Characterization of the canine CLCN3 gene and evaluation as candidate for late-onset NCL

Background The neuronal ceroid lipofuscinoses (NCL) are a heterogenous group of inherited progressive neurodegenerative diseases in different mammalian species. Tibetan Terrier and Polish Owczarek Nizinny (PON) dogs show rare late-onset NCL variants with autosomal recessive inheritance, which can not be explained by mutations of known human NCL genes. These dog breeds represent animal models for human late-onset NCL. In mice the chloride channel 3 gene (Clcn3) encoding an intracellular chloride channel was described to cause a phenotype similar to NCL. Results Two full-length cDNA splice variants of the canine CLCN3 gene are reported. The current canine whole genome sequence assembly was used for gene structure analyses and revealed 13 coding CLCN3 exons in 52 kb of genomic sequence. Sequence analysis of the coding exons and flanking intron regions of CLCN3 using six NCL-affected Tibetan terrier dogs and an NCL-affected Polish Owczarek Nizinny (PON) dog, as well as eight healthy Tibetan terrier dogs revealed 13 SNPs. No consistent CLCN3 haplotype was associated with NCL. Conclusion For the examined animals we excluded the complete coding region and adjacent intronic regions of canine CLCN3 to harbor disease-causing mutations. Therefore it seems to be unlikely that a mutation in this gene is responsible for the late-onset NCL phenotype in these two dog breeds.


Background
Neuronal ceroid lipofuscinoses (NCL) represents a group of heritable neurodegenerative storage diseases in man, mice, and several domestic animals like cattle, sheep, goat, cat, and certain dog breeds [1]. NCL diseases are characterized by the accumulation of autofluorescent cytoplasmic storage bodies in cells of the brain and retina. NCL diseases cause neurological symptoms that progress relentlessly and culminate in a vegetative state in humans and premature death [2]. Canine late-onset NCL variants primarily affect Tibetan Terrier and Polish Owczarek Nizinny (PON) dogs. A monogenic autosomal recessive mode of transmission was suggested for those breeds [3,4]. NCL-affected dogs represent valuable animal models to study human late-onset NCL variants since human families segregating for adult NCL are infrequent. Human NCL is a genetically heterogeneous disease with six identified disease genes (PPT1, TPP1, CLN3, CLN5, CLN6 and CLN8) [5]. Causal mutations within the canine orthologs of the six known human NCL genes have not been identified in NCL-affected Tibetan Terrier and PON dogs [6][7][8][9]. Single point mutations in the coding regions of the canine CLN8 and CLN5 genes were found in affected English Setter and Border collie dogs, respectively, showing juvenile NCL [6,10]. There are still undiscovered loci causing NCL beside the six known human genes, as indicated by findings in NCL-affected domestic and laboratory animals. In White Swedish Landrace sheep a CTSD mutation was reported and a mutation within the ortholog canine CTSD gene was detected in NCL-affected American Bulldogs [11]. Recently, CTSD was excluded as candidate gene in NCL-affected Tibetan Terrier and PON dogs [12]. In mice the chloride channel 3 gene (Clcn3) encoding an intracellular chloride channel was described to cause a phenotype similar to NCL [13]. Clcn3-deficient mice are characterized by developmental retardation and higher mortality combined with neurological manifestations such as blindness, motor coordination deficit, and spontaneous hyperlocomotion similar to human and canine NCL. To evaluate whether the CLCN3 gene is involved in the NCLaffected Tibetan Terrier and PON dogs, we determined the full-length cDNA sequence, characterized the gene struc-ture, and analyzed the coding sequence of the canine ortholog.

Results and discussion
Sequence analysis RT-PCR from canine lung mRNA amplified two splice variants, which were verified by direct DNA-sequencing of the RT-PCR products. Similar to the human CLCN3 sequence the alternative usage of exon 12 produces the shorter CLCN3 and the longer CLCN3 isoform e, respectively ( Figure 1). Overlapping canine cDNA fragments containing all junctions between the exons were generated by RT-PCR, sequenced, and used for comparison with the genomic sequence. These analyses indicated that the canine CLCN3 gene consists of 13 exons separated by twelve introns. The canine CLCN3 gene spans 52 kb (Figure 1) compared to 14 exons over 100 kb in human CLCN3 (NCBI build 35.1)) because in dog no untranslated 5'-exon is used. All splice donor/splice acceptor sites conform to the GT/AG rule. The experimentally verified existence of the two alternative splice variants is in agreement with the initially identified canine 5'-EST sequences (

