ATP8A2 is a P4-ATPase that is highly expressed in the retina, brain, spinal cord and testes. In the retina, ATP8A2 is localized in photoreceptors where it uses ATP to transport phosphatidylserine (PS) and phosphatidylethanolamine (PE) from the exoplasmic to the cytoplasmic leaflet of membranes. Although mutations in ATP8A2 have been reported to cause mental retardation in humans and degeneration of spinal motor neurons in mice, the role of ATP8A2 in sensory systems has not been investigated. We have analyzed the retina and cochlea of ATP8A2-deficient mice to determine the role of ATP8A2 in visual and auditory systems. ATP8A2-deficient mice have shortened photoreceptor outer segments, a reduction in photoresponses and decreased photoreceptor viability. The ultrastructure and phagocytosis of the photoreceptor outer segment appeared normal, but the PS and PE compositions were altered and the rhodopsin content was decreased. The auditory brainstem response threshold was significantly higher and degeneration of spiral ganglion cells was apparent. Our studies indicate that ATP8A2 plays a crucial role in photoreceptor and spiral ganglion cell function and survival by maintaining phospholipid composition and contributing to vesicle trafficking.
P4-ATPases are a family of P-type ATPases that use the energy from ATP hydrolysis to translocate or ‘flip’ aminophospholipids from the exoplasmic to the cytoplasmic leaflet of membranes. Flipping of aminophospholipids has been implicated in establishing phospholipid asymmetry in biological membranes and generating the membrane curvature required for many cellular processes (Coleman et al., 2013; Graham and Kozlov, 2010; Puts and Holthuis, 2009; Sebastian et al., 2012; van der Mark et al., 2013).
ATP8A2 is a member of the P4-ATPase family of phospholipid transporters which is highly expressed in the retina, brain, spinal cord and testis (Cacciagli et al., 2010; Coleman et al., 2009; Zhu et al., 2012). It associates with CDC50A, the β-subunit crucial for the expression and flippase activity of ATP8A2 (Coleman and Molday, 2011; van der Velden et al., 2010). In the retina, a significant portion of the ATP8A2–CDC50A complex localizes to the light-sensitive photoreceptor outer segments. Biochemical studies indicate that the ATP8A2–CDC50A complex transports phosphatidylserine (PS) and to a lesser extent phosphatidylethanolamine (PE) across photoreceptor disc membranes (Coleman et al., 2009).
Mutations in ATP8A2 are known to cause a severe neurological disorder characterized by cerebellar axatia, mental retardation and dysequilibrium syndrome (Cacciagli et al., 2010; Emre et al., 2012). Recently, the wabbler-lethal (wl) mouse was found to harbor a deletion in exon 22 of Atp8a2, which results in an inactive protein and distal axonal degeneration in spinal motor neurons (Zhu et al., 2012). Knockdown of ATP8A2 in PC12 cells indicates that the lipid transport activity of ATP8A2 influences neurite length (Xu et al., 2012). Mice deficient in ATP8A1, a homolog of ATP8A2, show an increase in externalization of PS on the plasma membranes of hippocampal cells and a deficiency in hippocampus-dependent learning (Levano et al., 2012). The function of ATP8A1 appears to partially overlap with the function of ATP8A2 because loss of activity of both transporters results in neonatal lethality (Zhu et al., 2012).
Exposure of PS on the surface of cells serves as a signal for phagocytosis (Wu et al., 2006) and is thought to play a role in outer segment turnover (Ruggiero et al., 2012). In yeast, P4-ATPases function in vesicle budding and trafficking (Sebastian et al., 2012). Little is known about the role of mammalian P4-ATPases in vesicle transport. However, the pathophysiology of P4-ATPase-associated disorders suggests that this might be an evolutionarily conserved function.
The role of P4-ATPases in sensory systems has not been previously investigated. In this study we have examined the retina and cochlea of Atp8a2-knockout and wl/wl mice to begin to define the role of ATP8A2 in the visual and auditory systems. Here, we show that ATP8A2 deficiency causes an alteration in phospholipid composition, a shortening of outer and inner segments, a reduction in the photoresponse and loss of photoreceptor cells in the visual system, and a reduction in hearing and spiral ganglion cell survival in the auditory system. Our studies suggest that ATP8A2 phospholipid flippase activity plays a crucial role in vesicle trafficking, and neuronal function and survival in these sensory systems.
An Atp8a2-knockout mouse was generated by replacing exons 11–13 with a neo cassette (Fig. 1A). Deletion of exons 11–13, which encode transmembrane segment M4 and part of the P-domain of P-type ATPases crucial for ATP8A2 structure and function, is predicted to generate an unstable product. PCR genotyping was used to identify homozygous knockout mice from their wild-type (WT) and heterozygous littermates (Fig. 1B) and RT-PCR was used to demonstrate the absence of Atp8a2 gene expression in the retinas of homozygous knockout mice and reduced expression in heterozygous knockout mice (Fig. 1C). Western blots confirmed the absence of ATP8A2 and reduced levels of its β-subunit CDC50A in retina extracts from knockout mice (Fig. 1D). Compared with WT and heterozygous animals, Atp8a2-knockout mice were noticeably smaller (Fig. 1E), and exhibited labored movements and clasping of hind legs when held by the tail (Fig. 1F). The overall phenotype is highly similar to that of the wl/wl mouse (Zhu et al., 2012), suggesting that distal axonal degeneration of the spinal cord also occurs in the Atp8a2-knockout mouse. The mean survival time of 1–2 months for the Atp8a2-knockout mouse is also similar to that of the wl/wl mouse.
