Severe Neurodegeneration with Impaired Autophagy Mechanism Triggered by Ostm1 Deficiency*

Background: Although loss of the Ostm1 gene leads to the most severe form of osteopetrosis, Ostm1 is expressed in other tissues, including the CNS. Results: Independently of hematopoietic lineages, loss of Ostm1 results in acute neurodegeneration with enhanced autophagy. Conclusion: We present evidence for an Ostm1 cell-autonomous role in neurons. Significance: This study shows a novel molecular pathogenic mechanism for neurodegeneration-related diseases. Loss of Ostm1 leads to the most severe form of osteopetrosis in mice and humans. Because functional rescue of the osteopetrotic defect in these mice extended their lifespan from ∼3 weeks to 6 weeks, this unraveled a second essential role of Ostm1. We discovered that Ostm1 is highly expressed in the mouse brain in neurons, microglia, and astrocytes. At 3–4 weeks of age, mice with Ostm1 loss showed 3–10-fold stimulation of reactive gliosis, with an increased astrocyte cell population and microglia activation. This inflammatory response was associated with marked retinal photoreceptor degeneration and massive neuronal loss in the brain. Intracellular characterization of neurons revealed abnormal storage of carbohydrates, lipids, and ubiquitinated proteins, combined with marked accumulation of autophagosomes that causes frequent axonal swelling. Stimulation of autophagy was provided by specific markers and by significant down-regulation of the mammalian target of rapamycin signaling, identifying a cellular pathologic mechanism. A series of transgenic mouse lines specifically targeted to distinct central nervous system cell subpopulations determined that Ostm1 has a primary and autonomous role in neuronal homeostasis. Complete functional complementation demonstrated that the development of severe and rapid neurodegeneration in these mice is independent of the hematopoietic lineage and has clinical implications for treatment of osteopetrosis. Importantly, this study establishes a novel neurodegenerative mouse model critical for understanding the multistep pathogenic cascade of cellular autophagy disorders toward therapeutic strategy design.

Loss of Ostm1 leads to the most severe form of osteopetrosis in mice and humans. Because functional rescue of the osteopetrotic defect in these mice extended their lifespan from ϳ3 weeks to 6 weeks, this unraveled a second essential role of Ostm1. We discovered that Ostm1 is highly expressed in the mouse brain in neurons, microglia, and astrocytes. At 3-4 weeks of age, mice with Ostm1 loss showed 3-10-fold stimulation of reactive gliosis, with an increased astrocyte cell population and microglia activation. This inflammatory response was associated with marked retinal photoreceptor degeneration and massive neuronal loss in the brain. Intracellular characterization of neurons revealed abnormal storage of carbohydrates, lipids, and ubiquitinated proteins, combined with marked accumulation of autophagosomes that causes frequent axonal swelling. Stimulation of autophagy was provided by specific markers and by significant down-regulation of the mammalian target of rapamycin signaling, identifying a cellular pathologic mechanism. A series of transgenic mouse lines specifically targeted to distinct central nervous system cell subpopulations determined that Ostm1 has a primary and autonomous role in neuronal homeostasis. Complete functional complementation demonstrated that the development of severe and rapid neurodegeneration in these mice is independent of the hematopoietic lineage and has clinical implications for treatment of osteopetrosis. Importantly, this study establishes a novel neurodegenerative mouse model critical for understanding the multistep pathogenic cascade of cellular autophagy disorders toward therapeutic strategy design.
Among the different human bone diseases, one of the earliest and most debilitating disorders is osteopetrosis. Human osteopetrosis is a genetic disorder that results from hematopoietic cell defects affecting osteoclast lineage differentiation and/or maturation, responsible for bone resorption. Severe recessive forms of osteopetrosis in humans lead to rapid death, in contrast with mild dominant forms. Severe human osteopetrosis is frequently associated with several neuronal complications, anemia, and high susceptibility to infection (1). The neuronal complications have been secondarily ascribed to abnormal bone accumulation in osteopetrosis that compresses cranial nerves, resulting in optic atrophy and blindness as well as hearing impairment (2). However, there is recent evidence for primary central nervous system (CNS) defects in osteopetrotic patients with OSTM1 and ClC7 mutations (3)(4)(5)(6)(7). Of interest, OSTM1 mutations lead to both more severe osteopetrosis and CNS defects (8,9).
The human OSTM1 gene was identified subsequent to characterization of the genetic null mutation in the mouse Ostm1 gene that is responsible for the severe osteopetrotic gl/gl phenotype. These gl/gl mice die within 3 weeks of age with defects in the osteoclast hematopoietic lineage (8,10). The murine Ostm1 gene is expressed in several tissues at variable levels. The Ostm1 expression pattern revealed high levels in hematopoietic lineages, including osteoclast, B-cells, and T-cells (11), as well as in the CNS, particularly the hippocampus and cerebellum (8). Ostm1 was shown to play essential roles in intracellular hematopoietic cross-talk from distinct transgenic mice targeting (11). Ostm1 transgenic expression in hematopoietic lineages, in contrast to the osteoclast lineage, showed full hematopoietic and osteopetrotic complementation in gl/gl mice. Ostm1 was reported to interact with ClC7 (12,13); however, ClC7 function in hematopoietic lineages is restricted to osteoclasts (14). Further, ClC7 systemic or conditional ablation is much less severe than in the gl/gl mice, consistent with human studies (15,16). Besides the critical functions of Ostm1 demon-strated in the hematopoietic lineages, the role of Ostm1 in the CNS independent of osteopetrosis has not been characterized.
