We have cloned the Schizosaccharomyces pombe homologue of the human Batten disease gene, CLN3. This gene, btn1, encodes a predicted transmembrane protein that is 30% identical and 48% similar to its human counterpart. Cells deleted for btn1 were viable but had enlarged and more alkaline vacuoles. Conversely overexpression of Btn1p reduced both vacuole diameter and pH. Thus Btn1p regulates vacuole homeostasis. The vacuolar defects of btn1Δ cells were rescued by heterologous expression of CLN3, proving that Btn1p and CLN3 are functional homologues. The disease severity of Batten disease-causing mutations (G187A, E295K and V330F), when expressed in btn1 appeared to correlate with their effect on vacuolar pH, suggesting that elevated lysosomal pH contributes to the disease process. In fission yeast, both Btn1p and CLN3 trafficked to the vacuole membrane via early endocytic and pre-vacuolar compartments, and localisation of Btn1p to the vacuole membrane was dependent on the Ras GTPase Ypt7p. Importantly, vacuoles in cells deleted for both ypt7 and btn1 were larger and more alkaline than those of cells deleted for ypt7 alone, indicating that Btn1p has a functional role prior to reaching the vacuole. Consistently, btn1 and vma1, the gene encoding subunit A of the V1 portion of vATPase, showed conditional synthetic lethality, and in cells deleted for vma1 (a subunit of the vacuolar ATPase) Btn1p was essential for septum deposition during cytokinesis.
The neuronal ceroid lipofuscinoses (NCLs), also known as Batten disease, are a group of neurodegenerative diseases affecting children and characterised by the accumulation of autofluorescent material in the lysosomes of most cells (Santavuori, 1988). Clinical features include visual failure, seizures and progressive mental and motor deterioration leading to early death. The NCLs are inherited in an autosomal recessive manner and six human NCL genes have now been identified (International Batten Disease Consortium, 1995; Gao et al., 2002; Ranta et al., 1999; Savukoski et al., 1998; Sleat et al., 1997; Vesa et al., 1995; Vines et al., 1999; Wheeler et al., 2002). Despite a common cellular phenotype of disturbed lysosomal function, not all of the proteins causing NCL are located in this organelle (Heine et al., 2004; Isosomppi et al., 2002; Järvelä et al., 1999; Järvelä et al., 1998; Lonka et al., 2000; Lonka et al., 2004; Mole et al., 2004; Ranta et al., 2004).
Mutations in CLN3 cause juvenile onset NCL (JNCL) that can be readily distinguished from the other NCLs (Goebel et al., 1999), and so far 39 mutations have been defined (see, http://www.ucl.ac.uk/ncl). CLN3 encodes a 438 amino acid membrane protein that is glycosylated (Ezaki et al., 2003), phosphorylated (Michalewski et al., 1998; Michalewski et al., 1999) and probably farnesylated (Kaczmarski et al., 1999; Pullarkat et al., 1997). It is located in the endosome-lysosome membrane and in neuronal cells is additionally associated with synaptosomes and microvesicles (Ezaki et al., 2003; Haskell et al., 2000; Järvelä et al., 1999; Järvelä et al., 1998; Kyttälä et al., 2003; Luiro et al., 2001). Two targeting motifs for lysosomal location have been identified (Kyttälä et al., 2003; Kyttälä et al., 2005; Storch et al., 2004). CLN3 is predicted to be a transmembrane protein and probably spans the membrane six times with both termini projecting into the cytoplasm (Ezaki et al., 2003; Janes et al., 1996; Kyttälä et al., 2003; Mao et al., 2003).
The genes underlying two types of NCL (CLN1/PPT1 and CLN3) are conserved in eukaryotes (Korey et al., 2003; Mitchell et al., 2001; Porter et al., 2005) including both budding yeast (Saccharomyces cerevisiae, CLN3 only) (Pearce and Sherman, 1997) and fission yeast (Schizosaccharomyces pombe) (Cho et al., 2004) (this report). CLN1 encodes the enzyme palmitoyl protein thioesterase 1 but the function of the CLN3 gene product, which has no homology with other proteins or functional domains, is unknown. However, its conservation suggests that it performs a fundamental cellular function. In budding yeast, Btn1p, the yeast orthologue of CLN3, is a vacuolar protein (the vacuole carrying out some of the functions of the lysosomes of higher eukaryotic cells) and has been implicated in vacuolar homeostasis (Pearce et al., 1999b; Pearce et al., 1998) although the molecular basis for this has not been determined. In cells deleted for Btn1p, decreased arginine transport into the vacuole (Kim et al., 2003), which is dependent on a functioning vATPase and vacuole acidification, has been reported, but this is at odds with the decreased vacuolar pH also reported in budding yeast cells missing functional Btn1p (Chattopadhyay et al., 2000; Pearce et al., 1999b).