Localization of ATP8A2 and CDC50A in the retina of Atp8a2-knockout and wl/wl mice
The distribution of ATP8A2 and CDC50A in WT mice was compared with that in age-matched Atp8a2 knockout and wl/wl mice by immunofluorescence microscopy (Fig. 2A,B). Strongest immunolabeling of ATP8A2 was observed in the outer segment layer of P23 WT mice with weaker labeling in other retinal layers including the inner segment as previously reported (Coleman et al., 2009). Strong CDC50A immunoreactivity was also observed in the outer segments of WT mice with moderate labeling in other retinal layers. In contrast, no significant immunolabeling of ATP8A2 was observed in the retina of age-matched Atp8a2-knockout mice (Fig. 2A). CDC50A labeling was significantly diminished in the outer segment of the knockout mice, but was retained in other retinal layers, probably as a result of its interaction with other P4-ATPases in the retina. In the retina of the wl/wl mice, ATP8A2 was detected, but only in the inner segment of the photoreceptor cells, indicating that mutant ATP8A2 is expressed, but is retained in the endoplasmic reticulum (ER) as an inactive and misfolded protein (Fig. 2B).
Localization of photoreceptor proteins in Atp8a2-knockout mice
The effect of ATP8A2 deficiency on the localization of various proteins in photoreceptor cells was examined by immunofluorescence microscopy on retinal cryosections labeled with specific antibodies (Fig. 2C). In WT mice, rhodopsin, which comprises ∼80% of the total outer segment membrane protein, is present in both the disc and plasma membrane of rod photoreceptor cells (Molday and Molday, 1987; Papermaster and Dreyer, 1974); the alpha-subunit of the cyclic nucleotide-gated channel (CNGA1) is localized to the plasma membrane of rod outer segments (Cook et al., 1989); and peripherin-2 is distributed along the rim region of rod and cone disc membranes (Molday et al., 1987). These proteins correctly localized to shortened outer segments of the Atp8a2-knockout mice. Similarly, other outer segment proteins, including guanylate cyclase 1 (GC1), the ATP binding cassette protein ABCA4, and transducin were also correctly localized to the outer segment layer (data not shown). Cone arrestin used as a cone specific marker also showed normal localization to the shortened cone outer segments of the Atp8a2-knockout mice. Furthermore, the localization of various inner segment and synaptic proteins such as GM130 (Golgi), syntaxin 3 (vesicles), Rab11 (endosome), synaptotagmin (synapse), and Na+/K+-ATPase were found to be similar to that of WT mice (supplementary material Fig S1). These studies indicate that ATP8A2 deficiency does not impair the selective targeting of photoreceptor proteins to their final destination.
Photoreceptor morphology and degeneration in the retina of Atp8a2-knockout and wl/wl mice
The morphological characteristics of retinas from wl/wl and Atp8a2-knockout mice were compared with that of age-matched WT mice to determine the significance of ATP8A2 in photoreceptor structure and survival. At P14, when ATP8A2 is first expressed, the outer nuclear layer and outer segment length of the wl/wl mouse retina were indistinguishable from that of WT mice (Fig. 3A). At later ages, the number of photoreceptor nuclei and length of the outer segment were significantly reduced in wl/wl mice (Fig. 3A–C). At P30, the wl/wl mice showed a 10% reduction in the number of nuclei and the OS length was approximately one-half that of WT mice (11.7±0.3 versus 20.4±0.4 µm). At P60, 30–40% of photoreceptor cells had been lost and the length of the outer segments was one-third of that in WT mice (7±0.3 versus 21.5±0.4 µm). The Atp8a2-knockout mouse also showed a decrease in the number of photoreceptors and length of the outer segment layer throughout the eye, with a 15% reduction in outer nuclear layer thickness and 60% reduction (7.7±0.6 versus 18.3±0.9 µm) in outer segment length near the optic nerve at P23 (Fig. 3B).
Next, we examined the ultrastructure of ATP8A2-deficient mouse retina using transmission electron microscopy (Fig. 3C). Despite the reduced length of the outer segments and some disorganization of the outer segments as a result of ongoing photoreceptor degeneration, the morphology of the photoreceptor outer segments was remarkably normal for both the wl/wl and knockout mice. The rod disc membrane stacks appeared to be correctly aligned and the structural organization of the hairpin rim region of the rod photoreceptor discs and the adjacent plasma membrane was well preserved. The RPE cells of the Atp8A2-deficient mice showed the presence of some intracellular vesicles not found in the RPE of WT mice, but otherwise the morphological features were similar (see also Fig. 3A,C). The observed vesiculation might be due secondary effects on the RPE cells resulting from the degenerating photoreceptor cells because there is no evidence indicating that ATP8A2 is expressed in RPE cells.
Visual function of photoreceptors
Electroretinograms (ERGs) of the rod-mediated dark-adapted scotopic and cone-mediated light-adapted photopic response were recorded to determine the effect of ATP8A2 deficiency on visual function. Both the wl/wl and Atp8a2-knockout mice at P30 showed a large reduction in scotopic and photopic response (Fig. 4A,B). The scotopic a- and b-wave amplitudes of the Atp8a2-deficient mice were quantified and found to be reduced by 75% and 67%, respectively, compared with values in WT type and heterozygous mice (Fig. 4C,D).
