Introduction

Motor neuron diseases (MNDs) are a group of disorders involved with selective degeneration of upper motor neurons, and/or lower motor neurons. The incidence of MND varied from 0.6 per 100 000 person-years to 2.4 per 100 000 person-years in different ethnic human populations (Lee, 2012). Upper motor neurons originate in the cerebral cortex or the brain stem, and provides indirect stimulation of the target muscles, while LMN connect the brain stem and spinal cord to muscle fibers, and provide nerve signals between the upper motor neurons and muscles. Common characteristics of MND are muscle weakness and/or spastic paralysis. Signs of LMN lesions are represented as muscle atrophy, fasciculations and weakness when associated LMN degenerate. Symptoms such as spastic tone, hyperreflexia and upgoing plantar reflex signs appear as upper motor neurons degenerate (Lee, 2012). The weakness of respiratory or bulbar muscle can even cause death of MND patients due to respiratory insufficiency (Borasio et al., 1998). Based on the time of onset, types of motor neuron involvement and detailed clinical features, MND in humans can be classified into six types of diseases including amyotrophic lateral sclerosis, hereditary spastic paraplegia, primary lateral sclerosis, spinal bulbar muscular atrophy, spinal muscular atrophy and lethal congenital contracture syndrome. MNDs are considered incurable and no effective treatment is currently available. However, the gene therapy strategy of performing gene transfer to patients by utilizing adeno-associated virus vectors is a promising therapeutic approach (Nizzardo et al., 2011). Previous evidence has shown genetic variants within >30 genes might be responsible for MND (Dion et al., 2009). For example, mutations in the Cu/Zn-binding superoxide dismutase (SOD1) gene have been identified in about 20% of patients with familial amyotrophic lateral sclerosis (Rosen et al., 1993). However, genetic causes still remain unclear for more than half of human patients with MND.

In 1999, an extended family of Romney lambs was found to have a suspected hereditary MND, characterized by poor muscle tone and progressive weakness from about 1 week of age, leading to severe tetraparesis, recumbency and muscle atrophy. Lambs remained alert up to 4 weeks of age when they were euthanized. The muscles were atrophied and pale, and histologically had variation in muscle fiber diameter and small angular fibers. Large foamy macrophages were present in cerebrospinal fluid, and there were degenerate neurons in the spinal cord leading to a decrease in the number of motor neurons, consistent with LMN diseases. No histological changes were seen in the cerebellum (Anderson et al., 1999). Outcross and backcross breeding trials were consistent with this LMN disease having simple autosomal recessive inheritance.

To find the causative locus responsible for this disease, we performed a genome-wide homozygosity mapping process and then fine mapped the mutation using a positional candidate gene approach within the largest homogygous genomic region found in the affected sheep. Our goals of this research were to discover the genetic basis of this disease, to help sheep breeders avoid matings between carrier animals and to provide a potential animal model for gene therapy treatment of human LMN disease.

Methods

Animals

Approvals dealing with DNA or tissue collection from animals were obtained from the Massey University Animal Ethics Committee (approval number 99/135). Iowa State University Animal Care and Use Committee did not require additional approvals.

A single Romney putative carrier ram was mated in New Zealand to 130 crossbred Finnish Landrace ewes to produce 115 daughters that were subsequently backcrossed to their sire. The backcrosses produced 5 affected sheep out of 47 lambs in the first lambing, and 11 out of 71 in the second lambing. The observed rate of affected sheep was consistent with the one in eight expectation for a homozygous recessive disorder from this mating scheme. Any affected offspring and their dams now proven to be carriers, formed the basis of the experimental population. High-density SNP genotypes were obtained from 19 affected, which included 3 affected lambs from the original Romney flock, and 11 known carrier sheep. Fine mapping using novel SNPs was conducted on 52 Romney or Romney/Finn crosses including the above mentioned 30 sheep with clear disease statuses and 22 potential carriers, which were phenotypically normal but sired by the carrier ram. Moreover, 85 control sheep consisted of 14 normal Romney sheep, 25 crossbred Texel sheep and 46 crossbred Corriedale sheep were used for validation of the causative mutation. These sheep were produced from unrelated populations without evidence of LMN disease.

