A CLN8 Nonsense Mutation in the Whole Genome Sequence of a Mixed Breed Dog with Neuronal Ceroid Lipofuscinosis and Australian Shepherd Ancestry

Molecular Genetics and Metabolism 112 (2014) 302–309

Contents lists available at ScienceDirect

Molecular Genetics and Metabolism

journal homepage: www.elsevier.com/locate/ymgme

A CLN8 nonsense mutation in the whole genome sequence of a mixed breed dog with neuronal ceroid lipofuscinosis and Australian Shepherd ancestry

Juyuan Guo a, Gary S. Johnson a, Holly A. Brown b, Michele L. Provencher c, Ronaldo C. da Costa c, Tendai Mhlanga-Mutangadura a, Jeremy F. Taylor d, Robert D. Schnabel d,DennisP.O'Briena, Martin L. Katz e,⁎ a Department of Veterinary Pathobiology, University of Missouri College of Veterinary Medicine, Columbia, MO, USA b Metz Petz Veterinary Clinic at Shawnee, Lima, OH, USA c Department of Veterinary Clinical Sciences, The Ohio State University College of Veterinary Medicine, Columbus, OH, USA d Division of Animal Science, University of Missouri College of Agriculture, Food and Natural Resources, Columbia, MO, USA e Mason Eye Institute, University of Missouri School of Medicine, Columbia, MO, USA article info abstract

Article history: The neuronal ceroid lipofuscinoses (NCLs) are hereditary neurodegenerative diseases characterized by seizures Received 1 May 2014 and progressive cognitive decline, motor impairment, and vision loss accompanied by accumulation of auto- Received in revised form 28 May 2014 ﬂuorescent lysosomal storage bodies in the central nervous system and elsewhere in the body. Mutations in at Accepted 28 May 2014 least 14 genes underlie the various forms of NCL. One of these genes, CLN8, encodes an intrinsic membrane Available online 4 June 2014 protein of unknown function that appears to be localized primarily to the endoplasmic reticulum. Most CLN8 mu-

Keywords: tations in people result in a form of NCL with a late infantile onset and relatively rapid progression. A mixed breed Batten disease dog with Australian Shepherd and Blue Heeler ancestry developed neurological signs characteristic of NCL Dog model starting at about 8 months of age. The signs became progressively worse and the dog was euthanized at CLN8 mutation 21 months of age due to seizures of increasing frequency and severity. Postmortem examination of the brain Lysosomal storage disease and retinas identified massive accumulations of intracellular autofluorescent inclusions characteristic of the Retina NCLs. Whole genome sequencing of DNA from this dog identified a CLN8:c.585GNA transition that predicts a Brain CLN8:p.Trp195* nonsense mutation. This mutation appears to be rare in both ancestral breeds. All of our 133 archived DNA samples from Blue Heelers, and 1481 of our 1488 archived Australian Shepherd DNA samples tested homozygous for the reference CLN8:c.585G allele. Four of the Australian Shepherd samples tested heterozygous and 3 tested homozygous for the mutant CLN8:c.585A allele. All 3 dogs homozygous for the A allele exhibited clinical signs of NCL and in 2 of them NCL was confirmed by postmortem evaluation of brain tissue. The occurrence of confirmed NCL in 3 of 4 CLN8:c.585A homozygous dogs, plus the occurrence of clinical signs consistent with NCL in the fourth homozygote strongly suggests that this rare truncating mutation causes NCL. Identification of this NCL-causing mutation provides the opportunity for identifying dogs that can be used to establish a canine model for the CLN8 disease (also known as late infantile variant or late infantile CLN8 disease). © 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction severity [1,2]. The majority of the NCLs have childhood onsets and are ultimately fatal. The genetic bases of most cases of NCL have been iden- The neuronal ceroid lipofuscinoses (NCLs) are inherited progressive tified and include a variety of mutations in at least 14 different genes [3]. neurodegenerative disorders characterized by accumulations of auto- Some of these genes encode soluble lysosomal enzymes of known func- fluorescent lysosomal storage material in the central nervous system tion, but the functions and even the subcellular localization of other NCL as well as in many other organs and tissues [1,2]. Among the symptoms proteins are unknown or are poorly understood. of NCL are progressive cognitive decline and loss of motor functions, sei- In addition to occurring in people, naturally occurring NCLs have zures, and vision loss. The NCLs vary in age at onset of neurological signs been identified in many other species including cattle, sheep, cats and from infancy to adulthood, as well as in rates of disease progression and dogs [1,2]. Canine genes harboring NCL-causing mutations have been identified in American Bulldogs (CTSD), Australian Shepherds (CLN6), Border Collies (CLN5), English Setters (CLN8), American Staffordshire ⁎ Corresponding author at: Mason Eye Institute, School of Medicine, University of Missouri, One Hospital Dr., Columbia, MO 65212, USA. Fax: +1 573 884 4100. Terriers (ARSG), Tibetan Terriers (ATP13A2), and Dachshunds (both E-mail address: [email protected] (M.L. Katz). TPP1 and PPT1) [4–10]. Mutations in the human orthologs of all these