The biology of S. pombe and S. cerevisiae are distinct in many respects, including the organisation of the vacuole (Moreno et al., 1991). Fission yeast has large numbers of small vacuoles (in the region of 50) (Bone et al., 1998) in contrast to budding yeast that contains a small number of large vacuoles. Fission yeast vacuoles fuse in response to hypotonic stress (Bone et al., 1998), a process dependent on the Ras GTPase Ypt7p (Bone et al., 1998). The fusion of isolated vacuoles from S. cerevisiae has been extensively studied (Seeley et al., 2002; Wickner, 2002) and many essential components for this and vesicle-mediated protein delivery to the vacuole in budding yeast are conserved in fission yeast (Takegawa et al., 2003). Since Batten disease is a lysosomal storage disease, we tested whether the fission yeast homologue of CLN3, Btn1p, was implicated in vacuole integrity and, if so, whether fission yeast was likely to provide a good model system for elucidating the basic function of Btn1p. The absence and overexpression of both wild-type and mutated btn1 was shown to have significant effects on vacuole homeostasis and Btn1p also appeared to be functional in pre-vacuolar compartments. We therefore concluded that the use of fission yeast may prove to be pivotal in determining the function of both Btn1p and CLN3.
Materials and Methods
Identification of a S. pombe orthologue to CLN3
btn1 was identified by searching the Sanger Centre Fission Yeast Genome Project (www.sanger.ac.uk/Projects/S_pombe) database for predicted proteins with homology to that encoded by the human disease gene, CLN3 (U32680). The btn1 gene resides on chromosome I, is contained in cosmid 607 (SPAC607.09c), and is retrievable from the EMBL database (accession number CAB63796). Sequences were aligned using ClustalX and shaded using MacBoxshade 2.15. The position of transmembrane domains are based on original predictions (Janes et al., 1996), recent work (Ezaki et al., 2003; Kyttälä et al., 2003; Mao et al., 2003) and the assumption that these will be conserved between mammalian and yeast species.
S. pombe strains and cell growth
Strains used in this study are listed in Table 1. Media, growth, maintenance of strains and genetic methods were as described previously (Moreno et al., 1991). Cells were grown in rich medium (YES) or minimal medium (MM) containing appropriate supplements. For protein expression, cells were grown overnight in MM plus thiamine (4 μM). Cells were washed three times in MM lacking thiamine and grown for 18 hours in the same medium. Repression of the nmt1 promoter (Maundrell, 1993) was carried out by the addition of 4 μM thiamine to log phase cells (usually 2×106 cells/ml). D-(-)-threo-2-amino-1-[p-nitrophenyl]-1,3-propainediol (ANP) was incorporated into MM agar at 1.0 mM.
Gene deletion and construction of expression plasmids
The coding region of btn1+ (all 396 amino acids plus the stop codon of the coding sequence) was replaced with the leu2+ gene. A vector derived from pSL1180 was used to subclone the leu2 gene (AscI fragment) flanked by a 1 kb 5′ upstream region of btn1 (amplified as a MluI-PstI fragment using primers aaa acg cgt ttc ata atg atc cca tta ctg t and aac tgc aga aga ttc gta cta tat tta gta) and a 1.5 kb 3′ downstream region of btn1 (amplified as a ApaI-NcoI fragment using primers aat tgg gcc caa aat agt gag tgc gca cat ct and cat gcc atg gca gca cca agc gaa aac cca). The plasmid (pSMr39) was linearised and transformed into a wild-type diploid strain. Transformants were selected for leucine auxotrophy, sporulated in MM without nitrogen and tetrads dissected using a Singer MSM Micromanipulator (Singer Instruments, Wachet, UK). Spores were inspected for growth on MM lacking leucine. Correct integration of the deletion cassette in selected colonies was confirmed by PCR using flanking primers to btn1 and by Southern blotting of genomic DNA digested with EcoRI and probed with randomly primed radiolabelled btn1 (see supplementary material, Fig. S1).
The btn1 gene (1.2 kb) was subcloned into pREP1 (Maundrell, 1993) and pREP41GFP or pREP42GFP (Craven et al., 1998) as a SalI-BamHI fragment amplified from S. pombe genomic DNA using primers 5′-acgcgtcgaccatgattaaattgaggttaac-3′ and 5′-cgggatcccgtcaagttaaggcacaccaat-3′, to allow expression from the nmt promoter. cDNA corresponding to the human CLN3 gene was subcloned into pREP42GFP as a SalI-BamHI fragment using primers 5′-gtcgacaatgggaggctgtgcaggct-3′ and 5′-cgcggatccgcgtcaggagagctggcagagga-3′. ypt7 was subcloned into pREP42GFP as a NdelI-BamHI fragment amplified from S. pombe genomic DNA using primers 5′-gggaattccatatggccggcaaaaagaagca-3′ and 5′-cgcggatccgcgttaacagtaacatgaagtt-3′. Disease-causing mutations were introduced into clone pREP42GFPBtn1, which contained GFP fused to the N terminus of Btn1p, using the QuickChange site directed in vitro mutagenesis kit and following the manufacturer's instructions (Stratagene, La Jolla, CA, USA). Forward and reverse primers for each mutation are as follows: (i) human disease mutation G187A (GFPBtn1G136A) 5′-tccggaacagccttggccggtctc-3′ and 5′-gagaccggacaaggctgttccgga-3′; (ii) mutation E295K (GFPBtn1E240K) 5′-cttgtatacttctcaaaatatactatcaatattg-3′ and 5′-caatattgatagtatattttgagaagtatacaag-3′; (iii) mutation V330F (GFPBtn1V278F) 5′-gtttaccagattggttttttcctatcgcgatc-3′ and 5′-gatcgcgataggaaaaaaccaatctggtaaac-3′. All constructs were verified by sequence analysis.
