Genetic Ablation of PLA2G6 in Mice Leads to Cerebellar Atrophy Characterized by Purkinje Cell Loss and Glial Cell Activation

Infantile neuroaxonal dystrophy (INAD) is a progressive, autosomal recessive neurodegenerative disease characterized by axonal dystrophy, abnormal iron deposition and cerebellar atrophy. This disease was recently mapped to PLA2G6, which encodes group VI Ca2+-independent phospholipase A2 (iPLA2 or iPLA2β). Here we show that genetic ablation of PLA2G6 in mice (iPLA2β-/-) leads to the development of cerebellar atrophy by the age of 13 months. Atrophied cerebella exhibited significant loss of Purkinje cells, as well as reactive astrogliosis, the activation of microglial cells, and the pronounced up-regulation of the pro-inflammatory cytokines tumor necrosis factor-α (TNF-α) and interleukin-1β (IL-1β). Moreover, glial cell activation and the elevation in TNF-α and IL-1β expression occurred before apparent cerebellar atrophy. Our findings indicate that the absence of PLA2G6 causes neuroinflammation and Purkinje cell loss and ultimately leads to cerebellar atrophy. Our study suggests that iPLA2β-/- mice are a valuable model for cerebellar atrophy in INAD and that early anti-inflammatory therapy may help slow the progression of cerebellar atrophy in this deadly neurodegenerative disease.


Introduction
The phospholipase A 2 (PLA 2 ) family is a diverse group of enzymes that catalyzes the hydrolysis of the sn-2 fatty acyl bond of phospholipids to liberate free fatty acids and lysophospholipids [1]. Among the family members, group VIA PLA 2 (iPLA 2 b) has unique structural features, including eight N-terminal ankyrin repeats, caspase-3 cleavage sites, an ATP-binding domain, a serine lipase consensus sequence (GXSXG), a bipartite nuclear localization sequence, and a C-terminal calmodulin-binding domain [1]. The human iPLA 2 b gene, PLA2G6, maps to chromosome 22q13.1 and encodes several isoforms [2].
Morgan et al. [3] reported that mutations in PLA2G6 underlie human neurodegenerative disorders, infantile neuroaxonal dystrophy (INAD) and neurodegeneration with brain iron accumulation (NBIA). INAD is a neurodegenerative disease with infantile onset and death as a teenager or in early adulthood. It is characterized by pathologic axonal swelling and spheroid bodies in the peripheral and central nervous systems (CNS) [4,5,6]. Mutations in PLA2G6 have recently been identified in many INAD patients, suggesting that iPLA 2 b plays an essential role in the CNS and the development of the disorder.
We used homologous recombination to generate iPLA 2 b-knockout mice (iPLA 2 b -/-) [7]. These mice developed severe motor dysfunction associated with the prominent formation of spheroids with tubulovesicular membranes remarkably similar to those seen in human INAD [8,9]. Consistent with the recessive nature of INAD, mutagenesis analysis demonstrated that the catalytic function of the INADassociated mutated PLA2G6 is impaired [10]. Thus, iPLA 2 b -/mouse is a useful animal model of the pathogenesis of INAD.
The neuroradiological hallmark of INAD is the cerebellar atrophy. Marked cerebellar atrophy is the most characteristic MRI feature of INAD patients [11,12,13,14] and observed in almost all INAD patients with mutations in PLA2G6 [3,4,5,6,15,16,17,18], suggesting a neuropathologic role of dysfunctional iPLA 2 b in the cerebellar atrophy in the INAD patients. Here, we report that iPLA 2 b -/mice develop severe cerebellar atrophy -characterized by the loss of Purkinje neurons, the activation of glial cells, and the elevation of pro-inflammation cytokines -by the age of 13 months. Our findings provide insight into the pathological mechanisms of cerebellar atrophy in human INAD.

