene–phenotype associations in yeast. The number of known single gene–phenotype associations in yeast (gray solid; left y-axis) has surpassed those known in humans (gray dashed; left y-axis). Similarly, many human genes are functionally replaceable in yeast (red; right y-axis) [data obtained from SGD, YeastMine (Balakrishnan et al., 2012)]. (B) Humanizable essential genes in yeast overlaid with genes from the Online Mendelian Inheritance in Man (OMIM) database. Of the 386 essential yeast genes that can be functionally replaced with human orthologs, 157 have disease association in the OMIM database. Similarly, of the 386 functionally replaceable essential yeast genes, 135 overlap with rare disease-associated genes in the Orphanet database [data obtained from OMIM and Orphanet (Amberger et al., 2015; Weinreich et al., 2008)]. See Table S2 for complete gene lists.

Fig. 3.

Deciphering the impact of human genetic variation in yeast. Experimental pipeline for high-throughput screening to interpret human genetic variation in yeast. A library of mutant human genes is generated by either error-prone PCR or by using a mutated oligonucleotide pool, cloned in yeast expression vectors and transformed into a suitable yeast strain in which the orthologous yeast gene can be inactivated by conditional repression or knocked out. The transformed yeast cells are grown under various conditions that control which allele is expressed, either the endogenous yeast one only (green culture) or the human one only – at low/medium expression levels (yellow culture), high expression levels (orange culture) or neither (blue culture). These four gene expression conditions will have different effects on the growth of individual yeast cells harboring unique human gene variants. The culture that expresses yeast alleles only serves as the ‘baseline’ readout of the overall variant human pool. The human gene expression conditions select for human gene variants based on whether their activity translates into their ability to sustain the growth of the yeast cells. Yeast cells that express deleterious human gene variants are eliminated from the pool. After selection, the plasmid pool from each growth condition is sequenced. Computational analysis of reads corresponding to each variant in the pool quantifies the fitness of each variant. The frequency of a particular variant relative to the baseline readout identifies deleterious mutations likely associated with human disease. NGS, next-generation DNA sequencing.

Fig. 4.

Systematic humanization of yeast reveals the properties critical for functional replaceability. (A) Yeast and human genomes share several thousand orthologs that belong to different classes. 1:1 orthologs are shared genes that have acquired no observable duplications in either lineage, whereas 1:2 or 1:>2 refers to orthologs that have undergone duplication in humans (Laurent et al., 2020). Our previous work shows that ∼40% of the tested human genes can functionally replace their yeast counterparts comprising 1:1, 1:2 or 1:>2 orthologs (Kachroo et al., 2015; Laurent et al., 2020). (B) Large-scale replaceability assays identify critical features of shared genes important for functional complementation in yeast. In 1:1 orthologs, genetic modularity is the best predictor of replaceability, followed by transcription rate and amino acid sequence identity. By contrast, in 1:2 or 1:>2 orthologs, the top predictors are divergence of the human or yeast genes, conserved interactions and similar sub-cellular localization. The x-axis represents the predictive power calculated as area under the curve (AUC) or receiver operator curve plots [computed from data in Kachroo et al. (2015) and Laurent et al. (2020)].

Fig. 5.

Genetic modules govern functional replaceability. (A) Lack of functional complementation by a shared human gene in yeast could be attributed to the inability of a human gene to perform critical genetic interactions or protein–protein interactions (PPIs) in yeast. Using a reverse evolution approach and modifying a non-replaceable human gene to complement the yeast ortholog should allow the discovery of critical interactions or other factors, such as diverged mechanisms or regulation between humans and yeast. (B) Genetic modularity is a feature that strongly predicts replaceability and allows researchers to test whether higher-order humanizations of yeast are possible. Some modules can be humanized because most of the individual genes within the module are replaceable, either sequentially or by expressing all humanized components simultaneously. However, non-replaceable modules represent a major challenge and could be humanized if the entire yeast genetic module is replaced simultaneously [as in Truong and Boeke (2017)]. (C) Several yeast genes are functionally replaceable by their human equivalents one gene at a time, but many are not. For example, genes encoding components of the transcription and translation machinery, the proteasome complex, and the sterol and heme biosynthesis pathways are mostly replaceable. By contrast, modules such as the splicing complex, the origin recognition complex (ORC), minichromosome maintenance (MCM) complex and the chaperone-containing TCP-1 (CCT) complex are largely non-replaceable. This distribution of replaceable or non-replaceable human genes in pathways or complexes suggests that these yeast processes are likely humanizable in their entirety, even when individual genes are non-replaceable. The module maps were generated using Cytoscape (version 3.9.1) (Shannon et al., 2003) with data from Garge et al. (2020), Kachroo et al. (2015) and Laurent et al. (2020), and are meant to illustrate the broad spectrum of functional replaceability across different cellular processes.

Fig. 6.

Expanding the scope of humanized yeast. (A) Systematic screening of human gene expression- or activity-mediated toxicity: these screens can identify novel therapeutics and genetic modifiers that restore optimal growth or phenotypes. In such screens, (over)expression of many human genes (of ∼15,000 genes in the human ORFeome collection) is typically toxic, resulting in a lethal yeast phenotype. However, applying chemical compounds or mutating the interacting genetic modifiers that abrogate the human gene's function restores yeast cell viability, allowing the discovery of potential therapeutics and likely genetic modifiers of disease. (B) Novel chemistry to assay human gene activity: measurable readouts, such as fluorescence, could identify disease-causing human gene variants in yeast. The strategy assays human gene function in yeast irrespective of orthology and essentiality. In Click-Seq, an example of such a strategy, the fully functional (wild-type) human gene product efficiently cleaves the substrate, resulting in intense green fluorescence of the probe. Sub-optimal cleavage activity of the mutated human gene product results in less-intense fluorescence, and the expression of a non-functional human gene results in an inactive protein that is unable to cleave the substrate and thus does not generate a fluorescent signal. (C) Functional replaceability testing: simply using growth as a readout, 667 human genes with clear, identifiable 1:1 fission yeast orthologs (essential) can be tested for functional replaceability [data obtained from InParanoid (Sonnhammer and Östlund, 2015) and Kim et al. (2010)].

Fig. 7.

Diverse humanized models beyond the yeast. Despite millions of years of separate evolution from the last eukaryotic common ancestor (LECA) and last universal common ancestor (LUCA), humans share several thousand orthologs with many organisms, including fish, flies, worms and prokaryotes [data obtained from InParanoid (Sonnhammer and Östlund, 2015)]. In the future, scientists can explore humanizations in these diverse model organisms, which diverged several million years ago (mya), to directly measure human gene function.