For measurement of cell length and septation index, cells in log phase growth were fixed in 10% formaldehyde for 15 minutes, washed three times in 1× PBS and stored at 4°C. DNA was stained with DAPI, septa were visualised with calcofluor (Moreno et al., 1991) and cell length at division determined. For measurement of cell cycle length, exponentially growing cell were diluted to 0.5×106 cells/ml, a growth curve determined over 8 hours, and cell cycle length measured, as described previously (Alfa et al., 1993). FM4-64 dye (Molecular Probes) was used to label vacuoles if required. 1 μl FM4-64 (1 μg/μl in DMSO) was added to 1 ml cells. Cells were rotated at room temperature for 30 minutes, washed in MM and the dye incorporation chased for 30 minutes in MM. To monitor the rate of uptake of FM4-64 by endocytosis, 1 μl FM4-64 (1 μg/μl) was added to cells mounted in MM containing 1% agarose on a glass slide. To measure total fluorescent dye uptake, 6 μl of FM4-64 was added to 6 ml cells and incubated with agitation at 29°C. Uptake was arrested at the indicated time points by incubation at 0°C. A 1 ml aliquot of cells was washed with pre-cooled MM three times and the fluorescence was measured, at 620 nm, in triplicate samples of 200 μl each.
To monitor intravacuolar pH, 1 ml cells was stained with 1 μl carboxy-dichlorofluorescein diacetate (CDCFDA; 47 mM in DMSO; Molecular Probes) in 1 ml MM for 30 minutes, with rotation, at RT. Cells were washed three times in MM and incubated for 10 minutes in 1 ml MM, then placed on ice prior to viewing. To mark a subpopulation of cells prior to the start of the experiment, 1 ml of cells were pre-labelled with calcofluor and Hoechst stain (1 μl from stock solution) for 10 minutes, the cells were then washed three times in MM and mixed with 1 ml of a second population of cells marked only with calcofluor. All cells were exposed to 1 μl each of CDCFDA and FM4-64 and incubated for 15 minutes before observation. For quantitative CDCFDA hydrolysis, 6 ml exponentially growing cell were labelled with 10 μl of dye. The reaction was stopped after specified intervals by exposure to 0°C. A 1 ml aliquot of cells was washed with pre-cooled MM medium three times, and the fluorescence was measured at 520 nm in triplicate samples of 200 μl each. Measurement of CDCFDA against GFP (for relative pH measurements of Btn1p mutant proteins) was possible since the fluorescence of GFP was much lower than that of CDCFDA, such that GFP was not detectable at the short exposures needed for CDCFDA (0.01 seconds at 50% UV intensity), whereas detection of GFP needed an exposure of 1 second at full UV. Background fluorescence of GFP (even though effectively negligible) was subtracted where applicable.
Images were taken using either a Hamamatsu digital camera C4742-95 fitted to a Zeiss Axioskop microscope with plan-Apochromat 63× 1.4 NA oil immersion objective or a Hamamatsu C4742 CCD camera fitted to a Zeiss Axiophot microscope with a 64× 1.4 NA objective, and were recorded using OpenLab 3.15 software (Improvision Ltd., Coventry, UK), downloaded to Microsoft Excel for analysis and to Adobe Imageready 3 for assembly into montages. A FITC filter was used for CDCFDA detection.
Visualisation, fusion and measurement of vacuoles
To visualise vacuoles, cells were labelled with 1 μl N-(3-triethylammoniumpropyl)-4-(6-(4-(diethylamino) phenyl) hexatrienyl) pyridinium dibromide (FM4-64) (Molecular Probes, 1 μg/ml in DMSO) with rotation for 30 minutes, washed in MM and then incubated for 30 minutes in MM. Vacuole fusion was assessed by labelling 1 ml log phase cells with FM4-64 for 30 minutes, which was then chased for 30 minutes. Cells were washed in H2O and resuspended in 1 ml H2O on ice for 2 hours, or overnight to ensure that full fusion had occurred. The diameter of every vacuole visible in one focal plane of a cell was measured using OpenLab 3.15 software and downloaded to Microsoft Excel for analysis (more than 300 vacuoles were counted for each data set). To determine the relationship between vacuole size and btn1 expression, individual cells expressing GFP-Btn1 were assigned an expression level according to arbitrary units given by densitometry measurements, and 20 cells from each expression level were analysed for vacuole size. To correlate vacuole size with pH, cells were grown to log phase in MM. 10 ml cells were transferred to a small flask and labelled with FM4-64, as described. After the chase, 1 ml aliquots of the cells were placed into fresh Eppendorf tubes. Cells were pelleted and resuspended in MM combined with the appropriate pH buffer: 40 mM potassium hydrogen phthalate-HCl (pH range 3.5-4); 40 mM potassium hydrogen phthalate-NaOH (pH range 4.5-5.7); 40 mM potassium dihydrogen phosphate-dipotassium hydrogen phosphate (pH range 5.8-7.0) or 40 mM Hepes-NaOH (pH range 7.0-7.9), and grown for 1 hour. Vacuole size was determined at each pH value (n=100).
