Mutations in a P-type ATPase gene cause axonal degeneration

Abstract

Neuronal loss and axonal degeneration are important pathological features of many neurodegenerative diseases. The molecular mechanisms underlying the majority of axonal degeneration conditions remain unknown. To better understand axonal degeneration, we studied a mouse mutant wabbler-lethal (wl). Wabbler-lethal (wl) mutant mice develop progressive ataxia with pronounced neurodegeneration in the central and peripheral nervous system. Previous studies have led to a debate as to whether myelinopathy or axonopathy is the primary cause of neurodegeneration observed in wl mice. Here we provide clear evidence that wabbler-lethal mutants develop an axonopathy, and that this axonopathy is modulated by Wld(s) and Bax mutations. In addition, we have identified the gene harboring the disease-causing mutations as Atp8a2. We studied three wl alleles and found that all result from mutations in the Atp8a2 gene. Our analysis shows that ATP8A2 possesses phosphatidylserine translocase activity and is involved in localization of phosphatidylserine to the inner leaflet of the plasma membrane. Atp8a2 is widely expressed in the brain, spinal cord, and retina. We assessed two of the mutant alleles of Atp8a2 and found they are both nonfunctional for the phosphatidylserine translocase activity. Thus, our data demonstrate for the first time that mutation of a mammalian phosphatidylserine translocase causes axon degeneration and neurodegenerative disease.

Figure 1. Phenotypic characterization of wl/wl mice on a C57BL/6J background.

(A) wl/wl mice are noticeably smaller at twenty one days after birth (P21) than their wild type (+/+) litter mates. (B) wl/wl mice fall behind their wild type (+/+) litter mates in weight gain from seven days after birth (P7) onward, and are clearly distinguishable from wild type at about 12 days of age (n = 13 for +/+; n = 9 for wl/wl). For animals at P7 or older, we used male animals. For younger animals, we used both males and females. The weight gain is consistent with previous reports of the wl mutation on other genetic backgrounds . (C) Compared to littermate controls (green line), mutant mice (orange line) had a much lower survival rate. Even with the addition of a soft maintenance diet, 20% of wl/wl mutants (blue line) die by 65 days and all wl/wl mutants die by 130 days of age (n = 46 for +/+; n = 44 for wl/wl). (D; E) Phenotypic consequences of the wl mutation: wl/wl mice drag their hind feet when walking as evident by ink traces (D: blue, front paws; red, rear paws), and (E) show clasping of the hind limbs when picked up by the tail, characteristic of a neurological deficit. (F) Hematoxylin and eosin staining of deep cerebellar nuclei in wild type (+/+) and wl/wl mice of two months of age show central chromatolysis in wl/wl mice (pale neurons) but not in wild type mice. Arrows indicate healthy neurons in +/+ mice and abnormal neurons in wl/wl mice. Black boxes in the top panels indicate areas that have been magnified in the bottom panels. Scale bar is 50 µm.

Figure 2. Characteristics of axonopathy in wl/wl mice.

(A) Diagram showing the structures of the spinal cord and associated peripheral nerves. (B,C) H&E staining of lumbar spinal cord sections. Note the central chromotalysis in the wl/wl mutant motor neurons (arrow). (D, E) Semi-thin (1 µm) sections of the L4 ventral root stained with toluidine blue. No morphological differences between wild type and mutant nerves are apparent. This is reflected in the similar number of axons present in the wild type and mutant nerves (H; n = 4 for each genotype). (F, G) Semi-thin (1 µm) sections of the femoral nerve motor branch stained with toluidine blue clearly showed a morphological difference between wild type and mutant nerves, with a reduced number of axons in the wl/wl mice. The average axon number in the motor branch was significantly decreased in wl/wl mice compared to wild type mice (p<0.01; n = 4 for both genotypes). (J, K) Semi-thin (1 µm) sections of the sciatic nerve stained with toluidine blue in 60 day old wild type and wl/wl mutant mice. (L) Quantification of the mean axonal diameter of these axons shows that wl/wl mice have far fewer large diameter axons compared to wild type mice. Average axon diameter was significantly decreased in wl/wl mice compared to controls (**, p<0.01). The axon diameter for over 400 axons from wild type and wl/wl mice were analyzed by TEM. Scale bar, 50 µm.

Figure 3. Phosphorylated neurofilament (pNF) accumulates in somas of spinal cord neurons and retinal ganglion cells in wl/wl mice.

