The neuronal ceroid lipofuscinoses (NCLs) are common neurodegenerative disorders of childhood and are classified as lysosomal storage diseases since affected cells exhibit lysosomes containing ceroid and lipofuscin-like material. CLN3 is the most widely conserved NCL gene, suggesting that it has a basic eukaryotic cell function; its loss might be expected to cause the earliest onset and/or most severe disease. However, mutations in CLN3 are linked to juvenile NCL (JNCL), the latest onset and mildest form of NCL in children. We sought to explain this paradox. Almost all patients with JNCL are homozygous or heterozygous for an intragenic 1 kb deletion within CLN3, hitherto presumed to be a null mutation. We hypothesized that the 1 kb mutation may allow CLN3 residual function. We confirmed the presence of CLN3 transcripts in JNCL patient cells. When RNA silencing was used to deplete these transcripts in cells from JNCL patients, the lysosomes significantly increased in size, confirming the presence of functional protein in these cells. Consistently, overexpression of mutant CLN3 transcript caused lysosomes to decrease in size. We modelled the JNCL mutant transcripts and those corresponding to mouse models for Cln3 in Schizosaccharomyces pombe and confirmed that most transcripts retained significant function as we predicted. Therefore, we concluded that the common mutant CLN3 protein does indeed retain significant function and that JNCL is a mutation-specific disease phenotype. This finding has important consequences for recognition and diagnosis of disease caused by mutations in CLN3 and for the development of therapy for JNCL.
The neuronal ceroid lipofuscinoses (NCLs), also known as Batten disease, are the most common neurodegenerative disorders of childhood, inherited in an autosomal recessive manner, and characterized by an accumulation of autofluorescent material within lysosomes in most cells (1,2). They are therefore classed as lysosomal storage disorders. At least nine human gene loci have been implicated, each with a characteristic age of onset (3–5). The prevalent NCL world-wide is juvenile NCL (JNCL). This has the latest onset and mildest form in children but is linked to the gene, CLN3 (6), that is the most widely conserved of the NCL genes (7), suggesting that it has the most basic eukaryotic cell function of this family. The loss of CLN3 might therefore be expected to cause the earliest onset and/or most severe disease. We sought to explain this paradox.
Most JNCL patients world-wide carry an intragenic 1 kb deletion in CLN3 (8) which results in the deletion of two adjacent exons and is predicted to produce a transcript where the RNA sequence is out of frame after only153 of the total 438 amino acid residues. It has generally been assumed that this mutant protein would either be non-functional and/or rapidly degraded, and early evidence had suggested that it was not trafficked correctly (9). We hypothesized that this mutation did not completely abolish CLN3 function. If so, such a transcript would produce an active protein, and this could change the interpretation of data arising from most models of the disease if they also did not completely lack CLN3 function. Previously, two mouse models (Cln3Δex7/8 and Cln3Δex7/8insneo) lacking the same two exons of the mouse orthologue Cln3 were found to express Cln3 transcripts (10,11). A third model (Cln3Δex1–6) that is often considered a knockout model retains the promoter region and the 5′-UTR of exon 1 and exons 7–15 of Cln3 (12). Since these three mouse models, which were readily derived, have similar although not identical phenotypes, we predicted that each expressed mutant Cln3 proteins that retained functionality. In contrast, a recent mouse model in which exons 1–8 of Cln3 were replaced with the β-galactosidase gene (Cln3Δex1–8inslacZ) (13) had no detectable Cln3 transcripts. However, eight backcrosses were required to derive a mouse homozygous for the mutated Cln3 allele, and the investigators think that this process allowed co-segregation of a modifying function that rescued the mouse from embryonic lethality caused by absence of functional Cln3 protein (B. Davidson, personal communication).
In a mammalian system, RNA silencing can target a specific gene transcript for destruction. This technique can therefore be exploited to determine whether mutant transcript, such as that arising from the 1 kb deletion of CLN3, retains some function, by examining any effect arising from its loss. We used this approach to compare cells from healthy individuals, cells from JNCL patients and those in which the mutant CLN3 transcript had selectively been destroyed, paying particular attention to the lysosomes since this is where the autofluorescent material characteristic of the NCLs accumulates.