SNP genotyping

DNA was extracted from blood using the QIAamp DNA Blood Mini Kit (QIAGEN, Valencia, CA, USA) and the Roche MagNA Pure automated analyzer (Roche Diagnostics N.Z., Ltd. Mt, Wellington, Auckland, New Zealand). The genotyping of 54 241 evenly distributed polymorphic SNP markers across the genome was performed using the Ovine SNP50 BeadChip (Illumina, San Diego, CA, USA) at the commercial company GeneSeek Inc., Lincoln, NE, USA. Standard procedures were performed with a PCR and ligation-free protocol, which routinely achieves high-average call rates and accuracy.

Genome-wide homozygosity mapping

To define the molecular genetic basis for the LMN disease in these Romney sheep, a search for putative homozygous-by-descent (also named IBD) regions was initiated by performing a genome-wide scan using an in-house UNIX script as described in our previous publication (Zhao et al., 2011). The SNP genotypes were ordered according to their genomic location as provided in the Illumina manifest, and used to identify regions of consecutive homozygous SNP loci that were common to all affected animals. The same procedure was applied to known carriers to ensure that the identified region was not simply fixed across all animals. A threshold of 10 SNPs defining a candidate consecutive homozygous region was chosen after calculating expected IBD fragment size for 19 affected lambs produced through the outcross and backcross experiments assuming exactly one crossover event occurred every meiosis on the chromosome carrying the causal mutation with uniform probability along the chromosome expressed in linkage units (cM). The homozygous fragment containing the causative mutation would have exceeded 0.5 cM or about 10 consecutive SNP with probability>95%. The expected size of the IBD region that contains the causative mutation depends in part on the number of meioses that separate the common ancestor from the affected lambs, and the number of affected individuals. The IBD region that included the causative mutation had 23% probability of being shorter than 2 cM and a 32% probability being longer than 5 cM. The mean expected length of the IBD region was about 4 cM. Genes in the identified IBD regions were examined for potential involvement using comparative homology between ovine and bovine genomes as provided through the Ovine Genome Assembly v1.0 (http://www.livestockgenomics.csiro.au/perl/gbrowse.cgi/oar1.0/#search).

Positional candidate genes sequencing

Primer3 software (version 0.4.0, Whitehead Inst, Cambridge, MA, USA and Howard Hughes Medical Inst, Chevy Chase, MD, USA) and the reference sequences from the Ovine Genome Assembly v1.0 were used to design PCR primers flanking the coding regions of ovine AGTPBP1 (Supplementary Table S1). Among the many genes in the identified homozygous regions, AGTPBP1 was considered to be the best candidate gene because of its role in neural functions as reported by several publications (Harris et al., 2000; Fernandez-Gonzalez). PCR was performed using a 10-μl cocktail mixture and a standard program as described previously (Zhao et al., 2011), with an annealing temperature of 58 or 60 °C. Negative controls were performed to check for the presence of contaminants in all reactions. PCR products were verified by 1.5% agarose gel electrophoresis, and treated with ExoSAP-IT (Affymetrix, Cleveland, OH, USA), before pooling them from two affected and two carrier sheep. The pooled PCR products were sequenced commercially using a 3730xl DNA Analyzer (Applied Biosystems, Pleasanton, CA, USA). The Sequencher software (version 3.0, GeneCodes Corp., Ann Arbor, MI, USA) was used for sequence alignment and comparison.