To calculate vacuole pH, 1 ml exponentially growing cells were exposed to 5 μl of LSDND/DND-189 (1 mM in DMSO; Molecular Probes) in MM + 40 mM Hepes pH 7.9 for 5 minutes at 25°C, and washed three times in MM. Quantitative fluorescence was determined using a Labsystem Fluoroskan Ascent FL microplate fluorometer with excitation at 450 nm and fluorescence reading at 520 nm for DND-189 (Cousin et al., 1997). Data from three independent experiments were exported to Excel for analysis. The vacuole pH was calculated from a standard curve, constructed by calibrating LSDND fluorescence against pH. Wild-type cells were loaded with LSDND, then permeabilised in the presence of bafilomycin (10 nM; Sigma), an inhibitor of vacuole ATPase, in 1 M sorbitol, 20% DMSO, 5 mM DTT 0.5% Triton X-100 for 4 minutes [protocol adapted from Li et al. (Li et al., 2000)] in the presence of buffers of different pH (50 mM sodium acetate pH 3.5, 3.8, 4; 50 mM MES pH 4.3, 4.58, 5, 5.5 and 6.15; 50 mM sodium phosphate pH 4.1 5, 5.6, 6, 6.5 and 7).
Identification of a S. pombe orthologue of CLN3
The fission yeast homologue of CLN3, btn1, encodes a predicted transmembrane protein of 396 amino acids. Btn1p is 30% identical and 48% similar to its human counterpart (Fig. 1). Residues conserved between S. pombe Btn1p and vertebrate CLN3 proteins are also conserved in S. cerevisae Btn1p, notably in all proposed transmembrane segments as well as other regions including two of the predicted intra-organelle loops (CLN3 residues 161-211 and 304-370) and the C terminus. Several regions conserved between vertebrate CLN3 proteins are absent in the yeast proteins (CLN3 residues 1-25; 65-92 inclusive) and residues in at least one of the motifs important for trafficking of CLN3 are conserved (CLN3 residues M409(X)9G419). Amino acids affected by missense mutations that cause NCL (L101P, A158P, L170P, G187A, E295K, V330F, R334C, R334H, D416C) (see, http://www.ucl.ac.uk/ncl) are also conserved in S. pombe (L48, A106, L118, G136, E240, V278, R282, D363).
Cells deleted for btn1 have larger vacuoles
As a first step to determining the function of btn1, the gene was deleted by targeted replacement. Deletion of btn1 was confirmed by Southern blotting of genomic DNA from transformants (supplementary material, Fig. S1A) and by PCR (not shown). Cells deleted for btn1 (btn1Δ) were viable (Fig. 1B) albeit slower growing (Fig. 1C) and slightly longer at division (16±2.0 μm) than wild-type cells (13±1.7 μm) (Fig. 1B) when grown under normal laboratory conditions. btn1Δ cells also had a significantly higher septation index (13.1%) and an increase in the number of binucleate cells (6.3%) compared with wild type (8.6% septation, 2.2% binucleate) suggesting a perturbation of cell cycle progression. These pleiotropic effects could be complemented by reintroducing a plasmid containing the wild-type btn1 gene fused at the N terminus with GFP (12.8±2.0 μm mitotic cell length, 8.0% septation and 2.6% binucleate), and human CLN3. Since Btn1p in budding yeast is a vacuolar protein we investigated vacuolar integrity in btn1Δ cells. Wild-type and btn1Δ cells were incubated in the dye FM4-64, which is taken up into fission yeast cells via the endocytic pathway (Brazer et al., 2000; Iwaki et al., 2004b) (Gachet and Hyams, 2005) eventually residing in the vacuoles, allowing their visualisation. Uptake of FM4-64 was similar in both wild-type and btn1Δ cells (supplementary material, Fig. S2A). Vacuoles in wild-type cells had a mean diameter of 0.92±0.26 μm whilst in btn1Δ the mean was 1.34±0.39 μm. This increase in vacuole size could be complemented by reintroducing a plasmid containing the human CLN3 gene fused at the N terminus with GFP (supplementary materials Fig. S2B). In both cases, vacuoles fused in response to hypotonic stress but in btn1Δ the fused vacuoles were again larger than those of wild-type cells (Fig. 2A,B). Cells lacking btn1 had a larger total vacuolar volume than wild-type cells, both before and after fusion (data not shown).