Analysis of rhodopsin and retinal in outer segments of Atp8a2-knockout mice
The relative amount of opsin in P23 Atp8a2-knockout mice was compared with age-matched WT mice by western blotting. A reduction of 50% was observed for the knockout mice (Fig. 5A). This is generally consistent with the reduction in the length of the outer segments in these mice as observed by microscopy (Figs 2,3). Rhodopsin levels were measured by difference spectrophotometry before and after bleaching. Detergent-solubilized lysates from knockout mice contained 75% lower levels of rhodopsin compared with results in WT mice (Fig. 5B). Rhodopsin content was also determined by retinoid analysis because essentially all the 11-cis-retinal is bound to opsin in the dark-adapted mice. A 75% reduction in 11-cis retinal was also observed by retinoid analysis. A representative chromatogram is shown in Fig. 5C.
Phospholipid composition was investigated using preparations of hypotonically lysed outer segment membranes from WT and Atp8a2-knockout mice containing similar concentrations of opsin and a similar protein profile as analyzed by SDS gel electrophoresis (Fig. 5D). A representative two-dimensional thin layer chromatogram of outer segment lipids is shown in Fig. 5E. Phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylserine (PS) constitute the majority of the membrane phospholipids in the outer segments. For WT mice, the composition of these lipids was similar to that previously reported for rat, cow and frog (Boesze-Battaglia et al., 1994; Fliesler and Anderson, 1983; Mason et al., 1973; Poincelot and Abrahamson, 1970). The WT composition of PC∶PE∶PS was 49±2%∶36±2%∶15±1% (Fig. 5F). The knockout mouse contained significantly higher levels of PC and lower amounts of PE and PS (66±4%∶25±2%∶9±2%).
Phagocytosis of outer segments and PS
Exposure of PS on the surface of cells has been associated with phagocytosis in several cell types (Darland-Ransom et al., 2008; Ruggiero et al., 2012). Accordingly, it is possible that the observed shortened outer segments in ATP8A2-deficient mice result from an increase in phagocytosis of outer segments by adjacent RPE cells owing to increased exposure of PS on outer segment surface membranes. To investigate this possibility, we compared the levels of phagocytosis in the Atp8a2-knockout mice with WT mice by Toluidine Blue staining for light microscopy (LaVail, 1976). No significant difference in phagocytosis was observed between WT and knockout mice 1 hour after light onset when phagocytosis of outer segments by RPE cells reaches a maximum (Fig. 6A–C). At 6 hours after light onset, phagocytosis was significantly reduced, but no significant difference was observed between the Atp8a2-knockout and WT mice. We also measured outer segment phagocytosis in primary cultures of RPE cells isolated from WT mice using an established assay (Gibbs et al., 2003). Phagocytosis of outer segments from Atp8a2-knockout mice was lower than that for outer segments from WT mice, but the ratio of outer segments internalized to outer segments bound to the cell surface was the same for both genotypes of mice (Fig. 6D,E).
Fluorescent-labeled annexin V, a protein that specifically binds PS, was used to determine PS exposure on the extracellular side of the plasma membrane of outer segments (Fig. 6F). Annexin V labeled a significant amount of PS exposed on the tips of the outer segments as visualized in whole-mounted retina collected ∼1 hour after light onset, in agreement with the finding that PS becomes exposed at the tips of outer segments as a signal for RPE phagocytosis (Ruggiero et al., 2012). However, no significant differences were observed in whole-mounted Atp8a2-knockout versus WT retinas (Fig. 6F).
Localization of ATP8A2 to the Golgi and endosomes in photoreceptors and PC12 cells
Because removal of the tips of the outer segments by phagocytosis appears normal in the Atp8a2-knockout mouse, we considered the possibility that ATP8A2 plays a role in outer segment morphogenesis by vesicle trafficking within the inner segment of photoreceptor cells. To explore this possibility, we examined in more detail the localization of ATP8A2 in photoreceptor cells of WT mice by confocal scanning microscopy. Immunofluorescence labeling of ATP8A2 was seen just above the ONL which partially colocalized with the Golgi marker GM130 (Fig. 7A) indicating that a fraction of ATP8A2 resides in or very close to the Golgi complex of photoreceptors.
We further examined the distribution of ATP8A2 in transfected and differentiated PC12 cells, which exhibit long neuron-like axons and dendrites (neurites). In agreement with a recent report (Xu et al., 2012), co-transfection of NGF-differentiated PC12 cells with ATP8A2 and CDC50A promoted neurite outgrowth (94±7 versus 62±5 µm). A large fraction of the expressed ATP8A2 was found to reside within the Golgi complex and at the tips of the neurites (Fig. 7B). ATP8A2 colocalized with CDC50A (data not shown) and Rab11, indicating that the ATP8A2–CDC50A complex is present in endosomes as well as the trans-Golgi network (Fig. 7B).
Auditory brainstem response
In addition to the visual system, we assessed hearing in wl/wl mice. Homozygous mutant wl/wl mice on a C57BL/6J (B6) strain background exhibited a diminished auditory startle response. Therefore, we used auditory brainstem response (ABR) threshold measurements to assess hearing in wl/wl mice. Analysis of representative ABR recordings revealed that the ABR threshold of mutant mice was much higher than that of control animals (Fig. 8A,B). At 2 months of age, the 16 kHz ABR thresholds of wl/wl mice were on average 37 dB higher than that of WT (Fig. 8C).
To understand the basis of this auditory defect, we analyzed the inner ear morphology of the wl/wl mice by light microscopy. Cross sections through the basal turns of cochleae showed that although the overall structure of the organ of Corti appeared normal and the outer and inner hair cells were intact (Fig. 8F,G), a substantial number of the spiral ganglion cells had been lost (21±2 versus 7±2; n = 4) in 2-month-old wl/wl mice (Fig. 8D,E).