Mutation analyses

As annotation of the ovine genome has not yet been completed, the bovine AGTPBP1 reference mRNA sequence (GenBank: XM_001252536.3) was used as a query in cross-species BLAST searches to identify sheep expressed sequence tag. Both the bovine AGTPBP1 mRNA and corresponding ovine expressed sequence tag served as references to search for the coding regions within ovine AGTPBP1 gene. The AGTPBP1 protein sequence was predicted using successfully amplified and resequenced exonic regions of the gene from the sheep population, the bovine AGTPBP1 reference mRNA sequence and ovine expressed sequence tag. To understand the functional annotation of this protein, the conserved domains within this sheep protein sequence were searched by blasting the complete protein sequence against the Conserved Domains database (database: CDD-40526 PSSMs) in NCBI (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) while setting an option of 500 as the maximum number of hits. Multiple sequence alignments provide a basis for conserved domain models. The specific hits and super families were given for the queried protein. Moreover, the conserved protein residues involved in conserved features such as binding or catalysis were annotated when the query sequence residues were in the Entrez Protein database. The sequence alignments of protein AGTPBP1 from multiple species, including human, mouse, chicken, frog, purpuratus, sea squirt, trichoplax, zebrafish and sheep, were displayed to show the conserved feature sites. The fold structure of the conserved domain of AGTPBP1 in sheep (residues 850–1150) and mouse (residues 800–1110) were predicted using the 3D_PSSM program (Kelley et al., 2000; http://www.sbg.bio.ic.ac.uk/3dpssm/index2.html). The known crystal structures of carboxypeptidases in the PDB database help to generate the predicted structure of AGTPBP1. Sequence alignment was performed on the ClustalW2 server (http://www.ebi.ac.uk/Tools/msa/clustalw2/) among sheep and mouse AGTPBP1, human (PDB ID: 1DTD) carboxypeptidase A2 and cattle (PDB ID: 1HDU) carboxypeptidase A. This helped us to be more confident about the conserved features of AGTPBP1 secondary structure and the amino acids critical for carboxypeptidase catalysis. A three-dimensional macromolecular structure of the conserved domain was searched and downloaded using homologous sequences with confirmed crystal structure (http://www.ncbi.nlm.nih.gov/Structure/MMDB/mmdb.shtml). Each SNP variant was analyzed to check whether it could induce missense or nonsense mutations. The identified SNPs were then genotyped using restriction enzyme digestion (PCR-restriction fragment length polymorphism) on the affected and carrier sheep as well as on other normal controls.

Restriction enzyme digestion (PCR-restriction fragment length polymorphism)

The sequences of PCR primer ‘LMN_new’ (Supplementary Table S1) were 5′ TGTTAGCACCTCCTCTTTGCT 3′ and 5′ AGCAGACCCTTGGCATGATA 3′. DNA fragments from the PCR reactions were subjected to restriction enzyme digestion with AciI (New England Biolabs Inc., Ipswich, MA, USA). The missense mutation in AGTPBP1 removes a recognition site of the AciI enzyme. The 10-μl system reaction mix contained 1 μl of NEB buffer 3, 0.5 units of AciI and 2 μl of PCR product. The reaction mix was incubated at 37 °C overnight, and the products were analyzed on a 2.0% (w/v) ultra-pure agarose gel.