To directly address the role of the Ostm1 gene in extrahematopoietic tissues, we investigate the phenotype of the hematopoietic targeted Ostm1 transgenic mice on the gl/gl background (PU.1-Ostm1 gl/gl) (11) that have an extended lifespan but still undergo premature death. Cellular characterization of the CNS in transgenic PU.1-Ostm1 gl/gl mice showed severe inflammatory responses by astrogliosis and microglia activation and showed major neuronal cell death associated with autophagosome accumulation, leading to rapid progression and widespread neurodegeneration. This study also establishes that Ostm1 has a neuronal cell-autonomous role, as determined from two series of specific CNS cell-targeted transgenic mouse lines with functional complementation in the hematopoietic rescued PU.1-Ostm1 gl/gl mice. Therefore, Ostm1 is essential in hematopoietic lineages and, subsequently, solely in neurons for CNS homeostasis throughout the mouse lifespan.

EXPERIMENTAL PROCEDURES
Mice-The mouse strain GL/Le dl J ϩ/ϩ gl was obtained from the Jackson Laboratory (Bar Harbor, ME). Animal use complied with the guidelines of the Canadian Committee for Animal Protection and was approved by the local institutional animal care committee.
Immunohistological Analysis-Anesthetized mice were perfused, and brains were postfixed overnight, followed by paraffin embedding. For cytoarchitectural studies, sections (4 m) were stained by hematoxylin and eosin (H&E), periodic acid-Schiff, or Oil Red O on paraffin-embedded or frozen sections. For immunohistochemistry, sections were subjected to protease antigen retrieval (0.1% trypsin in 0.1% CaCl 2 , 20 mM Tris-HCl, pH 8.0) and peroxidase quenching and immunolabeled with the R.T.U. VECTASTAIN Universal ABC kit using the DAB substrate kit (Vector Laboratories) and counterstained with hematoxylin. For immunofluorescence, brain sections were exposed to trypsin, blocked for 1 h in 10% normal goat serum and 0.1% Triton X-100 in phosphate-buffered saline (PBS), and incubated overnight at 4°C with the primary antibody and then with a goat anti-mouse Alexa-488 and a goat anti-rabbit Alexa 546 IgG (H ϩ L) (Molecular Probes) for 1 h. The following primary antibodies were used: monoclonal mouse anti-NeuN (Chemicon), monoclonal mouse anti-calbindin-D-28K, monoclonal mouse anti-Gfap (Sigma), Isolectin Ib4 Alexa Fluor dye conjugates (Invitrogen), and rabbit polyclonal anti-IbaI (Wako) and rabbit polyclonal anti-ubiquitin (Dako). Apoptosis was analyzed by a TUNEL assay using the FragEL DNA fragmentation detection kit (Calbiochem) according to the manufacturer's recommendations. Sections were counterstained with methyl green and analyzed using Axiophot (Carl Zeiss) and Northern Eclipse acquisition software.
Ultrastructural Analyses-Anesthetized mice were perfused with PBS followed by fixative (1% paraformaldehyde and 2.5% glutaraldehyde), and brain tissue was postfixed for 2 h in the same fixative. Floating sections (50 m) were postfixed in 1% OsO 4 in cacodylate buffer and dehydrated in ethanol series before embedding and polymerization in LR White Resin (Electron Microscopy Sciences). Hippocampal, cerebellar, and cortical brain ultrathin sections were stained on nickel grids with 1% uranyl acetate in 70% ethanol. For immunogold staining, ultrathin cortical brain sections were labeled with an anti-ubiquitin antibody, followed by incubation with an anti-rabbit antibody coupled to 18-nm gold beads (Jackson ImmunoResearch Laboratories) before staining with 1% uranyl acetate and lead citrate. Grids were viewed on a Jeol 1200-Ex transmission electron microscope (Jeol, Inc., Peabody, MA).
Neuronal and Astroglial Cell Quantification-Sagittal cerebral cortex sections (n ϭ 6) immunostained with NeuN antibodies were used to quantify the neuronal population. Gfapimmunostained sections of hippocampus (three fields; magnification, ϫ40) and cortex (six fields; magnification, ϫ40) were used to evaluate astrocyte number and Gfap immunoreactivity using the Color Picker MatLab application (Mathworks) to determine the percentage of Gfapstained pixel number per area. Data from at least three mice at 3 weeks of age and seven mice at 5 weeks of age were analyzed and submitted to statistical analysis. All images were captured via a live video camera (MicroPublisher version 3.3 RTV, QImaging) mounted onto an optic microscope (Axiophot, Carl Zeiss) or with a confocal microscope Axiovert 100M (Carl Zeiss). During this analysis, all parameters, including lamp intensity, video camera setup, and calibration, were maintained constant.
Magnetic Resonance Imaging (MRI)-For MRI structural imaging, each mouse was anesthetized with isoflurane 1-2%, placed in the supine position onto the scanner bed, covered with a warm water-circulating blanket, and loaded into the 7T Bruker Pharmascan MRI scanner. IR-3D Turbo RARE images were acquired for deformation-based morphometric analysis with a total scan time of 50 min. During the entire procedure, animals were under continual anesthesia inhalation and monitored for respiration, cardiac rate, and rectal temperature (SA Instruments Inc.). At the end of image acquisition, each mouse was monitored for full recovery from anesthesia.
