Behead and live long or the tale of cathepsin L

Abstract In recent decades Saccharomyces cerevisiae has proven to be one of the most valuable model organisms of aging research. Pathways such as autophagy or the effect of substances like resveratrol and spermidine that prolong the replicative as well as chronological lifespan of cells were described for the first time in S. cerevisiae. In this study we describe the establishment of an aging reporter that allows a reliable and relative quick screening of substances and genes that have an impact on the replicative lifespan. A cDNA library of the flatworm Dugesia tigrina that can be immortalized by beheading was screened using this aging reporter. Of all the flatworm genes, only one could be identified that significantly increased the replicative lifespan of S.cerevisiae. This gene is the cysteine protease cathepsin L that was sequenced for the first time in this study. We were able to show that this protease has the capability to degrade such proteins as the yeast Sup35 protein or the human α‐synuclein protein in yeast cells that are both capable of forming cytosolic toxic aggregates. The degradation of these proteins by cathepsin L prevents the formation of these unfolded protein aggregates and this seems to be responsible for the increase in replicative lifespan.

In the current study we try to combine the power of the highly developed yeast genetics with the power of a spectacular model organism that is still not in the focus of research. Flatworms are known to have an incredibly high regeneration capacity. Macrostomum lignano, a marine, free-living flatworm, can be cut into pieces and the head can fully regenerate the complete posterior part of the body. The only body part that cannot be regenerated is the head. Beheading and recultivation of the head is even having a positive side effect: M. lignano lives for about 10 months, but Egger et al. were able to demonstrate that a series of regenerations of the flatworms' body more than doubled the lifespan of this marvelous organism. It is quite possible that this organism is made immortal by beheading (Egger, Ladurner, Nimeth, Gschwentner, & Rieger, 2006;Egger, Gschwentner, & Rieger, 2007).
As will be demonstrated below, we successfully established an aging reporter in S. cerevisiae that enables us to easily measure the replicative lifespan of yeast cells. In this way we can screen for substances and genes that are capable of prolonging the lifespan. In the present study we combined our aging reporter with a cDNA library from Dugesia tigrina (another tubellaria congeneric to M. lignano). By testing all flatworm genes in S. cerevisiae we identified the gene cathepsin L-like that greatly improves the replicative lifespan of yeast cells. Owing to the high sequence identity between cathepsin L-like from Dugesia and its homologue in Hydra, this finding could contribute to the understanding of Hydra's immortality.

BY4741
YCplac111-HOprom-GFP p416GPD or BY4741 YCplac111HOprom-GFP p416GPD-cDNA (D. tigrina) was grown overnight in SC-Leu-Ura medium (buffered with 100 mM BES at pH 7.5 to improve GFP fluorescence). This culture was then diluted to an OD 600 0.1 in 500 mL SC-Leu-Ura and grown for 2 days to stationary phase in SC-Leu-Ura. The elutriation was performed with the Beckman elutriation system and rotor JE-6B with a standard elutriation chamber. Prior to elutriation cells were centrifuged at 3000 rpm for 10 min, resuspended in 10 mL 1 × PBS and sonicated to separate mother from daughter cells. The cells were then loaded into the elutriation chamber with a rotor speed of 2700 rpm and a flow rate of 10 mL/min. Reduction of the rotor speed to 1350 rpm yielded fractions III-V and these three fractions were inoculated again in SC-Leu-Ura to generate cells of higher age. After 2 days a second elutriation was performed with altered parameters. The cells were loaded at 3200 rpm and a reduction of the rotor speed to 2700 rpm yielded fraction II.

| Mutation of cathepsin L
In Dugesia cathepsin L a cysteine was replaced with a serine using mutagenic primers. Two PCRs were performed using p416GPD as a matrix.

| Replicative lifespan
The replicative lifespan was analysed as described previously (Pichova, Vondrakova, & Breitenbach, 1997). For the determination of the replicative lifespan a Singer MSM microscope and micromanipulator were used. The lifespan study was performed with at least 45 virgin cells and the daughters produced were counted and removed. During day the plates were incubated at 28°C, and during night at 4°C. The lifespans were performed on either SC-glucose media or SC-Ura-Leu media. The significances in lifespan differences were calculated with log-rank statistics at the 98% confidence level using an available online tool (http://www.evanmiller.org/ab-testing/survival-curves.html).