http://dx.doi.org/10.1016/j.ymgme.2014.05.014 1096-7192/© 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 303 genes except ARSG are known to cause NCL in people. The canine disor- therapeutic range, the dog's anxiety continued to worsen and by ders can serve as NCL models for studying disease pathogenesis and for 51 days post-presentation the dog was exhibiting progressively more evaluating therapeutic interventions. For example, evaluation of en- frequent and severe seizure activity that was not ameliorated by zyme replacement therapy in Dachshunds with a TPP1 null mutation Diazepam administration. As a result of the intractable seizures, the laid the groundwork for a human clinical trial to determine the efﬁcacy dog was euthanized at this time. of this approach in children with NCL resulting from mutations in the The affected dog was conceived from a breeding between two litter- orthologous gene [5]. mates (Fig. 1) one of whose parents was an Australian Shepherd and the Because of the potential value of canine models for developing treat- other of which was a Blue Heeler (breed also referred to as Australian ments for the NCLs, we undertake studies to determine whether dogs Cattle Dog). The owners of both the grandparents and parents of the af- exhibiting clinical signs that may be indicative of NCL harbor mutations fected dog were unwilling to provide DNA samples from their dogs and in any of the known NCL genes. One such dog of mixed Australian we were unable to locate littermates of the affected dog. Shepherd and Blue Heeler ancestry was recently reported to us. Studies were undertaken to conﬁrm that this dog suffered from NCL and to 2.2. Magnetic resonance imaging identify the causative mutation. The patient's brain was imaged with a 3 T Philips magnetic reso- 2. Materials and methods nance imaging instrument (Achieva, Cleveland OH). Several sequences

2.1. Subject dog — clinical signs and ancestry

A 20 month old 26.3 kg spayed female dog with mixed Australian Shepherd/Blue Heeler ancestry presented to Metz Petz Veterinary Clinic after a 3 week history of severe vision loss culminating in blindness. The owners noted that since puppyhood the dog had exhibited signs of visual impairment characterized by difficulty in catching and locating tossed treats and bumping into objects, even under bright light conditions. The owners reported that over a period of 10 to 12 months prior to presentation the affected dog exhibited a loss of housetraining, decreased responsiveness to voice commands, and increased sensitivity to noises. She also exhibited compulsive circling and pacing behavior and a loss of the ability to navigate stairs. The dog became progressively more ataxic and exhibited periods of trance-like staring behavior. At presentation, the patient was very anxious. She lacked a menace response in both eyes. Pupillary light responses, both direct and consensual, were normal in both eyes, and dazzle reflexes were positive in both eyes. The rest of her cranial nerve and physical examination were within normal limits. The pet was referred to the Ohio State University (OSU) Veterinary Medical Center. While being transported the patient began exhibiting signs of focal seizures with fly biting behavior, pawing at her face, shak- ing her head, falling, and biting the air near her flank region. Ophthalmic examination revealed no retinal abnormalities upon fundic examination and normal tear film production and intraocular pressures. Electroretinogram responses were within normal limits. Within 24 h of presentation, the patient became very aggressive with progressive seizure activity and high pitched barking. She presented to an emergency clinic overnight and was started on Phenobarbital IV. The patient continued to exhibit seizure activity laying in lateral recum- bency and paddling the following day. She presented to the OSU Neurology Service 48 h after the initial clinical presentation for examination, magnetic resonance imaging (MRI) and spinal cerebrospinal fluid collection and analysis. A complete neurologic examination revealed inappropriately anxious behavior, with absent menace responses, and decreased proprioceptive position- ing bilaterally in both pelvic limbs. The remainder of the neurologic examination was normal. The complete blood chemistry analysis was unremarkable with only mild alanine aminotransferase elevation. No abnormalities were observed in the analysis of the cerebrospinal fluid. MRI revealed diffuse brain atrophy but no localized lesions. To amelio- rate seizure activity, treatment with Phenobarbital was continued. She did well at home initially with no seizures but was dragging her feet more, stumbling, and had difficulty keeping her balance to urinate and defecate. The patient was polyuric, polydipsic, with an increased appetite — known effects of phenobarbital administration. At 27 days after initial presentation, the dog was again brought in for evaluation due to increased anxiety, decreased responsiveness to sound stimuli, Fig. 1. (A) Littermates produced by breeding between an Australian Shepherd and a Blue and severe behavioral changes including aggression, disorientation, Heeler. A breeding between the dogs shown in (A) produced the dog shown in (B) that and ataxia. Despite blood levels of phenobarbital being within the was characterized in this study. 304 J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 were obtained including T2- and T1-weighted images (pre- and post- (Sigma, St. Louis, MO) at a 1:1000 dilution for 60 min in place of the contrast administrations) in the sagittal, transverse and dorsal planes. primary antibody. The detection system was a rabbit anti-goat horse Additionally FLAIR (fluid attenuated inversion recovery) and gradient radish peroxidase (HRP) kit (PK6105; Biocare Medical) with 30-min echo sequences were obtained in the transverse plane. Slice thickness incubations in link and label steps followed by a Tris-buffered rinse. was 3 mm for all sequences with no interslice interval. The sections were then incubated in romulin red (Biocare Medical) for 10 min. Slides were counterstained in hematoxylin (Biocare Medical) 2.3. Histopathology and electron microscopy at a 1:10 dilution for 5 min followed by a Tris-buffered rinse, dehy- drated, coverslipped, and then imaged with light microscopy. The affected dog was euthanized at approximately 20 months of age due to intractable seizures. The eyes were enucleated and the corneas 2.4. Molecular genetic analyses were removed before placing one of the globes in EM fixative (2% glutarladehyde, 1.12% paraformaldehyde, 130 mM sodium cacodylate, A DNA sample from the affected dog was used with the Illumina