Cells deleted for btn1 have more alkaline vacuoles
Vacuole size has recently been shown to relate to vacuolar pH (Iwaki et al., 2004a). We therefore investigated whether the increased vacuolar size in btn1Δ cells reflected increased vacuolar pH and whether the vacuole size was regulated in response to changes in external pH. We first monitored vacuolar pH using the proton sensitive probe CDCFDA which is membrane permeable until the acetate group is hydrolysed in the vacuole, whereupon it forms more highly charged compounds whose fluorescence intensity increases with decreased vacuole pH (Pringle et al., 1989). btn1Δ cells had larger vacuoles and increased vacuolar pH (reduced CDCFDA fluorescence) than wild-type cells when grown in standard medium (Fig. 3A). btn1Δ cells were marked by staining nuclear DNA with Hoechst and then mixed with unstained wild-type cells, and vice versa (data not shown), to ensure that both strains were subjected to identical growth conditions. In both cases, calcofluor, which stains septa, was also used to show that similar fluorescence levels were observed in cells at different stages of the cell cycle. The size of vacuoles and their pH was clearly distinct between cells deleted for btn1 and wild-type cells (Fig. 3B). The fluorescence intensity of CDCFDA increased linearly over time in both wild-type and btn1Δ cells (Fig. 3C). However, the intensity of fluorescence in btn1Δ cells was approximately 55% that of wild-type cells. This decrease could be rescued, at least partially, by reintroducing the btn1 gene fused at the N terminus with GFP (Fig. 3C). We also used LSDND as an indicator of vacuolar pH since its fluorescence does not depend on uptake into the vacuole. The fluorescence intensity of LSDND increases as pH decreased (there is no fluorescence of LSDND at or above pH 6). Again btn1Δ cells had larger vacuoles and increased vacuole pH (reduced LSDND fluorescence) than wild-type cells (Fig. 3D). In btn1Δ cells, intravacuolar pH was estimated to be 5.1 in contrast to 4.14 in wild-type cells (Fig. 3E). btn1Δ cells also had increased sensitivity to growth on the chemical D-(-)-threo-2-amino-1-[p-nitrophenyl]-1,3-propainediol (ANP) (Pearce et al., 1999a), consistent with increased vacuole pH (supplementary material, Fig. S3).
Taken together, these results show that cells lacking functional Btn1p have larger and more alkaline vacuoles. We therefore investigated whether altered vacuolar pH was able to change vacuole size. Vacuole size and vacuole pH of wild-type and btn1Δ cells changed when exposed to media of different pH (pH 3.0-7.5) (Fig. 3F), presumably through a homeostatic mechanism to maintain the pH of the cytosol (Bone et al., 1998). Lower extracellular pH resulted in smaller vacuoles with increased CDCFDA fluorescence in both strains. Conversely, higher extracellular pH resulted in larger vacuoles and decreased CDCFDA fluorescence. Thus the vacuolar size defect of btn1Δ cells can be rescued by decreasing extracellular pH.
Btn1p traffics to the vacuole
Lack of functional Btn1p clearly affects the size and pH of the vacuole. We determined the location of Btn1p by introducing a wild-type btn1 gene fused to GFP and under the control of the inducible nmt42 promoter (Maundrell, 1993), into wild-type and btn1Δ deletion strains. This construct rescued the pleiotropic effects of btn1Δ cells (data not shown) as well as decreased intravacuolar pH (Fig. 3C). GFP-Btn1 localised to the vacuolar membrane (confirmed by pre-staining vacuoles with FM4-64) and to a variety of smaller, presumably pre-vacuolar compartments (Iwaki et al., 2004b) (Fig. 4A). To follow the trafficking of Btn1 and to identify its final location the nmt42 promoter was repressed by the addition of thiamine. GFP-Btn1 was first visible in finger-like projections and cytoplasmic compartments mainly at the poles or septum of the cells, the major sites for endocytosis in S. pombe cells (Gachet and Hyams, 2005). These compartments co-stained early with FM4-64, consistent with a route that is identical or overlaps with an endocytic route via the plasma membrane (Fig. 4B). GFP-Btn1 was observed, over a 3-hour period, to traffic through theses small pre-vacuolar compartments to the vacuolar membrane (Fig. 4A).
Over-expression of Btn1p decreases vacuolar size
To investigate whether Btn1 directly influenced vacuolar size, we expressed GFP-Btn1 in btn1Δ cells and correlated vacuolar size with the expression level of GFP-Btn1p by measuring the intensity of GFP fluorescence. We measured the size of vacuoles in cells with identical levels of GFP-Btn1 expression (Fig. 4C) and in cells following induction of GFP-Btn1 expression (Fig. 4D). Increasing Btn1p levels correlated with decreasing vacuolar size and, by extrapolation from earlier results, probably also with decreasing vacuolar pH.
btn1 and vma1 show conditional synthetic lethality
The major contributor to the acidic pH of the vacuole is the activity of the vacuolar ATPase complex (vATPase) (Mellman et al., 1986; Takegawa et al., 2003; Iwaki et al., 2004a). S. pombe cells deleted for subunits of vATPase (Iwaki et al., 2004a) have extremely large, alkaline vacuoles and significantly reduced endocytosis. Surprisingly, GFP-Btn1p trafficked directly to the vacuole membrane in cells deleted for vma1 (or vma3, unpublished data), the gene encoding subunit A of the V1 portion of vATPase, and caused a striking reduction in vacuole size (supplementary material, Fig. S4A). This suggests that Btn1p can utilise a second trafficking route to the vacuole that bypasses the endocytic pathway. Since Btn1p and vATPase affect vacuole pH we examined the effect of combining btn1Δ and vma1Δ deletions. btn1Δvma1Δ cells were viable at 25°C and there was no difference in vacuole pH or ability to grow on media of varying pH between cells lacking vma1 or cells lacking vma1 and btn1 (supplementary material, Fig. S4B). However, Btn1 protein was essential for survival at any temperature over 30°C (Fig. 5A). After incubation for 2 hours at 30°C, 86% of btn1Δvma1Δ cells were binucleate and showed no septum deposition, and after 4 hours at the non-permissive temperature, 98% of cells were elongated and multinucleated without any septum deposition (Fig. 5B) compared to vma1Δ cells (6%) or btn1Δ cells (8.2%). Thus at higher temperatures, cells deleted for both btn1 and vma1 were elongated and swollen, and unable to initiate septum deposition, suggesting a severe defect in cytokinesis.
