The cellular prion protein, a major player in the neuropathology of prion diseases, is believed to control both death and survival pathways in central neurons. However, the cellular and molecular mechanisms underlying these functions remain to be deciphered. This chapter presents cytopathological studies of the neurotoxic effects of infectious prions and cellular prion protein-deficiency on cerebellar neurons in wild-type and transgenic mice. The immunochemical and electron microscopy data collected in situ and ex vivo in cultured organotypic cerebellar slices indicate that an interplay between apoptotic and autophagic pathways is involved in neuronal death induced either by the infectious prions or by prion protein-deficiency.
- prion protein
1.1 Prion diseases
Transmissible spongiform encephalopathies (TSEs) or “prion diseases” are fatal neurodegenerative disorders in humans (Creutzfeldt-Jakob disease (CJD), Gerstmann-Straüssler-Scheinker syndrome (GSS), variant CJD (vCJD), fatal familial insomnia (FFI) and kuru) and in animals (bovine spongiform encephalopathy (BSE), transmissible mink encephalopathy (TME), chronic wasting disease of cervids (CWD), camel prion disease (CPD), and scrapie of sheep and goats) [1, 2, 3, 4]. Prevailing over a viral etiology, the conformational corruption of host-encoded cellular prion protein (PrPc) by a pathogenic isoform (PrPTSE) is now widely accepted as underlying prion transmission and pathogenesis in TSEs [5, 6, 7].
1.2 PrPc functions
Prnp-knockout mice were generated in order to investigate the physiological functions of PrPc. In either mixed C57BL/6 j x129/Sv(ev) (Zurich I, ZrchI, Prnp ZH1/ZH1, ) or pure 129/Ola (Npu, Edinburgh, Edbg, ) or C57BL/6J (Zurich III, ZrchIII, Prnp ZH3/ZH3, ) genetic backgrounds, the first Prnp null mouse strains produced were viable with no clear abnormality except for their resistance to prion infection  and absence of obvious neurodegeneration. Similar absence of neurodegeneration or histopathology resulted from depletion of neuronal PrPc in adult conditional Prnp-knockout NFH-Cre/tg37 mice . Thus, a physiological function of PrPc that is essential for life seemed to be ruled out unless it is highly redundant or is compensated. Nevertheless, looking at different neuronal and other cell functions in PrPc-ablated mice has revealed a number of differences that can be attributed to the physiological functions of PrPc (see  for review).
PrPc has been implicated in neurotransmission, olfaction, proliferation and differentiation of neural precursor cells, neuritic growth, neuronal homeostasis, cell signaling, cell adhesion, myelin maintenance, copper and zinc transport, as well as neuroprotection against toxic insults, such as oxidative stress and excitotoxicity (see [14, 15] for reviews). Increasing evidence links prion protein misfolding and accumulation to neurodegeneration in prion diseases. Accordingly, several nonexclusive mechanisms of prion-mediated neurotoxicity are currently under investigation (see  for review). PrPc has been localized in three major sites: enriched in lipid rafts, anchored in the outer plasma membrane leaflet by its GPI tail , and intracellularly in the Golgi apparatus early and late endosomes [18, 19]. Since lipid rafts are pivotal microdomains for signal transduction, PrPc is likely triggering intracellular signaling pathways [20, 21]. The first evidence that PrPc might mediate extracellular signals was the caveolin-1-dependent coupling of PrPc to the tyrosine-protein kinase Fyn . From this pioneering work, accumulating data suggested that PrPc functions as a “dynamic cell surface platform for the assembly of signaling molecules,” partnering with other membrane proteins to transduce cellular signaling .
1.3 Synaptic PrPc
Whereas, PrPc is highly expressed in both neurons and glial cells of the CNS [19, 23, 24], it is preferentially localized in the pre- and postsynaptic terminals of neurons [19, 24, 25]. Immunocytochemical studies of primate and rodent brains [25, 26] including an EGFP-tagged PrPc in transgenic mice, showed that PrPc is enriched along axons and presynaptic terminals [27, 28, 29], and undergoes anterograde and retrograde transport [30, 31]. Such a synaptic targeting of PrPc suggests that it could be involved in preserving synaptic structure and function. Indeed, synaptic dysfunction and loss are early prominent events in prion diseases [32, 33]. However, a functional role of PrPc at synapses is not consistently supported by functional data and still remains contentious.