1mMCaCl2, pH 7.4) and the other in immuno ﬁxative (0.05% glutaral- TruSeq DNA PCR-Free Sample Preparation Kit to prepare 2 paired-end dehyde, 3.5% paraformaldehyde, 120 mM sodium cacodylate, 1 mM libraries: one with a fragment size of approximately 350 bp and the

CaCl2, pH 7.4). The brain was then removed from the skull. Slices of other with a fragment size of approximately 550 bp. The 2 libraries the cerebral cortex parietal lobe and of cerebellum were each placed were sequenced with an Illumina HiSeq 2000 sequencer at the Univer- in vials of the same fixatives. The remainder of the brain was placed in sity of Missouri DNA Core Facility. Adapter sequences were trimmed 10% phosphate-buffered formalin (Fisher Scientific SF93-4, Fair Lawn, from the sequence reads with custom Perl scripts and the adapter- NJ). trimmed reads were error corrected using MaSuRCA v1.9.5 software After approximately 24 h at room temperature, the immuno-fixed [12]. NextGENe software (Soft Genetics) was used to align the error samples of cerebral cortex and cerebellum were transferred to 170 mM corrected reads to the CanFam3.1 reference genome assembly and to sodium cacodylate, pH 7.4, in which they were stored at 4 °C until further categorize sequence variants. Variant calls were further processed processing. At the same time, the irises, lenses and vitreous bodies were using custom Perl scripts to remove likely false positives. The remaining removed from the eyes, the remainders of which were then incubated sequence variants were uploaded to a custom PostgreSQL database for another 24 h in the fixatives. The immuno-fixed eyecup was then which contained the variant calls for an additional 101 canid sam- transferred to 170 mM sodium cacodylate, pH 7.4, and stored at 4 °C ples. These control whole genome sequences included: 43 genomes until further processing. The EM-fixed tissues were incubated at room from our group; 15 wild canid genomes provided by the University of temperature for at least 48 h and then stored in the fixative at 4 °C California, Los Angeles; 28 genomes from the Institute for Translational until further processing. After incubation at room temperature for ap- Genomic Research; 10 genomes provided by investigators at North proximately 48 h, the formalin-fixed portion of the brain was stored at Carolina State University; and, 5 genomes provided by investigators at 4 °C until further processing. the University of Pennsylvania. Custom SQL scripts were used to identify Slices of the immuno-fixed cerebral cortex, cerebellum, and retina variants that fit an autosomal recessively inherited rare disease model from the posterior pole of the immuno-fixed eye were processed, such that the case is homozygous for an allele not observed in either embedded and frozen for cryostat sectioning as previously described heterozygous or homozygous state in any of the similarly generated [11]. Cryostat sections of each of these tissues were cut at a thickness 101 canid whole genome sequences in our data set. The candidate of 8 μm, mounted on Superfrost Plus slides (Fisher Scientific, Fair variant list was further filtered to include only variants predicted to Lawn, NJ) in 170 mM sodium cacodylate. The sections were examined alter the primary structure of gene products. A detailed description of and photographed using fluorescence microscopy as previously de- the data processing pipeline is in preparation and will be published scribed [11]. elsewhere. After at least 48 h of incubation at room temperature, pieces of An apparent CLN8:c.585GNA sequence variant (GenBank accession the EM-fixed cerebral cortex, cerebellum, and retina from the posterior NM_001012343) was verified by PCR amplification with primers 5′- pole of the EM-fixed eye were post-fixed and embedded in epoxy resin. GGCAGTAAGTCTTCCTCAAAGTG-3′ and 5′-TGCTGACCAGACCGTCCCA-3′ Sections of the embedded tissue were cut on an ultramicrotome at and automated Sanger sequencing. A TaqMan allelic discrimination thicknesses of 0.5 to 0.8 μm, mounted on glass slides and stained with assay [13] was used to genotype DNA samples from individual dogs at toluidine blue. Areas of interest were identified by microscopic exami- CLN8:c.585GNA. For this assay, the PCR primer sequences were 5′- nation of these sections, and the blocks were trimmed to remove GGTCCGAGTCTCTGTTTTGGAA-3′ and 5′-CGGCAGTGGAACATGTGGAT-3′ tissue from outside the areas of interest. Sections were then obtained and the competing probes were 5′-VIC-TGAACCAGTGGCTGATG-MGB-3′ from the trimmed blocks at thicknesses of 70 to 90 nm. The latter sec- (reference allele) and 5′-FAM-TGAACCAGTGACTGATG-MGB-3′ (mutant tions were mounted on 200 mesh copper thin-barred grids, stained allele). with uranyl acetate and lead citrate, and were then examined and photographed using a JEOL 1400 transmission electron microscope. 3. Results Slices of the formalin-fixed cerebellum were embedded in paraffin. For comparison, similar slices of cerebellum from two one-year-old 3.1. Magnetic resonance imaging normal Beagles were collected, fixed in formalin, and embedded in paraffin. Four μm-thick sections of these tissues were mounted on T-2 weighted MR imaging of the brain revealed ventriculomegaly, as positively charged glass slides, deparaffinized, and immunostained well as abnormally prominent cortical sulci (Fig. 2). The folia of the cer- with an antibody directed against glial fibrillary acid protein (GFAP). ebellum were also abnormally prominent. No localized brain lesions For immunolabeling, the slide-mounted sections were steam treated were observed. in a decloaking chamber (Biocare Medical, Concord, CA) at 98 °C for 30 min in citrate target retrieval solution (Dako, Carpinteria, CA), cooled 3.2. Histopathology and electron microscopy at room temperature for 10 min, rinsed with distilled water, then placed on an IntelliPATH FLX autostainer for staining. Slides were treated with Adefining feature of the NCLs is the accumulation of autofluorescent

