Comparison of endogenously expressed fluorescent protein fusions behaviour for protein quality control and cellular ageing research

The yeast Hsp104 protein disaggregase is often used as a reporter for misfolded or damaged protein aggregates and protein quality control and ageing research. Observing Hsp104 fusions with fluorescent proteins is a popular approach to follow post stress protein aggregation, inclusion formation and disaggregation. While concerns that bigger protein tags, such as genetically encoded fluorescent tags, may affect protein behaviour and function have been around for quite some time, experimental evidence of how exactly the physiology of the protein of interest is altered within fluorescent protein fusions remains limited. To address this issue, we performed a comparative assessment of endogenously expressed Hsp104 fluorescent fusions function and behaviour. We provide experimental evidence that molecular behaviour may not only be altered by introducing a fluorescent protein tag but also varies depending on such a tag within the fusion. Although our findings are especially applicable to protein quality control and ageing research in yeast, similar effects may play a role in other eukaryotic systems.

www.nature.com/scientificreports/ of studied processes are significantly restricted by maturation times and photostability of fluorescent protein.
Moreover, such tags significantly increase the overall size of the protein construct, which might affect its natural molecular conformations, hence, its behaviour and function. Therefore, the choice of the most suitable fluorescent tag largely depends on the type of the experiment as well as information one wants to obtain. To date, the vast majority of studies characterising fluorescent fusions mainly focus on the fluorophores (brightness, lifetime, maturation, etc.) and are conducted on exogenous constructs [15][16][17][18][19] .
Here, we performed a comparative assessment of endogenously expressed Hsp104-fluorescent protein fusions under control of the native HSP104 promoter with the focus on Hsp104 behaviour and function. We report comparisons of protein fusions containing one of the following fluorophores: a version of an enhanced GFP used in the GFP-tagged protein library (GFP) 20 , a monomeric form of GFP (mGFP) and two recently described bright monomeric fluorescent proteins: green mNeonGreen (derived from a tetrameric fluorescent protein from cephanolochordate Branchiostoma lanceolatum) 21 and red mScarlet-I 22 . We provide comprehensive data that can be utilised for choosing a fluorescent protein for in vivo and in vitro experiments on the budding yeast in protein quality control and ageing research. Our results illustrate a range of phenotypical variations which may appear depending on a chosen fluorophore. This can help researchers in selecting individual fluorescent proteins or using several of them, as well as comparing the results across studies.

Results
Fluorescent tags do not affect growth characteristics. Fluorescent tags have been reported to affect yeast growth fitness via altering protein function and intracellular localisation 23 . We first examined whether fluorescent labelling of Hsp104 had any impact on growth rates of yeast cultures. We compared the wild type and four strains with endogenously labelled Hsp104: Hsp104-GFP, Hsp104-mGFP, Hsp104-mNeonGreen and Hsp104-mScarlet-I, within the same BY4741 background. The hsp104Δ strain was also used as a reference to Hsp104 dysfunction. The cultures were monitored for 72 h and the growth was measured based on the absorbance at 600 nm. The growth curve profile was identical for all strains tested (Fig. 1a). No significant difference (Student's t-test) between doubling times was observed between all cultures (Fig. 1b, top). We estimated the length of the lag phase to be similar (Student's t-test) and slightly over 4 h across all strains tested (Fig. 1b, bottom).