Mutation analysis
A total of 13 sequence polymorphisms (Table 1) were found in the examined dogs as compared to the CLCN3 reference sequence of the current dog genome assembly (boxer genome assembly 1.1). The codon in exon 2 with the A/G transition codes for glycine in both SNP variants and the codon in exon 9 with the A/G transition codes for proline in both SNP variants. None of the 11 intronic polymorphisms did affect splice sites in the CLCN3 gene. For the single PON dog there were seven polymorphisms compared to the Boxer reference sequence (Table 1).
Assuming linkage disequilibrium between the polymorphic loci six different haplotypes could be constructed for the SNP genotypes in the Tibetan Terrier dogs (Table 1). Four out of six haplotypes occurred in both, affected and unaffected dogs, respectively ( Table 2). Only the haplotypes 3 and 4 occurred in a single heterozygous NCLaffected Tibetan Terrier dog ( Table 2). Due to the assumption of a single recessive founder mutation within this breed we expect homozygosity in affected individuals. Table 2 summarizes the haplotype distribution among the NCL-affected and the clinical unsuspicious dogs. The chisquare statistic for testing these haplotypes for association with disease status in the Tibetan Terrier dogs was calculated as 5.5786 with degree of freedom 1, which had a pvalue of 0.80, indicating no significant association.

Conclusion
The presented data indicate that the detected polymorphisms in the coding and adjacent intronic regions of canine CLCN3 can be excluded as disease harboring mutations in the examined dogs. Therefore it seems to be likely that the entire CLCN3 can be excluded as a candidate gene for the late-onset NCL phenotype in Tibetan Terrier and PON dogs. As the candidate gene approach did not reveal the causative gene in Tibetan Terrier and PON dogs it might be indicated to perform a genome wide linkage scan using NCL segregating families to map the canine chromosome region harboring the deleterious gene.

Sequence analysis
The human reference CLCN3 mRNA [Gen-Bank:NM_001829] was used as query in cross-species BLAST searches against the dog genome assembly (Boxer genome assembly 1.1). A single canine genomic contig of 577,638 bp was isolated [GenBank:AAEX01020012]. The human mRNA sequence was used to identify putative  T  T  T  T  T  T  T  C  e x o n 2 Numbering refers to the position of the polymorphic nucleotide within the given exon or intron respectively. 2 Nucleotide refers to the publicly available dog genome sequence (AAEX01020012) from the boxer named Tasha. Briefly, in RLM-RACE uncapped RNAs were dephosphorylated before the cap of full-length messenger RNAs (mRNAs) was removed enzymatically. After this step an RNA oligonucleotide adaptor was ligated to the 5'-end of the decapped mRNAs. As only full-length RNAs carried a 5'-phosphate group, the adaptor was expected to ligate exclusively to full-length mRNAs, while the dephosphorylated other RNAs were not able to undergo a ligation reaction. RT-PCR using two pairs of nested gene-specific (Table 3) and adaptor-specific primer pairs (Ambion) were then used to amplify the complete 5'-end of the CLCN3 cDNA according to the instructions of the manufacturer. Similarly, the 3'-end was amplified using two pairs of nested gene-specific and 3'-adaptor-specific primers. 5'-and 3'-RACE products and an additional 1885 bp RT-PCR product using sense and antisense primers from exon 1 and 10 (   respectively. The exact canine genomic structure was determined using the mRNA-to-genomic alignment program Spidey [14].

Mutation analysis
Genomic DNA was isolated from a single NCL-affected PON dog, six unrelated NCL-affected Tibetan Terrier dogs, and eight unrelated clinical unsuspicious Tibetan Terrier dogs (> 8 years old). Clinical neurologic, behavioral, and ophthalmologic evaluations were performed on each dog by a single external consultant veterinarian [4,15]. The phenotypes of the affected animals have been confirmed by detection of autofluorescent cytoplasmic inclusions within neurons throughout the retina and brain after necropsy.

Authors' contributions
AW did the mutation screen and drafted parts of the manuscript. OD proposed the idea and was responsible for funding. CD performed the RACE experiments, analyzed the sequence data, and performed manuscript editing.