3% H2O2 for 15 min, washed in buffer, and treated with Sniper block lysosomal storage material in the central nervous system as well as in (Biocare Medical) for 20 min. They were then incubated in primary many nonneuronal tissues and organs [1,2]. Unless the causative muta- antibody (rabbit anti-GFAP, DakoCytomation, product # Nr.Z 0334) for tion is known and can be screened for, the only way to deﬁnitively diag- 60 min. Control sections were treated with nonimmune rabbit IgG nose NCL in a dog exhibiting signs of these disorders is to examine J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 305

neural tissues for the presence of this autfluorescent material. In the dog that was evaluated for this study, massive amounts of autofluorescent material with the fluorescence spectral characteristics of NCL-specific storage substance were observed in the cerebellum, cerebral cortex, and retina (Fig. 3). The storage bodies that accumulated in the brain and retina con- sisted of membrane-bounded intracellular inclusion bodies (Fig. 4). The most prominent feature of the contents of these inclusion bodies from all cell types examined were multilamellar structures composed of stacks of membrane-like rectilinear structures (Fig. 4). A minor component of the storage bodies in the retinal ganglion cells had a cur- vilinear appearance (Fig. 4C). The storage bodies along the retinal outer limiting membrane appeared to be present primarily in the Mueller cells. The NCLs are characterized by astrogliosis throughout the central nervous system as demonstrated by the presence of large numbers of astrocytes that exhibit immunohistochemical staining for GFAP [14–17]. GFAP immunohistochemical staining of the cerebellum from the affected dog revealed the presence of a dramatically larger number of GFAP-positive cells, primarily in the medulla, than were present in the same brain areas of two normal one-year-old Beagles that were examined (Fig. 5). In those cells that did exhibit GFAP immunostaining in the cerebellums of the normal dogs, the staining intensity was much lower than in cells from the same region of the cerebellum of the affected dog (Fig. 5).

3.3. CLN8 mutation identiﬁcation

DNA from the NCL-affected dog was used to generate a 20-fold Fig. 2. T2 weighted MRI in the sagittal (A), dorsal (B) and transverse (C) planes from the average coverage whole genome sequence which contained 6.1 million affected dog. Cerebrospinal ﬂuid (CSF) is white and brain parenchyma gray. Diffuse sequence variants (differences from the canine genome reference brain atrophy is indicated by increased CSF surrounding the folia of the cerebellum sequence). Of these, 83 sequence variants remained after ﬁltering to (A arrow), enlargement of the lateral ventricles (B & C asterisks), and widening of the retain only the variants that were homozygous in the target sequence, sulci of the cerebral cortex (B & C arrowheads). predicted to alter the amino acid sequence of the gene product, and absent from 101 other canid whole genome sequences (Table S1). Among

Fig. 3. Fluorescence micrographs of cryostat sections of the cerebellum (A), parietal lobe of the cerebral cortex (B), and the retina (C). Accumulation of autofluorescent storage material was abundant in all 3 tissues. In the cerebellum the Purkinje cells (PC) showed the most pronounced accumulation of storage material. In the retina autofluorescent storage material content was most abundant in the ganglion cells (GC), along the outer limiting membrane (OLM), and in the retinal pigment epithelium (RPE). Bar in (B) indicates magnification for panels (A) and (B). 306 J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309