A fission yeast model for JNCL has already provided important clues to both the function of CLN3 and the cellular consequences of JNCL mutations. Complete deletion of the fission yeast Schizosaccharomyces pombe orthologue to the CLN3 gene, btn1, causes an increase in the size of the yeast vacuoles (equivalent to mammalian lysosomes), a phenotype that can be rescued by ectopic expression of Btn1p or CLN3p but not by expression of Btn1p carrying three disease-causing missense mutations (7). This system is therefore amenable to testing whether mutant Btn1 protein(s) mimicking the CLN3 transcripts present in cells from a JNCL patient homozygous for the common 1 kb deletion retain function that affects the yeast vacuole. Interestingly, no change in vacuole size has been reported when the btn1 gene is deleted or overexpressed in Saccharomyces cerevisiae, although other phenotypes have been described (14–18).
These complementary approaches, utilizing two different model systems, were used to determine whether the common 1 kb deletion completely abolished CLN3 function, as had previously been assumed, or allowed production of mutant protein retaining significant function.
Lysosome size correlates with CLN3 expression
The NCLs are lysosomal storage diseases which suggests that lysosomes in cells from patients are dysfunctional, consistent with previous data showing that in most types of NCL the pH of the lysosome was less acidic (19). We had previously shown using S. pombe that vacuole size inversely correlates with the level of expression of Btn1p, the orthologue to CLN3 (7) (Fig. 1A and C). Low levels of expression of btn1 were associated with large vacuoles, whereas high expression levels of btn1 caused vacuoles to be smaller than those of wild-type cells. Significantly, the human CLN3 gene was able to fully rescue the enlarged and less acidic vacuole phenotype of cells deleted for btn1 (7) as well as other observed differences. We therefore investigated whether wild-type CLN3 affected lysosomes of mammalian cells in a manner similar to fission yeast since this would provide a readily detectable marker phenotype for lysosome dysfunction arising from mutations in CLN3. First, we examined the effect of depletion of CLN3 protein by siRNA in HeLa cells and found that lysosome size in these cells was increased (Fig. 1B and D). Next, we examined the effect of overexpression of CLN3 protein in HeLa cells, and found that lysosome size in these cells was decreased (Fig. 1B and D). In mammalian cells, therefore, it appears that lysosome size inversely correlates with the level of expression of CLN3, as in fission yeast cells. Thus, the size of vacuoles in fission yeast cells and the size of lysosomes in mammalian cells are sensitive assay systems for Btn1p or CLN3 function, allowing the effects of lower and higher levels of expression compared to endogenous protein to be observed. Unfortunately, the normal levels of endogenous expression of CLN3 were too low to detect using already available and newly derived antibodies (data not shown) so that we were unable to directly correlate CLN3 protein levels with lysosome size. However, this system was suitable for investigating whether any function was associated with the mutant protein arising from the most common CLN3 mutation, the 1 kb intragenic deletion.
Depletion of CLN3 in JNCL cells increases lysosome size
We examined the size of lysosomes in cultured skin fibroblast cells from six healthy controls and from four patients with JNCL and found that, although the size of the lysosomes was very variable in all cell lines, there was no striking difference between control and patient cells (data not shown). We next examined the cellular effects of inactivation of CLN3 function in mammalian cells. When we used RNA silencing (siRNA) to inhibit expression of CLN3 in cultured cells from a healthy individual, we observed an increase in the size of the lysosomes (Fig. 2A and C). Inhibition of CLN3 expression in cells from a patient homozygous for the 1 kb deletion also caused an increase in the size of the lysosomes (Fig. 2B and C). We checked the specificity of the siRNA inhibition by using four different silencing oligonucleotides to CLN3: those directed to part of the CLN3 gene outside of the 1 kb deletion caused enlargement of lysosomes, but two oligonucleotides targeted to the missing exons had no effect on cells from patients but caused enlargement of lysosomes in normal cells (Fig. 2D). These results suggested that cells from patients homozygous for the 1 kb deletion produce a mutant CLN3 protein that can influence the size of lysosomes in a manner similar to wild-type protein.
