Polyglutamine toxicity assays highlight the advantages of mScarlet for imaging in Saccharomyces cerevisiae

Introduction

Following the development of the green fluorescent protein (GFP) from the jellyfish Aqueaora victoria (Chalfie et al., 1994), several other FPs with various spectral properties have been characterized (Thorn, 2017), allowing simultaneous detection of multiple fluorescent reporters. Among the most popular alternatives to GFP are the red fluorescent proteins (RFPs) isolated from Anthozoa coral and anemone species. One of the drawbacks of RFPs is that Anthozoa derived FPs are obligate tetramers (Baird et al., 2000; Verkhusha & Lukyanov, 2004). While development of RFPs into monomeric versions has been successful, it is often associated with reduced brightness of the fluorescent signal (Campbell et al., 2002) and therefore reduced overall performance of the resulting monomeric FPs. Moreover, RFPs such as TagRFP and mRuby2 reported as monomeric by passing purified proteins through sizing columns still display high tendency to oligomerize in living mammalian cells (Costantini et al., 2015; Costantini et al., 2012). Thus, under specific circumstances, FPs reported as monomeric can still be prone to oligomerization. Unwanted formation of oligomers could potentially significantly alter the function/localization of the protein of interest fused to the FP and render reporters unreliable (Costantini et al., 2015; Snapp et al., 2003; Zacharias et al., 2002). Indeed, various RFPs (mCherry, mKate2, mRuby, mKO2, mApple, TagRFP-T) have been shown to have differential effects on localization of cdc12 in yeast (Lee et al., 2013). Thus, being able to assess the behavior of fluorescent reporter in a given organism and/or cellular compartment is critical to help optimize fluorescent reporter design (Snapp, 2009).

We recently established a method to rapidly compare the behavior of FPs against a monomeric variant of superfolder GFP (msfGFP) in yeast (Jiang et al., 2017). The assays exploit the ability of polyglutamine expansions associated with Huntington’s disease (HD) to form toxic aggregates in yeast cells. The cause of HD can be traced back to abnormal expansion of a polyQ stretch within the first exon of the gene encoding the Huntingtin protein (Htt^ex1) resulting in chorea and cognitive defects in patients (Gusella & MacDonald, 1995; Huntington, 1872; Penney et al., 1997). Expansion over 36 repeats is known to cause the Htt protein to misfold and aberrantly accumulate into detergent-insoluble amyloid-like aggregates in the cytoplasm of striatal neurons (Penney et al., 1997). Expression of expanded Htt^ex1 in yeast results in severe polyQ aggregation and growth defect (Duennwald, 2013; Krobitsch & Lindquist, 2000; Mason & Giorgini, 2011; Meriin et al., 2002). Interestingly, the nature of the sequences flanking the polyQ regions (in this case fluorescent or epitope tags) greatly affects the propensity of the polyQ expansions to aggregate and to display significant growth defects in yeast (Duennwald et al., 2006). Using polyQ toxicity assays in yeast, we previously showed that a yeast-optimized version of mCherry (termed yemRFP (Keppler-Ross et al., 2008)) displays only a mild growth defects compared to yeast-optimized msfGFP (ymsfGFP) (Jiang et al., 2017). These results lead us to exploit the polyQ toxicity and aggregation assays to explore to effects of two of the most recently available RFPs. Here, we focused on FusionRed, a red monomeric fluorescent variant of mKate2 known for its low cytotoxicity in cells (Shemiakina et al., 2012) that displays low propensity to oligomerize in mammalian cells (Costantini et al., 2015). We also included mScarlet, a monomeric synthetic RFP that was recently shown to outperform other RFPs in terms of brightness of the fluorescent signal (Bindels et al., 2017). Both have yet to be characterized for expression in yeast.

Methods

Fluorescent microscopy

Under the different experimental conditions, cells were diluted 10x in growth media and plated in Lab-tek (Thermo Inc.) imaging chambers and processed for fluorescent microscopy. Images in Figure 1 were acquired using a Zeiss AxioVert A1 wide field fluorescent microscope equipped with a 63X NA 1.4 Plan Apopchromat objective, a 560 to 600nM excitation/630 to 705 nm emission bandpass filter and Zeiss Axiocam 506 mono camera. Images presented in Figure 2 were collected using a Zeiss 800 confocal microscope equipped with 488 nm and 561 nm diode lasers and a 63x PlanApochromat NA 1.4 objective.

Figure 1. Comparison of red fluorescent proteins (RFPs) fluorescent intensities in yeast.

(A) yemRFP, yFusionRed and ymScarlet were introduced into centromeric vectors under the control of a constitutive GPD promoter. (B) Representative images from 3 fields of yeast cells expressing different RFPs. Imaging conditions were kept constant between samples to allow direct comparison of fluorescent intensities. Inverted black and white images are shown for clarity. Bar: 5µm (C) Yeast cells expressing the different RFPs were analyzed by flow cytometry and compared to cells carrying an empty vector. (D) Median fluorescent intensities of the various RFPs were calculated from fluorescent data acquired using flow cytometry. *p<0.05.