Fig. 4. Electron micrographs of the storage material from a cerebellar Purkinje cell (A), a cerebral cortical neuron (B), a retinal ganglion cell (C), and a cell adjacent to the retinal outer limiting membrane (D). these 83 sequence variants, a CLN8:c.585GNA transition that predicts a accumulates in many postmitotic cells during normal aging, but the CLN8:p.Trp195* nonsense mutation (Fig. 6A) was considered most like- amount of this material in the brain and retina of the young affected ly to be causal because it was the only variant in a canine ortholog of a dog was dramatically higher than would have accumulated as a result gene associated with human NCL and because it is predicted to encode of aging alone. In the retina, lipofuscin normally accumulates only in a truncated protein missing the 93 C-terminal amino acids. The CLN8: the RPE; the presence of autofluorescent material in the ganglion and c.585GNA transition was verified by automated Sanger sequencing Mueller cells and elsewhere in the neural retina is diagnostic of NCL (Fig. 6B). [5,9,19,21]. In addition, the mulitlamellar rectilinear ultrastructural fea- Since the affected dog was a member of the F2 generation from tures of the storage material in the affected dog brain are similar to an Australian Shepherd crossed with a Blue Heeler, we genotyped ar- those of the storage materials that accumulate in the cells of human chived DNA samples from each breed to identify the most likely source subjects with some late infantile forms of NCL [22–25] and does not of the CLN8 nonsense mutation. All of our 133 DNA samples from Blue contain lipid-like components that are typical of lipofuscin [26].Thela- Heelers tested homozygous for the reference CLN8:c.585G allele. Of mellar appearance of material within the storage bodies is suggestive of our 1488 Australian Shepherd samples, 1481 tested homozygous for membrane fragments which may not be properly degraded in this the CLN8:c.585G allele, 4 were heterozygous, and 3 tested homozygous disease. for the mutant CLN8:c.585A allele. One of the CLN8:c.585A homozygous Because the canine NCLs are rare recessive diseases, we searched the DNA samples came from an Australian Shepherd with a history of sei- whole genome sequence from the affected dog for homozygous se- zures. No postmortem histopathological data are available from that quence variants that were predicted to alter the amino acid sequences dog. The other two DNA samples had been extracted from paraffinem- of the gene products but absent from the whole genome sequences of bedded formalin-fixed brain from NCL-affected littermates character- 101 control canids. Among the 83 sequence variants that met these ized in an earlier study [18]. criteria, the only one that involved an ortholog of a gene associated with human NCL was CLN8:c.585GNA, which predicts a nonsense muta- 4. Discussion tion, p.Trp195*. The function of CLN8, the protein encoded by CLN8,has not been clearly identified, although there is some evidence that CLN8 We investigated a mixed-breed dog with a progressive neurodegen- may be involved in ceramide biosynthesis [27] and vesicular and erative disease and a medical history that included blindness, seizures, membrane trafficking [28]. CLN8 appears to be an intrinsic membrane cognitive decline, behavioral changes, ataxia, and brain atrophy similar protein localized primarily to the endoplasmic reticulum (ER) and ER- to the disease signs seen in human and in other canine NCLs [1,2].In Golgi intermediate compartment [29]. It belongs to a family that in- addition to these signs, the presence of autofluorescent cytoplasmic cludes 15 other mammalian proteins encoded by CLN8 paralogs, each inclusions in unstained sections of the dog's brain and retina and containing a TRAM-LAG-CLN8 (TLC) domain comprised of 5 putative astrogliosis in the cerebellum supported a diagnosis of NCL [4,5,7,9,10, transmembrane helices [30,31].TheCLN8:c.585GNA transition predicts 19,20]. Storage material with similar fluorescence properties (lipofuscin) a CLN8:p.Trp195* nonsense mutation and a truncated protein product J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 307

affected dog would be benign and very likely that it is the molecular genetic cause of the dog's NCL. Our assertion that the homozygous CLN8:c.585A allele is a cause of canine NCL was further supported when we genotyped more than 1600 archived Australian Shepherd and Blue Heeler DNA samples at CLN8:c.585. Only 3 of these samples were from c.585A homozygous dogs. Little is known about the clinical history of one of these dogs except that it had recurrent seizures (a prominent clinical sign of the mixed breed dog). The other two CLN8:c.585A homozygous samples were extracted from formalin-fixed brains that were collected from Australian Shepherd littermates with well-documented NCL of unknown etiology [18]. The clinical histories and neuropathologic lesions of the Australian Shepherd littermates were almost identical to those of the mixed-breed dog reported here. Thus, 3 of the 4 dogs known to be homozygous for the rare CLN8 nonsense mutation also shared a rare clinical condition (and the fourth dog had a scant clinical history suggestive of this condition). We conclude that the CLN8 nonsense mutation is almost certainly the cause of the neurological disorder in these dogs. Many different CLN8 mutations have been identified in human NCL patients (http://www.ucl.ac.uk/ncl/CLN8mutationtable.htm) [32, 33]. These patients have shown either of two distinct disease pheno- types. A large cohort of Finnish patients with a CLN8:c.70CNG missense mutation exhibited an atypical juvenile-onset NCL sometimes called Northern epilepsy [34]. In these patients, the first seizures occur at 5 to 10 years of age [34,35]. A progressive psychomotor deterioration becomes apparent a few years later; however, disease progression is markedly slower in Northern epilepsy patients compared to patients with the other forms of childhood-onset NCL. Northern epilepsy patients have a near normal life expectancy and retain their eyesight [34,35].Morethan10differentCLN8 mutations have been found in the other disease phenotype, often referred to as late infantile CLN8 disease, or variant late infantile NCL (vLINCL) (http://www.ucl.ac.uk/ncl/ CLN8mutationtable.htm) [32,33]. In these patients, seizures may begin Fig. 5. GFAP immunostained sections of the cerebellar white matter (medulla) from a one- as early as 3 years of age. There is rapid psychomotor deterioration year-old normal Beagle (A) and from the dog with NCL (B). The density of GFAP-positive and visual impairment or loss [32,36,37]. The NCL of the CLN8:c.585A cells and the amount GFAP staining per cell was much higher in the affected dog than in the dog that did not suffer from NCL. homozygous dogs more closely resembles the vLINCL disease phenotype in that there is visual loss, rapid disease progression and a markedly decreased life expectancy. missing the 93 C-terminal amino acids. Truncated CLN8 is unlikely to be In our survey of CLN8:c.585GNA genotypes in archived DNA, all of the functional because it lacks 2 of the 5 predicted transmembrane helicies samples from Blue Heelers were homozygous for the reference CLN8: of the TLC domain. A potentially functional C-terminal endoplasmic c.585G allele, whereas 4 of the Australian Shepherd samples were retriculum retrieval signal [29] is also predicted to be missing from c.585A/G heterozygotes and 3 were homozygous for the mutant CLN8: the truncated CLN8. Even missense mutations both within and out- c.585A allele. This distribution of genotypes could have occurred be- side of the TLC domain of CLN8 result in late infantile onset NCL in peo- cause 92% of the samples in the survey were from Australian Shepherds. ple [32], so it is unlikely that the nonsense mutation observed in the While we cannot be certain about the source of the c.585A alleles in the