In this study we have analyzed two lines of mice deficient in ATP8A2-catalyzed phospholipid transport. The phenotype of the newly generated Atp8a2-knockout mouse was similar to the wl/wl mouse, which harbors a deletion of 7 highly conserved amino acids (TAIEDRL) in the nucleotide-binding domain of ATP8A2 (Zhu et al., 2012). Both mice were smaller than their littermates, walked with an abnormal gait, survived for only 2 months and exhibited similar aberrations in the visual system.
A prominent feature of ATP8A2-deficient mice is the progressive reduction in the length of the retinal photoreceptor outer segments, with no detectable abnormality in their ultrastructural organization. Outer segments are dynamic structures which turnover every 10 days (LaVail, 1976; Sung and Chuang, 2010). Adjacent RPE cells engulf the distal tenth of an outer segment every 24 hours. This is compensated by the addition of newly synthesized membrane at the proximal end of the outer segment via vesicle trafficking from the inner segment. The balance between phagocytosis and disc morphogenesis maintains a constant length of the outer segment. We reasoned that the progressive reduction in outer segment length in the ATP8A2-deficient mice might result from either an increase in phagocytosis or a decrease in outer segment morphogenesis. Because increased exposure of PS on the extracellular surface of cells has been implicated in the recognition step for a number of phagocytic processes (Ruggiero et al., 2012; Wu et al., 2006), we first investigated whether Atp8a2-knockout mice showed an increase in PS on the surface of outer segments and higher levels of phagosomes in RPE cells. However, annexin V labeling of outer segments and the number of phagosomes in the knockout and WT mouse were similar, indicating that ATP8A2 deficiency does not directly affect PS cell surface exposure or phagocytosis by RPE cells. Other mechanisms appear to control PS exposure on the outer segment plasma membrane as part of the phagocytic process.
The shortened outer segments in ATP8A2-deficient mice more likely arise from decreased vesicle trafficking to the outer segment required for disc morphogenesis. A number of studies have implicated ATP8A2 and related P4-ATPases in protein-mediated vesicle trafficking. Overexpression of ATP8A2 and CDC50A increases the length of neurite outgrowths in NGF-differentiated PC12 cells and primary cultures of rat hippocampal neurons, whereas depletion of ATP8A2 or CDC50A reduces neurite length (Xu et al., 2012). In this study, we have confirmed that overexpression of the ATP8A2–CDC50A complex increases neurite length in PC12 cells and have further shown that this complex localizes to the Golgi complex and endosomes, as well as the tips of the neurites of transfected PC12 cells. Drs2p, the yeast ortholog of ATP8A2, is a key regulator of the vesicle budding from the trans-Golgi network (Sebastian et al., 2012). ArfGEF binds to Drs2p and stimulates its lipid flippase activity. ArfGEFs recruit activated Arf proteins and phospholipid flipping drives vesicle bud formation. Activated Arf proteins recruit adaptors and coat proteins. We have shown here the existence of a population of ATP8A2 in the inner segment of photoreceptor cells with a significant fraction colocalizing with the Golgi marker GM130, in agreement with ATP8A2 localization in transfected PC12 cells and other cells (Coleman and Molday, 2011; van der Velden et al., 2010). On the basis of these studies, we propose that the reduction in outer segment length observed in the Atp8a2 mutant mice is due to a decrease in vesicle budding and trafficking in the inner segment required for efficient disc morphogenesis. The fact that outer segments are formed in these mutant mice suggests that another P4-ATPase, probably ATP8A1, partially compensates for the loss of ATP8A2 within the inner segment, allowing vesicle budding and trafficking to occur in ATP8A2-deficient mice, but less efficiently, thereby giving rise to shortened outer segments. Photoreceptor degeneration observed in the ATP8A2-deficient mice is most likely a consequence of a reduction in outer segment formation because it has been shown previously that genetic defects that affect normal outer segment morphogenesis result in photoreceptor degeneration (Hawkins et al., 1985; Humphries et al., 1997). ATP8A2-mediated vesicle budding and trafficking also appears to be directly involved in neurite outgrowth in PC12 and hippocampal cells. Axonal degeneration observed in wl/wl mice deficient in ATP8A2 might also result from defective vesicle trafficking (Zhu et al., 2012).
Another feature of ATP8A2-deficient mice is the marked loss in visual function as measured by the ERGs. The 50% reduction in outer segment length could contribute to the observed decrease in a-wave amplitude of the ERG response, but this is likely to be a relatively small contribution. Heterozygous rhodopsin-knockout mice with an outer segment length and volume that were 60% and 40%, respectively, that of WT mice, show only a 15% reduction in the a-wave amplitude (Liang et al., 2004). Instead, the large reduction in the photoresponse observed in ATP8A2-deficient mice might be due to the altered lipid environment experienced by opsin and other membrane-associated proteins, which in turn impacts phototransduction. It is not known why the absence of ATP8A2 alters the lipid composition of the outer segment. One possibility is that ATP8A2 in the Golgi and endosomes concentrates PS and PE in budding vesicles through bilayer coupling resulting in higher PS or PE levels in the transport vesicles destined for the outer segment. Accordingly, in the absence of ATP8A2, a lower level of these lipids would be present in the outer segments of the Atp8a2-knockout mouse. Another possibility is that the biosynthesis of these lipids is downregulated to compensate for the absence of the transporter to maintain their membrane distribution. Finally, it cannot be ruled out that the difference in phospholipid contents arises from membrane contaminants in the outer segment preparation of the knockout mice, although analysis of the protein content by SDS gel electrophoresis does not reveal significant differences in protein content between outer segments of WT and knockout mice. Interestingly, a reduction in PS and PE composition has been previously reported for yeast lacking the P4-ATPase Drs2p (Pomorski et al., 2003). It remains to be determined whether alteration of phospholipid composition is a common feature of biological systems displaying a deficiency in other P4-ATPases, or if this alteration is specific for ATP8A2 and Drs2p deficiency.