Trafficking of Btn1p to the vacuole is Ypt7 dependent
The Rab GTPase Ypt7p is essential for heterotypic (vesicle-) and homotypic (vacuole-) vacuole fusion in S. pombe (Bone et al., 1998; Iwaki et al., 2004b; Murray et al., 2001). GFP-Btn1 failed to localise to the small vacuoles of ypt7Δ cells (as defined by FM4-64 staining) but was retained in pre-vacuolar compartments (defined by lack of FM4-64 staining following its chase to vacuoles; Fig. 6A) indicating that Btn1p requires Ypt7p-dependent heterotypic fusion to traffic to the vacuole, consistent with our earlier observations on trafficking routes of Btn1p. However, GFP-Ypt7 localised correctly in btn1Δ cells (supplementary material, Fig. S5) and vacuoles were enlarged when GFP-Ypt7 was overexpressed in both wild-type and btn1Δ cells, suggesting that overexpression of Ypt7p increased vacuole fusion and overrode the effect of deleting btn1. Cells lacking ypt7 have smaller vacuoles (0.39±0.13 μm) than wild-type cells (Fig. 6A-C) because of the intrinsic defect in fusion. Just as deleting btn1 from wild-type cells resulted in larger vacuoles (1.34±0.39 μm; Fig. 2A,B), so the vacuoles of cells lacking both ypt7 and btn1 were larger than those of ypt7 cells (0.68±0.18 μm; Fig. 6B,C). The vacuoles of cells lacking ypt7 or ypt7 and btn1 did not fuse in response to hypotonic shock (Fig. 6B). In addition the fluorescence of CDCFDA in vacuoles was more intense in ypt7Δ cells than in cells lacking both ypt7 and btn1 (Fig. 6D), consistent with ypt7Δbtn1Δ cells having more alkaline vacuoles than ypt7 cells. This indicates that in cells deleted for ypt7 only, Btn1p was affecting vacuole size and pH from its location in the prevacuolar compartment. These results support a functional role in a pre-vacuolar compartment for Btn1p. Finally ypt7 and btn1 showed conditional synthetic lethality at 30°C (data not shown).
The human CLN3 gene can functionally compensate for btn1
We determined the intracellular location of CLN3 fused to GFP. Heterologous expression of this chimeric protein resulted in rapid trafficking of GFP-CLN3 to the vacuole membrane, such that immediately following promoter repression all GFP-CLN3 was present in the vacuole (Fig. 7A). Thus Btn1p and CLN3 both traffic to the vacuole, albeit with different kinetics. We examined whether Btn1p and CLN3 could restore the vacuolar sise and pH defects of btn1Δ cells. Overproduction of both GFP-Btn1 and GFP-CLN3 protein resulted in a reduced vacuole size (Fig. 7B,C) and also decreased pH (Fig. 7C) as monitored by increased CDCFDA fluorescence. Thus, CLN3 can functionally compensate for lack of Btn1 in terms of defects in vacuolar pH and size.
Effect of CLN3 disease mutations on vacuole homeostasis
Since CLN3 can substitute for Btn1p, we investigated whether mutations in Btn1p, which mimicked those in CLN3 causing human disease, affected its intracellular location and vacuolar function. There are ten missense mutations in CLN3 (see, http://www.ucl.ac.uk/ncl) that change conserved amino acid residues, most of which cause classic JNCL disease. However, some result in an altered disease course (Munroe et al., 1997) suggesting that the mutant protein retains some residual function. We first examined the location of three GFP-Btn1 proteins carrying different missense mutations 3 hours after repression of their expression in btn1Δ cells. GFP-Btn1G136A (CLN3 mutation G187A) was retarded in the endoplasmic reticulum (Pidoux et al., 1993) whereas GFP-Btn1E240K (E295K) and GFP-Btn1V278F (V330F) trafficked to the vacuolar membrane through the prevacuolar compartments (Fig. 7D), although slightly more slowly that wild-type Btn1. We next examined the effect of expressing these mutations on the vacuolar size of btn1Δ cells. The vacuoles of cells expressing all three mutant proteins remained enlarged in contrast to those expressing wild-type Btn1, which were reduced in size (Fig. 7B). Cells overexpressing GFP-Btn1E240K were also strikingly elongated (mean cell length at division was 22.1 μm) with an increased number of binucleate cells (10.5%) and increased septation index (19.0%) (Fig. 7D) suggesting a significant perturbation in the cell cycle. Cell length at mitosis and duration of septation of cells expressing GFP-Btn1G136A and GFP-Btn1V278F did not differ significantly from cells lacking the btn1 gene, although the number of binucleate cells was slightly increased (7.9% for GFP-Btn1G136A and 8.2% for GFP-Btn1V278F). We then monitored CDCFDA fluorescence to compare the vacuolar pH of the mutants expressed in btn1Δ cells (Fig. 7C). For GFP-Btn1G136A, vacuolar pH was similar to that in btn1Δ cells. For GFP-Btn1E240K and GFP-Btn1V278F, the pH defect was partially rescued (Fig. 7C), though not as well as by wild-type Btn1 or CLN3 protein. Thus, mutant Btn1p protein that can traffic to the vacuole may be capable of restoring vacuole pH but not vacuole size. In addition, in fission yeast it appears that control of vacuole size and pH regulation can be dissociated.
