Heavy metal contamination is a severe environmental problem (Vareda et al., 2019; Hu et al., 2020). Cadmium is one of the primary risks to human health due to the pollution of food crops and drinking water (Nordberg et al., 2018). Saccharomyces cerevisiae is a model organism to study cadmium toxicity, and related mechanisms include glutathione biosynthesis, stress response, vacuole transportation and sorting, metal ion homeostasis, and chromatin remodeling (Wysocki and Tamas, 2010). Notably, cadmium tolerance mechanisms are conserved between S. cerevisiae and plants, including gene overexpression of glutathione reductase (Kim et al., 2012), metallothionein (MT) (Wei et al., 2016; Ansarypour and Shahpiri, 2017), and phytochelatin (Cahoon et al., 2015; Yang et al., 2017).

Physiochemical and biological remediation are common strategies to treat heavy metal contamination. For example, chemical precipitation is frequently utilized, but the main limitations include secondary waste generation and complex infrastructure requirements (Fu and Wang, 2011; Li et al., 2019; Vardhan et al., 2019). Phytoremediation by “hyperaccumulator” plants is a promising alternative, but it suffers from the long life cycle and engineering challenges of plants (Rascio and Navari-Izzo, 2011; Reeves et al., 2018). On the other hand, microbial bioremediation may provide a scalable and cost-effective solution. In particular, S. cerevisiae has been recently engineered as a heavy-metal hyperaccumulator when equipped with selected membrane transporters and enhanced vacuolar compartmentalization (Sun et al., 2019, 2020). Furthermore, S. cerevisiae can be utilized to ferment the biomass of phytoremediation plants (Jing et al., 2020). However, S. cerevisiae is relatively sensitive to cadmium, and it is desirable to improve its cadmium robustness of S. cerevisiae as a future bioremediation agent.

Chemical tolerance is a complex phenotype with hundreds of genetic determinants (Almeida et al., 2007). Due to the lack of mechanistic understanding, it often requires genome-wide screening to identify mutations that confer resistance (Warner et al., 2010; Si et al., 2015b; Chen et al., 2020; Liu et al., 2020). For such endeavors, gene-deletion strain libraries are often utilized in S. cerevisiae, but most identified mutations lead to cadmium sensitivity (Jin et al., 2008; Ruotolo et al., 2008; Serero et al., 2008; Thorsen et al., 2009). Genetic overexpression libraries are less studied, and hence beneficial mutations are relatively scarce (Hwang et al., 2009; Fang et al., 2016). Notably, existing screening campaigns for cadmium resistance are confined to certain laboratory strains derived from the S288c background. Moreover, only single genetic mutations are investigated, whereas combinatorial optimization of multiplex gene targets is often needed for resistance engineering (Si et al., 2015a; Jiang et al., 2020).

Here we performed genome-wide screening in S. cerevisiae using a comprehensive cDNA plasmid library that was previously constructed (Si et al., 2017). We utilized a widely used industrial strain, CEN.PK2-1c, whose cadmium resistance has not been studied. Both known and new gene targets were identified to confer increased resistance. Further improvement in cadmium tolerance was achieved via the combinatorial introduction of individual mutations. Dosage-dependence and genetic interactions were also observed among different gene targets.

Here we performed the first genome-wide overexpression screening to engineer cadmium resistance in an industrial CEN.PK-derived S. cerevisiae strain. Possible mechanisms underlying identified targets may be deduced from GO analysis (Supplementary Table 4) and known functions reported in literature (Supplementary Table 5). Particularly, identifying well-established targets such as CAD1 and CUP1 indicated our approach was effective. These results also suggested that the corresponding mechanisms are conserved among different S. cerevisiae backgrounds for cadmium tolerance. For example, CAD1 encodes an AP-1-like basic leucine zipper (bZIP) transcriptional activator, which increased cadmium resistance when expressed on a multicopy plasmid (Wu et al., 1993; Hirata et al., 1994). Cad1p was imported to the nucleus from cytoplasm upon cadmium stress, upregulating FRM2 for oxidative stress attenuation, as well as activating gene expression of SLT2, RLM1, and CHS1 for cell-wall maintenance (Fernandes et al., 1997; Bilsland et al., 2004; Azevedo et al., 2007; Mazzola et al., 2015). In addition to nuclear import, our qPCR results implied that upregulation of CAD1 induced by cadmium may also participate in these processes (Figure 3). Moreover, we confirmed another known target, CUP1, which was reported to increase copper and cadmium resistance when constitutively expressed on high-copy episomal plasmids (Ecker et al., 1986).

Several less studied overexpression targets were also isolated. For example, Nrg1p is a Rim101-mediated transcriptional repressor. NRG1 overexpression enhanced transcriptional repression of TAT1, whose deletion also increased cadmium tolerance (Ruotolo et al., 2008) (Supplementary Table 5). Moreover, PPH21 encodes a catalytic subunit of protein phosphatase 2A (PP2A), which mediates cadmium-induced repression of class I transcription by targeting the formation/dissociation of the Pol I-Rxn3 complex in yeast (Zhou et al., 2013). It was reported that PP2A homolog prevented cadmium-induced cell death via inactivation of Erk1/2 and JNK and participated in MT expression regulation through dephosphorylation of MTF-1 in mammalian cells (Chen et al., 2008, 2014) (Supplementary Table 5). Also, BMH1 encodes a member of 14-3-3 proteins that function in various biological processes such as stress response (Kumar, 2017), and yeast cells expressing the human 14-3-3β/α showed enhanced cadmium resistance (Clapp et al., 2012) (Supplementary Table 5).