Interestingly, this change in lipid composition does not affect the morphological features of photoreceptor outer segments, but does affect rhodopsin content. A decrease in opsin content of 50% in 1-month-old Atp8a2-knockout mice is expected on the basis of the 50% reduction in outer segment length. However, rhodopsin content as measured by spectral bleaching and retinoid analysis is only 25% of the WT value, indicating that only 50% of the opsin contains a bound 11-cis retinal chromophore. This suggests that either a portion of opsin is misfolded, preventing the binding of 11-cis retinal or the supply of 11-cis retinal to outer segments is impaired. Rhodopsin activation has been reported to be highly dependent on membrane environment with meta II rhodopsin being formed more readily in the presence of PE and PS (Gibson and Brown, 1991). Helix 8 of rhodopsin acts as a membrane switch, causing adoption of a helical conformation upon PS binding (Krishna et al., 2002). Helix 8 has also been shown to be important for protein folding, binding of 11-cis retinal and transducin activation (Natochin et al., 2003). Accordingly, the significant reduction in the a-wave amplitude of the ERGs observed in Atp8a2-deficient mice might result from the effect of altered phospholipid composition on phototransduction.
A principal function of ATP8A2 is to actively transport aminophospholipids across the lipid bilayer to generate phospholipid asymmetry in biological membranes. Because ATP8A2 is present in outer segment disc membranes, it might be expected to induce aminophospholipid asymmetry. A number of earlier studies have focused on phospholipid asymmetry in bovine photoreceptor disc membranes. Although initial reports suggested that disc membranes were highly asymmetrical with respect to the transbilayer distribution of PS and PE in disc membranes (Miljanich et al., 1981; Wu and Hubbell, 1993), subsequent studies have indicated that PC and PE are symmetrically distributed across the bilayer with PS exhibiting an ATP-independent asymmetrical distribution resulting from the high density and orientation of rhodopsin (Hessel et al., 2000; Wu and Hubbell, 1993). The recently reported scramblase activity of rhodopsin (Menon et al., 2011) might overwhelm the ATP-dependent phospholipid flippase activity of ATP8A2, resulting in a random distribution of phospholipids across the disc membrane. Although the ATP8A2 phospholipid transport activity might not play an important role in establishing the bulk phospholipid transbilayer distribution, it could play a role in generating transient local aminophospholipid asymmetry in discs, which is important for phototransduction. It remains to be determined whether localization of ATP8A2 to the tips of the neurite outgrowth and possibly synaptic vesicles (as observed in PC12 cells) plays an important role in the generation of phospholipid asymmetry and neuronal function.
In addition to affecting visual function, ATP8A2 deficiency was found to cause a loss in hearing and degeneration of cochlea spiral ganglion cells. This is the second P4-ATPase associated with hearing deficiency. Previously, it was reported that deficiency in functional ATP8B1 in humans and mutant mice causes hearing loss, associated with progressive degeneration of cochlear hair cells consistent with the localization of ATP8B1 in the sterocilia of hair cells (Stapelbroek et al., 2009). These combined studies highlight the crucial roles played by distinct P4-ATPases in different auditory cells. The mechanism by which deficiency of ATP8A2 causes loss in auditory function and neuronal degeneration remains to be determined. It is possible that these phospholipid transporters are important in generating and maintaining PS asymmetry required for neuronal vesicle trafficking and auditory function.
Recently, a missense mutation (I376M), which abolishes ATP8A2 phosphatidylserine flippase activity (our unpublished data) in a highly conserved transmembrane segment of ATP8A2 has been reported in several members of a consanguineous family from Turkey (Emre et al., 2012). Affected individuals displayed severe neurological complications including mental retardation, mild cerebellar and cerebral atrophy and truncal ataxia. Unfortunately, members of this family declined neuro-ophthalmological examinations and hence the effect of ATP8A2 deficiency in the human visual and auditory sensory systems could not be determined.
In summary, our studies indicate that the ATP8A2–CDC50A phospholipid transporter plays a crucial role in the function and survival of photoreceptors and spiral ganglion cells and implicate ATP8A2 as a crucial mediator of vesicle trafficking in neuronal cells.
MATERIALS AND Methods
All animal protocols were approved by the Animal Care Committee of the University of British Columbia and conform to the Canadian Council on Animal Care guidelines. Mice were raised in cyclic lighting conditions with a 12 hour light and 12 hour dark cycle. The Association for Assessment and Accreditation of Laboratory Animal Care guidelines was followed for all animal procedures at the Jackson Laboratory and procedures were approved by the Institutional Animal Care and Use Committee of the Jackson Laboratory.
Knockout mice were produced by InGenious Targeting Laboratory (Stony Brook, NY). An 11.6 kb fragment spanning exons 10–16 used to construct the targeting vector was first subcloned from a positively identified C57BL/6 BAC clone into the pSP72 vector (Promega, Madison, WI). The neomycin resistance cassette replaced 3.86 kb of the gene including exons 11–13. This targeting vector was linearized by NotI and transfected by electroporation of C57BL/6 × 129/SvEv hybrid embryonic stem cells. Homologously recombined clones were selected by neomycin, screened by PCR and confirmed by Southern blotting. Targeted stem cells were microinjected into C57BL/6 blastocysts. Resulting chimeras were mated to WT C57BL/6 mice to generate F1 heterozygous offspring. Heterozygous mice were crossed and matings producing homozygous wl/wl mutants or knockout were kept for strain production. Thus, the knockout mice had a mixed 129SvEv and C57BL/6 background.