Results

Detection of identical by descent (IBD) regions

The whole-genome scan discovered 209 homozygous fragments consisting of three or more consecutive single-nucleotide polymorphisms (SNPs) shared by all 19 affected lambs. The size distribution of common homozygous fragments in these lambs is shown (Figure 1). Only one large homozygous segment exceeded the 10-SNP threshold and it consisted of 136 consecutive SNP loci identified in all 19 affected sheep, whereas the 11 carriers exhibited heterozygosity in this segment. The region started from SNP s56017.1 and extended to SNP OAR2_36428027.1. This segment is physically positioned on the small arm of ovine chromosome 2 (OAR 2) from 29 284 415 to 36 428 027 bp, covering a region of 7 Mb (Figure 2a and b). We found this segment likely encompassed 36 genes (Figure 2d and Supplementary Table S2) based on the bovine reference sequences. A gene called ATP/GTP-binding protein, transcript variant 1 (AGTPBP1), also named Nna1 (nervous system nuclear protein induced by axotomy protein 1 homolog) was the most plausible candidate because of its involvement in axonal regeneration and its causal relationship with mouse Purkinje cell degeneration (pcd) phenotype (Harris et al., 2000, Fernandez-Gonzalez et al., 2002). The ovine AGTPBP1 gene spans about 0.17 Mb and encodes a 1225-amino-acid intracellular protein that contains a predicted zinc carboxypepetidase domain near its C-terminus (Harris et al., 2000). This gene was initially cloned from regenerating spinal cord neurons of the mouse (supplied by OMIM 606830).

Figure 1
figure 1

The distribution of identical homozygous fragments in affected lambs. Based on the different number of SNPs spanning the homozygous fragment, the counts of identical homozygous fragments in all affected lambs were shown. Only one homozygous fragment spanning 136 SNPs was identified to be common to all 19 affected lambs when the ‘larger than 10-SNP’ threshold was applied.

Full size image
Figure 2
figure 2

Work flow chart for discovering the mutation locus of low MND in Romney sheep. (a) The sheep chromosome 2 (OAR 2). (b) The IBD region containing the causative locus is located on OAR 2 between 29 and 36 Mb. (c) SNP locations on the Ovine SNP50 BeadChip in the IBD region (d) The black boxes show reference genes encompassing the sheep IBD region based on the comparative genomic maps. The gene AGTPBP1 was depicted by a red lined box. (e) Sequence analyses of AGTPBP1 showed a point mutation in exon 21, resulting in an Arg to Pro substitution. This missense mutation is located in a highly conserved residue for the catalytic activity of this protein. The PCR-restriction fragment length polymorphisms show different patterns for affected and carrier groups. A full color version of this figure is available at the Heredity journal online.

Full size image

Gene coding region resequencing and mutation analyses

Functional mutations in candidate genes were examined by resequencing all available exons using primers based on the adjacent intronic sequences of these genes. The ovine AGTPBP1 gene was predicted to have 25 exons based on the bovine reference gene and protein sequences. All the exons except exons 11, 12 and 17 for this gene were successfully amplified and three exonic SNPs were identified. The first exonic SNP (called c.2909G>C) was located at the sixth base pair of exon 21 of the sheep AGTPBP1 gene. The other two were located at the ninety-third and ninety-ninth base pair of exon 22. An additional 17 SNPs were identified in either introns or the 3′-untranslated region on this gene. These were immediately excluded as causative loci as none were located either in the splicing donor/receptor regions or other important regulatory regions and were not in concordance with sheep phenotypes (data not shown). After comparing our findings with the bovine coding sequence of AGTPBP1, we observed that the first exonic SNP (c.2909G>C) introduced a missense transition (Arg>Pro) at amino-acid 970 (p.Arg970Pro; Figure 2e), and the third exonic SNP (c.3133T>C) also resulted as a missense mutation causing a Pro to Ser substitution (p.Pro1045Ser). The second exonic SNP (c.3126C>T) was a synonymous mutation and did not change any amino acid. A conserved domain named M14_Nna1 (CDD: cd06906) spanning from amino-acid residue 859–1136 on AGTPBP1 in sheep was predicted in this study using the ‘CD-search’ option available in NCBI. This domain is a zinc-binding carboxypeptidase domain, which hydrolyzes the peptide bonds at the C-terminal part of certain amino acids in polypeptide chains. The sequence alignment among multiple species demonstrated two specific feature hits including metallocarboxypeptidase activity and zinc binding. Nine conserved amino-acid residues were involved in these two features hits (Supplementary Figures S1 and S2; Gomis-Ruth et al., 1999; Aloy et al., 2001). Three conserved residues (His920, Glu923 and His1017) expressed as a motif ‘HXXE…H’ are responsible for zinc binding, and all nine conserved residues (His920, Glu923, Arg970, Asn979, Arg980, His1017, Gly1018, Met1027 and Glu1102) are important for substrate binding and catalysis of this protein (Supplementary Figure S1 and Figure 3a). In terms of the size and relative positioning of β-sheets and α-helices, the overall predicted secondary structures of AGTPBP1 in sheep and mouse were the same as that of other metallocarboxypeptidases in human and cattle by using the 3D-PSSM program. Features and conserved amino acids and motifs are located in similar positions. For example, the well-known zinc-binding HXXE…H motif lies between the second β-sheet and the first α-helix as reported (Wang et al., 2006). Six residues (His920, Glu923, Arg970, Asn979, Arg980 and Glu1102) remain conserved among these proteins (Supplementary Figure S3). The first exonic missense mutation (c.2909G>C) identified on exon 21 is the middle nucleotide within a codon ‘CGC’, which was translated and substituted from Arg970 (codon ‘CGC’), one of the conserved residues responsible for substrate binding and catalysis, to Pro970 (codon ‘CCC’; Figure 3b). Molecular modeling based on the crystal structure of the homologous protein, putative carboxypeptidase (PBDID: 3K2K), also demonstrated that Arg970 forms a part of a loop (the short yellow section of the middle loop in Figure 3c), which might be a key site for substrate binding or catalytic activity. The amino-acid residue encoded by the exonic missense mutation on exon 22 (c.3133T>C; p.Pro1045Ser) varies considerably among different species (Figure 3b). It indicates that this missense mutation may not affect the function of the protein AGTPBP1 as it is neither involved in zinc binding nor in substrate binding and catalysis.