Statistical Analysis-Values are expressed as mean Ϯ S.E. Unpaired two-sample Student's t test was used for statistical analysis, with p Ͻ 0.05 considered significant.

RESULTS
Extrahematopoietic Role of Ostm1-To examine the role of Ostm1 in the CNS, the gray-lethal gl/gl mice that develop hematopoietic and osteopetrotic defects were backcrossed to our transgenic mice expressing Ostm1 specifically in the early hematopoietic lineages with PU.1 regulatory sequences. The PU.1-Ostm1 bacterial artificial chromosome transgenic mice on the gl/gl background display full correction of the hematopoietic defects (11). Compared with the spontaneous gl/gl mice (8), the mean lifespan of PU.1-Ostm1 gl/gl mice was extended from 3 weeks to 5.4 weeks (Fig. 1A). This premature death of the PU.1-Ostm1 gl/gl mice revealed an additional role for the Ostm1 gene independent of hematopoiesis.
Because Ostm1 expression is normally highly expressed in wild-type brain but is undetectable in brain from gl/gl homozygous mice (Fig. 1B), we investigated whether Ostm1 plays an essential role in the CNS. Ostm1 subcellular expression was analyzed in microglia, neurons, and astrocytes in wild type and in PU.1-Ostm1 gl/gl mice. Because weak Ostm1 expression was detected in total brain of PU.1-Ostm1 gl/gl in comparison with gl/gl mice (11), we questioned whether the microglia, considered to be brain macrophages that derive from hematopoietic lineages (17), are targeted with the PU.1 regulatory elements. Indeed, using qPCR, we found that the PU.1-Ostm1 transgene was highly expressed in enriched microglial cells compared with wild-type controls, whereas the PU.1 endogenous expression was readily detectable and unaltered (data not shown). Neurons, astrocytes, and microglia, three predominant CNS cell populations, were isolated and enriched ex vivo by cell culture from wild type and gl/gl mice for expression analyses using qPCR. To validate the cell enrichment efficiency, expression of specific neuronal and astrocyte markers was monitored using neuronal nuclei (NeuN), glial fibrillary acidic protein (Gfap), and isolectin Ib4, respectively (data not shown). As shown in Fig. 1B, Ostm1 expression was high in neuronal cells, whereas it was lower in astrocytes and microglia relative to the internal ribosomal S16 control. Therefore, the absence of Ostm1 expression associated with early death of PU.1-Ostm1 gl/gl mice suggested that both cell populations, neurons and astrocytes, could be implicated in the phenotype and were thus characterized further. compared with controls PU.1-Ostm1 ϩ/ϩ (n ϭ 8), wild type ϩ/ϩ (n ϭ 8), and gl/gl (n ϭ 10) mice. The PU.1-Ostm1 gl/gl mice displayed a significant longer lifespan than the gl/gl mice but abrogated. B, histogram of Ostm1 expression analysis by qPCR from total brain and from enriched neuronal, astrocyte, and microglial cell populations of wild type and gl/gl mice. Ostm1 is expressed at high levels in neurons and moderately in astrocytes and microglia but is undetectable in gl/gl tissues or cells. C, analysis of the hippocampus CA3 layer from 3-and 5-week-old (3W and 5W) PU.1-Ostm1 gl/gl mice was performed with the microglial activation marker Ib4 (left panels; scale bar, 100 m) and structural marker Gfap (right panels; scale bar, 50 m) immunostaining. Compared with 3-week PU.1-Ostm1 ϩ/ϩ control mice, important microgliosis and astrogliosis were detected. D, quantification of astrocyte Gfap immunostaining area and cell population in hippocampus was carried out in non-transgenic (NT) wild type (ϩ/ϩ, black bars; n ϭ 3) and mutated (gl/gl, open bars; n ϭ 3) Ostm1 endogenous locus and in transgenic PU.1-Ostm1 with wild type (ϩ/ϩ, black bars; n ϭ 3) or null (gl/gl, open bars; n ϭ 3) endogenous Ostm1 gene, at 3 weeks of age. In addition, 5-week-old transgenic PU.1-Ostm1 with wild type (ϩ/ϩ, black bars; n ϭ 7) or null (gl/gl, open bars; n ϭ 7) Ostm1 gene were analyzed. *, p Ͻ 0.05; ***, p Ͻ 0.001. E, quantification of astrocyte cell population in cerebral cortex was performed in 3-week-old non-transgenic wild type (ϩ/ϩ, black bars; n ϭ 3) and null (gl/gl, open bars; n ϭ 3) Ostm1 endogenous locus and in 3-and 5-week-old transgenic PU.1-Ostm1 with wild type (ϩ/ϩ, black bars; n ϭ 3 and 7) and null (gl/gl, open bars; n ϭ 3 and 7) endogenous Ostm1 gene. *, p Ͻ 0.05; ***, p Ͻ 0.001. Error bars, S.E. MAY 16, 2014 • VOLUME 289 • NUMBER 20 CNS Inflammatory Response in PU.1-Ostm1 gl/gl Mice-To analyze the role of Ostm1 in the brain, the populations of astrocytes and microglia in the mouse brains of PU.1-Ostm1 gl/gl were compared with PU.1-Ostm1 ϩ/ϩ transgenic and/or wild type controls. Because one of the primary and frequent events occurring in abnormal brain phenotypes is an inflammatory response, we assessed the microglia and astrocyte population in these mice beginning at 2 weeks of age because no phenotype was detectable at 1 week. To determine the impact of the gl mutation on microglia, microglial