The wl mutation backcrossed to strain C57BL/6J was used (Zhu et al., 2012). As standard practice for wl/wl mutant mice, dry food was supplemented with a soft maintenance diet (DietGel 76A, ClearH2O, Portland, ME). Knockout mice were provided with dry food at the bottom of the cage.
Genotyping by PCR
Genomic DNA extracted from ear punches was amplified by PCR using primers for the Atp8a2 gene (forward, 5′-ATGCAGGGTCTGTGAGTAGTAGTC-3′) and (reverse, 5′-GTGGCCAGATGACAAGCATTCCCT-3′) and the neomycin resistance cassette (reverse, 5′-TGCGAGGCCAGAGGCCACTTGTGTAGC-3′). Amplification was performed using the REDExtract-N-Amp kit (Sigma) with the addition of 0.5 M Betaine (Sigma, Oakville, ON). The first cycle used 94°C for 3 minutes, followed by 35 cycles of 94°C for 30 seconds, 60°C for 30 seconds and 72°C for 1.5 minutes. To identify mice with the wl allele, the following primers were used: forward, 5′-TGAACTGTCCCTTAACTGATGGTA-3′ and reverse, 5′-TGGCTATGGTTTCTGGAACG-3′. This primer pair spans the 21 base-pair deletion in exon 22 in wl allele and produces a 108 bp amplicon in WT controls and an 87 bp amplicon in wl/wl mice.
Transmission electron microscopy and light microscopy
Eyes were fixed in 1% glutaraldehyde and 1% paraformaldehyde in phosphate buffer at 4°C for ∼1 week. Retinas were treated with OsO4 for 1 hour at a concentration of 40 mM in 0.1 M cacodylate buffer pH 7.4 containing 0.2% sucrose. Samples were dehydrated with ethanol and embedded in Epon 812-Araldite resin. Ultrathin sections (0.07 µm) were cut and stained with uranyl acetate and lead citrate solution. For light microscopy, thin sections (0.5 µm) were cut and stained with Toluidine Blue. Blue particles ≤1 µm were counted as phagosomes. For hematoxylin and eosin staining (H&E), eyes were fixed overnight in 1.22% glutaraldehyde and 0.8% paraformaldehyde in 0.08 M phosphate buffer, embedded in Technovit resin, cut in 1.5 µm sections. For cochlear tissue, anesthetized mice were perfused with PBS followed by 4% paraformaldehyde. Tissues were decalcified with Cal-EX solution for 24 hours, and embedded in paraffin. 5 µm sections were counterstained in H&E. Spiral ganglion cells were counted in an 80×80 µm square positioned over the center of the cochlear ganglion.
RNA was extracted from retinas using the RNeasy kit (Qiagen, Maryland, MA). Genomic DNA was removed by incubation of 2 µg of total RNA with 2 units DNaseI for 30 minutes at 37°C. Complementary DNA was prepared by reverse transcription of 1 µg of total RNA using the iScript cDNA Synthesis kit (Bio-Rad). Amplification of the cDNA was performed with the following primers specific to Atp8a2: forward, 5′-ACGAGGGACGTGCTCATGAAGC-3′ and reverse, 5′-CCTCAAGTGTCACCAGCAGGCT-3′; and for glyceraldehyde phosphate dehydrogenase (Gapdh): forward, 5′-ATCAAATGGGGTGAGGCCGGTG-3′ and reverse, 5′-CGGCATCGAAGGTGGAAGAGTG-3′. PCR was performed using Taq (New England Biolabs, Ipswich, MA) for 30 cycles for Atp8a2 and 25 cycles for Gapdh and the PCR products were visualized on 1.5% agarose gels.
Animals were kept in the dark for at least 2 hours before the experiment and anesthetized with an intraperitoneal injection of xylazine (7 mg/g body weight) and ketamine (15 mg/g body weight) in PBS. A heated water blanket was used to keep animal body temperature constant at 38°C. For ERG evaluation of mice, an Espion Visual Electrophysiology System (Diagnosys, Westford, MA) was used. Scotopic recordings used an intensity setting of 1.42 log scotopic troland. Photopic recordings were performed using an intensity setting of 1.13 log photopic troland.
Preparation of photoreceptor outer segments
Purified mouse photoreceptor outer segments were prepared according to the Optiprep method (Tsang et al., 1998). Briefly, 12 mouse retinas were vortexed for 1 minute in 120 µl of Ringer's buffer (10 mM HEPES, pH 7.4, 130 mM NaCl, 3.6 mM KCl, 2.4 mM MgCl2, 1.2 mM CaCl2 and 0.02 mM EDTA) containing 8% Optiprep (Sigma, Oakville, ON) and complete inhibitor (Roche, Laval, QU). The retinas were centrifuged at 200 g for 1 minute, the supernatant was collected, and the procedure was repeated five times. The supernatant was overlaid on a 10% and 18% Optiprep step gradient in Ringer's buffer and centrifuged for 30 minutes at 26,500 g. The outer segment membranes were collected on top of the 18% Optiprep with a needle, diluted with six volumes of Ringer's buffer, and centrifuged at 26,500 g for 30 minutes. The pellet was resuspended in Ringers buffer containing 8% Optiprep and complete inhibitor and stored at −30°C. Bovine outer segments were isolated as described (Papermaster and Dreyer, 1974).