Mutant CLN3 transcripts are present in JNCL cells
We determined the CLN3 transcripts present in cells from a JNCL patient homozygous for the 1 kb deletion to identify those that might encode protein that affected lysosome size. We detected two transcripts by RT–PCR and sequencing (Fig. 3). The major transcript contained exon 6 spliced to exon 9, which encoded a truncated CLN3 protein containing the first 153 amino acids of CLN3 plus an additional 28 novel amino acids due to out-of-frame RNA sequence at the novel splice site (GenBank accession no. EF587245). A second, minor, transcript contained exon 6 spliced to exon 10, which encoded a protein that is missing amino acids 154–263 of CLN3 as a result of exon skipping and the restoration of an in-frame RNA sequence at the novel splice site (GenBank accession no. EF587244).
Mutant JNCL transcripts retain function
Schizosaccharomyces pombe cells deleted for btn1, the fission yeast orthologue to CLN3, have larger vacuoles than wild-type cells (7). This phenotype can be rescued by expression of both Btn1p and human CLN3 protein (7), suggesting that both proteins perform an equivalent function. To confirm whether the mutant transcripts detected in cells from JNCL patients exerted a function that affects lysosome size, we expressed mutant Btn1 proteins that mimicked these two mutant transcripts (Fig. 4A) in the S. pombe model system. Both mutant proteins (GFP.Btn1p102fsX5, equivalent to a truncated CLN3 protein containing the first 153 amino acids of CLN3, and GFP.Btn1p102–208del, equivalent to a mutated CLN3 protein that is missing amino acids 154–263 of CLN3) rescued the vacuole phenotype as effectively as the wild-type protein (Fig. 4B). This indicates that both mutant Btn1 proteins retained significant function. Not all Btn1p mutants were indiscriminately able to rescue the vacuole phenotype: expression of a mutant Btn1p (GFP.Btn1pG136A) mimicking a very rare CLN3 missense mutation, p.Gly187Ala, that affects a single residue encoded within the missing exons did not rescue (Fig. 4B) (7) nor did previously published mutant Btn1 proteins mimicking the missense disease mutations p.Val330Phe or p.Glu295Lys (7).
Using the S. pombe system, we also showed that mutant proteins similar to those encoded by the three Cln3 mice models that expressed detectable Cln3 transcripts (Fig. 4A) were also capable of rescuing vacuole size. Mutant protein GFP.Btn1p102fsX5, equivalent to a truncated CLN3 protein containing the first 153 amino acids of CLN3, corresponds to the majority of the transcripts detected in the Cln3Δex7/8 mouse (10) and the Cln3Δex7/8insneo mouse (11). GFP.Btn1p103–396, equivalent to a mutant Cln3 protein encoded by exons 7–15, covers the longest transcript that could be present in the Cln3Δex1–6 mouse (12). Both mutant proteins restored vacuole size (Fig. 4B). Although no transcript was detected in the Cln3Δex1–8inslacZ mouse, the longest possible mutant CLN3 transcript could contain the equivalent of human exons 9–15 which was mimicked in construct GFP.Btn1p200–396. This construct did not rescue vacuole size consistent with this mouse model lacking a mutant Cln3 protein that exerted significant function.
Taken together, these results indicate that cells from patients homozygous for the 1 kb deletion produce a CLN3 protein that can function to influence the size of lysosomes and that three of the four Cln3 mouse models could also produce Cln3 protein that exerts a function affecting lysosome size.
Expression of mutant CLN3 encoding residues 1–153 decreases lysosome size
We have already shown that overexpression of wild-type CLN3 in normal cells causes a decrease in size of lysosomes (Fig. 1B and D). We next tested whether a function that affected lysosome size was exerted by a mutant CLN3 equivalent to that expressed in JNCL cells homozygous for the 1 kb deletion, i.e. CLN3 protein encoding amino acid residues 1–153 followed by 28 other amino acids. When such a construct was expressed in normal fibroblast cells and fibroblast cells from a JNCL patient homozygous for a 1 kb deletion, there was a reduction in lysosome size in both cell types (Fig. 5A and B) confirming that this mutant construct was functional. We checked the intracellular location of wild-type and mutant CLN3 expressed from the constructs used in our system. Wild-type CLN3 tagged with GFP at its N-terminus trafficked to the lysosome but the majority of the truncated CLN3 protein expressed by the 1 kb mutant construct was retained in the endoplasmic reticulum, consistent with a previous report (9) (Fig. 5C). To prove that the presence of mutant CLN3 within the ER was not itself affecting lysosome size, we also overexpressed a GFP-tagged mutant CLN3 protein containing the disease mutation p.Leu101Pro which was also retained in the ER and which we never observed in Lamp1-staining compartments in normal cells (Fig. 5C). Expression of CLN3 containing this missense mutation had no effect on lysosome size in both healthy and patient fibroblasts (Fig. 5B). We had previously shown in S. pombe that Btn1p affected vacuole size prior to it reaching the vacuole (7). We therefore concluded that the 1 kb mutant CLN3 protein may be able to affect lysosome size from the ER though some may be trafficking further and able to act en route to the lysosome.