cell activation was monitored using specific Ib4 isolectin immunostaining on brain sections from mice 3 and 5 weeks old. Interestingly, a significant and progressive increase in reactive microglial cell numbers was observed in the hippocampus and cerebral cortex of PU.1-Ostm1 gl/gl mice compared with PU.1-Ostm1 ϩ/ϩ mice (Fig.  1C). In addition to this activation, microglial cells in the hippocampus of PU.1-Ostm1 gl/gl mice displayed positive staining with the Iba1-Ca 2ϩ -binding protein at 5 weeks of age, indicating an active phagocytic state. This microglial physiologic response occurs despite Ostm1 expression, suggesting an effect from either inappropriate Ostm1 levels or a secondary response mechanism. In parallel, widespread astrogliosis was detected in the hippocampus and cerebral cortex from PU.1-Ostm1 gl/gl brain sections, as shown by Gfap immunostaining (Fig. 1C). Astrocyte Gfap immunostaining area in hippocampus was ϳ3and ϳ12-fold higher at 3 and 5 weeks, respectively, when compared with controls (Fig. 1D). The astrocyte cell populations from the hippocampus and cerebral cortex were quantified in PU.1-Ostm1 gl/gl and PU.1-Ostm1 ϩ/ϩ at 3 and 5 weeks, as well as in gl/gl mice at 3 weeks of age. The number of astrocytes in the hippocampus of 3-week-old gl/gl and PU.1-Ostm1 gl/gl was not significantly different from controls, and at 5 weeks, the cell population had increased by ϳ7-fold (Fig. 1D). This indicates that the gl/gl mutation in astrocytes leads to an increase in cell population and more intense immunoreactive signal, in part due to cellular expansion. In the cerebral cortex, the astrocyte cell population was also increased relative to controls by ϳ9-fold for gl/gl and ϳ4-fold for PU.1-Ostm1 gl/gl at 3 weeks of age and ϳ30-fold at 5 weeks of age in PU.1-Ostm1 gl/gl mice (Fig. 1E). The enlarged cellular astrocyte population in the cerebral cortex occurred with a more rapid temporal response than in the hippocampus. Thus, PU.1-Ostm1 gl/gl mice exhibit a rapid onset and progressive development of CNS gliosis, an inflammatory response that could be directly responsible for the phenotype or may result indirectly from neuronal damage.

Ostm1 and Neuronal Homeostasis
Severe and Massive Neurodegeneration in PU.1-Ostm1 gl/gl Transgenic Mice-To decipher whether high Ostm1 expression in neurons has a direct role and indirectly induces a severe inflammatory status, the CNS cytoarchitectural structure was analyzed upon loss of Ostm1 expression in PU.1-Ostm1 gl/gl transgenic mice. Immunostaining of brain sections for the neuron-specific NeuN protein showed mild cell loss in the hippocampus of PU.1-Ostm1 gl/gl mice by 2-3 weeks of age and massive neurodegeneration by 5 weeks of age (5W) (Fig. 2A), whereas gl/gl mice do not display any defect by 2 weeks of age. This defect was widespread, particularly affecting cells of the CA3 and CA2 layers and the structure of the CA1 and GCL cell layers. Similarly, neuronal loss was also detected in four cell layers of the cerebral cortex in PU.1-Ostm1 gl/gl relative to PU.1-Ostm1 ϩ/ϩ control mice (Fig. 2B). Quantification of the cortical neuronal population at 5 weeks of age revealed a substantial 30% reduction in PU.1-Ostm1 gl/gl mice (Fig. 2B). In addition, a major deficit of Purkinje cells in the cerebellum, undetectable at 5 weeks, was readily apparent in 6-week-old (6W) PU.1-Ostm1 gl/gl mice (Fig. 2C), and the cells had virtually disappeared by 7 weeks of age. Analyses of 6-week-old PU.1-Ostm1 gl/gl transgenic mice using MRI showed severe neurodegeneration with thinning of the corpus callosum and enlargement of the lateral ventricles, confirming the reduced cortex thickness and hippocampus degeneration detected by histologic analyses. To determine whether the cell loss occurred via an apoptotic cell death mechanism, TUNEL assays were performed on sections from PU.1-Ostm1 gl/gl and PU.1-Ostm1 ϩ/ϩ control mice at 3-5 weeks of age (data not shown). Interestingly, very few cells displayed TUNEL positivity, indicating that degenerating neurons are not eliminated through apoptosis, as shown for some human neurodegenerative diseases (18).
To determine whether neuronal loss was a general phenomenon, we investigated the retina of PU.1-Ostm1 gl/gl mice. At 3 weeks of age (3W), the gl/gl and the transgenic PU.1-Ostm1 gl/gl mice displayed considerable retinal degeneration with marked loss of photoreceptors, bipolar, amacrine, and horizontal cells (Fig. 2D). In fact, neuronal cell loss in the retina is the first evidence of pronounced cellular degeneration in PU.1-Ostm1 gl/gl mice. Degeneration of all photoreceptor layers causes major retinal atrophy with advancing age in PU.1-Ostm1 gl/gl mice (Fig. 2D). Thus, Ostm1 ablation leads to severe neuronal loss, starting about 3 weeks after birth, that progresses with age and affects all parts of the CNS, indicative of a potential intrinsic role for Ostm1 in neurons and excluding an early developmental defect.