Fig. 6. (A) Sequence reads from the mixed-breed dog whole-genome sequence aligned to a segment of canine chromosome 37 showing the homozygous GNA transition in CLN8. (B) Au- tomated Sanger sequence confirms the GNA transition in CLN8 in DNA from the mixed-breed dog with NCL. 308 J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 mixed breed dog, we know that this allele has occurred in purebred [10] D.N. Sanders, F.H. Farias, G.S. Johnson, V. Chiang, J.R. Cook, D.P. O'Brien, S.L. Hofmann, J. Lu, M.L. Katz, A frame shift mutation in canine palmitoyl protein thiesterase 1 Australian Shepherds. The samples from the 7 Australian Shepherds (PPT1) causes early onset neuronal ceroid lipofuscinosis in a Dachshund, Mol. harboring 1 or 2 c.585A alleles were acquired over a span of 10 years Genet. Metab. 100 (2010) 349–356. and the 2 homozygous siblings had the only known family relationship [11] B. Vuillemenot, M.L. Katz, J.R. Coates, P. Lobel, P. Tiger, S. Bunting, S. Kanazono, D. Kennedy, L. Tsuruda, C. O'Neill, Intrathecal tripeptidyl peptidase-1 reduces lysosom- among these 7 dogs. Thus, the c.585A allele appears to be rare, but wide- al storage in a canine model of LINCL, Mol. Genet. Metab. 104 (2011) 325–337. spread among Australian Shepherds. Previously, English Setters were [12] A.V. Zimin, G. Marçais, D. Puiu, M. Roberts, S.L. Salzberg, J.A. Yorke, The MaSuRCA ge- found to develop NCL as a result of a CLN8:c.491CNT transition that nome assembler, Bioinformatics 29 (2013) 2669–2677. fl ′ predicts a CLN8:p.Leu164Pro amino acid substitution [7].Dogswith [13] K.J. Livak, Allelic discrimination using uorogenic probes and the 5 nuclease assay, Genetic Analysis 14 (1999) 143–149. this mutation were maintained for many years in research colonies [14] S.L. Macauley, M. Pekny, M.S. Sands, The role of attenuated astrocyte activation in in- [38–40]; however, it is doubtful that any descendants of the English fantile neuronal ceroid lipofuscinosis, J. Neurosci. 31 (2011) 15575–15585. Setters with the CLN8:c.491CNT mutation still survive. The existence of [15] Y. Nakamoto, O. Yamato, K. Uchida, K. Nibe, S. Tamura, T. Ozawa, N. Ueoka, A. Nukaya, N A. Yabuki, M. Nakaichi, Neuronal ceroid-lipofuscinosis in longhaired Chihuahuas: dogs with the CLN8:c.585G A mutation may provide an opportunity to clinical, pathologic, and MRI findings, J. Am. Anim. Hosp. Assoc. 47 (2011) 64–70. establish a new canine NCL model to investigate the functions of CLN8 [16] J. Tyynela, J.D. Cooper, M.N. Khan, S.J. Shemilts, M. Haltia, Hippocampal pathology in and evaluate therapies that could prove helpful to human vLINCL pa- the human neuronal ceroid-lipofuscinoses: distinct patterns of storage deposition, neurodegeneration and glial activation, Brain Pathol. 14 (2004) 349–357. tients with CLN8 mutations. [17] R.C. Riis, J.F. Cummings, E.R. Loew, A. de Lahunta, Tibetan terrier model of canine In conclusion, we have identified a CLN8 nonsense mutation that is ceroid lipofuscinosis, Am. J. Med. Genet. 42 (1992) 615–621. most likely the cause of NCL in a mixed breed dog and also the cause [18] D.P. O'Brien, M.L. Katz, Neuronal ceroid lipofuscinosis in three Australian shepherd littermates, J. Vet. Intern. Med. 22 (2008) 472–475. of previously described NCL in unrelated Australian Shepherds. This [19] M.L. Katz, J.R. Coates, J.J. Cooper, D.P. O'Brien, M. Jeong, K. Narfström, Retinal pathol- missense mutation was found by prioritizing the sequence variants in ogy in a canine model of late infantile neuronal ceroid lipofuscinosis, Invest. the whole genome sequence of a single affected dog. Similar whole ge- Ophthalmol. Vis. Sci. 49 (2008) 2686–2695. [20] J.D. Cooper, The neuronal ceroid lipofuscinoses: the same, but different? Biochem. nome sequencing strategies have been used to identify CUBN mutations Soc. Trans. 38 (2010) 1448–1452. in Border Collies and Beagles with Imerslund-Gräsbeck syndrome, a [21] T. Bensaoula, H. Shibuya, M.L. Katz, J.E. Smith, G.S. Johnson, S.K. John, A.H. Milam, COL11A2 mutation in Labrador Retrievers with disproportionate dwarf- Histopathologic and immunocytochemical analysis of the retina and ocular tissues – ism, a SUV39H2 mutation in Labrador Retrievers with hereditary nasal in Batten disease, Ophthalmology 107 (2000) 1746 1753. [22] H.H. Goebel, K.E. Wisniewski, Current state of clinical and morphological features in parakeratosis, and a KCNJ10 mutation in Jack Russell Terriers with human NCL, Brain Pathol. 14 (2004) 61–69. spinocerebellar ataxia and myokymia [41–45]. As the costs for generat- [23] S. Ranta, M. Savukoski, P. Santavuori, M. Haltia, Studies of homogeneous popula- – ing whole genome sequences decrease and software for data analysis tions: CLN5 and CLN8, Adv. Genet. 45 (2001) 123 140. [24] M. Haltia, The neuronal ceroid-lipofuscinoses, J. Neuropathol. Exp. Neurol. 