Results
Cerebellar atrophy and Purkinje cell loss in iPLA 2 b -/mice Cerebellar ataxia is an early symptom of INAD patients with PLA2G6 mutations [6]. We found that iPLA 2 b -/mice exhibited ataxia and irregular strides by the age of 13 months, and that these conditions became more severe with age ( Fig 1A). Furthermore, the cerebella of the iPLA 2 b -/mice were noticeably smaller than that of the age-matched WT littermates (Fig. 1B) and they weighed nearly 25 percent less (Fig 1C). These findings indicated that iPLA 2 b -/mice develop cerebellar atrophy by the age of 13 months that resembles the atrophy seen in humans with INAD.
We next analyzed cerebellar sections from the iPLA 2 b -/mice to determine the mechanism of cerebellar atrophy. In agreement with a previous report [8], parasagittal views of nissl-stained mouse cerebella revealed no apparent anatomical differences between the iPLA 2 b -/mice and their WT littermates ( Figure S1). However, when we examined the expression of calbindin, a widely used immunohistochemical marker of Purkinje neurons in the cerebellum [19], we found that, in comparison to the WT controls ( Fig. 2A-B), the Purkinje cell layer and the molecular layer in the iPLA 2 b -/mice are disrupted (Fig. 2C-F). By stereological counting of calbindin-positive cells ( Figure S2), we found that iPLA 2 b -/mice have significantly fewer Purkinje cells that the WT controls (Fig. 2J). Our data suggest that Purkinje cell loss is a pathologic characteristic of cerebellar atrophy in iPLA 2 b -/mice.
Glial cell activation in the cerebella of iPLA 2 b -/mice Cerebellar atrophy observed in INAD patients is often associated with cerebellar gliosis, which is characterized by signal hyperintensity in the cerebellar cortex on T2-wighted images of MRI [6]. Since reactive astrogliosis has long been recognized as a ubiquitous feature of CNS pathologies [20], we determined whether reactive astrogliosis occurs in the cerebella of the iPLA 2 b -/mice by examining the expression of glial fibrillary acidic protein (GFAP), of which up-regulation is the best known hallmark of reactive astrogliosis in CNS [20]. We found that, with the exception of the white matter, GFAP levels were very low in the cerebella of the WT mice (Fig. 3A). However, GFAP levels were markedly increased in the molecular and granule cell layers of the cerebella of the iPLA 2 b -/mice ( Fig. 3B and 3C). In consistent with the immunostaining, Western blot analysis ( Fig 3D) and real time quantitative RT-PCR (Fig. 3E) also showed the significant increase in GFAP expression in the cerebella of the iPLA 2 b -/mice. This pronounced increase of GFAP in the molecular and granule cell layers of the cerebella of the iPLA 2 b -/mice indicate that reactive astrogliosis is a characteristic feature of cerebellar atrophy in iPLA 2 b -/mice.
Microglial activation plays a central role in reactive astrogliosis [21]. To determine whether microglial activation also occurs in the cerebella in the iPLA 2 b -/mice, we examined the expression of ionized calcium binding adaptor molecule 1 (Iba-1) -a 17-kDa EF hand protein that is specifically expressed in brain microglia and is up-regulated during their activation [22]. Iba-1 immunostaining revealed ramified microglia with small cellular bodies in the molecular layer and granule cell layer (

Elevation of proinflammatory cytokines
Activation of microglia often results in the release of proinflammatory cytokines such as TNF-a and IL-1b [24,25]. We found that although the levels of TNF-a and IL-1b in the serum of iPLA 2 b -/and WT mice do not differ ( Fig. 5A-B), the cerebella of iPLA 2 -/mice expressed significantly higher levels of these cytokines ( Fig. 5C-D) and their mRNA transcripts ( Fig. 5E-F). Our findings demonstrate that Purkinje cell degeneration is tightly associated with inflammation in the cerebella of the iPLA 2 b -/mice.
Glial cell activation and TNF-a and IL-1b production without apparent cerebellar atrophy Cerebellar atrophy in the iPLA 2 b -/mice does not become apparent until the age of 13 months. However, we found that microglia (Fig. 6B) and astrocytes (Fig. 6D) are activated in the cerebella of iPLA 2 b -/mice as young as 10 months (Fig. 6A, 6C). These mice also exhibit significantly increased levels of IL-1b (Fig. 6E) and TNF-a ( Fig. 6F) in their cerebella, which further increase with age. These elevated cytokine levels can be attributed to local protein production, as demonstrated by elevated mRNA levels of IL-1b (Fig. 6G) and TNF-a ( Fig. 6H) in the iPLA 2 b -/mice. Since we observed no significantly difference in the Purkinje cell layer between the WT and iPLA 2 b -/mice at this age, our findings suggest that inflammation in the cerebella of iPLA 2 b -/-  mice may be one contributing factor leading to Purkinje cell loss and subsequent cerebellar atrophy.