Ostm1 Modulates Neuronal Metabolism-The wide spectrum of neuronal cell loss in different regions of the CNS prompted us to analyze the molecular mechanism responsible for neuronal cell loss. To further define the potential pathological mechanism of PU.1-Ostm1 gl/gl in the CNS, cellular analyses of neurons and astrocytes were performed. Based on different degenerative brain pathologies, such as Parkinson and Huntington disease (19), that display cytosolic accumulation of ubiquitin-positive proteins, we monitored the protein-processing mechanism by assessing free and/or conjugated ubiquitin levels. Cortical sections of PU.1-Ostm1 gl/gl and PU.1-Ostm1 ϩ/ϩ mice were co-immunostained for Gfap and ubiquitin from 3 weeks onward. As shown in Fig. 3A, astrocytes identified by Gfap-positive staining in PU.1-Ostm1 gl/gl sections from 4-week-old mice did not display significant ubiquitin accumulation, suggesting no apparent cellular metabolic defect. In contrast, neurons from brain sections of 3-or 4-week-old PU.1-Ostm1 gl/gl mice co-immunostained with NeuN/ubiquitin and displayed strong cytosolic ubiquitin accumulation compared with controls (Fig. 3A). Abundant ubiquitin accumulation in neurons from the hippocampus and cerebral cortex correlated with major cell degeneration. These analyses provide evidence that loss of Ostm1 results in abnormal protein turnover in neurons.
In an attempt to define the type of neurodegenerative disease caused by Ostm1 among the various subgroups, we questioned whether cellular carbohydrate and lipid metabolism could be altered in addition to the abnormal protein accumulation observed in PU.1-Ostm1 gl/gl. To examine glycogen metabolism in the transgenic PU.1-Ostm1 gl/gl and gl/gl neurons, brain sections from 2-7-week-old mice were stained with periodic acid-Schiff to identify glycol conjugates. At 3 weeks of age (3W), these mice displayed enhanced intracellular storage of glycogen, glycolipids, and/or glycoproteins in pyramidal cells compared with controls (Fig. 3B). Extensive glycogen accumulation was also noted in all layers of the cerebral cortex and hippocampus, particularly in the CA3 layer, suggesting a general and important deregulation of cellular carbohydrate storage. Furthermore, lipid metabolism was also analyzed in PU.1-Ostm1 gl/gl brain sections at 3 weeks of age and onward using Oil Red O, which stains neutral lipids. Both the cerebral cortex and hippocampus showed important neuronal intracellular lipid accumulation by 5 weeks of age in PU.1-Ostm1 gl/gl mice relative to control (Fig. 3C), suggesting also altered lipoprotein accumulation. In addition, extracellular lipid droplet build-up was also present in the CA3 layer of the hippocampus overlying the region where numerous pyramidal neurons are lost (Fig. 3D). Together, enhanced neuronal cytoplasmic storage of ubiquitin, carbohy-  MAY 16, 2014 • VOLUME 289 • NUMBER 20 drates, and lipid conjugates in PU.1-Ostm1 gl/gl mice indicates that Ostm1 plays a major role in regulating neuronal cell metabolism and suggests an impaired autophagy mechanism.

Ostm1 and Neuronal Homeostasis
To further characterize the intracellular impact of Ostm1associated metabolic defects, ultrastructural analyses of brain sections was undertaken in 3-week-old gl/gl and PU.1-Ostm1 gl/gl mice. Unlike neurodegenerative disorders of the lysosomal storage subgroups, single membrane lysosomes in these mice were normal in appearance and numbers. This observation was confirmed by unaltered expression level of the lysosomal trans- membrane structural proteins Lamp1 and Lamp2 and by unnoticeable autofluorescence (data not shown) that differs from neuronal ceroid lipofuscinosis lysosomal storage disorders (20). In contrast, there was an important accumulation of numerous electron-dense inclusions in neuronal perikaryon and cytosol (Fig. 3E). Such accumulation extended into the axons of cortical and hippocampal neurons, frequently causing axonal swelling (Fig. 3E). Importantly, higher magnification of PU.1-Ostm1 gl/gl neuronal sections, similar to gl/gl, revealed that these inclusions consisted predominantly of double membrane vesicular structures typical of autophagosomes (Fig. 3F). With advancing age, autophagosome accumulation was further exacerbated in PU.1-Ostm1 gl/gl mice. To identify whether these neuronal autophagosome inclusions are composed of ubiquitin-targeted substrates, anti-ubiquitin immunogold staining was performed on PU.1-Ostm1 gl/gl sections. Most of the cytoplasmic osmophilic inclusions were ubiquitin-positive, consistent with large protein accumulation in autophagosomes (Fig. 3G). This early response of Ostm1-defective cell metabolism with protein, carbohydrate, and lipid accumulation is concomitant with the formation of autophagosomal structures in neurons, possibly as a protective response that precedes massive neurodegeneration.