Btn1p is the functional homologue of CLN3
We identified btn1, the fission yeast homologue of the human Batten disease gene, CLN3, on the basis of sequence homology and functional complementation of a btn1 null strain by heterologously expressed CLN3. Many residues are conserved between vertebrate and fungal species, suggesting that these are critical to structure and/or function. These include residues important for trafficking of CLN3 (Kyttälä et al., 2003) and those affected by missense mutations (see, http://www.ucl.ac.uk/ncl). Other regions are absent or reduced in lower organisms, suggesting that these are not critical for function in these species. Btn1p is predicted to be a transmembrane protein and a GFP-Btn1 fusion protein localised to the vacuole membrane and to various pre-vacuolar compartments, in agreement with the observed location of CLN3 (Ezaki et al., 2003; Haskell et al., 2000; Järvelä et al., 1999; Järvelä et al., 1998; Luiro et al., 2001). Promoter repression experiments resulted in the accumulation of the fusion protein exclusively in the vacuole membrane, which we concluded was the final destination of Btn1p. In budding yeast, overexpression of its CLN3 homologue gave a location either in the matrix or in the membrane of the vacuole (Croopnick et al., 1998). The fission yeast btn1 null strain is viable but has vacuolar defects that are rescued by GFP-Btn1p and GFP-CLN3 chimeras, consistent with their location, and shows subtle and reproducible defects in cell cycle progression (both G2 and cytokinesis are extended) that are probably secondary effects. We are confident therefore that we have correctly identified the cellular location of Btn1p. A vacuole location for Btn1p, or an effect on vacuole function, is consistent with accumulation of storage of material in the lysosome in the human disease. However, since trafficking of GFP-Btn1 to the vacuole took significantly longer than that reported for other vacuolar membrane proteins in S. pombe (Bellemare et al., 2002; Gaits et al., 1999; Iwaki et al., 2003), Btn1p may be specifically retarded in its trafficking and exerting a functional effect en route to the vacuole.
Btn1p acts in vacuole pH homeostasis
Two significant finding of these studies were that vacuolar pH was elevated by approximately 1 pH unit in btn1Δ cells and that overexpression of Btn1p reduced vacuole pH, strongly supporting a role for Btn1p in vacuole homeostasis. This is opposite to that reported for btn1Δ strains of budding yeast (Pearce, et al., 1998), and CLN3 overexpression studies in HEK293 cells (Golabek et al., 2000), but in agreement with observations made in cells from most types of NCL, including JNCL (Holopainen et al., 2001), supporting the use of fission yeast as a model for Batten disease. Our observation that the severity of mutations in Btn1p that mimicked those causing JNCL correlated with the increased vacuole pH further supports the use of fission yeast as a model for Batten disease.
Btn1p is functional in a pre-vacuolar compartment
Trafficking of Btn1p to the vacuole appeared to follow or overlap with the endocytic route since it was first visible in early FM4-64-staining compartments and was dependent on Ypt7p, which is required for vacuole fusion (Bone et al., 1998). However, trafficking of Btn1p to the vacuole, like other vacuole membrane proteins, was not dependent on a functioning vATPase, required for acidification of the endocytic pathway and for endocytosis (Iwaki et al., 2004b), suggesting that an alternate route to the vacuole can be exploited. Vacuole size in ypt7Δ cells increased when btn1 was also deleted, indicating that Btn1p exerted an effect on vacuole size separate from that within the vacuole membrane, i.e. it has a functional role in a pre-vacuolar compartment. Cells deleted for vma1 required Btn1p for cytokinesis. This implies either that the non-vacuolar role of Btn1p is required for cytokinesis and is independent of its contribution to vacuole function or that the function of Btn1 in a vacuole lacking a functional vATPase is essential for cytokinesis.
