Yeast proteinopathy models are valuable tools to assess the toxicity and aggregation of proteins implicated in disease. Here, we present methods for screening Hsp104 variant libraries for toxicity suppressors. This protocol could be adapted to screen any protein library for toxicity suppressors of any protein that is toxic in yeast.
Many protein-misfolding disorders can be modeled in the budding yeast Saccharomyces cerevisiae. Proteins such as TDP-43 and FUS, implicated in amyotrophic lateral sclerosis, and α-synuclein, implicated in Parkinson’s disease, are toxic and form cytoplasmic aggregates in yeast. These features recapitulate protein pathologies observed in patients with these disorders. Thus, yeast are an ideal platform for isolating toxicity suppressors from libraries of protein variants. We are interested in applying protein disaggregases to eliminate misfolded toxic protein conformers. Specifically, we are engineering Hsp104, a hexameric AAA+ protein from yeast that is uniquely capable of solubilizing both disordered aggregates and amyloid and returning the proteins to their native conformations. While Hsp104 is highly conserved in eukaryotes and eubacteria, it has no known metazoan homologue. Hsp104 has only limited ability to eliminate disordered aggregates and amyloid fibers implicated in human disease. Thus, we aim to engineer Hsp104 variants to reverse the protein misfolding implicated in neurodegenerative disorders. We have developed methods to screen large libraries of Hsp104 variants for suppression of proteotoxicity in yeast. As yeast are prone to spontaneous nonspecific suppression of toxicity, a two-step screening process has been developed to eliminate false positives. Using these methods, we have identified a series of potentiated Hsp104 variants that potently suppress the toxicity and aggregation of TDP-43, FUS, and α-synuclein. Here, we describe this optimized protocol, which could be adapted to screen libraries constructed using any protein backbone for suppression of toxicity of any protein that is toxic in yeast.
Yeast proteinopathy models have been developed for protein-misfolding disorders including amyotrophic lateral sclerosis (ALS) and Parkinson’s disease (PD)1-3. Expression of the proteins TDP-43 and FUS, which misfold in ALS patients, are toxic and mislocalize to form cytoplasmic aggregates in yeast1,2. Similarly, expression of α-synuclein (α-syn), which is implicated in PD, is toxic and mislocalizes to form cytoplasmic aggregates in yeast3. These features recapitulate phenotypes in patients with these disorders4,5. Thus, yeast models provide a useful platform for screening for proteins or small molecules that prevent or reverse these phenotypes2,6-13. We are interested in the development of proteins that are capable of reversing aggregation and toxicity due to TDP-43, FUS, and α-syn. We focus on Hsp104, an AAA+ protein from yeast that is uniquely capable of disaggregating proteins both from amorphous aggregates and amyloid in yeast, yet it has no human homologue14,15. Hsp104 is finely tuned to disaggregate endogenous yeast prions and has only limited ability to disaggregate substrates implicated in human neurodegenerative diseases, which it never ordinarily encounters16,17. Thus, we aim to engineer enhanced versions of Hsp104 that are able to efficaciously disaggregate these human substrates. To do so, we construct large libraries of Hsp104 variants using error-prone PCR; these libraries can be screened using the yeast proteinopathy models17. We have adopted a domain-targeted approach to constructing and screening libraries, as Hsp104 is very large17. We initially focused on the middle domain (MD) of Hsp10417, though similar approaches can be employed to screen other domains. These models enable screening for disaggregase activity directly, as opposed to alternative techniques such as surface display, which is restricted to use for monitoring binding18.
Our protocol is based on two screening steps (Figure 1). First, Hsp104 variants that suppress the toxicity of the disease substrate in yeast are selected. To do so, the Hsp104 variants and disease-associated substrate are cotransformed into ∆hsp104 yeast. We employ ∆hsp104 yeast to explore Hsp104 sequence space in the absence of wild-type (WT) Hsp10417. Importantly, deletion of Hsp104 does not affect α-syn, FUS, or TDP-43 toxicity in yeast, and expression of Hsp104WT provides minimal rescue1,13,17. The yeast is then plated on inducing media to induce expression of both proteins. Yeast harboring Hsp104 variants that suppress toxicity of the disease-associated substrate confer growth of the colony. These variants are selected for further analysis, while colonies maintaining variants that do not suppress toxicity die. However, false positives are a substantial problem in this screen. Expression of TDP-43, FUS, and α-syn are highly toxic, which creates a strong selective pressure for the appearance of spontaneous genetic suppressors of toxicity unrelated to the Hsp104 variant being expressed. Thus, we have used a secondary screen that is also relatively high throughput to eliminate these nonspecific toxicity suppressors17. In this secondary screen, selected yeast are treated with 5-Fluorootic Acid (5-FOA) to counter select for the Hsp104 plasmid19. The strains are then assessed for substrate (TDP-43, FUS, or α-syn) toxicity via spotting assay to ensure that the toxicity of the substrate is restored after loss of the Hsp104 plasmid. Thus, yeast in which toxicity is restored in this secondary screen presumably originally displayed toxicity suppression due to the presence of the Hsp104 variant. These yeast are designated as ‘hits’ and the Hsp104 plasmid should then be recovered and sequenced to identify the mutations in the Hsp104 gene17 (Figure 1). Any hits should then be reconfirmed by constructing the mutation independently using site-directed mutagenesis and then retesting for toxicity suppression. The potential applications for this protocol are broad. Using these methods, libraries of any type of protein could be screened for variants that suppress toxicity of any substrate protein that is toxic in yeast.