Induction of an Enhanced Autophagic Response in the Absence of Ostm1-Based on the abundance of autophagic vesicles in PU.1-Ostm1 gl/gl neuronal cells detected by electron microscopy, stimulation of the autophagy pathway was monitored in the CNS of gl/gl and PU.1-Ostm1 gl/gl mice using specific autophagic molecular markers and signaling effectors. The expression ratio of the autophagosomal lipidated form of LC3 (LC3-II) relative to cytosolic LC3-I was monitored using Western blot analyses of total brain protein extracts. In PU.1-Ostm1 gl/gl lysates, this ratio was significantly increased (by ϳ1.8-fold) compared with controls by 3 weeks of age (Fig. 4A). This result is significant, considering that neurons constitute a minor fraction of the brain cell population, and further shows accumulation of autophagosomes, consistent with our cellular EM analyses (Fig. 3, E-G).  4 -6). C, analysis of brain extracts of 3-week-old PU.1-Ostm1 gl/gl mice (lanes 1-3) displayed significantly decreased expression levels throughout the mTOR signaling cascade, including active phospho-mTOR (P-mTOR), phospho-Akt (p-Akt), and downstream effector phospho-p70S6K (P-p70S6K), in comparison with three PU.1-Ostm1 ϩ/ϩ controls (lanes 4 -6). *, p Ͻ 0.05. Error bars, S.E. MAY 16, 2014 • VOLUME 289 • NUMBER 20

JOURNAL OF BIOLOGICAL CHEMISTRY 13919
To investigate the Ostm1 autophagy signaling cascade in the neurodegeneration of PU.1-Ostm1 gl/gl mice, the two pathways normally associated with autophagy, Beclin 1 and Akt/ mammalian target of rapamycin (mTOR), were analyzed. Although similar expression of Beclin-1 was detected in PU.1-Ostm1 gl/gl total brain protein extracts compared with controls (Fig. 4B), modulation of the mTOR pathway was evidenced by a significant reduction of ϳ30 and ϳ40% in activated phospho-Akt and phospho-mTOR, respectively, in PU.1-Ostm1 gl/gl brain samples (Fig. 4C). This modulation further correlated with an ϳ50% decrease in activation of the mTOR target phospho-p70S6K kinase (Fig. 4C), consistent with observations in Huntington's disease (21). These data indicate overall inhibition of the neuronal mTOR pathway in PU.1-Ostm1 gl/gl mice that results in stimulation of autophagy, consistent with accumulation of autophagosomes and probably ineffective clearance by the unstimulated lysosomal compartment. Therefore, build-up of proteins, carbohydrates, and lipids in Ostm1-null neurons deregulate the autophagy mechanism and actively contribute to neuronal cell loss.
CNS Defects Persist with Ostm1 Astrocyte-specific Expression-Because it has been shown that astrocytes can play a role in neuronal disease progression and/or protection (22,23), we directly addressed whether Ostm1-deficient astrocytes contribute to the PU.1-Ostm1 gl/gl neurodegeneration via a complementation strategy. Three transgenic hGFAP-Ostm1 mouse lines were generated by specifically targeting Ostm1 expression in astrocytes using the human hGFAP promotor (24) (Fig. 5A) and were analyzed for expression. As shown in Fig. 5B, one of the hGFAP-Ostm1 transgenic lines showed enhanced Ostm1 levels in total transgenic brain tissue as well as in transgenic primary astrocyte cells compared with controls (Fig. 5B). Transgenic hGFAP-Ostm1 mice developed normally and did not display any abnormal phenotype. We produced hGFAP-Ostm1-PU.1-Ostm1 gl/gl mice by successive matings to gl/ϩ mice. Analyses of these double transgenics on gl/gl background at 6 -7 weeks of age displayed a major loss of neurons in the CA3 layer of hippocampus, atrophy of the cerebral cortex and corpus callosum, enlargement of the lateral ventricles, and severe astrogliosis (Fig. 5, C and D). In addition, the neuronal autophagic responses determined by the LC3-II/LC3-I and phospho-mTOR/mTOR CNS ratios were significantly increased and decreased, respectively, in these mice and comparable with that detected in PU.1-Ostm1 gl/gl mice (Fig. 5E). The double hGFAP-Ostm1-PU.1-Ostm1 gl/gl transgenic mice had a similar lifespan and phenotype as the PU.1-Ostm1 gl/gl transgenic mice. The Ostm1 astrocyte inflammatory response could correspond to an indirect role in the neurodegeneration process.
Ostm1 Autonomous Role in Neuronal Degeneration-Because neuronal loss in PU.1-Ostm1 gl/gl does not result from a unique defective glial cellular compartment, we investigated whether Ostm1 has an intrinsic role in neurons alone or in both glial and neuronal cells. Using a similar complementation strategy as for astrocytes, we produced transgenic mice that specifically targeted Ostm1 expression in neurons using the rat synapsin 1 (Syn1) gene promotor (25). Two Syn1-Ostm1 transgenic mouse lines (2 and 10 copies of the transgene) (Fig.  6A) were generated and exhibited brain-specific expression of the transgene based on qPCR (Fig. 6B). Because these mice did not display any abnormal phenotype, each mouse line was first crossed to gl/ϩ mice to generate Syn1-Osm1 gl/ϩ mice and subsequently to PU1-Ostm1 gl/ϩ mice to generate double transgenic Syn1-Ostm1-PU.1-Ostm1 gl/gl progeny to test for phenotypic rescue in vivo. The low and high Syn1-Ostm1 expressors were mated to PU.1-Ostm1 lines (11), and each line of double transgenic gl/gl mice showed the same phenotype. In contrast to the severe neurodegeneration detected by MRI in 6-week-old (6W) single PU.1-Ostm1 gl/gl transgenic mice, all double Syn1-Ostm1-PU.1-Ostm1 gl/gl transgenic progeny exhibit a CNS structure indistinguishable from controls (Fig.  6C). The functional rescue also included the hippocampus, cortex, and retina that showed normal distribution of inner and outer photoreceptor segments in double transgenics relative to the severe degeneration observed in single PU.1-Ostm1 gl/gl transgenic mice (Figs. 2 (A-D) and 6E). Complementation of the neuronal autophagic response was also observed, with LC3-II/LC3-I and phospho-mTOR/mTOR ratios reaching basal control levels in double Syn1-Ostm1-PU.1-Ostm1 gl/gl transgenic mice (Fig. 6F). Noticeably, these double transgenic mice had an absence of astrogliosis, indicating that the CNS inflammatory response arises as a secondary mechanism. Consistently, the Syn1-Ostm1-PU.1-Ostm1 gl/gl transgenic mice had normal lifespan expectancy. The single Syn1-Ostm1 gl/gl transgenic mice showed appropriate CNS structural organization but died by ϳ3 weeks of age from hematopoietic defects, similar to the gl/gl mice (Fig. 6D). These results demonstrate full rescue of Ostm1-associated neurodegeneration in double Syn1-Ostm1-PU.1-Ostm1 gl/gl transgenic mice.