Two different cell systems have been used to investigate whether the most common mutation in the CLN3 gene, which is associated with disease onset late in childhood, allows production of a partially functional protein. We defined the RNA transcripts from the CLN3 gene present in cells from a JNCL patient homozygous for the 1 kb deletion and discovered two transcripts that are both predicted to encode the first 153 amino acids of CLN3, with one also containing residues 264–438. Using RNA interference to specifically deplete these transcripts in cells from a JNCL patient resulted in a significant increase in the size of the lysosomes in these cells suggesting that one or both of these mutant CLN3 transcripts was biologically active. Supporting this conclusion is our observation that constructs expressed in mammalian cells or in S. pombe that mimic the common 1 kb deletion mutation of CLN3 are biologically active and cause a decrease in the size of lysosomes or vacuoles, the opposite effect to that of depleting CLN3 transcripts or deleting the S. pombe CLN3 orthologue btn1.
CLN3 is the most highly conserved of the NCL genes with homologues identified to date in 46 diverse eukaryotic species with the sequence of 50 CLN3 proteins available (Caenorhabditis species have three cln-3 genes). Sequence homology extends from residue 41 of the human CLN3 protein in large stretches to the end of the protein, with certain residues, including most disease-causing missense mutations, identical or similar across all species (data not shown). Significantly, the region of CLN3 deleted by the most common mutation (amino acids 154–227) is conserved in sequence and length suggesting that it is important structurally or functionally, although it contains only three of the 10 known disease-causing missense mutations. Our results suggest that JNCL is caused by loss of function associated with this region of CLN3. Other studies have deposited sequences of transcripts reported to be from patients with the ‘56’ genetic haplotype that is in linkage disequilibrium with the 1 kb deletion (6). However, in those confirmed to be lacking exons 7 and 8, some (GenBank accession nos AF077964 and AF077968) were predicted to express protein products identical to one of the mutant transcripts that we report.
We have shown that lysosome size can be affected by a truncated protein that contains residues 1-153 or, by extrapolation from the yeast data (i.e. Btn1p2–102del), residues 154–438 of CLN3. Only nine JNCL patients have been defined who do not carry the 1 kb deletion on either disease allele. Eight of these are homozygous for different mutations (c.424delG, c.558delAG, c.586-587insG, c.622-623insT, c.631C>T, c.944-945insA, 6 kb deletion with undefined boundaries) (6,8) (http://www.ucl.ac.uk/ncl/cln3.shtml). One is heterozygous for a missense mutation and c.1056+3A>C (http://www.ucl.ac.uk/ncl/cln3.shtml). All [if, as likely, the 5′ boundary of the 6 kb deletion is downstream of exon 6 (6,8)] of these nine patients are predicted to produce a transcript from one disease allele that either includes CLN3 residues encoded by the first six exons or, as a result of alternative splicing, residues encoded by exons 7–15 (http://www.ucl.ac.uk/ncl/cln3.shtml), and which therefore should exhibit a function, like the 1 kb mutant protein, that affects lysosome size.
Our results indicate that the common 1 kb deletion present in most patients diagnosed with JNCL does not completely abolish CLN3 function. This is in contradiction with a recent report (20) that did not take into account the possibility that the 1 kb deletion disease allele may provide some CLN3 function. We conclude that JNCL, one of the most common types of NCL, is actually a mutation-specific disease arising from a partial loss of CLN3 function. Similar genotype–phenotype correlations for other genes are increasingly being recognized (21–24) including one other example within the NCLs: the mutation p.Arg24Gly in the CLN8 gene causes a distinct disease phenotype, progressive epilepsy with mental retardation or northern epilepsy (25), whereas other known mutations in the same gene are associated with variant late infantile NCL (3).
