Association of Mutations in a Lysosomal Protein with Classical Late-Infantile Neuronal Ceroid Lipofuscinosis

See allHide authors and affiliations

Science  19 Sep 1997:
Vol. 277, Issue 5333, pp. 1802-1805
DOI: 10.1126/science.277.5333.1802


Classical late-infantile neuronal ceroid lipofuscinosis (LINCL) is a fatal neurodegenerative disease whose defective gene has remained elusive. A molecular basis for LINCL was determined with an approach applicable to other lysosomal storage diseases. When the mannose 6-phosphate modification of newly synthesized lysosomal enzymes was used as an affinity marker, a single protein was identified that is absent in LINCL. Sequence comparisons suggest that this protein is a pepstatin-insensitive lysosomal peptidase, and a corresponding enzymatic activity was deficient in LINCL autopsy specimens. Mutations in the gene encoding this protein were identified in LINCL patients but not in normal controls.

The neuronal ceroid lipofuscinoses (NCLs) are a group of related hereditary neurodegenerative disorders that occur at a frequency of between 2 and 4 in 100,000 live births (1, 2). Most forms of NCL afflict children, and their early symptoms and disease progression tend to be similar. Diagnosis is frequently based on visual problems, behavioral changes, and seizures. Progression is reflected by a decline in mental abilities, increasingly severe and untreatable seizures, blindness, and loss of motor skills. There is no effective treatment for NCL, and all childhood forms are eventually fatal. Several forms of NCL are differentiated according to age of onset, pathology, and genetic linkage. These are infantile NCL (INCL,CLN1), classical late-infantile NCL (LINCL,CLN2), juvenile NCL (JNCL, CLN3), adult NCL (CLN4), two variant forms of LINCL (CLN5 andCLN6), and possibly other atypical forms (1,3). The defective genes in INCL and JNCL have recently been identified by positional cloning (4, 5), but the molecular basis for LINCL has remained elusive although the defect has recently been mapped to chromosome 11p15 (3). There are reasons, however, to suspect that the CLN2 gene product could have a lysosomal function. First, LINCL, like other forms of NCL, is characterized by an accumulation of autofluorescent lysosome-like storage bodies in the neurons and other cells of patients. Second, several other related disorders are caused by lysosomal deficiencies, for example, palmitoyl protein thioesterase in INCL, neuraminidase in sialidosis, and β-hexosaminidase A in Tay-Sachs disease. Third, continuous infusion of lysosomal protease inhibitors into the brains of rats induces an accumulation of ceroid-lipofuscin in neurons that resembles NCL (6, 7).

We have identified a protein that is deficient in LINCL by means of a biochemical approach that relies on the fact that newly synthesized soluble lysosomal enzymes contain a modified carbohydrate, mannose 6-phosphate (man-6-P). Man-6-P functions as a targeting signal in vivo as it is recognized by man-6-P receptors (MPRs), which direct the intracellular vesicular targeting of newly synthesized lysosomal enzymes from the Golgi to a prelysosomal compartment (8). Purified cation-independent MPR can be used as an affinity reagent for the detection of immobilized man-6-P glycoproteins in a protein immunoblot-style assay or can be coupled as an affinity chromatography reagent for the purification of man-6-P glycoproteins (9-11).

If LINCL results from the absence or deficiency of a lysosomal enzyme, then its corresponding man-6–phosphorylated form should also be absent or decreased. To test this possibility, we fractionated detergent-soluble extracts of LINCL patient and normal control brain autopsy samples by two-dimensional gel electrophoresis, transferred them to nitrocellulose, and detected man-6-P glycoproteins with an iodinated fragment of the MPR (Fig. 1). Normal brain contains ∼75 distinct spots representing multiple isoforms of various man-6-P–containing glycoproteins (Fig. 1, top). LINCL brain is similar, except that one prominent spot is absent (Fig. 1, bottom). The corresponding normal spot is ∼46 kD and has an isoelectric point (pI) of ∼pH 6.0. Three other LINCL specimens were examined with the consistent observation that this protein was missing.

Figure 1

A protein deficient in LINCL. Detergent-solubilized extracts of gray matter (50 μg of protein) from normal (top) or LINCL (bottom) brain autopsy specimens were fractionated by isoelectric focusing and SDS-PAGE, transferred to nitrocellulose, and man-6-P glycoproteins detected with 125I-labeled MPR. The man-6-P glycoprotein that is absent in LINCL extracts is indicated with an arrow.