1. Library Generation
2. Transformation of the Hsp104 Library
3. Screening for Suppression of Proteotoxicity
4. 5-FOA Counterselection and Spotting to Eliminate False Positives
5. Sequencing the Hsp104 Variants by Colony PCR
We have constructed a library of Hsp104 variants randomized in the middle domain and screened it for suppression of TDP-43 toxicity. The library was transformed and plated onto glucose and galactose plates (Figure 2) to assess the stringency of the screen. Single colonies were selected and the strains were counter selected using 5-FOA to eliminate the Hsp104 plasmid. These strains were then assessed to confirm that toxicity was due to TDP-43 alone, without the Hsp104 variants. Spotting assays of a subset of the variants selected in the initial screen showed that of the 4 colonies selected, 2 displayed Hsp104-mediated TDP-43 toxicity suppression, 1 was a false positive, and 1 displayed enhanced toxicity following 5-FOA treatment (Figure 3). The 2 true hits were then sequenced by colony PCR to identify the middle domain mutations (Figure 4). Once these mutations were identified, the Hsp104 mutation was constructed afresh in the parental Hsp104 plasmid by site-directed mutagenesis to confirm toxicity suppression. Variants selected using these methods were later confirmed to suppress aggregation in yeast, clear preformed aggregates in biochemical assays, and suppress dopaminergic neurodegeneration in a C. elegans model of PD17.
Figure 1: Flow-chart for isolating potentiated Hsp104 variants. Hsp104 libraries (URA3 marker, GAL1 promoter) are transformed in yeast containing the disease substrates (HIS3 marker, GAL1 promoter) and screened for toxicity suppressors by plating on inducing media. Potential hits are then screened again using a 5-FOA counterselection step to eliminate nonspecific toxicity suppressors. Variants selected in this second step are then sequenced by colony PCR. Please click here to view a larger version of this figure.
Figure 2: Screening for toxicity suppressors. Yeast cotransformed with pAG303GAL-TDP-43 and the pAG416GAL-Hsp104 library were plated onto glucose (repressing) or galactose (inducing) media. TDP-43 is highly toxic, so very few Hsp104 variants are capable of suppressing this toxicity, and this screen is very stringent. Single colonies are selected as hits from the galactose plate for the 5-FOA secondary screen. Both large and small colonies are typically obtained, but we have not observed any trends that dictate how colony size correlates with activity.
Figure 3: 5-FOA secondary screen. Yeast treated with 5-FOA to counter select for the Hsp104 plasmid were assessed by spotting assay. Strains were grown in SRaff-His media, serially diluted 5 fold, and spotted in duplicate onto SD-His and SGal-His plates. Hsp104A503V is a true hit that we have previously verified and show here as a control along with vector alone17. Rows 3 and 4 are library hits that retain TDP-43 toxicity following loss of the Hsp104 suppressor. Row 5 shows a false positive, where TDP-43 is no longer toxic, so a nonspecific toxicity suppressor is present. Row 6 shows a strain that is more toxic than TDP-43 typically is, possibly due to 5-FOA toxicity. This strain could still be sequenced but may not be a valid hit.
Figure 4. Analyzing sequencing results. Selected yeast often contains multiple plasmids. Therefore, following sequencing of the colony PCR products, care must be taken to ensure any mutations are not overlooked in the chromatograms. In this chromatogram, two potential mutations (denoted by *) could be overlooked. In the first instance, there is a mixture of A and T and in the second, a mixture of T and C.
Here we present our approach to isolating potentiated Hsp104 variants that suppress the toxicity of disease-associated substrates using yeast proteinopathy models. Using this approach, large libraries of variants can be screened in high throughput, with the only limitation being the number of variants that pass the 5-FOA secondary screen. By performing these steps in 96 well format, we routinely screen up to 200 hits at a time in the 5-FOA step over the course of 1-2 weeks. The number of hits obtained during the initial screening step will vary largely depending upon substrate toxicity and the makeup of each particular library. When sequencing hits, it is essential to carefully analyze all sequencing chromatograms since mixtures of plasmids are typically present in each cell.
We have sometimes noted a large percentage of Hsp104WT being isolated as ‘hits’. This is likely due to the presence of spontaneous genetic suppressors or very weak activity. To circumvent this problem, we have found that screening at elevated temperatures (e.g. 34 °C) can decrease the percentage of ‘hits’ that are Hsp104WT. It is also essential to re-clone any sequenced hits to independently assess the activity of any novel variants and to ensure plasmid purity. Once new hits are identified, they can be assessed for suppression of aggregation in yeast and characterized biochemically.
We have used these methods to isolate potentiated Hsp104 variants against TDP-43, FUS, and α-synuclein and verified their activity using pure protein biochemistry assays17. Ultimately, these methods could be used to screen protein libraries against any substrate of interest that is toxic in yeast.
The authors have nothing to disclose.
We thank Sue Lindquist, Aaron Gitler, and Martin Duennwald for kindly sharing reagents. Our studies were supported by: an American Heart Association Post-Doctoral Fellowship (M.E.J); NIH Director’s New Innovator Award DP2OD002177, NIH grants R21NS067354, R21HD074510, and R01GM099836, a Muscular Dystrophy Association Research Award (MDA277268), Packard Center for ALS Research at Johns Hopkins University, Target ALS, and an Ellison Medical Foundation New Scholar in Aging Award (J.S.).
Table of Specific Materials/Equipment | |||
Name | Company | Catalog Number | Comments |
GeneMorphII EZClone Domain Mutagenesis Kit | Agilent | 200552 | |
150mm Petri dishes | Falcon | 351058 | |
5-Fluorootic Acid | Research Products International | f10501-5.0 | |
96-DeepWell 2mL Plates | Eppendorf | 0030 502.302 | |
96 bold replicator tool | V&P Scientific | vp-404 | |
ExoSAP-IT | Affymetrix | 78200 |