62 (2003) improves, we expect whole genome sequencing to become more widely 1–13. used for identifying the mutations responsible for many other heritable [25] G. Anderson, M. Elleder, H.H. Goebel, Morphological diagnostic and pathological con- diseases in domestic animals. siderations, in: S.E. Mole, R.E. Williams, H.H. Goebel (Eds.), The Neuronal Ceroid Lipofuscinoses (Batten Disease), Oxford University Press, Oxford, UK, 2011, pp. 35–49. Supplementary data to this article can be found online at http://dx. [26] D. Harman, Lipofuscin and ceroid formation: the cellular recycling system, Adv. Exp. doi.org/10.1016/j.ymgme.2014.05.014. Med. Biol. 266 (1989) 3–15. [27] S.E. Haddad, M. Khoury, M. Daoud, R. Kantar, H. Harati, T. Mousallem, O. Alzate, B. Meyer, R. Boustany, CLN5 and CLN8 protein association with ceramide synthase: Acknowledgments biochemical and proteomic approaches, Electrophoresis 33 (2012) (3978-3809). [28] R. Passantino, C. Cascio, I. Deidda, G. Galizzi, D. Russo, G. Spedale, P. Guarneri, Iden- Our thanks to Joyce and John Van Dyke for donating tissues from tifying protein partners of CLN8, and ER-resident protein involved in neuronal ceroid lipofuscinosis, Biochim. Biophys. Acta 1833 (2013) 529–540. their dog Annie for this study and to Liz Hansen for assisting with [29] L. Lonka, A. Kyttala, S. Ranta, A. Jalanko, A.E. Lehesjoki, The neuronal ceroid DNA sample collections and archiving. This work was supported lipofuscinosis CLN8 membrane protein is a resident of the endoplasmic reticulum, – in part by the Batten Disease Support and Research Association, the Hum. Mol. Genet. 9 (2000) 1691 1697. [30] E. Winter, C.E. Ponting, TRAM, LAG1 and CLN8: members of a novel family of lipid- University of Missouri PRIME Fund, the University of Missouri Research sensing domains? Trends Biochem. Sci. 27 (2002) 381–383. Board, and Mizzou Advantage. [31] Y. Pewzner-Jung, S. Ben-Dor, F. A.H., When do lasses (longevity assurance genes) become CerS (ceramide synthases)?: Insights into the regulation of ceramide synthe- sis, J. Biol. Chem. 281 (2006) 25001–25005. References [32] K. Reinhardt, M. Grapp, K. Schlachter, W. Bruck, J. Gartner, R. Steinfeld, Novel CLN8 mutations confirm the clinical and ethnic diversity of late infantile neuronal ceroid [1] S.E. Mole, R.E. Williams, H.H. Goebel, The Neuronal Ceroid Lipofuscinoses (Batten lipofuscinosis, Clin. Genet. 77 (2010) 79–85. Disease), Oxford University Press, Oxford, UK, 2011. [33] M. Kousi, A.E. Lehesjoki, S.E. Mole, Update of the mutation spectrum and clinical cor- [2] K. Wisniewski, N. Zhong, Batten Disease: Diagnosis, Treatment, and Research, relations of over 360 mutations in eight genes that underlie the neuronal ceroid Academic Press, NY, 2001. lipofuscinoses, Hum. Mutat. 33 (2012) 42–63. [3] V. Warrier, M. Vieira, S.E. Mole, Genetic basis and phenotypic correlations of the [34] A. Hirvasniemi, H. Lang, A.E. Lehesjoki, J. Leisti, Northern epilepsy syndrome: an neuronal ceroid lipofuscinoses, Biochim. Biophys. Acta 1832 (2013) 1827–1830. inherited childhood onset epilepsy with associated mental deterioration, J. Med. [4] T. Awano, M.L. Katz, D.P. O'Brien, J.F. Taylor, J. Evans, S. Khan, P. Lobel, I. Sohar, G.S. Genet. 31 (1994) 177–182. Johnson, A mutation in the cathepsin D gene (CTSD)inAmericanbulldogswithneu- [35] R. Herva, J. Tyynela, A. Hirvasniemi, M. Syrjakallio-Ylitalo, M. Haltia, Northern ronal ceroid-lipofuscinosis, Mol. Genet. Metab. 87 (2006) 341–348. epilepsy: a novel form of neuronal ceroid-lipofuscinosis, Brain Pathol. 10 (2000) [5] M.L. Katz, F.H. Farias, D.N. Sanders, R. Zeng, S. Khan, G.S. Johnson, D.P. O'Brien, A mis- 215–222. sense mutation in canine CLN6 in an Australian shepherd with neuronal ceroid [36] S. Ranta, M. Topcu, S. Tegelberg, H. Tan, A. Ustubutun, I. Saatci, A. Dufke, H. Enders, K. lipofuscinosis, J. Biomed. Biotechnol. (2011), http://dx.doi.org/10.1155/2011/198042 Pohl, Y. Alembik, W.A. Mitchell, S.E. Mole, A.E. Lehesjoki, Variant late infantile neu- (Article ID 198042). ronal ceroid lipofuscinosis in a subset of Turkish patients is allelic to Northern epi- [6] S.A. Melville, C.L. Wilson, C.S. Chiang, V.P. Studdert, F. Lingaas, A.N. Wilton, A muta- lepsy, Hum. Mutat. 23 (2004) 300–305. tion in canine CLN5 causes neuronal ceroid lipofuscinosis in Border collie dogs, Ge- [37] N. Zelnik, M. Mahajna, T.C. Iancu, R. Sharony, M. Zeigler, A novel mutation of the CLN8 nomics 86 (2005) 287–294. gene: is there a Mediterranean phenotype? Pediatr. Neurol. 36 (2007) 411–413. [7] M.L. Katz, S. Khan, T. Awano, A. Shahid, A.N. Siakotos, G.S. Johnson, A mutation in the [38] N. Koppang, The English setter with ceroid-lipofuscinosis: a suitable model for the CLN8 gene in English Setter dogs with neuronal ceroid-lipofuscinosis, Biochem. juvenile type of ceroid-lipofuscinosis, Am. J. Med. Genet. (Suppl. 5) (1988) 117–126. Biophys. Res. Commun. 327 (2005) 541–547. [39] N. Koppang, English setter model and juvenile ceroid-lipofuscinosis in man, J. Med. [8] M. Abitbol, J.L. Thibaud, N.J. Olby, C. Hitte, J.P. Puech, M. Maurer, F. Pilot-Storck, B. Genet. 42 (1992) 594–599. Hedan, S. Dreano, S. Brahimi, D. Delattre, C. Andre, F. Gray, F. Delisle, C. Caillaud, F. [40] A.N. Siakotos, G.D. Hutchins, M.R. Farlow, M.L. Katz, Assessment of dietary therapies Bernex, J.J. Panthier, G. Aubin-Houzelstein, S. Blot, L. Tiret, A canine arylsulfatase G in a canine model of Batten disease, Eur. J. Paediatr. Neurol. 5A (2001) 151–156. (ARSG) mutation leading to a sulfatase deficiency is associated with neuronal ceroid [41] M. Owczarek-Lipska, V. Jagannathan, C. Drogemuller, S. Lutz, B. Glanemann, T. Leeb, lipofuscinosis, Proc. Natl. Acad. Sci. U.S.A. 107 (2010) 14775–14780. P.H. Kook, A frameshift mutation in the cubilin gene (CUBN)inBorderCollieswith [9] T. Awano, M.L. Katz, I. Sohar, P. Lobel, J.R. Coates, S. Khan, G.C. Johnson, U. Giger, G.S. Imerslund-Gräsbeck syndrome (selective cobalamin malabsorption), PLoS One 16 Johnson, A frame shift mutation in the canine ortholog of human CLN2 in a juvenile (2013) e61144. dachshund with neuronal ceroid lipofuscinosis, Mol. Genet. Metab. 89 (2006) [42] M. Drogemuller, V. Jagannathan, J. Howard, R. Bruggmann, C. Drogemuller, M. 254–260. Ruetten, T. Leeb, P.H. Kook, A frameshift mutation in the cubilin gene (CUBN)in J. Guo et al. / Molecular Genetics and Metabolism 112 (2014) 302–309 309