Figure 2. Unlike yFusionRed, ymScarlet displays a toxic polyQ phenotype similar to ymsfGFP.

(A) Fluorescent proteins (FPs) were cloned in frame with FLAG-Htt^ex1 into a centromeric vector carrying a GAL1 inducible promoter. (B) Yeast cells carrying an empty vector or 25/68Q Htt^ex1 fused to either ymsfGFP, yemRFP, yFusionRed or ymScarlet were grown to saturation overnight in glucose (control) or galactose (polyQ induced) containing media. The next days, cell concentrations were equalized to OD₆₀₀ 0.2 and 5 fold serial dilutions of the cell suspension spotted on synthetic complete agar media plates containing either glucose or galactose. (C) Yeast cells carrying an empty vector or 25/68Q Htt^ex1 fused to ymsfGFP, yemRFP, yFusionRed or ymScarlet or carrying an empty vector were induced overnight in galactose containing media and protein levels analyzed by dot blot using either an anti-FLAG (detection of fluorescent fusions) or anti-Pgk1 antibody (loading control). (D) Representative fluorescent images from 3 fields of yeast cells expressing 25/68Q Htt^ex1 fused to ymsfGFP, yemRFP, yFusionRed or ymScarlet after overnight induction in galactose-containing media.

Results and discussion

To analyze the performance of the three different RFPs in yeast, we first generated codon optimized versions of both FusionRed and mScarlet (termed yFusionRed and ymScarlet, respectively) (Table 2). Centromeric plasmids encoding yFusionRed, yemRFP and ymScarlet under the control of the constitutive GPD promoter were transformed in yeast (Figure 1A). Fluorescent intensities were compared using wide-field fluorescence microscopy (Figure 1B). Median fluorescent intensity (MFI) was then quantified using flow cytometry. Quantification revealed that yFusionRed was significantly dimmer than yemRFP (Figure 1C and D). This result was surprising given that previously published data reported a slightly increased brightness for FusionRed when compared to mCherry (Shemiakina et al., 2012). However, it is known that fluorescent brightness of FPs expressed in yeast can differed from the ones registered for pure purified proteins (Lee et al., 2013). As opposed to yFusionRed, ymScarlet displayed the strongest fluorescent signal (Figure 1C and D). These results are in agreement with previous studies reporting increased brightness of mScarlet compared to other RFs variants (Bindels et al., 2017). Based on the intensity of the fluorescent signal, ymScarlet appears to be the optimal RFP for imaging in yeast.

Next, we sought to determine how the three different FPs affect their fusion partners in living yeast. To this end, we employed the polyQ toxicity assays. Each RFP was cloned in frame with a galactose inducible version of Htt^ex1 carrying either 25Q (non-pathological length) or 68Q (HD-associated) (Figure 2A). 25Q constructs show no growth differences across the different FPs in both uninduced (glucose media) and polyQ-induced (galactose media) conditions indicating that expression of the different constructs results in similar growth phenotypes. When fused to 68Q Htt^ex1, yFusionRed displayed no significant toxicity when compared to the non-toxic 25Q fusion (Figure 2B). Interestingly, ymScarlet displayed severe toxicity, showing a slow growth phenotype comparable to what was observed for ymsfGFP (Figure 2B). In addition, assessment of protein abundance for each constructs using dot blot revealed that both 25 and 68Q yFusionRed fusions were present at lower levels compared to other fluorescent counterparts Figure 2C). The cause of this phenotype is unclear and could result from increased degradation of the fusions. Based on these observations, we then investigated the effects of the different RFPs on polyQ aggregation using fluorescent microscopy. We found that yemRFP displayed robust 68Q aggregation similar to ymsfGFP as we previously described (Jiang et al., 2017). It is important to note that while prone to aggregation, yemRFP polyQ proteins were shown to form aggregates with different biophysical properties (increased detergent solubility) that can account for their moderately toxic nature (Jiang et al., 2017). In accordance with the absence of toxicity noted in the growth assay, 68Q-FusionRed did not form visible aggregates, while ymScarlet displayed strong aggregation propensity (Figure 2D). The absence of aggregation for the 68Q-yFusionRed is consistent with our previous observation showing that yomTagBFP2, a blue fluorescent proteins similarly does not form toxic aggregates (Jiang et al., 2017). In fact, both FusionRed and mTagBFP2 (Subach et al., 2011) are evolved versions of the wild-type RFP from sea anemone Entacmaea quadricolor (Merzlyak et al., 2007). In the case of mScarlet, the protein was evolved from a synthetic template design for generating a monomeric protein. Therefore, based on our data, mScarlet appears to be an attractive alternative to mCherry, which minimizes the effect of the FP on its fusion partner.

Data availability

Dataset 1: Raw data behind Figure 1 and Figure 2. DOI, 10.5256/f1000research.15829.d213385 (Albakri et al., 2018).

Both p415 GPD-yFusionRed and p415 GPD-ymScarlet are available from addgene (#111916/11917). All plasmids are available upon request from the corresponding author.