Note the white boxes in A, C, F and H define areas that have been magnified in B, D, G and H, respectively. (A, B) At P30 pNF localized to only axons of lumbar motor neurons in wild type mice, but was not found in the somas of neurons. (C, D) At P30 in wl/wl mutant animals pNF was present in the axons of motor neurons, but also accumulated in the somas of some motor neurons. (E) Lumbar motor neuron sections stained with pNF antibody were used to count the number of soma in which pNF accumulated. There was a significant increase in the number of somas accumulating pNF in the motor neurons of wl/wl mice (12±0.4) compared to wild type mice (0.2±0.2). (F, G) In P30 wild type mice pNF labeled only ganglion cell axons in the retina, but in wl/wl mutants (H, I) pNF was also present in a subset of ganglion cell somas, especially in the peripheral region of the retina. In addition, focal swelling was observed in some axons. An enlargement of a focal swelling is shown in the upper right hand corner of I, which corresponds to the box in I. (J) Quantification of pNF+ somas confirms a massive increase in the number of somas accumulating pNF in wl/wl mice (364±7) compared to controls (2±0.5). **, P<0.01; Scale bar, 50 µm.

Figure 4. The Wld^s mutation delays axonal degeneration in wl (wl/wl) mice.

(A–C) Femoral nerves of wild type (+/+), wl/wl and wl/wl Wld^s (wl mice hemizygous for a Wld^s allele) were examined for axonal degeneration at two months of age. Severe axonal degeneration occurred in wl mutant mice (B), while the Wld^s allele significantly delays axonal degeneration (C). Axon morphology wl Wld^s was similar to that in wild type control mice. (D–I) Wld^s prevented accumulation of pNF (green) in spinal cord soma of wl mutant mice (note, white box in D–F indicates area magnified is G–I). Abnormal accumulation of pNF was observed in the soma of wl mice (E, H) but not wl Wld^s mice (F, I). (J) Total number of myelinated axons in the femoral nerve of mice of each genotype at two months of age. Hemizygosity (hemi) for Wld^s mice prevented axon loss by this age (n = 4 for each genotype assessed). Number of axons ± sem for mice of each genotype were 537±3 (+/+); 396±3.4 (wl/wl) and 532±4.4 (wl/wl Wld^s). (K) Nerve conduction velocity (NCV) was also rescued by Wld^s. In wl mice, increased distal latency indicated a reduction in nerve conduction velocity. In the presence of Wld^s, NCV was restored to almost control levels (n = 4, for control mice; n = 5 for wl/wl mutant; n = 6 for wl/wl Wld^s animals). **, P<0.01; Scale bar, 50 µm.

Figure 5. Genetic ablation of Bax protects axons from degeneration in wl mice.

(A–D) Femoral nerves of mice of the indicated genotypes were examined at two months of age (genotypes above panels are for Atp8a2). The severe axonopathy in wl mutant mice (B), was significantly delayed by the Bax mutation (D). (E) Quantification of axon number in nerves from wl mutants either wild type or null for Bax (Bax^+/+ and Bax^−/− respectively) showed that Bax deficiency prevented the axon loss in wl/wl mutants. wl mice had significantly fewer axons than wild type control animals (420±12 axons versus 535±8). In contrast, wl mice that are Bax deficient do not have any axon loss compared to Bax deficient mice (note Bax deficient mice have more neurons because of lack of normal developmental neuronal death; Bax−/− 734±12 axons, wl/wl Bax ^−/− 731±9). Four mice of each genotype were examined. **, P = 0.01. Scale bar is 50 µm.

Figure 6. Positional cloning of the wl mutation.