Intracellular localisation of Hsp104 depends on a fluorophore. According to the Yeast GFP Fusion
Localization Database 20 , under standard conditions, the Hsp104 protein is located in the cytoplasm. However, several studies report both nuclear and cytoplasmic localisation of Hsp104 in unstressed cells 24 . We performed live cell imaging of cell in stationary phase to determine whether the subcellular localisation of Hsp104 differs depending on the fluorescent tag. By eye, Hsp104-GFP and Hsp104-mSc-I were cytoplasmic, but Hsp104-mGFP and Hsp104-mNG exhibited also nuclear localisation (Fig. 2). We quantified the percentage of total fluorescence intensity displayed by the nucleus and the rest of the cell (Fig. 2 bottom). Indeed, nuclear fluorescent fraction is larger for Hsp104-mGFP (12.1 ± 0.96%) and Hsp104-mNG (11.4 ± 1.3%) compared to that in Hsp104-GFP (9.8 ± 1.1%) and Hsp104-mSc-I (7.3 ± 1.4%) strains.
While the GFP-labelled strain originates from a previously created library 20 , other fluorescent strains were created by introducing the fluorescent protein encoding sequence into the genome via homologous recombination. To obtain the insertion fragments, we amplified the DNA fragments by the polymerase chain reaction (PCR), a method which may provide errors in the final product due to thermal damage of DNA or editing errors during DNA copying 25 . Such mutations may cause changes in protein sequence, hence affect its behaviour, which could www.nature.com/scientificreports/ result in altered localisation. To explore this possibility, we sequenced DNA of the wild type and newly designed fluorescent fusions of Hsp104. Sequencing results did not indicate any PCR-induced mutations within cells. Therefore, subcellular distribution of Hsp104 fusions seems to be defined by fluorophores.
Hsp104-mScarlet-I fusion enhances aggregate clearance. The Hsp104 chaperone is widely used in ageing and proteopathy research as a marker of misfolded or damaged protein aggregates. In yeast, S. cerevisiae, Hsp104 binds to stress-induced misfolded proteins and assists their refolding. It had been previously reported that introduction of a fluorescent chaperone fusion affects the chaperone function itself 31 . To test whether the rate of Hsp104-dependent protein aggregate removal differs and depends on the fluorescent tag, we performed a clearance assay of heat-induced aggregates. After 30 min at 42 °C, all the cells showed clear response with high numbers of fluorescent foci corresponding to heat stress-induced damaged protein aggregates. While all strains showed significant decrease in the amounts of fluorescent spots 60 min after the cells were brought to their standard growth conditions (30 °C, 180 rpm), only Hsp104-mSc-I strain exhibited almost 100% clearance as opposed to 30-40% for other strains (Fig. 3a). Interestingly, while in other strains aggregate removal continues at 90 min recovery, by this point the number of cleared cells expressing Hsp104-mSc-I significantly decreases (Student's t-test, p < 0.05). More detailed analysis revealed that the main source of this artefact was mother cells while the daughter cells stayed aggregates free (Fig. 3b). Like other proliferating cells, yeast undergo asymmetric cell division which allows for asymmetric segregation of damaged proteins meaning the mother cell retains protein aggregates, and a new daughter cell is produced free of damage and with a full replicative potential 2 . Thus, one possible mechanism of how daughter cells are cleared from aggregated proteins is by dragging them back into the mother cell. To further investigate the Hsp104-mSc-I behaviour during the recovery, we performed timelapse microscopy to follow the fluorescent foci within the Heat tolerance is dependent on a fluorophore within the Hsp104 fusion. To test whether the differences in heat-induced foci formation is due to the fact that a fluorescent tag alters protein expression, we examined Hsp104 levels in cells subjected to the heat stress followed by recovery at the standard conditions (30 °C, 180 rpm). Our data show a drastic increase of the Hsp104 expression upon exposure to the heat stress ( Fig. 4a, top, full-length blots are presented in Supplementary Fig. 1) which is consistent with our expectations based on previous reports 26,27 . Fluorescently tagged Hsp104 exhibited lower initial levels of protein expression (Fig. 4a, bottom). While this could be an effect of potentially lower antibody binding efficiency to Hsp104 fusions, the Hsp104 protein fold change expression in response to the high temperature was higher for fluorescent Hsp104 compared to that of the unlabelled Hsp104 in the wild type strain (Fig. 4b). Overall, fluorescent fusions did not disrupt the Hsp104 function in cell survival after the heat insult (Fig. 4c). As a control for the total loss of function, we used the hsp104Δ strain which failed to survive at high temperatures even with a prior treatment at 37 °C. Such pretreatment has been suggested to be essential for cellular recovery after the heat insult.
Interestingly, while recovery rates for strains with green fluorophores were lower compared to the unlabelled strain, cells carrying the Hsp104-mSc-I fusion survived as well as the wild type.