To identify this potential candidate for CLN2, we purified man-6-P–containing glycoproteins from normal brain by affinity chromatography on a column of immobilized MPR and, after fractionation by SDS–polyacrylamide gel electrophoresis (PAGE) and transfer to a polyvinylidene difluoride membrane, the band that was absent in the LINCL specimens was isolated and sequenced. The NH2-terminal sequence was compared with predicted translation products from the expressed sequence tag (EST) database (dbEST) with the tBLASTN program. The initial search detected a murine clone encoding a sequence identical to the peptide in 16 of 20 positions, and later releases of dbEST contained human clones identical to the peptide in 19 of 20 positions. By iterative database searching and sequencing select clones, a nearly full-length sequence for the human CLN2 candidate was assembled (12). The 5′ end of the human cDNA was obtained from a human cortex cDNA library (13). This composite sequence (14) was confirmed from a genomic clone and genomic DNA from LINCL patients and controls.

The location of polyadenylate tracts on various human EST cDNA clones indicates that there are two transcripts, with the polyadenylate tail starting after nucleotide (nt) 2503 for the short transcript and nt 3487 for the long transcript. This was confirmed by RNA blot analysis, which revealed two transcripts of ∼2700 and 3700 nt (Fig.2). The mRNA was detected in all tissues examined with highest levels in heart and placenta and relatively similar levels in other tissues.

Figure 2

CLN2 expression in various human tissues. An RNA blot of polyadenylated [poly(A)+] human RNA (Clonetech, Palo Alto, California) containing 2 μg of poly(A)+ RNA was probed with the 32P-labeled insert of EST37588. Hybridization with two transcripts of ∼2.7 and 3.7 kb is apparent in all tissues. The mRNA was also detected in spleen, thymus, prostate, testis, ovary, small intestine, colon, and peripheral blood leukocytes (16). The ubiquitous distribution of this mRNA as indicated by RNA blotting is confirmed by the existence of highly related clones in many different cDNA libraries as found by database searches. S. muscle, skeletal muscle.

The CLN2 transcript long open reading frame encodes a 563-residue protein, predicted to contain a 16-residue signal sequence (Fig.3). There are no methionines between the putative initiation codon and the start of the sequence determined by Edman degradation at residue 195, indicating that the CLN2 precursor contains a long pro-region or consists of an NH2-terminal light and a COOH-terminal heavy chain. Because all five potential glycosylation sites reside COOH-terminal to the cleavage site, if a light chain were present in the mature protein, it would not have been detected by the man-6-P glycoprotein assay.

Figure 3

Amino acid sequence of CLN2. Aligned sequences of the human CLN2 protein, Pseudomonassp. 101 pepstatin-insensitive carboxyl proteinase (PsCP), andXanthomonas sp. T-22 pepstatin-insensitive carboxyl proteinase (XaCP). Shading indicates regions of amino acid conservation: heavy shading indicates identical amino acids and light shading indicates similar amino acids. Predicted and known peptide cleavage sites are indicated by unfilled and filled arrows, respectively, and are labeled C, X, and P for CLN2, XaCP, and PsCP. Asterisks indicate amino acids that are mutated in LINCL patients, and the horizontal bar indicates the chemically determined peptide sequence of the 46-kD mature form heavy chain. XaCP has a 192–amino acid COOH-terminal extension that is proteolytically removed. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

The predicted physical properties of the protein are in accordance with the observed properties of the protein that is missing in LINCL patients [apparent relative molecular mass (Mr) of 46,000 and pI of 6.0]. The calculated Mr of the mature protein heavy chain is 39,700. Assuming all glycosylation sites are utilized and an average Mr of 1800 for each oligosaccharide, the total Mr would be ∼48,000. The calculated pI is 6.13 without considering posttranslational modifications (for example, man-6-P residues) that would shift the pI toward the acidic range.