(A) Genotyping 688 affected F2 mice (1354 meiotic events) allowed the wl locus to be mapped to a 773 kb region on mouse chromosome 14 between DLM14-10 (60.6 Mb) and DLM14-21 (61.4 Mb). This region contains 10 genes. Solid lines represent parts of the chromosome containing B6 sequences, while open rectangles symbolize parts of the chromosome containing wl sequences in different mice. Numbers on the left side refer to the number of independent recombinant chromosomes obtained. (B) Sequence analysis of genomic DNA from mutant (wl/wl) and wild type (+/+) mice revealed a 21 bp deletion (ATCGAAGACCGTCTTCAAGCC) in exon 22 of Atp8a2, while the rest of the coding sequence is still in frame. Nucleotide bases immediately flanking the deletion region are boxed for easy comparison. The amino acids (TAIEDRL) encoded by the deleted base pairs are given at the top of the sequences. (C) Diagram showing the position and nature of the wl mutation and two additional wl alleles, wl^vmd (vmd) and wl^3J (3J). vmd genomic DNA harbors a 9,167 bp deletion, resulting in loss of exon 32 from the RNA transcript (Figure S10). 3J has a 641 bp deletion starting from tenth base pair of exon 30. 3J also contains a duplication of 10 bp (TCTTTGGTGT) in exon 32 (Figure S11). (D) Protein domains of ATP8A2 with the location of the three wl allele mutations indicated. ATP8A2 contains 10 putative transmembrane domains (M1–M10), an actuator domain (A-domain), a nucleotide-binding domain (N-domain), and a phosphorylation domain (P-domain). The position of wl mutation resulting in the deletion of seven highly conserved amino acids (TAIEDRL) in the N-domain is indicated by a green triangle; the position of the vmd mutation resulting in the deletion of 32 amino acids spanning the ninth transmembrane domain is indicated by a purple triangle, while the area of the protein affected by the 3J deletion is indicated by a red bar (the small 3J duplication overlaps the vmd mutation). The location of seven highly conserved amino acids (DKTGTLT) is indicated with a black line. (E) Conservation of the seven amino acid DKTGTLT motif, which is highly conserved among all ATPases, as well as the seven amino acid sequence (TAIEDRL) deleted in wl, between mouse, human, C. elegans (TAT1) and yeast (DSR2P).

Figure 7. ATP8A2 expression and localization.

(A) Reverse transcript (RT)-PCR analysis showed that Atp8a2 is expressed in the cerebrum, cerebellum, spinal cord, testis and retina (Upper panel). The control for RT-PCR efficiency was β-actin (Actb; lower panel). (B) HEK293T cells were transfected either with mouse Atp8a2 cloned into pCMV6-AN-DDK vector, or empty vector in order to characterize the newly generated anti-ATP8A2 polyclonal rabbit antibody. Total protein (10 µg) isolated from these cells was used in western blot analysis with commercial anti-Flag antibody, or the anti-ATP8A2 antibody. Both Flag (left panel) and ATP8A2 antibodies (right panel) detected a protein of the expected 130 kDa in lysates from Atp8a2 cDNA transfected cells. (C) Total brain proteins were fractioned into membrane and cytosol fractions and subjected to SDS-PAGE and western blot analysis. ATP8A2 antibody detected the 130 kDa band only in the brain membrane fraction of wild type (+/+), wl (Atp8a2^wl/wl) and vmd (Atp8a2^vmd/vmd) mice. PSD95 was used as a membrane protein marker; TUBULIN was used as a marker for cytosolic proteins.

Figure 8. ATP8A2 contains phosphatidylserine translocase activity.

(A) Internalization of NBD phospholipids by UPS-1 cells transfected with a control vector (Empty) or plasmid expressing Atp8a2. Expression of ATP8A2 leads to a population of UPS-1 cells with increased NBD-PS uptake (Top right panel). Representative numbers of NBD-PS-labeled UPS-1 cells are shown (Bottom panel). The X-axis represents NBD-PS fluorescence intensity of GFP-positive cells; cells transfected with pcDNA62-Atp8a2 vector shows increased fluorescence intensity (bottom right panel), compared to cells transfected with empty vector (bottom left panel). (B) ATP8A2 specifically translocates NBD-PS across the plasma membrane of UPS-1 cells. Lipid translocation activity is shown as a percentage of NBD-lipid fluorescence intensity relative to control empty vector (defined as 1). Results are representative data from three independent experiments. No translocation activity was observed for NBD-phospholipids PE- phosphatidyletholamine, PC- phosphatidylcholine, or PG- phosphatidylglycerol. Four independent samples were assessed for each group. (C) Mutant proteins encoded by wl and vmd mutant mice are non-functional for PD translocation. Lipid translocation activity of ATP8A2^wl and ATP8A2^vmd encoded by wl and vmd mutant animals are similar to the vector control. A single D→A point mutation (ATtp8a2^D388A) in the conserved DKLTG motif completely abolishes the lipid translocation activity of ATP8A2. A chemical ATPase inhibitor sodium vanadate significantly reduces ATP8A2 activity. (D) ATP8A2 protein levels were similar in all transfected cells (other than those transfected with empty vector) as assessed by western blotting analysis using the ATP8A2 antibody.