Figure 3
figure 3

The conserved domain structures of partial protein sequence of AGTPBP1. (a) The conserved domains includes zinc-binding sites and putative active sites shared by M14_Nna1-like superfamily genes. (b) Alignment of partial sequence of sheep protein AGTPBP1 with eight most dissimilar protein sequences from species of human, mouse, chicken, frog, purpuratus, sea squirt, trichoplax and zebrafish. Alignment of the full sequence is shown in the Supplementary information. The residues labeled by hash marks on the top are known to be involved in substrate binding or catalytic activity. The two missense mutations are labeled on the protein sequence. Upper case amino acids are aligned and a read to blue color scale represents the degree of conservation, with red showing highly conserved residues. Lower case, gray amino acids are unaligned. Dashes indicate variations in sequence length among multiple sequence alignment. (c) Three-dimensional molecular modeling of the conserved domain based on the crystal structure of putative carboxypeptidase (PBDID: 3K2K). The short yellow section in the middle loop represents the position of the mutated amino-acid residue.

Full size image

The Arg970Pro mutation was genotyped by PCR-restriction fragment length polymorphism (Figure 2e). The 611 bp length of the ‘LMN_new’ PCR fragment amplified from normal sheep had one restriction site, which could be recognized by the restriction enzyme AciI and cut into two fragments with lengths of 519 and 92 bp. However, the restriction site was removed due to the Arg970Pro mutation, and the 611 bp fragment amplified from the affected lambs was intact after digestion. Therefore, the PCR fragment amplified from sheep with a genotype of ‘GG’ was cut into two fragments with lengths of 519 and 92 bp; the one with a genotype of ‘GC’ had three fragments with lengths of 611, 519 and 92 bp; the one with a genotype of ‘CC’ had a single fragment with length 611 bp. Our genotyping results showed all 19 affected sheep were ‘CC’, the 11 known carriers were ‘GC’ and the 22 phenotypically normal but possible carrier sheep were either ‘GC’ or ‘GG’ genotypes. In contrast, 85 control (unrelated) sheep were only ‘GG’ genotypes (Table 1). This mutation was 100% concordant with the recessive pattern of inheritance in affected, carrier, phenotypically normal and unrelated normal individuals. The two exonic SNP (c.3126C>T and c.3133T>C) on exon 22 were genotyped by direct sequencing. However, there is no perfect concordant rate for these two SNPs based on the counts for each of the genotypes (Table 1). At this stage, we excluded these two SNP as causative mutations for this defect due to their genotyping results and mutation analyses. Taken together, our results strongly suggest that the Arg970Pro substitution is the causative locus for LMN disease in Romney sheep and it could act by decreasing the function of protein AGTPBP1 to generate LMN disease in these lambs.