Notably, two new overexpression targets were identified in this study. We demonstrated plasmid-overexpression of CRS5 increased cadmium tolerance in S. cerevisiae for the first time (Figures 2B,C). CRS5 and CUP1 both encode MT in S. cerevisiae (Chatterjee et al., 2020), and Crs5p possesses a Cd(II)-binding ability (Pagani et al., 2007) (Supplementary Table 5). Gene deletion of crs5 did not lead to cadmium sensitivity (Culotta et al., 1994). Like CUP1, genomic overexpression of CRS5 was less effective than episomal expression (Supplementary Figures 3, 4). In addition, both CUP1 and CRS5 transcription levels were upregulated upon cadmium exposure (Figure 3), suggesting dosage effects of MT may be common when regarding cadmium tolerance (Jeyaprakash et al., 1991; Adamo et al., 2012). Furthermore, QCR6 encodes the subunit 6 of the ubiquinol cytochrome-c reductase complex (complex III). Complex III is a component of the mitochondrial inner membrane electron transport chain, which is suggested as a substantial source of reactive oxygen species (ROS) induced by cadmium in mammalian tissues (Wang et al., 2004) (Supplementary Table 5). Here we showed QCR6 overexpression weakly increased cadmium tolerance in yeast (Figures 2B,C).

In addition to single targets, we also explored multiplex overexpression optimization for improving cadmium resistance in S. cerevisiae. The most resistant strain overexpressing both CAD1 and NRG1 on plasmids exhibited a 3.6-fold improvement in biomass accumulation compared with WT under 200 μM cadmium, and 1.1-fold and 3.1-fold improvement relative to the mutant strains containing individual overexpression of CAD1 and NRG1, respectively. Furthermore, additive and synergistic effects were observed among overexpression targets. Additive effects for NRG1 and BMH1 were observed in CRS5-, CUP1-, NRG1-, or PPH21-integrated strains (Figure 4), possibly mediated via the cooperation of different tolerant mechanisms. For example, the gene products of CRS5, CUP1, NRG1, and PPH21 are metal-binding proteins, whereby NRG1 and BMH1 both encode transcription factors (Supplementary Table 4). Synergistic tolerance improvement was achieved by CRS5-NRG1, CUP1-NRG1, and CRS5-CUP1 co-expression (Figure 4), suggesting the presence of genetic interactions that warrant future mechanistic investigation.

Certain discrepancies were observed between this study and previous literature. For example, overexpression of PPH21 improved cadmium tolerance (Figure 2), but it was down regulated in response to cadmium stress (Figure 3). It is not surprising, however, because the genes with substantially altered transcription and translation revealed by transcriptomic and proteomic profiling do not necessarily represent effective targets for genetic engineering (Momose, 2001; Vido et al., 2001; Fauchon et al., 2002; Jin et al., 2008; Ruotolo et al., 2008; Serero et al., 2008; Thorsen et al., 2009; Huang et al., 2016; Marmiroli et al., 2016). Indeed, little overlap was observed between transcriptomics and genome-deletion screening results for cadmium sensitivity (Serero et al., 2008). Moreover, different sets of gene targets were often identified during different genome-wide screening campaigns targeting cadmium resistance (Thorsen et al., 2009), and the main reasons may include variations in strain background, screening conditions, and the formats of genetic mutations. For example, we observed different effects of the same gene target between genomic and episomal expression (Supplementary Figure 3), as well as between liquid and agar media (Figure 2). Therefore, it is necessary to perform genome-wide screening and multiplex optimization in a particular target strain under conditions that best reflect the real-world application scenario. Considering a range of 1–100 mg/L (∼8.9–890 μM) cadmium in wastewater (Huang et al., 2020), existing efforts on engineering cadmium-resistant yeast for bioremediation has only achieved moderate success. For example, upon heterologous introduction of TaPCS1 that encodes a phytochelatin synthase, S. cerevisiae was engineered as a plant-like cadmium hyperaccumulator in the presence of around 100 μM cadmium (Sun et al., 2019). In this study, we engineered a yeast strain that can tolerate up to 400 μM cadmium by co-overexpression of CAD1 and NRG1 (Supplementary Figure 6). But further endeavor is warranted to further improve the resistance to cadmium and other inhibitors in waste water for real-world applications using S. cerevisiae.

Another limitation of the current approach is the inability to explore functional potential within the S. cerevisiae pan-genome due to the use of a cDNA-derived library of a select yeast strain. For example, PCA1 encodes a cadmium efflux transporter. Whereas some natural S. cerevisiae isolates harbor a functional PCA1 gene that confers cadmium resistance, most laboratory S. cerevisiae strains including CEN.PK2-1c used in this study contain a nonfunctional Pca1p due to a G970R missense mutation (Adle et al., 2007; Wei et al., 2014; Wong et al., 2021). Therefore, it is unlikely to identify PCA1 as a positive hit via genome-scale screening using a CEN.PK cDNA-derived overexpression library. Moreover, the cDNA-templated library can only cover endogenous genes. To engineer CEN.PK yeast strains as bioremediation agents, it is necessary to screen heterologous genes to further improve its cadmium tolerance in future studies.

Using a genome-wide approach, we identified new and known targets and their dual combinations that substantially improved cadmium resistance of an industrial S. cerevisiae strain. Genome-wide search and optimization of more than two targets are desirable and underway for achieving superior cadmium robustness via iterations of the current workflow either manually (Si et al., 2015a) or using an automated biofoundries (Si et al., 2017). This workflow can also be extended to engineer robustness toward other chemical inhibitors in S. cerevisiae.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

YC, TS, and BW conceived this study. YC designed and performed genome-scale screening, gene expression analysis, combinatorial optimization, and drafted the manuscript. JL measured growth inhibition of the WT strain by cadmium. ZC assisted with mutant screening and confirmation. TS and BW supervised the research. All authors proofread the manuscript before submission.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.