The disease severity of disease mutations may correlate with their effect on vacuole pH
Overexpression of three different disease-causing mutations in Btn1p in a btn1 null background caused enlargement of vacuoles but had different effects on vacuolar pH. In humans, the CLN3 mutation G187A causes a classic disease course, and expression of GFP-Btn1G136A did not rescue the pH defect of btn1Δ cells. Residue G187 (G136 in Btn1p) is located within a highly conserved stretch of amino acids that is predicted to project into the lumen of the organelle (Ezaki et al., 2003; Janes et al., 1996; Kyttälä et al., 2003; Mao et al., 2003). Its mutation could interfere with the conformation of Btn1p, consistent with its retarded trafficking and retention within the ER. Disease mutation E295K is associated with a markedly protracted disease course (Lauronen et al., 1999; Munroe et al., 1997; Wisniewski et al., 1998) suggesting that the mutant protein retains significant residual activity. Consistently, GFP-Btn1E240K protein targeted the vacuole membrane and partially rescued the vacuolar pH defect of btn1Δ cells, further supporting a link between disease severity and vacuole pH in this model system. The same mutant CLN3 protein trafficked to the lysosome (Järvelä et al., 1999). Mutation V330F, although thought to be associated with a classical disease course (Munroe et al., 1997), when expressed in S. cerevisiae clearly allowed residual activity of the mutant protein (Haskell et al., 2000), consistent with the ability of GFP-Btn1V278F to traffic to the vacuole and partially rescue the pH defect of btn1Δ cells. It has not been possible to confirm whether the patient originally reported as carrying the mutation V330F did indeed have a more protracted disease course. Mutant E295K was also associated with a marked increase in cell length, indicative of a significant delay in the cell cycle that must be connected with the residual Btn1p activity associated with this mutant, and may be linked to the requirement for Btn1p for cytokinesis in vma1Δ cells and the higher septation index and slower growth of btn1Δ cells.
The molecular basis of NCL
How does absence or mutation in CLN3 cause disease? Our results are consistent with observations that the pH of lysosomes in cells from a JNCL patient are more alkaline. This would affect the activity of lysosomal enzymes and membrane transporters and may explain the accumulation of undegraded material as the disease progresses, with the build-up of subunit c of mitochondrial ATP synthase due to a defect in the final stages of autophagocytic degradation (Nakamura et al., 1997). In addition to other types of NCLs (Holopainen et al., 2001), another lysosomal storage disorder, mucolipidosis IV, is associated with enlarged lysosomes and increased lysosomal pH (Bach et al., 1999). The MCOLN1 gene encodes a transient receptor potential (TRP)-related channel that is pH regulated and involved in fusion of membranes within the endosome-lysosome system (LaPlante et al., 2004; Raychowdhury et al., 2004). Altered lysosomal pH might therefore exert a secondary effect on trafficking and fusion of compartments of the endocytic pathway and other vesicular routes. In S. pombe, we have also shown that in cells already compromised in the acidification of vesicular compartments (i.e. cells deleted for vma1) Btn1p is essential for septum formation (Wang, H. et al., 2002), and in cells lacking btn1 the process of septation is lengthened. In the brain, a defect in trafficking could have a significant impact on transmitter movement, release and uptake and may be the molecular basis for the neurodegeneration that occurs in JNCL. However it is also possible that these effects are secondary to the loss of function of CLN3, and whilst they may contribute significantly to the disease mechanism, the primary function of CLN3 has yet to be defined.
Fission yeast lacking the S. pombe orthologue of another human NCL gene, CLN1, were also retarded in their growth and were sensitive to raised extracellular pH and to sodium orthovanadate (Cho et al., 2004). This implicates Ppt1p in activities that may affect vacuole pH or homeostasis. The severity of the sensitivity to raised extracellular pH was more marked in S. pombe cells expressing mutant Ppt1p than those with mutant Btn1p, in line with the severity of the human disease, and supports the hypothesis that part of the pathogenesis of NCL may be due to increased lysosomal pH. In S. cerevisiae, palmitoylation is known to be important in vacuole fusion (Dietrich et al., 2004; Wang, Y. et al., 2001) suggesting that depalmitoylation may also be important, either as a switch or during protein turnover. Thus inactivity of a depalmitoylating enzyme such as Ppt1p, or of the protein Btn1p, may interfere with vacuole integrity through defects in vacuole fusion and pH homeostasis. Just as the use of budding yeast is proving fruitful as a model for other neurodegenerative disorders (Malathi et al., 2004; Outeiro et al., 2004) we conclude that the use of fission yeast, with its many vacuoles, is particularly appropriate to model Batten disease, and the advantages associated with such a genetically tractable organism may prove vital in determining the function of Btn1p/CLN3 and the biological roles of both Btn1p/CLN3 and Ppt1/CLN1.
We thank Kaoru Takegawa and John Armstrong for supplying strains, and Chris Knapp and Vasanti Amin for excellent technical assistance. Y.G. and S.C. were supported by the Wellcome Trust, UK (grants 066043 to J.H. and 054606 to S.M.). We thank the Children's Brain Diseases Foundation, USA and the Batten Disease Support and Research Association, USA for additional financial support.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/118/23/5525/DC1
↵* Present address: Institut d'Exploraton Fonctionelle des Génomes, Université Paul Sabatier, Toulouse 31062, France
↵‡ These authors contributed equally to this work
↵§ Present address: MRC Laboratory for Molecular Cell Biology, University College London, Gower Street, London, WC1E 6BT, UK
↵¶ Present address: Institute of Molecular Biosciences, Massey University, Palmeston North 11222, New Zealand
- Accepted August 16, 2005.
- © The Company of Biologists Limited 2005