Table 1 The genotyping results of three exonic SNPs
Full size table

Discussion

The AGTPBP1 gene encodes a protein that belongs to a member of the cytosolic carboxypeptidase subfamily (M14D subfamily; Kalinina et al., 2007, Rodriguez de la Vega et al., 2007). This protein has a role in protein turnover by cleaving proteasome-generated peptides into amino acids (Berezniuk et al., 2010), enzymatically removing gene-encoded glutamic acids or tyrosine from the C-terminus of proteins (Rogowski et al., 2010, Kalinina et al., 2007) and also has a role in neuronal bioenergetics at least in mouse brains (Chakrabarti et al., 2010). Similar to other cytosolic carboxypeptidases, the mouse Nna1 protein is broadly expressed in the brain, pituitary, eye, testis and other tissues (Kalinina et al., 2007). Mutations or abnormal expression of Nna1 have been found to cause a pcd phenotype in mice characterized by degeneration of Purkinje cells, mitral cells of the olfactory bulb, thalamic neurons and retinal photoreceptor cells as well as degeneration of sperm (Fernandez-Gonzalez et al., 2002, Mullen et al., 1976). A total of 12 alleles in mouse mutants related to Agtpbp1 have been summarized by the Jackson Laboratory (http://www.informatics.jax.org/searches/allele_report.cgi?_Marker_key=79447). Among which, loss-of-function of Agtpbp1 was indicated as being responsible for pcd3J and pcd5J mouse mutants (Fernandez-Gonzalez; Wang and Morgan, 2007; Chakrabarti et al., 2008). Expression of full-length Nna1 successfully rescued Purkinje cell loss and ataxic behavior in pcd3J, and retinal photoreceptor loss and pcd in pcd5J mice (Wang et al. 2006, Chakrabarti et al., 2008).

In this study, the missense mutation c.2909G>C on exon 21 of AGTPBP1 induces an Arg to Pro substitution (p.Arg970Pro) at amino-acid 970 (R970). Based on the crystal structure of human carboxypeptidase A2, it has been reported that arginine 127 (R127) in carboxypeptidase A2 is an electrophile site crucially responsible for the catalytic activity along with glutamate 270 (Mangani et al., 1994; Garcia-Saez et al., 1997; Christianson et al., 1989). Similarly, the homologous arginine residue R127 (numbered according to mature bovine carboxypeptides A1) in Caenorhabditis elegans AGBL4 (AGP/GTP-binding-like protein 4) has been predicted to stabilize the oxyanion hole in the S1’ site of the carboxypeptidase domain (Rodriguez de la Vega et al., 2007). Sequence comparison and structural modeling of sheep AGTPBP1 with other carboxypeptidases predicted R970 was located in a region known as the substrate-binding pocket and catalytic center of metallocarboxypeptidase as mouse Nna1 (Supplementary Figure S3; Wang et al., 2006). These results enforce the view that R970 in sheep is a very conserved site being critical for its function of catalytic activity. Therefore, we concluded the Arg970Pro substitution could be a candidate causative mutation possibly altering the characteristic of the AGTPBP1 protein and very likely impairing its function of catalytic activity.