| ImmunoBlot
Western Blot analysis was performed as described in Bischof et al. The cut-off for GFP-high cells was defined based on cells transformed with an empty control plasmid.

| Fluorescence microscopy
The fusion proteins (either GFP or RFP fusions) were analysed with a × 100 objective (NA = 1.4) using either a Carl Zeiss AG Axioscope (Oberkochen, Germany) or a Nikon (Tokyo, Japan) Eclipse Ni-U equipped with a DS-Fi2 digital camera. The grade of co-localization was quantified as described in Rinnerthaler et al.  using the Co-localization Finder plugin as part of NIH ImageJ software.

| GFP quantification
An aliquot of 5 × 10 6 cells in a volume of 200 μL PBS was pipetted in a black microwell plate (Nunc). The fluorescence was then measured in an Anthos Zenyth 3100 (Anthos Labtec Instruments) plate reader with an excitation wavelength of 485 nm and an emission wavelength of 595 nm.

| RNA isolation
An aliquot of 5 × 10 9 cells was resuspended in Peqlab TriFast (VWR) and broken by vortexing. Total RNA was isolated according to the manufacturer's instructions.
2.15 | cDNA synthesis cDNA was synthesized as described in . A 0.5 μg aliquot of RNA was reversed transcribed using 2 μg oligo-dTprimer and the MMLV High Performance Reverse Transcriptase (Epicenter, Madison, WI, USA) according to the manufacturer's instructions.

| Real-time PCR
The following primers were ordered from Sigma-Aldrich Co. LLC:

| Aging reporter
In a first step a 'replicative aging' reporter was created that allows for the screening of life-prolonging drugs and longevity genes. This reporter is based on the HO endonuclease that is cleaving DNA at the MAT locus, a prerequisite for the mating type switch. HO is tightly repressed during the whole cell cycle but is expressed at the G1/S transition (Jarolim et al., 2004). Both GFP and the promoter of the HO gene were cloned into the centromeric vector YCPlac111  extrachromosomal rDNA circles, thus increasing the replicative lifespan of yeast cells (Defossez et al., 1999). This strain was transformed with YCplac111-HOprom.-GFP. In fact the GFP fluorescence of old cells increased from~3-fold (comparing young and old cells) in the control strain to more than 3.6-fold in the Δfob1 strain (Figure 2d). The plasmid DNA was isolated, digested with BamHI and XhoI (to prove the presence of an insert) and sequenced using a plasmid-specific primer (5′-AGGTATTGATTGTAATTCTG-3′). Of all the sequenced genes (all of them sequenced for the first time), only one was overrepresented (25.9% of all clones). The sequence is presented in Figure 4(a). A blast search revealed that this gene is most probably coding for a cathepsin L-like protein. The closest homologues are found in the freshwater planarian Schmidtea mediterranea (78% identical) and Hydra vulgaris (72% identical) (Figure 4b).