The revelation that JNCL arises when partial CLN3 function remains has a number of important implications. One is that total loss of CLN3 function almost certainly causes a more severe disease. We therefore addressed whether any data existed that did not support this conclusion. Our data predict that mice that lack functional Cln3 would have severe disease or be unviable, whereas mice that express Cln3 containing exons 1–6 would have a phenotype similar to JNCL. Two mouse models for Cln3 exist that express detectable mutant transcripts (10,11) which, even without exon skipping to extend the ORF, would encode proteins that contain the equivalent of residues 1–153 of CLN3, so these mice almost certainly have partial Cln3 function. The mouse model that lacks most of exons 1–6 (12) is often referred to as a ‘knockout’ model because it was assumed that any mutant transcripts produced would not be functional. However, the expression of a transcript in yeast equivalent to one containing exons 7–15 is able to rescue vacuole size. We therefore suspect that this mouse produces a transcript and does express a mutant Cln3 protein that retains significant function. The most recent Cln3 mouse model lacks exons 1–8 and no Cln3 transcript was detected in this mouse (13). We showed that a Cln3 transcript encoding exons 9–15 was unlikely to be functional, since the equivalent residues in Btn1p were not able to rescue vacuole size. For our hypothesis to be correct, the mild phenotype of this mouse must arise via another mechanism. Those responsible for deriving this mouse consider that it expresses a modifying function that rescues observed embryonic lethality since multiple backcrosses were required to produce viable progeny homozygous for the deleted allele. Certainly, genes that modify the pathogenic process of the Cln3Δex7/8 mouse that is most similar to JNCL exist since specific phenotypic aspects of this mouse vary according to genetic background (10). Available data are therefore consistent with our hypothesis. Additionally, this suggests that the rescue of lysosome size in cells lacking CLN3 function may be critical to survival.
Another implication from our data is that other mutations in CLN3 may not always result in the recognized hallmarks of classic JNCL (2), which include fingerprint profiles of the storage material visible using electron microscopy, vacuolated lymphocytes (26), and visual failure. Indeed, some mutations in CLN3 may cause diseases not currently classed as NCL, including embryonic death, or severe diseases that present in infancy, as well as milder and more protracted forms of NCL, some of which may not be inherited in a recessive manner. The true level of diseases caused by mutations in CLN3 will have to be re-assessed since this may include genetic forms of visual failure (27,28), perinatal death (29) and late-onset dementia (30), in which many of the underlying genes are still unknown. To date, most variability in JNCL progression is associated with missense mutations. Ten missense mutations are known in CLN3, some of which are associated with a milder disease course although disease onset is not changed. Nine of these are found heterozygous with the 1 kb deletion, and one is found heterozygous with a mutation (c.1056+3A>C) also predicted to produce a similar transcript to that of the 1 kb deletion. Thus, patients with a milder disease course have at least one transcript that rescues lysosome size (the 1 kb deletion) and one (the missense mutation) that presumably retains some other function of CLN3. In JNCL patients heterozygous for the 1 kb deletion mutation, and particularly if the disease progression is not following a classical course, every effort to identify the other disease allele should be made to add to the list of known mutations in CLN3 (http://www.ucl.ac.uk/ncl/cln3.shtml), so that the effect of these mutations on CLN3 function can be studied experimentally. Mutation analysis by complete sequencing of CLN3 should now also be routinely considered for atypical cases of NCL, of both early and late onset, and the absence of the 1 kb deletion should not be taken as indicative of lack of any mutation in CLN3.
Our findings suggest that CLN3 contributes an important and complex cellular function and that lysosome size is affected by the level of CLN3 expression. Even before the CLN3 gene was identified, it had been noted that carriers of the gene that caused JNCL, who almost certainly were carriers for the 1 kb deletion and therefore expressed a fully functional CLN3 protein from one allele and a partially functional protein from the 1 kb allele, had subclinical eye changes (31). Given that most patients with JNCL present with visual problems, and for a few patients, this was their major clinical symptom for decades; this is consistent with retinal cells being most sensitive to malfunctioning or lower levels of CLN3 protein. The derivation of sophisticated vertebrate models in which the complete abolition of CLN3 function can be induced will be required to assess whether our prediction of embryonic lethality is correct, and the use of these and simpler models may be vital to understand the role of CLN3 in embryonic and brain development.