Discussion
With the development of fluorescent microscopy, the use of fluorophores as protein fusions or individual particles within cells became an invaluable tool in modern cell and molecular biology research. While allowing for direct visualisation of proteins of interest directly inside the cell, fluorescent tags increase the overall protein size and www.nature.com/scientificreports/ might alter its function through, for example, disrupting native protein conformation or changing accessibility of essential domains and regions to other molecules 13 . Despite being widely used, very few studies examined biophysical characteristics of fluorescent proteins within living systems [28][29][30] . Fluorescent proteins within protein fusions are even more sparsely characterised. Here we show that the behaviour of the protein of interest cannot only be altered by a tag but also varies depending on a fluorophore. While we did not observe any effect on cell cultures growth characteristics, we report alterations in Hsp104 subcellular localisation. Several C-terminal point mutations from lysine to alanine that inhibit nuclear localisation of Hsp104 have been suggested 24 . Experimental and computational studies on amino acid substitutions indicate their drastic effects on protein folding, stability and protein-protein interactions 32 . However, we did not identify any previously reported HSP104 mutations. Therefore, fluorescent tags themselves may alter protein conformation which may affect behaviour and localisation. While we show that proteins behaviour and localisation changes depending on a fluorescent label, a detailed structural analysis is beyond the scope of this study, but it is required to provide a definitive answer on how the conformation of a protein of interest is modified within fluorescent fusions in vivo.
In yeast cultures, cellular rejuvenation and a life-span reset is achieved during cellular division when the mother cells maintain all misfolded and damaged proteins, and a new daughter cell is created free of damage. However, to deal with stress-induced damage, cells possess a cohesive protein quality control (PQC) system, where the Hsp104 protein disaggregase assists damaged protein aggregates clearance via their disassembly and protein refolding 33,34 . Such protein refolding and reactivation is essential for longevity of living organisms, thus, overexpression of the yeast Hsp104 has been shown to prolong the lifespan in mice models 35 . Our data clearly indicate heat-induced aggregate clearance within 90 min after the exposure to the high temperature. Interestingly, we could observe Hsp104-mGFP cells free from aggregates, mainly daughter cells, immediately after the stress conditions were removed. This corresponds to higher level of the Hsp104 protein expression in this strain at this time point compared to other strains carrying fluorescently labelled Hsp104. Having similar protein expression levels, only Hsp104-mSc-I construct showed complete aggregate clearance already one-hour post stress. Whether this fusion possesses increased Hsp104 activity or it triggers other components that protect young cells from damage, remains to be investigated. Surprisingly, 90 min post stress we observe an increasing number of mother cells (ca 30%) with one aggregate. While we did capture protein aggregates being dragged from a daughter cell at earlier times after the heat stress, those newly appeared fluorescent foci have their origin within the mother cells. Prior to our experiments, we estimated a dark immature fraction of fluorescent proteins within endogenously expressed fluorescent fusions by performing maturation assessments. To overcome low Hsp104 expression level at normal growth conditions, we created and analysed strains expressing fluorescent   Fig. 2a), we estimated 10% of mScarlet-I dark fraction, and its maturation time about 30 min ( Supplementary Fig. 2b) which is consistent with previously reported data and is longer than that of GFP, mGFP and mNeonGreen 36 . Therefore, newly appearing foci are likely to be age-related, hence while enhancing damage clearance efficiency in daughter cells, the Hsp104-mSc-I construct seems to provide more stress within the mother cell. Fast daughter cells clearance seems to be the reason of a better and comparable to the wild type heat insult tolerance of Hsp104-mSc-I compared to other fluorescent strains tested. This work highlights the importance of characterising the effects that fluorescent tags can have on protein function, specifically for the field of proteostasis and ageing, which largely relies on fluorescence microscopy. Our data clearly indicate that the behaviour and function of the protein of interest can be severely affected by fluorescent labels. We discuss the issues that researchers may consider upon choosing fluorescent proteins for their experimental approach. In addition to monitoring endogenous proteins using the general disaggregase Hsp104, many misfolding reporters are available to be introduced into the cell. These are known to be affected by fluorescent tagging as well, which should be taken into account when selecting tools to study protein quality control 34 . Although our findings are especially applicable to protein quality control and ageing research in the budding yeast S. cerevisiae, similar effects and points may play a role in other eukaryotic systems.