Discussion
In this study, we used a mouse model of INAD [8,9] to investigate potential mechanisms of cerebellar atrophy -a hallmark of INAD patients with PLA2G6 mutations [3,4,5,6,15,16,17,18]. We found that, by the age of 13 months, iPLA 2 b -/mice exhibited a loss of Purkinje cells and their dendrites in the molecular layer of the cerebella. Our results indicate that Purkinje cell loss is a pathologic characteristic of cerebellar atrophy in iPLA 2 b -/mice.
An important finding in our study is that the cerebellar atrophy observed in iPLA 2 b -/mice is associated with reactive astrogliosis, overactivation of microglial cells, and pronounced upregulation of pro-inflammatory cytokines including IL-1b and TNFa -all features of neuroinflammation associated with a vast array of CNS pathologies [26]. Importantly, these features were observed not only in the cerebella of 13-month-old iPLA 2 b -/mice with apparent neurological abnormality, but also in younger mice prior to the onset of overt disease. Our findings suggest that inflammation may play a pathological role in the development of cerebellar atrophy in INAD.
Microglia typically exist in a resting state characterized by ramified morphology (Fig. 3C) monitoring the brain environment and they are usually the first cells to be activated by proinflammatory stimuli, such as lipopolysaccharide (LPS), or in response to neuronal damage [24]. In their activated state, they can serve diverse beneficial functions essential to neuron survival including cellular maintenance, innate immunity, and regulating brain development by enforcing the programmed elimination of neural cells [24]. Indeed, activation of microglia has been reported to promote the death of developing Purkinje cells during brain development [27]. However, under certain circumstances, microglia become overactivated and release excess TNF-a and IL-1b, which further activate astrocytes and spark a perpetuating cycle of neuron death [21].
What could cause the overactivation of microglia in the cerebella of iPLA 2 b -/mice? It is known that the oxidation of mitochondrial phospholipids, particularly cardiolipin, sets cytochrome c free and release to trigger apoptosis [28]. Mitochon- dria are the primary source of ROS in the CNS [29] and the inner mitochondrial membranes are particularly susceptible to ROS attack because they contain a high proportion of cardiolipin, which is very rich in polyunsaturated fatty acids [30,31]. Accumulating evidence suggests that iPLA 2 b-mediated deacylation of phospholipid is a critical step in maintaining optimal mitochondrial membrane composition by participating in cardiolipin remodeling and in repairing of oxidative modifications caused by mitochondrial ROS production [32,33,34,35,36,37,38,39]. When iPLA 2 b is dysfunctional or missing, Purkinje cells in the cerebellum may be sensitized to apoptotic induction.
Microglia are phagocytes of myeloid origin that function as the principal immune effectors in the CNS [40]. They work to promptly clear apoptotic neuronal cells and prevent the secondary necrosis and inflammation that can further damage neighboring neural cells [41]. Previous studies suggest iPLA 2 b can be irreversibly activated by caspase-3 activation during apoptosis [1,42,43] and this activation produces the chemoattractant lysophosphatidylcholine to recruit professional phagocytes to engulf the apoptotic cells [43]. Effective clearance of apoptotic cells is an important physiologic homeostatic mechanism that is required to prevent secondary necrosis and inflammation [44,45]. In the absence of iPLA 2 b, apoptotic Purkinje cells may not be promptly cleared, which can cause overactivation of microglia [24]. These overactive microglia can produce highly neurotoxic factors, including superoxide, TNF-a, and IL-1b, which are likely to perpetuate neuronal death and lead to overt cerebellar atrophy in iPLA 2 b -/mice.
In summary, Purkinje cell loss is a pathological characteristic of cerebellar atrophy of the iPLA 2 b -/mice. The loss of functional iPLA 2 b in vivo may result in the increase in apoptosis and the decrease in apoptotic cell clearance, which instigate inflammation. The presence of reactive astrogliosis, the activation of microglial cells, and the up-regulation of the proinflammatory cytokines TNF-a and IL-1b in the cerebella of the iPLA 2 b -/mice suggest that inflammation occurs and inflammation-mediated neurotoxicity may play a role in the pathogenesis of cerebellar atrophy in INAD. These findings suggest that in vivo microglial imaging by positron emission tomography (PET) may allow the assessment of disease progression [24] and that early intervention with antiinflammatory therapy may help slow, if not prevent or reverse, the cerebellar atrophy in INAD.

Materials and Methods
Animals iPLA 2 b -/mice (C57/BL6 background) and their wild type (WT) littermates were obtained by mating heterozygote iPLA 2 b +/mice. Mice were housed in a temperature-and humidity-controlled facility with a 12-hour light/dark cycle with standard chow diet and water ad libitum. All animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the Mount Sinai School of Medicine.