The function of CLN3 has remained a mystery for more than a decade since the gene was first identified. It was reported recently that CLN3 may possess a novel Δ9 palmitoyl desaturase activity (32) that was initially investigated due to low-level homology of the N-terminus with a Pfam domain and possible conservation of His residues encoded by exon 3. However, the His residues in question are not conserved in S. pombe Btn1p. Extensive work using the S. cerevisiae model has suggested that Btn1p affects vacuole homeostasis including arginine transport and the activity of the V-type H+ ATPase (14–18) but surprisingly with the opposite effect on vacuole pH to that in S. pombe (7) and lysosome pH in patients’ cells (19), and no effect on vacuole size has been reported. Our work in the S. pombe system has indicated that Btn1p (and therefore probably CLN3) has multiple functions (7), one of which impacts on the size of the vacuole and has been investigated here. That two regions of CLN3/Btn1p affect lysosome or vacuole size indicate the complexity of CLN3 function. Future work will be able to show whether the decrease in Δ9 fatty acid desaturase activity correlates with the effect of mutant CLN3 on vacuole size or with some other aspect of its function. We are continuing to use the S. pombe model system to investigate the biological activities of the Btn1 and CLN3 proteins.
Finally, since we can say that most JNCL patients retain partial CLN3 function, then the mutant protein produced by the 1 kb deleted allele is capable of staving off early onset of disease. Thus, therapeutic approaches that either enhance this function or mimic the effect of this function in neurons may considerably extend and increase the quality of life of patients. Importantly, it seems that manipulation of an intrinsic intracellular activity can offset the loss of Cln3 in a mouse (13). However, the most effective therapeutic approach will depend on the discovery of the true function or indeed functions of this elusive protein.
MATERIALS AND METHODS
Schizosaccharomycespombe strains and cell growth
Strains used in this study are listed are as follows: wild-type 972 (h−); YG660 (btn1Δ) h+, btn1::leu2, ura4-D18, leu1-32, his2; RH11 (YG660+pREP42GFPBtn1102fsX5) h+, ura4-D18, leu1-32, his2; RH13 (YG660+pREP42GFPBtn1102–208del) h+, ura4-D18, leu1-32, his2; RH14 (YG660+pREP42GFPBtn1103–396) h+, ura4-D18, leu1-32, his2; RH15 (YG660+pREP42GFPBtn1200–396) h+, ura4-D18, leu1-32, his2; SC12 (7) (YG660+pREP42GFPBtn1G136A) h+, ura4-D18, leu1-32, his2. Media, growth and maintenance of strains were as described (7, 33). Cells were grown in minimal medium containing appropriate supplements.
Construction of yeast expression plasmids
Deletions were introduced into clone pREP42GFPBtn1, which contained GFP fused to the N-terminus of Btn1p (7), using inverse PCR (34) with Pfu DNA Polymerase (Promega, Madison, WI, USA) and the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA). Forward and reverse primers 5′-ATGTACTTTTTTGTGTTGCCAG-3′ and 5′-CAAGCATTTTCGAAGGGACG-3′ were used to produce the construct pREP42GFPBtn1102–176delinsD that encodes Btn1p lacking amino acids 102–176 with an aspartate (d) residue inserted between residues 101 and 177. Mutagenesis forward primer 5′-GACTTTTTTGTGTAGCCAGAGTCTGAATCTACATCACC-3′ and reverse primer 5′- GGTGATGTAGATTCAGACTCTGGCTACACAAAAAAG -3′ (changed bases are underlined) were then used on pREP42GFPBtn1102–176delinsD to introduce a stop codon soon after Btn1p amino acid 101, resulting in construct pREP42GFPBtn1102fsX5. Forward and reverse primers 5′-TTCAATTTCGTTAATTCTTTGAAG-3′ and 5′-AAGCATTTTCGAAGGGACGG-3′ were used on pREP42GFPBtn1 to produce pREP42GFP-Btn1102–208del. Reverse primer 5′-CATGGTCGACATATGTTTGTATAG-3′ and forward primers 5′-GTATCTCTGGCAGCCATTTCATCGAGTTTTG-3′ and 5′-TCTATTGATCTTCGAGCAGGGC-3′ were used to produce pREP42GFP-Btn1103–396 and pREP42GFP-Btn1200–396, respectively. The construction of pREP42GFPBtn1G136A has been described previously (7). All constructs were verified by sequence analysis. It is possible, since each Btn1p protein was tagged with GFP at its N-terminus to allow us to confirm its intracellular expression and trafficking, that this tag differentially interfered with wild-type, mutant or truncated proteins.