Beagles with Imerslund-Gräsbeck syndrome (selective cobalamin malabsorption), mutation in the SUV39H2 gene in Labrador Retrievers with hereditary nasal Anim. Genet. 48 (2014) 148–150. parakeratosis (HNPK) provides insights into the epigenetics of keratinocyte differen- [43] M. Frischknecht, H. Niehof-Oellers, V. Jagannathan, M. Owczarek-Lipska, C. tiation, PLoS Genet. 9 (2013) e1003848. Drögemüller, E. Dietschi, G. Dolf, B. Tellhelm, J. Lang, K. Tiira, H. Lohi, L. T., A [45] D. Gilliam, D.P. O'Brien, J.R. Coates, G.S. Johnson, G.C. Johnson, T. Mhlanga- COL11A2 mutation in Labrador retrievers with mild disproportionate dwarﬁsm, Mutangadura, L. Hansen, J.R. Taylor, R.D. Schnabel, A homozygous KCNJ10 mutation PLoS One 8 (2013) e60149. in Jack Russell Terriers and related breeds with spinocerebellar ataxia with [44] V. Jagannathan, J. Bannoehr, P. Plattet, R. Hauswirth, C. Drogemuller, M. Drogemuller, myokymia, seizures, or both, J. Vet. Intern. Med. (2014) (in press). D.J. Wiener, M. Doherr, M. Owczarek-Lipska, A. Galichet, M.M. Welle, K. Tengvall, K. Bergvall, H. Lohi, S. Rüfenacht, M. Linek, M. Paradis, E.J. Müller, P. Roosje, T. Leeb, A