Polyclonal ATP8A2 antibody
A DNA fragment corresponding to amino acids 369–644 of mouse ATP8A2 (NM_016529.4) was cloned in-frame with glutathione S-transferase (GST) in the pGEX-4T-1 vector using the EcoRI and XhoI restriction sites. This fragment was subcloned into pMAL-c2 using EcoRI and SalI in-frame with maltose binding protein (MBP). ATP8A2 antibodies were raised in rabbits immunized seven times with 500 µg of the GST-fusion protein (YenZym Antibodies, Burlingame, CA). ATP8A2-specific antibodies were purified from 50 ml of serum in PBS on an affinity column consisting of MBP-fusion protein coupled to Sepharose 2B by CNBr and eluted with 0.1 M glycine (pH 2.5).
Proteins were separated on 9% polyacrylamide SDS gels and stained with Coomassie Blue or transferred to Immobilon FL membranes (Millipore, Bedford, MA) in buffer containing 25 mM Tris-HCl, pH 8.3, 192 mM glycine, 10% methanol. Membranes were blocked with 1% milk in PBS for 30 minutes. Primary antibody cell culture supernatants were diluted in PBS and 0.1% milk at the following concentrations: Atp6C11 (1∶10), Cdc50-7F4 (1∶10), 3F4 (1∶20), 1D1 (1∶20), 5H2 (1∶10), and 1D4 (1∶1000). The ATP8A2 polyclonal and the β-actin polyclonal (ab8227, Abcam) antibodies were diluted at a concentration of 0.3 µg/ml and 0.2 µg/ml respectively. Blots were incubated with primary antibodies for 40 minutes, washed with PBS containing 0.05% Tween 20 (PBST), incubated for 40 minutes with secondary antibody (goat anti-mouse or anti-rabbit conjugated with IR dye 680 or 800 (LI-COR, Lincoln, NE) diluted 1∶20,000 in PBST containing 0.5% milk, and washed with PBST prior to visualization on a LI-COR Odyssey imager (LI-COR, Lincoln, NE).
Cryosections of mouse retinas were prepared by fixing eyes in 4% paraformaldehyde, 100 mM phosphate buffer (PB) (pH 7.4) for 1–3 hours as previously described (Cheng et al., 2013). Cryosections were cut using a cryostat at a thickness of 10 µm. Labeling with various antibodies was performed as previously described (Cheng et al., 2013; Coleman et al., 2009; Coleman and Molday, 2011; Kwok et al., 2008). PC12 cells were cultured and differentiated according to (Xu et al., 2012). Sections were blocked and permeabilized with 10% normal goat serum and 0.2% Triton X-100 in phosphate buffer for 30 minutes. Labeling of primary antibodies was carried out overnight at room temperature or in the case of opsin Rho-1D4 labeling, for 1 hour. Primary antibodies (cell culture supernatant and purified antibodies) were diluted in phosphate buffer containing 2.5% normal goat serum and 0.1% Triton X-100 at the following concentrations: ATP8A2 polyclonal antibody (purified 0.3 µg/ml), Cdc50-7F4 monoclonal antibody (supernatant 1∶2), Per5H2 monoclonal antibody to peripherin-2 (supernatant 1∶10), PMc 1D1 monoclonal antibody to CNGA1 channel subunit (supernatant 1∶20), Rho1D4 monoclonal antibody to rhodopsin (supernatant 1∶1000), Rho 4D2 monoclonal antibody to rhodopsin (supernatant 1∶100), cone arrestin polyclonal antibody (purified 0.5 µg/ml, AB15282, Millipore, Bedford, MA), GM130 monoclonal monoclonal antibody (2.5 µg/ml, 61-0822, BD Biosciences, Mississauga, ON) and Rab11 monoclonal antibody (2.5 µg/ml, 610656, BD Biosciences, Mississauga, ON). Sections were washed with PB and labeled for 1 hour with Alexa-Fluor-488- or Alex-Fluor-594-labeled goat anti-mouse or anti-rabbit Ig secondary antibody (diluted 1∶1000) and counterstained with DAPI.
Retinal whole mounts were prepared as previously described (Cheng et al., 2013). Briefly, mouse eyes were lightly fixed for 1 hour after light onset with 4% paraformaldehyde for 15 minutes. The retina was dissected from the retinal pigment epithelial layer and fixed for another 15 minutes. After washing in 10 mM HEPES, pH 7.4, 140 mM NaCl and 2.5 mM CaCl2, the retinas were labeled with annexin-V–Alex-Fluor-594 (Molecular Probes, Eugene, OR) for 1 hour in 10 mM HEPES, pH 7.4, 140 mM NaCl, 2.5 mM CaCl2. The retinas were then washed and relabeled with the Rho 4D2 antibody (supernatant 1∶100). The Rho 4D2 antibody is against the N-terminus of rhodopsin and labels rhodopsin exposed on the extracellular surface of rod outer segments without permeabilization (Hicks and Molday, 1986; Laird and Molday, 1988). After labeling with the secondary antibody, the retinas were mounted on slides with the photoreceptor outer segments facing up and visualized on a Zeiss LSM 700 confocal scanning microscope.