Visualization of vacuoles
To visualize vacuoles, cells were incubated for 1 h at 25°C with 4 µl/ml N-(3-triethylammoniumpropyl)-4-(6-(4-(diethylamino) phenyl) hexatrienyl) pyridinium dibromide (FM4-64) (Molecular Probes, Invitrogen, Paisley, Scotland), with shaking, washed in minimal media and chased for 30 min under the same conditions. The diameter of every vacuole visible in one focal plane of a cell was measured using OpenLab 3.5.1 software and downloaded to Microsoft Excel for analysis (more than 500 vacuoles were counted for each data set, and the data presented are the mean of three independent experiments). Only data above the resolving limit of the microscope used are presented. GraphPad Prism 4.0 software was used to perform Student’s unpaired t-tests to assess the significance of a reduction in vacuole size compared with cells deleted for btn1.
Images of yeast cells were visualized as described previously (7). Tissue culture cells grown on cover slips were fixed with 3% PFA, quenched and permeabilized with 50 mm NH4Cl and 0.2% saponin. After blocking in PGAS (PBS, 0.2% gelatin, 0.02% NaN3 and 0.02% saponin), the cells were incubated with primary antibody for 1 h, washed in PGAS, followed by incubation with a Texas red-conjugated secondary antibody for 1 h (Jackson ImmunoResearch, West Grove, PA, USA) together with Alexa Fluor 350 streptavidin conjugate (Invitrogen). After washing in PGAS, cover slips were mounted with ProLong antifade reagent and DAPI (Molecular Probes). The monoclonal Lamp1 antibody H4A3 developed by J.T. August/J.E.K. Hildreth was obtained from the Developmental Studies Hybridoma Bank. Cells expressing GFP constructs were mounted directly after fixing. Images were taken by a ZEISS Axioskop inverted microscope fitted with a Hamamatsu Orca Camera using a 100× oil-immersion objective. In a representative set of cells the perimeter of lysosomes that were clearly distinguishable were measured using OpenLab 3.5.1 software and downloaded to Microsoft Excel for analysis (more than 50 lysosomes were measured for each data set). The significance of the change in lysosome size was assessed performing Student’s t-test.
Construction of cln3 expression constructs
CLN3 cDNA was subcloned into pcDNA3.1/NT-GFP-TOPO (Invitrogen) and pGEM-T (Promega) following amplification from pREP42GFP-CLN3 (7) using forward and reverse primers 5′-ATGGGAGGCTGTGCAGGCTCG-3′ and 5′-TCAGGAGAGCTGGCAGAGG-3′. From pGEM-T-CLN3 the gene was subcloned into pCI-neo (Promega) as an EcoRI fragment to give pCI-neo-CLN3. The 1 kb deletion was introduced into this vector using forward and reverse primers 5′-CTATTTCTTGTTCGTCAC-3′ and 5′-CACACAGGCTGGTCCCC-3′ and intramolecular annealing. The 1 kb fragment was subcloned into pEGFP-C3 (BD Biosciences, Franklin Lakes, NJ, USA) as a XhoI–XbaI fragment. Mutagenesis primers 5′-GCTGCTGTGCCCCTGGCGGACATCCTC-3′ and 5′-GAGGATGTCCGCCAGGGGCACAGCAGC-3′ (changed bases are underlined) were used to introduce the disease-causing mutation L101P into pcDNA3.1/GFP-CLN3 using QuikChange® II XL Site-Directed Mutagenesis Kit (Stratagene).
For localization studies, constructs were nucleofected into fibroblasts using 5 mg of DNA per 1 × 105 cells. Acute overexpression in HeLa cells and fibroblasts was achieved by microinjecting DNA at a concentration of 100 ng/µl together with biotin-dextran (5 µg/µl) into about 50 cells grown on cover slips. It is possible, since each CLN3p protein was tagged with GFP at its N-terminus to allow us to confirm its intracellular expression and trafficking, that this tag differentially interfered with wild-type, mutant or truncated proteins.