Comparing to the phenotype of the original pcd mouse model, the affected sheep do not have apparent degeneration of Purkinje cells but show their specific characteristics including degeneration of neurons in ventral horns of the spinal cord and brain stem, Wallerian degeneration of motor nerve fibers and atrophy of skeletal muscles. The phenotype of these sheep is more similar to the symptoms in the Sod1 transgenetic mice (mSOD1), which were constructed as animal models for human amyotrophic lateral sclerosis patients (Gurney et al., 1994). It is interesting to note that the lysyl oxidase (Lox) gene, encoding a protein having an important role in the formation and repairing of extracellular matrix, was found to be upregulated in both mSOD1 mice and Agtpbppcd-sid mice (Li et al., 2004; Li et al., 2010; Lucero and Kagan, 2006). The Agtpbppcd-sid mutant was characterized as a deletion of Agtpbp1 exon 7 that produced a stop codon in exon 8. The absence of protein Agtpbp1 will decrease dendrite development of cerebellar Purkinje cells by changing the expression of downstream components, including Lox propeptide and the NF-κB signaling pathway in mice (Li et al., 2010). Therefore, we assume the p.Arg970Pro substitution might impair the catalytic activity of AGTPBP1 and decrease its ability of protein turnover of lysyl oxidase. The overabundant lysyl oxidase would cause abnormal development of dendrite in LMN and then develop phenotypes shown in these sheep.

To ascertain that the Arg970Pro substitution was really a causative mutation, measuring mRNA expression of the genes AGTPBP1 and LOX and their enzymatic activities in the spinal cord, brain stem and cerebellar tissues would be desired. It could exclude the possibility that the Arg970Pro mutation in protein AGTPBP1 is only a coincidental finding from the genetic mapping study or this mutation disrupts expression of another gene in the same region of the genome. Nevertheless, a point mutagenesis of mouse Nna1 gene at arginine 962 (R962), the homologous residue of R970 in sheep, to generate transgenic mice and then using the intact full-length Nna1 gene to rescue phenotypes in mice will provide further evidence for the Arg970Pro substitution being a causative mutation in sheep.

The missense mutation c.2909G>C (p.Arg970Pro) in AGTPBP1 appears to be a spontaneous one arising from a single sheep family. To avoid spreading out this disease throughout the sheep population, a simple genetic test can be easily designed based on our results to identify and remove carriers with the defective c.(2902C) allele, and to decrease the risk of economic losses in the future.

Regenerating motor neurons expressing protein NNA1 in human nervous tissues indicated that NNA1 was a key factor for motor neuronal degeneration (Harris et al., 2000). To date, no human cases have been reported with spontaneous mutations in AGTPBP1 that would implicate the gene as a cause for a LMN disease. Neither have they been reported to occur in other large mammalian species. Mice and hamsters are convenient animal models for studies of human neural diseases (Mashimo et al., 2009; Akita et al., 2007). However, there are still significant structural and functional differences between rodent and large mammalian species. Even among rodents themselves, the gene distributions could be different. Hamsters with undetectable Nna1 mRNA in the brain did not exhibit distinctive neurodegenerative features as serious as mice mutants, which could possibly be explained due to the different distributions of Nna1 and other Nna 1-like genes in the brains of mice and hamsters (Kalinina et al., 2007; Akita et al., 2007; Akita and Arai, 2009). Sheep have large brain size, long life span and long gestation period relative to rodents. Our finding in sheep, a large mammalian species, raises the possibility that human patients with similar etiology may also result from this gene or other genes in the same pathway of neuronal development. As there is no effective treatment available for MNDs, the identification of molecular pathogenetic targets will shed light on the development of gene therapy for this type of diseases (Nizzardo et al., 2011). Here, we reported that LMN disease originating in a Romney ram was likely caused by a mutation in a gene locus AGTPBP1 (NNA1). The affected sheep could be used as a valuable animal model for gene therapy if human patients are successfully diagnosed in the future.

Data archiving

Data deposited in the Dryad repository: doi:10.5061/dryad.8fp2pc88.