DISCUSSION
This study establishes that Ostm1 in the CNS plays an essential role in neurons, as shown by extensive and severe neurodegeneration in null mice. This function of Ostm1 was revealed by the limited increased lifespan upon functional rescue of hematopoietic lineage defects. Ostm1-deficient CNS first developed gliosis with increased astrocyte cell population and microglia activation in parallel with retinal degeneration at 2-3 weeks of age. Within 1 week, the phenotype progressed to a widespread and marked neuronal cell metabolic defect with autophagosome accumulation via the mTOR pathway. Subsequent to autophagy stimulation, these mice demonstrated swift and massive neurodegeneration and brain atrophy that resulted in premature death at a median age of 5.4 weeks. Moreover, our analyses of specific targeted transgenic expression in distinct CNS cell subpopulations determined that Ostm1 has a primary and autonomous role in neuron homeostasis. Thus, the direct role of Ostm1 in CNS neurons correlates with our patient studies and reveals Ostm1 as a potential key effector or modulator in different neuropathologies.
Evidence of histopathologic inflammation with microglia activation and an increased astrocyte population indicates a primary role or an indirect functional cellular physiologic response induced by the loss of Ostm1. The microglia-increased inflammation in the osteopetrosis-rescued PU.1-Ostm1 gl/gl at 3 weeks of age, similar to the gl/gl mice, sug-gested that this process most likely results from a secondary reaction because Ostm1 is expressed in microglia. Prominent stimulation of the astrocyte cell population parallels expansion of hematopoietic macrophages and osteoclast cell populations upon loss of Ostm1 function (11). Accordingly, Ostm1 appears to be a negative modulator of cell proliferation. Such activation of astrocytes could be directly responsible for neuronal death by increased toxicity (26). Alternatively, astrogliosis could be a consequence of neuronal loss. In addition to this inflammation, photoreceptor degeneration in the retina of 3-week-old gl/gl and PU.1-Ostm1 gl/gl mice revealed a genuine role for Ostm1 in the retina that is not secondary to excess bone from gradual occlusion of skull foramina. Fulgurant neuronal loss in the next 2 weeks occurred in all regions of the PU.1-Ostm1 gl/gl brain and resulted in major neurodegeneration with cerebral atrophy. This finding highlights lower susceptibility in the hippocampus, cortex, and cerebellum relative to retina upon absence of Ostm1. This difference suggests the existence of a factor with partially redundant function to Ostm1 in the brain relative to the retina or a greater protective role of astrocytes that delay disease progression in the brain, as in amyotrophic lateral sclerosis (22). Consistently, astrogliosis and CNS inflam-FIGURE 5. Neurodegeneration persists with Ostm1 targeted to astrocytes. A, representation of the transgene used to generate hGFAP-Ostm1 mice. The human GFAP regulatory sequence (hGFAP) from Ϫ2,163 to ϩ47 was cloned upstream of the Ostm1 (1.055 kb) open reading frame, followed by an intron and poly(A) signal of the mouse protamine-1 gene (mP-1). B, expression analysis of the GFAP-Ostm1 transgene (TR) relative to non-transgenic (NT) in total brain and liver as tissue control (left) and enriched primary astrocytes (right). Expression is relative to S16 as internal control. C, analysis of cerebral cortex from double hGFAP-Ostm1-PU.1-Ostm1 gl/gl transgenic mice (n ϭ 6) relative to hGFAP-Ostm1 ϩ/ϩ controls by Gfap immunostaining (left panels) exhibits significant stimulation of astrogliosis (ϳ7-fold; right). Scale bar, 50 m. D, severe neurodegeneration in 7-week-old double hGFAP-Ostm1-PU.1-Ostm1 gl/gl transgenic mice (right) compared with single hGFAP-Ostm1 ϩ/ϩ transgenic control (left). The hGFAP-Ostm1-PU.1-Ostm1 gl/gl sections displayed loss of CA3 hippocampus cell layer (top panels, H&E staining; scale bar, 1 mm) and decreased cerebral cortex thickness by NeuN immunostaining (bottom panels; scale bar, 50 m). E, Western blots and histograms of the LC3-II/LC3-I and phospho-mTOR (P-mTOR)/mTOR ratios from brain extracts of GFAP-Ostm1-PU.1-Ostm1 ϩ/ϩ (lanes 1-3), GFAP-Ostm1gl/gl (lanes 4 -6), and PU.1-Ostm1 gl/gl mice (lanes 7-9). From Western blot quantification, the GFAP-Ostm1 gl/gl and PU.I-Ostm1 gl/gl mice show a similar increase in the LC3-II/LC3-I and decrease in phospho-mTOR/mTOR ratios, compared with GFAP-Ostm1-PU.1-Ostm1 ϩ/ϩ controls. **, p Ͻ 0.01. Error bars, S.E. matory response prior to acute neuronal loss is frequent in several human neurodegenerative diseases (27) but could also be responsible for the phenotype.