CLN3 gene silencing
HeLa cells, and fibroblasts derived from a patient with JNCL and homozygous for the 1 kb common deletion or from a healthy control, were maintained in modified Eagle's medium alpha or Dulbecco’s modified Eagle's medium, respectively, supplemented with 10% fetal calf serum at 37°C with 5% CO2. For gene silencing in fibroblasts 1 × 105 cells were transiently transfected with 100 pmol of specific siRNAs using nucleofection and an NHDF nucleofection kit with programme U-23 (Amaxa, Cologne, Germany). Cells were grown on cover slips for 2 days. About 50 cells/cover slip were then microinjected with 0.1 µg/µl of the same siRNA together with biotin-dextran to visualize injected cells (data not shown). Cells were again left to grow for 2 days and then used for immunofluorescence labelling. For gene silencing in HeLa cells, cells were grown on cover slips and about 50 cells were microinjected with siRNA and left to grow for 2 more days before being used for immunoflourescence. Small interfering RNAs were from QIAGEN (Crawley, UK) and designed to target the following specific mRNA sequences: CLN3-2 (GCUGGUACCAGAUGCUGUA), CLN3-3 (GUGGUCUUCGCUAGCAUCU) and CLN3-4 (GCCGUGAUCUCCUGGUGGU), where CLN3-3 and CLN3-4 are directed against sequences in exons 7 and 8 of the gene. To verify the efficiency of the different siRNAs, 2 × 105 cells were nucleofected twice at 2-day intervals with 100 pmol of siRNA as above, harvested and the amount of CLN3 mRNA determined. RNA was extracted using QIAshredder and the QIAGEN RNAeasy kit. RNA was reverse translated to cDNA with the SuperScript III first-strand cDNA synthesis system (Invitrogen) using random hexamer primers. To assess the amount of CLN3 mRNA in the different samples, quantitative real-time PCR was performed in a DNA Engine Opticon 2 system (MJ Research) using DyNAmo SYBR green qPCR kit (Finnzymes) and CLN3 specific primers (Hs_CLN3_1_SG QuantiTect Primers from QIAGEN). The C(T) data were normalized with a GAPDH internal control and analysed using the 2–C(T) method (35). Primers for GAPDH amplification were 5′-CAGCCTCAAGATCATCAGCA-3′ and 5′-GTCTTCTGGGTGGCAGTGAT-3′. In HeLa cells, the amount of CLN3 mRNA remaining compared with mock-treated cells for each siRNA was: 13% for CLN3-2, 8% for CLN3-3 and 17% for CLN3-4.
CLN3 transcript analysis
Cells from JNCL patients and healthy controls were grown in 3.5 cm dishes and harvested when confluent. cDNA was obtained as described above. In a first round of PCR amplification, primers in the 5′ and 3′ region of the CLN3 mRNA were used (CLN3-5′ 5′-CCCTGGACCCTCGGGGGACC-3′ and CLN3-3′ 5′-CCTGCGTCCTGAGGATCCC-3′). The amplified product was used as a template for subsequent PCR reactions. The sequence of products obtained by using primers in exon 4 (6172 5′-CTGTCTCTACGGCTGCTGTG-3′) and exon 13 (R11 5′-AGGTGAAACGGATGCGAC-3′) of the gene was confirmed by MWG-Biotech (Ebersberg, Germany) using the R11 primer. PCR products were separated by electrophoresis in a 1% agarose gel in TAE.
This work was supported by the Medical Research Council, UK (to D.F.C., studentship to R.L.H., core support to the LMCB), the European Commission (503051 to S.E.M.); the Wellcome Trust, UK (067991 to S.E.M.); Batten Disease Family Association, UK (to S.E.M.).
We thank Professor Martin Raff for helpful discussion. We also thank the many families and clinicians who have contributed to our collection of NCL cell lines, particularly Professor A. Kohlschütter for the JNCL cell lines.
Conflict of Interest statement. The corresponding author confirms that neither she nor any of the co-authors have any conflicts of interest to declare.
- eukaryotic cell
- lysosomal storage diseases
- neurodegenerative disorders
- neuronal ceroid-lipofuscinoses
- rna interference
- protein overexpression
- juvenile neuronal ceroid lipfuscinosis
- symptom onset