Measurement of outer nuclear layer and outer segment layers
Eyes from WT or knockout mice were removed, marked on the nasal side for orientation, cut into cryosections, and labeled with DAPI and the Rho1D4 antibody as described above. The outer nuclear layer and outer segment layers were defined by DAPI and Rho1D4 labeling, respectively. Three measurements of the outer nuclear layer and outer segment layers were taken every 200 µm from the optic nerve and averaged. The optic nerve was defined as 0 µm. In the case of wl/wl mice, H&E stained retinas were used to count the rows of photoreceptors in the outer nuclear layer and measure the outer segment length.
The procedure for growing RPE cells on a Transwell filter was performed exactly as described (Gibbs et al., 2003). Briefly, RPE cells were isolated from 10- to 15-day-old WT mice. Eyes were digested using 2% Dispase (Sigma, Oakville, ON) and sheets of RPE cells were isolated and dispersed into growth medium containing DMEM, 10% bovine FCS, 1% penicillin/streptomycin, 2.5 mM L-glutamine, 1× MEM non-essential amino acids. All cell culture materials were obtained from Invitrogen. RPE cells were grown until confluent at 37°C, 5% CO2. Equal amounts of WT and knockout outer segments were added to the RPE cells in growth medium and incubated at 37°C, 5% CO2 for 60 minutes. RPE cells were washed three times with growth medium and incubated for an additional 60 minutes and then fixed in 4% paraformaldehyde. Bound outer segments were labeled with the rhodopsin Rho-4D2 antibody followed by an Alexa-Fluor-488-conjugated goat anti-mouse antibody. RPE cells were permeabilized with 47.5% ethanol. Internalized outer segments were labeled with the rhodopsin 4D2 antibody (Laird and Molday, 1988) and an Alexa-Fluor-594-conjugated goat anti-mouse antibody.
Rhodopsin and retinoid quantitative analysis
All procedures were conducted under dim red light as described (Liang et al., 2004). Typically two eyes were used for each experiment. The lens was removed and the eyes were cut into several small pieces and homogenized in a Dounce homogenizer ∼20 times in 1 ml of 20 mM HEPES, pH 7.5, 150 mM NaCl, 5 mM hydroxylamine and 1% DDM and stirred for 1 hour at 4°C. Insoluble material was pelleted at 100,000 g for 10 minutes. The absorption spectra were acquired before and after a 10 minute bleach with light. The rhodopsin concentration was determined by the decrease in absorbance at 500 nm using a molar extinction coefficient of 40,000 M−1 cm−1. In parallel, an aliquot was taken for analysis of opsin by western blotting. Retinoids were extracted from two eyes by homogenization six times in 2 ml of 50 mM MOPS, pH 6.5, 10 mM hydroxylamine and 50% ethanol and incubated for 30 minutes to allow formation of retinal oximes. Retinoids were extracted three times with 4 ml of hexane, dried with N2, dissolved in 100 µl of hexane, and analyzed on an Agilent HPLC 1100 (Agilent Technologies, Mississauga, ON) using a silica column (Supelcosil LC-Si, 150 × 4.5 mm). Retinoids were separated by normal phase with 10% ethyl acetate, 90% hexane at a flow rate of 1.0 ml/min. Retinoids were identified by comparison with known standards.
Quantification of phospholipids
Lipids were extracted according to the method of Bligh and Dyer (Bligh and Dyer, 1959), separated by thin layer chromatography in CHCl3∶MeOH∶NH4OH (65∶25∶4) followed by a second dimension in CHCl3∶CH3COOH∶MeOH∶H2O (75∶25∶5∶2.2) and quantified by measurement of the phosphorus content (Zhou and Arthur, 1992).
Auditory brainstem response (ABR)
Hearing in mice was assessed by ABR threshold analysis. Mice were anesthetized with an intraperitoneal injection of tribromoethanol (2.5 mg tribromoethanol/10 g body weight) and then placed on a heating pad in a sound-attenuating chamber. Recording electrodes (Model F-E2, Astro-Med, Inc.) were placed just under the skin, with the active electrode placed between the ears just above the vertex of the skull, the ground electrode between the eyes, and the reference electrode underneath the left ear. High-frequency transducers were placed just inside the ear canal and computer-generated sound stimuli were presented to both ears at defined intervals. Stimulus-evoked signals were recorded in an ABR recording system (Intelligent Hearing System, IHS, Miami, FL). Thresholds were determined for broadband click and 8 kHz, 16 kHz and 32 kHz pure-tone stimuli by increasing the sound pressure level (SPL) in 10 dB increments followed by 5 dB increases and decreases to determine the lowest level at which a distinct ABR wave pattern could be recognized.
The authors would like to thank Andrew Metcalfe and Jennifer Ryan for expert technical assistance with ERG recordings, Jesse Hammer for assembly of the auditory figure and Dr Wayne Vogl for assistance with light microscopy. We would also like to acknowledge Scientific Services at The Jackson Laboratory.
J.A.C., X.Z., H.R.D., L.L.M., R.S.S., R.T.L. carried out the experiments. J.A.C., X.Z., R.S.M. and S.W.M.J. designed the experiments. J.A.C., R.S.M. and S.W.M.J. wrote the manuscript.
This work was supported by grants from the Canadian Institutes for Health Research [grant number MOP-106667 to R.S.M.]; the National Institutes of Health [grant number EY02422 to R.S.M.; EY11721 to S.W.M.J.]; Pew Charitable Trust [S.W.M.J.]; National Science Foundation of China [grant number 81271007 to X.Z.]. J.A.C. was supported by a National Sciences and Engineering Council predoctoral studentship. R.S.M. holds a Canada Research Chair in Vision and Macular Degeneration. S.W.M.J. is an Investigator of the Howard Hughes Medical Institute. Deposited in PMC for release after 6 months.
The authors declare no competing interests.