Based on molecular and cellular characteristics, the absence of Ostm1 resulted in a distinct type of brain disorder. Accumulation of cellular ubiquitin in the neurons specifically of PU.1-Ostm1 gl/gl mice suggested protein misfolding or misprocessing. Such mechanisms occur in many types of human neurodegeneration, such as mucolipidosis and Huntington, Parkinson, and Alzheimer diseases (20). Altered cell metabolism in Ostm1 null neurons, with intracellular storage of ubiquitinated proteins, carbohydrates, and lipids from most regions of the brain, was concomitant with the presence of numerous electron-dense, double-membrane vesicular structures or autophagosomes. This autophagic response in Ostm1-deficient brains correlated with increased LC3II/LC3I and decreased phospho-mTOR/mTOR ratios. The accumulation of autophagosomes was so severe that it resulted in axonal swelling. Similar features with impaired vesicular trafficking were characterized in the dynactin/dynein motor mouse mutants and in Alzheimer disease mouse models (28 -31). Evidence for stimulation of autophagy is provided by significant down-regulation of the mTOR pathway. Consistently, both the downstream target phospho-p70S6K and phospho-Akt are also repressed in Ostm1-deficient brains (32). Of interest, the accumulation of autophagosomes was most likely exacerbated by the unstimulated lysosomal compartment, determined from the unchanged transmembrane Lamp1 and Lamp2 protein levels. This supports a potential role for Ostm1 upstream of the lysosomal compartment, such as endolysosomal transport on microtubules or in the fusion of autophagosome with lysosomes to form autophagolysosomes (33). Nonetheless, the accumulation of autophagosomes in the absence of Ostm1 appears to result from both enhanced autophagy and unstimulated lysosomal degradation process. Hence, this impaired physiologic sequence in the autophagy to lysosomal mechanism leads to massive CNS neuronal loss (Fig. 6F).
Neurodegeneration arises in a number of mouse mutants. The Ostm1 mouse mutant has some similarities with the autophagy gene Atg5 and Atg7 null mice (34,35), the lysosomal enzyme Cln3 ⌬ex7/8 (36) and cathepsin D null mice (37), and the chloride channel ClC7 null mice (14), all of which have an accumulation of toxic metabolites. The Cln3 ⌬ex7/8 and cathepsin D null mice like PU.1-Ostm1 gl/gl showed activation of the mTOR-dependent and Beclin-1-independent autophagic signaling pathways but with evidence of autofluorescence as in a lysosomal storage disease. ClC7 null neurons display increased carbohydrates in lysosomes (38,39) with pronounced autofluo-rescence, a hallmark of neuronal ceroid lipofuscinosis (40). In Ostm1 null neurons, accumulation of lipids and carbohydrates is, however, detected in autophagosomes without lysosomal autofluorescence that differs from the ClC7 typical lysosomal storage disease. Further, the neurodegeneration in the PU.1-Ostm1 gl/gl transgenic mouse model is much more severe and leads to a shorter lifespan than the ClC7 null and forebrainconditional mice (16) or even the Cln3 ⌬ex7/8 and cathepsin D null mice. The earlier onset and more severe progression of Ostm1 phenotype in comparison with CLC7 suggest additional functions of Ostm1 and/or the existence of other interacting partners.
The implication of microglia, astrocytes, and neurons in Ostm1 brain pathology indicates that Ostm1 could be important in only one cell type or in a combination of these cell types, as observed for multiple hematopoietic lineages (11). Given that specific microglial Ostm1 transgene expression in null Ostm1 mice did not improve the CNS pathology, the phenotype is either a secondary event or involves Ostm1 in both microglia and another cell type. Analyses of the role of astrocytes revealed that astrocyte-targeted Ostm1 expression in Ostm1-deficient brain was insufficient by itself to rescue the phenotype or delay neuronal death. This result supports two possibilities: 1) Ostm1 could play a direct role in astrocytes conjointly with an additional cell type, or 2) astrogliosis is in response to neuronal cell death for functional protection. The critical role of Ostm1 in neurons was highlighted by complete rescue of the CNS defects in Ostm1-deficient brain with specific neuron-targeted expression of Ostm1. Full correction of the phenotype, including astrogliosis and microgliosis, demonstrates that Ostm1 plays a cell-autonomous role in neurons and excludes a primary role for Ostm1 in astrocytes as responsible for neurodegeneration. Thus, Ostm1 appears to directly target the neurons rather than a combination of cells in the CNS.
Our characterization of severe and extensive neurodegeneration in Ostm1-deficient brains independently of hematopoietic defects identified an essential factor in neuronal homeostasis and physiology. Ostm1 plays a key function in the intracellular trafficking mechanism of organelles in the late endosome, autophagosome, and lysosome cascade. Importantly, we demonstrated a neuronal cell-autonomous role for Ostm1 in CNS pathology that indirectly induced axonal swelling and widespread inflammatory gliosis. In summary, these mice provide a highly relevant model to understand and alleviate the molecular pathologic mechanism of this disorder and potentially neurodegeneration-related diseases.