| Cathepsin L in S. cerevisiae
The isolated plasmid, harbouring the cathepsin L-like gene, was retransformed again into the strain BY4741 YCplac111-HOprom.-GFP.
After elutriation fraction V cells were analysed with FACS BD Aria III.
A robust increase in GFP fluorescence compared with control cells was reproducibly observed (see Figure S1 in the Supporting Information).
The results of the FACS analysis were confirmed with micromanipulation/replicative lifespan analysis as established previously (Kennedy, Austriaco, Zhang, & Guarente, 1995). As a test the BY4741 background harbouring no plasmid was analysed on SC media The presence of cathepsin L from D. tigrina significantly (p < 0.001) FIGURE 4 Identification of cathepsin L. the sequence of cathepsin L from Dugesia tigrina is presented (a). A multiple sequence alignment comparing cathepsin L from D. tigrina, Schmidtea mediterranea and Hydra vulgaris using Clustal omega (http://www.ebi.ac.uk/ Tools/msa/clustalo/) was performed. The high similarity between these three proteins is very obvious. An asterix stands for a conserved residues, a colon stands for highly similar residues and a period stands for a weakly similar residues [Colour figure can be viewed at wileyonlinelibrary.com] FIGURE 5 Cathepsin L and the replicative lifespan. A mother cell specific lifespan with the strain BY4741 was performed using a singer micromanipulator (a). Transformation of BY4741 with the two plasmid YCplac111-HOprom.-GFP p416GPD (continuous line; diamonds) leads to a decreased replicative lifespan that increases greatly and significantly (p-value <0.001) upon expression of cathepsin L (p416GPD-cathepsin L) (dotted lines; triangles) increased the mean (11.9 generations) as well median lifespan (12 generations) (Figure 5b).
This~34% increase in replicative lifespan can most probably be attributed to the endoprotease cathepsin L. In mammals this cysteine cathepsin is primarily found in the lysosome and the extracellular space, but it was also found in the cytoplasm and nucleus and is involved in the terminal degradation of intracellular proteins (Adams-Cioaba, Krupa, Xu, Mort, & Min, 2011;Sudhan & Siemann, 2015;Sullivan et al., 2009).
Among the targets of this protease are collagen, elastin, histone H3, the transcription factor CUX1 and many more (Adams-Cioaba et al., 2011;Liu et al., 2006). Cathepsin L is also capable of degrading α-synuclein amyloid fibrils that are believed to be a main contributor to Parkinson's disease (McGlinchey & Lee, 2015) and polyglutamine-containing protein aggregates that cause several neurodegenerative diseases (Bhutani, Piccirillo, Hourez, Venkatraman, & Goldberg, 2012). The alignment ( Figure 4b) clearly shows that the N-terminal part in Dugesia is missing that is present in nearly all other organisms. Cathepsin L in humans is encoded as a pre-proenzyme and the N-terminal domain is inhibiting its activity (Coulombe et al., 1996). Because of the absence of this regulatory domain, it has to be concluded that this enzyme is of great importance for D. tigrina and is constitutively active. This leads also to the assumption that the enzyme is active in yeast cells without further regulation.