Strain construction.
We created a number of novel yeast strains expressing fluorescent Hsp104 by introducing mGFP-HIS3, mNeonGreen-HIS3 and mScarlet-I-LEU2 fragments flanked on their 5′-and 3′-ends with ~ 50 bp sequences up-and downstream of the Hsp104 STOP codon, respectively. The mGFP-HIS3 fragment was amplified from the pmGFP-S plasmid 37 . pmNG-S and pmScI-S plasmids were created for this study by introducing mNeonGreen and mScarlet-I sequences into YDp-H and YDp-L vectors, respectively. PCR reaction mixes with amplified fragments were introduced into the yeast genome by standard LiAc protocol 38 . Successful transformants were verified by confirmation PCR and standard fluorescence microscopy. Full list of strains and plasmids used in this study is presented on Tables 1 and 2, respectively.
Growth curves. For the growth curve experiments, strains were pre-grown overnight in YNB medium supplemented with 2% glucose. Cells were then sub-cultured to OD 600 ~ 0.01 and placed onto a 24-well plate for monitoring the growth. Absorbance measurements at 600 nm were taken every 60 min for more than 50-72 h. Doubling times and the length of the Lag phase were estimated using the PRECOG software 40 . www.nature.com/scientificreports/ Mature fraction of fluorescent proteins. Yeast strains were pre-grown overnight in YNB complete medium supplemented with 2% glucose. Cells were then sub-cultured to OD 600 ~ 0.2 and grown for 4 h (until mid-logarithmic phase). To inhibit protein synthesis, the cultures were subjected to 100 µg/ml cycloheximide (Sigma Aldrich) for 1 h at room temperature, protected from light). The cells were placed onto a 1% agarose pad prepared using Geneframes (Thermo Scientific) supplemented with 1 × CSM, 1 × YNB, 2% glucose and 250 µg/ ml cycloheximide. Images were acquired at room temperature. Mature fraction of fluorophores within protein fusions was estimated as described previously by diving the recovered intensity after complete bleaching by the initial, autofluorescence corrected pre-bleached intensity 41 . Maturation time was estimated using following equation: where I rec is the recovered intensity after bleaching, t mat is the maturation time, and I bleach is the remaining intensity after bleaching.
Aggregate clearance. Cells pre-grown to OD 600 ~ 0.4-0.5 were subjected to 42 °C heat shock for 30 min.
For recovery, cells were then placed into 30 °C, 180 rpm. To follow aggregates formation and clearance, samples were taken prior to the heat shock, immediately after, and after 30, 60 and 90 min of recovery. Cells were fixed in 3.7% formaldehyde (Scharlau) and imaged in Z-stacks. The number of aggregates per cell was scored manually using open ImageJ FiJi 2.1.0/1.53c software, Cell Counter Plugin. Efficiency of aggregate clearance was calculated for each time point after the heat shock as a percentage of cells without any fluorescent foci. Table 3 indicates numbers of cells counted per one representative experiment illustrated in Fig. 3 f (x) = I rec * 1 − exp − x t mat + I bleach ; Table 3. Numbers of mother and daughter cells counted within one representative experiment illustrated in Fig. 3 to assess efficiency of heat-induced protein aggregate clearance. www.nature.com/scientificreports/ Heat tolerance assay. Overnight pre-grown strains were sub-cultured and grown to OD 600 ~ 0.5. The strains were sampled for spot test as a 30 °C control. Remaining cultures were split into two groups and heatstressed in a 37 °C water bath. One of the sample groups was collected after 30 min as a 37 °C sample, the other group was further heat-treated in a 50 °C water bath for 30 min. All strains were placed onto YPD plates in 5 tenfold dilution 5 µl spots and allowed to grow at 30 °C for 3 days.

Mother cells Daughter cells
Epifluorescence microscopy. To determine Hsp104 localisation, strains were grown in YNB complete medium supplemented with 2% glucose (w/v) overnight until the stationary phase. Live cells were then stained with 4′,6-diamidino-2-phenylindole (DAPI) (final concentration 10 µg/ml) for 5 min and directly visualised using a Zeiss Axio Observer.Z1 inverted microscope with standard Zeiss high efficiency (HE) filter sets (38,45 and 49), an Axiocam 506 345 camera, and a Plan-Apochromat 100x/1.40 Oil DIC M27 objective. Fluorescence signal was acquired via 509, 517 and 593 nm emission channels for visualisation of GFP and mGFP, mNeon-Green and mScarlet-I, respectively. Fluorescence intensity was estimated using ImageJ FiJi 2.1.0/1.53c software as raw integrated density value of the entire cell and within the nucleus. Nuclear segmentation was performed using DAPI images as a reference. For heat stress recovery experiments, every field of view was visualised as a Z-stack of 10 images through the range of 7 µm. For timelapse microscopy, the strain expressing Hsp104-mSc-I was grown in YNB complete medium supplemented with 2% glucose, sub-cultured to OD 600 ~ 0.45 and subjected to 30 min heat-shock at 42 °C. The cells were then gently spun down and placed onto a 1% agarose pad perfused with YNB medium supplemented with 2% glucose (w/v) and sealed with a coverslip as described previously 43 . The sample was imaged in 11 Z-stacks every 5 min for 90 min at 30 °C using the TempModule S1 (Zeiss), Y-module S1 (Zeiss), Temperable insert S1 (Zeiss), Temperable objective ring S1 (Zeiss), and Incubator S1 230 V (Zeiss) to maintain the temperature. The focus was set manually and maintained using the software autofocus with SR101 (593 nm) as a reference channel. The timelapse images were processed with Fiji software using ImageJ plugin BleachCorrect V.2.0.2 with simple ratio and the manual drift correction plugin 44 . For representation, enhancements in brightness/contrast were used.