Gait stride test
Gait stride test was performed as described previously [46]. Briefly, mice in each group were given three training runs each day for three days, assisted by the home cage on the first day and unassisted on the next two days. On the fourth day (test), the mouse's paws were coated with ink (red ink on the hindpaws and blue on the forepaws), and paw prints were recorded on two consecutive runs. Records that displayed the clearest prints and the most consistent gait for five consecutive strides were chosen for analysis.

Cerebellar measurements
iPLA 2 b -/mice and their age-matched WT littermates (n = 15/ group) were deeply anesthetized by intraperitoneal injection of a mixture of 100 mg/kg of Ketamine and 20 mg/kg of Xylazine. Cerebella were carefully collected and weighted.

Cerebellum preparation and immunohistochemistry
Mice were anesthetized by intraperitoneal injection of a mixture of 100 mg/kg of Ketamine and 20 mg/kg of Xylazine and  perfused with 4% paraformaldehyde via puncture of the left ventricle. The cerebella were removed and further fixed in 4% paraformaldehyde at 4uC overnight, and then cryopreserved in 30% sucrose.

Purkinje cell and microglia counts
Purkinje cells in the cerebella were analyzed as described previously [47]. Briefly, Purkinje cells were quantified by stereologically counting calbindin positive-cell bodies in 106 images from six entire cerebellar sections of each animal (30 mm thick and 180 mm apart) were analyzed (control, n = 6; iPLA 2 b -/-, n = 8). Microglial cells in the molecular layer of the cerebellum were quantified by counting Iba-1-positive cells.

Quantitative Real-time RT-PCR analysis
Total RNA was extracted from the cerebella collected from the iPLA 2 b -/and age-matched WT mice using an RNeasy Mini Kit. according to the manufacturer's instructions on 7900HT Real-Time PCR Systems (Applied Biosystems, Foster City, CA). The mRNA levels were calculated by comparative CT methods (X Test / X GAPDH = 2 DDC T ) with GAPDH as the endogenous reference gene.

IL-1b and TNF-a measurements
The levels of cytokines in cerebellum homogenates were measured as described previously with a modification [48]. Briefly, mouse blood was collected before sacrificing. The mice were anesthetized by intraperitoneal injection of a mixture of 100 mg/ kg of Ketamine and 20 mg/kg of Xylazine. Then cerebella were removed, snap frozen in liquid nitrogen, and stored at 280uC until use. Cerebellar samples were placed in sterile PBS containing protease inhibitor cocktail (300 ml solution per 10 mg), homogenized using Tissuemiser (Fisher Scientific, Pittsburgh, PA), and centrifuged at 11,000 rpm at 4uC for 20 minutes. The supernatants or blood were aliquoted and cytokine levels were determined with mouse TNF-a and IL-1b ELISA Kits (eBioscience, Inc., San Diago, CA). Cytokine levels were normalized with total proteins and expressed as ng/mg proteins.

Statistics
Data were expressed as means 6 SEM. Significant differences were evaluated by unpaired two-tailed Student's t test by using PRISM. Significance levels are described in individual figure legends. Figure S1 The nissle staining analysis of mouse cerebellum. Nissle staining was performed according to standard methods. Briefly, the sagittal sections of cerebella from 13-monthold iPLA 2 b -/mice and the age-matched wild type littermates (WT) were mounted on slides, dehydrated and rehydrated in graded ethanols and xylenes, incubated in 1% cresyl violet for 30 sec, decolorized in acetic acid, and then dehydrated and coverslipped and analyzed by a a Zeiss Axioplan 2 microscope. The parasagittal views of the mouse cerebella of the nissl staining revealed no apparent anatomical differences between the iPLA 2 b -/and their wild type littermates with the exception of the less intensive staining in the granule cell layer of cerebella in iPLA 2 b -/mice. ml, molecular layer; gcl, granule cell layer; DCN, deep cerebellar nucleus. Scale bar, 100 mm. (TIF) Figure S2 The Calbindin staining analysis of mouse cerebellum. The cerebellum was cut into 20 mm sections by Leica Cryomicrotome (Wetzlar, Germany) and standard immunostaining was performed by rat anti-GFAP ( 1:1000; Invitrogen, Carlsbad, CA), rabbit polyclonal anti-calbindin-2 (1:100; abcam, Cambridge, MA), Goat anti-rat IgG with the Alexa FluorH dyes (Alexa FluorH 488, Invitrogen, Carlsbad, CA) were used for the secondary antibodies. Calbindin staining (green) showed the Purkinje cell lose and degeneration of PC dendrites in iPLA 2 b -/mice cerebella. ml, molecular layer; gcl, granule cell layer; DCN, deep cerebellar nucleus. Scale bar, 100 mm.