| Cathepsin L and Sup35
In a first attempt to characterize cathepsin L a GFP tagged version was created by cloning its open reading frame (ORF) into the vector pUG35. Fluorescence microscopy of the strain BY4741 pUG35cathepsin L revealed a cytosolic punctate pattern (Figure 6a). These patterns resemble protein aggregates. In the strain BY4741 p416GPD-cathepsin L the number of protein aggregates stained BY the dye Thioflavin S is reduced compared with the strain BY4741 p416GPD (b). In the strain BY4741 HSP104-GFP::HIS3 p416GPD-cathepsin L-RFP either a co-localization between Hsp104-GFP and cathepsin L-RFP was detected (c) or it was observed that Hsp104-GFP surrounds cathepsin L-RFP (d) One of the hallmarks of replicative aging is the formation of protein aggregates and these aggregates are one of the main factors that shorten the lifespan of yeast cells. As has been shown previously, these aggregates are already formed early in life and are not passed on to daughter cells but are retained in the mothers (Saarikangas & Barral, 2015). Getting rid of these aggregates would be beneficial and this could explain the life-prolonging effect of cathepsin L. Therefore protein aggregates (especially amyloid like proteins) were stained with the dye Thioflavin S (Kimura et al., 2003). As can be seen in Figure 4 (b), a strain overexpressing Dugesia cathepsin L shows a clear reduction in the number of protein aggregates. One of the proteins that are prone to forming aggregates and limiting the lifespan is the yeast prion Sup35.
A strain harbouring genomic integrated SUP35-GFP obtained from the yeast GFP clone collection (ThermoFisher Scientific Waltham, MA, USA; Huh et al., 2003) was transformed with either the plasmid p416GPD or p416GPD-cathepsin L. This vector leads to a constitutive expression of the endoprotease via the GPD (glyceraldehyde-3-phosphate dehydrogenase) promoter. The amount of GFP fluorescence was quantified using a fluorimeter. A clear decrease in fluorescence (more than 50%) was detected after expression of cathepsin L (see Figure 7a). We assume that cathepsin L leads to a fragmentation of Sup35. This effect seems to be specific for Sup35, because other proteins that are known to form protein aggregates were tested and these proteins (Ssa2-GFP, Vma2-GFP and Hsp104-GFP) showed no such decrease in GFP fluorescence after cathepsin L expression (data not shown). However, fluorescence microscopy revealed that cathepsin L expression changes the distribution of Hsp104-GFP. The disaggregating chaperone Hsp104 is believed to dissolve protein aggregates consisting of denatured proteins. There is also a certain risk that Hsp104 by breaking long amyloid filaments creates new foci on which prions can form (Cox, Byrne, & Tuite, 2007;Kryndushkin, Engel, Edskes, & Wickner, 2011). In the control strain (BY47141 p416GPD) Hsp104-GFP can be found in the cytosol as well as in certain distinct FIGURE 7 Sup35p and cathepsin L. Fluorometric measurements of the strains BY4741 SUP35-GFP::HIS3 p416GPD and BY4741 SUP35-GFP::HIS3 p416GPDcathepsin L that are either stressed (46°C) or unstressed. Expression of cathepsin L clearly and significantly (student t-test; pvalue = 0.0001) decreases the Sup35p-GFP fluorescence (a). Fluorescence microscopy of the same strains as in (a). Sup35-GFP is equally distributed in the cytosol. Co-expression of cathepsin L only leads to a very faint GFPsignal (b). Fluorescence microscopy of the same strains as in (a) after a heat shock (46°C; 10 min) was applied. Sup35-GFP clearly accumulates in cytosolic foci. These foci are nearly absent when cathepsin L is expressed. Western blot analysis of the same strains as in (a) with and without stress (46°C; 10 min) using an anti-GFP antibody. In the control strain Sup35-GFP is clearly detectable (with and without stress), whereas in the cathepsin L expressing strain only marginal protein levels are detectable [Colour figure can be viewed at wileyonlinelibrary.com] cytosolic foci. Overexpression of cathepsin L (p146GPD-cathepsin L) had no effect on the protein abundance of Hsp104-GFP, but led to a disappearance of these Hsp104-GFP foci in~60% of all cells ( Figure   S2). In a follow-up experiment we tried to co-localize cathepsin L with Hsp104, which is a marker of protein aggregates. Therefore an RFP tagged version was created (p416GPD-cathepsin L-RFP) and was transformed into the yeast strain BY4741 HSP104-GFP::HIS3. It has to be stated that in most cells either a GFP or a RFP signal in cytosolic foci was detected. In cells expressing both (Hsp104-GFP and cathepsin L-RFP) a co-localization was observed with a relatively high Pearson's correlation coefficient (Rr = 0.6; Figure 6c). In certain cases it even looks as though the Hsp104-GFP signal is surrounding the cathepsin L-RFP signal (Figure 6d).
In the case of SUP35-GFP the strain harbouring the empty plasmid aggregates ( Figure 9a). Co-transformation with cathepsin L (BY4741 pUG35-α-synculein pESC-HIS-cathepsin L) leads to a complete disappearance of α-synculein from the cytosol, whereas this human protein is still present in the plasma membrane ( Figure 9b). In aged cells obtained via elutriation, α-synculein detaches from the plasma membrane and is present in cytosolic foci. Overexpression of cathepsin L also completely removes these aggregates (Figure 9c).