The absence of this 46-kD lysosomal protein in LINCL patients makes it a likely candidate for CLN2. Strong support for this conclusion comes from the observation that the gene identified here maps to chromosome 11p15 (15), which is also the locus identified forCLN2 by genetic linkage analysis (3). Direct evidence for the identification of CLN2 came from sequence analysis of DNA from LINCL patients and unaffected family members (Table 1). The gene structure (16) of the CLN2 candidate was determined, thus allowing analysis of both intronic and exonic sequences from LINCL patient DNA with genomic DNA prepared from cell lines (17). Mutations were observed in two of the polymerase chain reaction (PCR) segments generated from the DNA of LINCL patients. Two unrelated LINCL patients contained mutations within the codon (TGT) encoding Cys365. In one case, a monoallelic transversion of T to C resulted in a Cys to Arg substitution; presumably, the defect in this patient is compound heterozygous and there is therefore an additional, as yet unidentified, mutation. Evidence that this substitution represents a deleterious mutation rather than a polymorphism is provided by the observation that another patient contains a different mutation in the same codon. In this case, a homozygous G to A transversion resulted in a Cys to Tyr substitution in the protein expressed from both alleles. Should this Cys prove to be involved in disulfide bonding, mutations are likely to be highly disruptive given the role of disulfide bonds in establishing and maintaining protein structure. Different compound heterozygous mutations were found in two affected siblings. A heterozygous C to T transversion resulted in the conversion of the codon (CGA) for Arg208 to an umber (TGA) stop codon. In the other allele, the conserved AG of the intronic 3′ splice junction sequence is mutated to AC, which is likely to result in incorrect splicing of the CLN2 candidate mRNA. Each parent possessed a single different mutant allele and an unaffected sibling had only the premature stop mutation in one allele, indicating conventional Mendelian inheritance of these mutations. None of these mutations was observed in the genomic clone, genomic DNA from four controls (one normal subject and three patients with unrelated diseases), or in any of the EST sequences that overlap these sites.

Table 1

Genotype analysis of LINCL patients.

View this table:

Sequence comparisons revealed significant similarities (18) between CLN2 with carboxyl peptidases fromPseudomonas (PsCP) (19) andXanthomonas (XaCP) (20). Multiple alignments between the CLN2 protein and the two bacterial proteases revealed significant blocks of sequence similarities and that both PsCP and XaCP have long propieces, with mature NH2termini located proximal to the known NH2-terminus of the mature heavy-chain CLN2 (Fig. 3). PsCP and XaCP are highly unusual carboxyl proteinases that are not inhibited by pepstatin, the classical inhibitor of pepsin, cathepsin D, and other aspartyl proteases.

Normal and LINCL brain extracts contained similar pepstatin-sensitive (cathepsin D) activities (Fig.4, right). In contrast, whereas normal brain contains an acid protease activity not inhibited by pepstatin and E-64, this activity was essentially absent from LINCL brains (Fig. 4, left). Pepstatin-insensitive carboxyl proteases have not, to date, been reported in mammals and would thus have been overlooked in earlier studies of lysosomal activities in LINCL patients. One characteristic of LINCL is the storage of mitochondrial ATP synthase subunit c in the lysosomes of patients (21-23), which may therefore indicate that subunit c represents a substrate for CLN2. Also, although the prominent neurological component of LINCL may be due to the susceptibility of neurons to metabolic insults, one intriguing possibility is that CLN2 is involved in processing of neuron-specific trophic factors.

Figure 4

Pepstatin-sensitive and -insensitive protease activities in extracts of normal and LINCL brain samples. Samples were homogenized in 50 volumes (w/v) of 0.15 M NaCl, 0.1% Triton X-100 and centrifuged at 14,000g for 25 min. Pepstatin-insensitive activity in the supernatant was measured with 1% bovine hemoglobin as a substrate in 25 mM formate buffer containing 2 μM pepstatin, 0.1 mM E-64, 0.15 M NaCl, and 0.1% Triton X-100 (pH 3.5). The trichloroacetic acid–soluble degradation products were quantitated with fluorescamine (24) in borate buffer (pH 8.6). As a control, in the absence of pepstatin, cathepsin D activity was detected in LINCL and normal extracts.

Identification of CLN2 will not only aid in the prevention of LINCL through genetic counseling but may reveal strategies and provide test systems for therapeutic intervention. These findings also demonstrate the utility of a general approach for determining the molecular bases for lysosomal disorders of unknown etiology.


Stay Connected to Science

Navigate This Article