| CONCLUSION
In this study we demonstrated the feasibility of S. cerevisiae to screen for longevity genes. As a test candidate the planarian flatworm D.
tigrina was chosen. By stimulating its regenerative capacity by beheading, this mortal plathelminth becomes potentially immortal. As the most probable candidate for this immortality, the enzyme telomerase is discussed in the literature (Tan et al., 2012). In animals the telomerase enzyme is mainly active in the germ line, but asexual forms of the flatworms can activate the telomerase after amputation, leading to a high somatic telomerase activity. To screen a cDNA library of D.
tigrina for longevity genes, we successfully created an aging reporter in S.cerevisiae. Of all~20,000 genes we found only one that significantly increased the replicative lifespan. This enzyme is not the 'immortality telomerase', but a cysteine protease: cathepsin L. To our knowledge this enzyme was sequenced for the first time in this study.
An overexpression of D. tigrina cathepsin L led to a 34% increase in the replicative lifespan of yeast cells.
Among the animal kingdom there are already hints that cathepsin L could be a longevity gene. In certain tumours cathepsin L is massively upregulated, thus promoting the malignancy of tumours. One reason to induce the expression of cathepsin L is very obvious. Secretion of cathepsin L by tumour cells leads to a degradation of extracellular matrix components such as collagen, fibronectin and laminin and enables the migration of tumour cells. However, it has also been demonstrated that cathepsin L inhibits apoptosis by stimulating the expression of Bcl-2 (Levicar et al., 2000). The main regulator of cathepsin L expression is the human transcription factor FOXO3a. The importance of FOXO3a for the aging process has already been demonstrated. For example the effect of dietry restriction seems to be closely related to the FOXO transcription factors. This was first shown in C. elegans. In this organism the nutrient sensing pathway upon its activation by the receptor DAF-2 leads to a phosphorylation and inactivation of the effector protein DAF-16, which is a fork-head box O (FOXO) transcription factor. Inactivating mutants in daf-2 leads to a translocation of DAF-16 to the nucleus and to the transcription of a large group of genes that are involved in stress response, fat metabolism, cellular protection and dauer formation (Tullet, 2015). This FOXO transcription factor is highly conserved and can be found in humans and Hydra with some spectacular aging phenotypes.
Hydra, a freshwater polyp, is probably one of the rare organisms that are potentially immortal. Loss of foxO in the immortal Hydra increases the number of differentiated cells and decreases the number of stem cells, thus limiting the lifespan of this polyp (Boehm et al., 2012). Furthermore, single nucleotide polymorphisms (SNPs) in the human transcription factor FOXO3A can increase the probability of FIGURE 9 α-Synuclein and cathepsin L. fluorescence microscopy of the strain BY4741 pUG35-α-synuclein pESC-HIS. In unstressed cells α-synuclein-GFP is closely associated with the plasma membrane. A heat shock (10 min; 46°C) leads to a detachment of α-synuclein from the plasma membrane and to an accumulation in cytosolic foci (a). Fluorescence microscopy of the strain BY4741 pUG35-α-synuclein pESC-HIS cathepsin L. similar to (a) α-synuclein-GFP is localized at the plasma membrane but forms no visible cytosolic aggregates (b). Fluorescence microscopy of the strains BY4741 pUG35-α-synuclein pESC-HIS and BY4741 pUG35-α-synuclein pESC-HIS cathepsin L after elutriation. With and without the expression of cathepsin L αsynuclein-GFP is localized at the plama membrane in young cells (fraction II). In old cells (fraction V) α-synuclein-GFP detaches from the plasma membrane and forms aggregates. No such aggregates can be seen after expression of cathepsin L and the protein is still visible in the plasma membrane [Colour figure can be viewed at wileyonlinelibrary. com] an increased lifespan, because different SNPs were found in people >100 years old; Flachsbart et al., 2009). We have shown that the closest homologue to cathepsin L from D. tigrina can be found in Hydra vulgaris. We were also able to predict FOXO response elements in the promotor of cathepsin L using JASPAR (http://jaspar.genereg. net/cgi-bin/jaspar_db.pl). It would be thrilling to see if a knock down of cathepsin L would limit the lifespan of H. vulgaris. In yeast cells we have been able to demonstrate that cathepsin L reduces the levels of the prion protein Sup35, thus limiting the formation of protein aggregates, one of the hallmarks of aging. This effect was also confirmed with Thioflavin S staining. After cathepsin L expression fewer protein aggregates were observed, although they were not completely removed. We could also demonstrate that cathepsin localizes to the same protein aggregates as the disaggregase Hsp104. Additionally we could show that overexpression of cathepsin L stimulates autophagy. Both reduction of protein aggregates and stimulation of autophagy are most probably how the increase in replicative lifespan is achieved after overexpression of cathepsin L. Our data also clearly indicate that cathepsin L from D. tigrina has the ability to dissolve Lewy bodies consisting of aggregated human α-synuclein molecules in yeast cells. It seems to us that cathepsin L recognizes and degrades unstructured proteins such as α-synuclein and SUP35, but has no effect on such proteins as Hsp104.