Insights into possible mechanisms by which PrPc modulates synaptic mechanisms and neuronal excitability at a molecular level have been provided by the documented interactions of PrPc with several ion channels including the voltage-gated calcium channels (VGCCs) , the N-methyl-D-aspartate glutamate receptors (NMDARs)  and the voltage-gated potassium channels Kv4.2 . PrPc has been shown to regulate NMDARs due to its affinity for copper that leads to inhibition of glutamate receptors and excitotoxicity [37, 38]. While interaction of PrPc with these channels may account for some of its functions, a toxic response can also be activated when PrPc misfolds. A structural change in cell surface PrPc has been proposed to simultaneously disrupt NMDAR function and plasma membrane permeability, leading to dysregulation of ion homeostasis and neuronal death [39, 40]. PrPc can also interact with kainate receptor subunits GluR6/7 , α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors subunits GluA1 and GluA2 [42, 43], and metabotropic glutamate receptors of group 1 mGluR1 and mGluR5 [44, 45]. PrPc can interact with the β-amyloid peptide (Aβ) and the later [45, 46] is believed to underlie the Aβ oligomer-induced disruption of LTP in Alzheimer’s disease . Thus, PrPc seems to behave as a cell surface receptor for synaptic oligomers of the Aβ peptide and, of other β-sheet-rich neurotoxic proteins .
1.4 PrPTSE-related neurotoxicity in prion diseases
The histopathological signature of TSEs notably relies on the aggregation of PrPTSE, vacuolation of the brain tissue, astrogliosis, and synaptic and neuronal loss. How neurons, the major targets of prions, die, remains a central question in prion diseases. The absence of neurodegenerative phenotypes after depletion of PrPc suggests that neurotoxicity is not due to a loss of PrPc function but rather results from a gain of toxicity upon its conversion to PrPTSE, which then acts on the central nervous system (CNS) . Although PrPc is required for propagation of infectious prions and PrPTSE-mediated toxicity , the mechanisms by which prions are lethal for neurons remain mostly unknown. Nevertheless, the endogenous PrPc conversion has been shown to cause neuronal dysfunction and death, rather than PrPTSE itself which does not seem to be directly neurotoxic. A precise understanding of the factors leading to neurotoxicity in prion infections is crucial to developing targeted therapies and investigating the role of PrPc in neurons should provide insight.
The conformational conversion of PrPc begins on the neuronal surface, where PrPc interacts with exogenous PrPTSE, and then proceeds within endogenous compartments suggesting that neurotoxicity may be triggered by PrPc misfolding both at the cell surface and inside the cell. In both acquired and genetic prion diseases, intracellular PrPc misfolding would ultimately alter synaptic proteostasis, either through an indirect unfolded protein response (UPR)-mediated mechanism , likely arising either from an impairment of the neuronal ubiquitin-proteasome system (UPS) , or a direct interference with secretory trafficking of PrPc-interacting cargoes . Common features associated with prion infections include Ca2+ dysregulation, release of reactive oxygen species, and induction of endoplasmic-reticulum (ER) stress, which has been recently suggested as an important player in pathogenesis . Prion-infected mice show brisk activation of the UPR and specifically of the PERK pathway, resulting in eIF2α phosphorylation and suppression of translational initiation. PERK inhibition protects mice from prion neurotoxicity, confirming an important pathogenic role of ER stress . Since UPR activation and/or increased eIF2α-P levels as well as UPS impairment are commonly seen in prion disorders and in Alzheimer’s and Parkinson’s diseases, translational control, and UPS stimulation strategies may offer a common therapeutic opportunity to prevent synaptic failure and neuronal loss in protein misfolding diseases [51, 54].
1.5 Loss of PrPc anti-inflammatory protective function in prion disease
A protective role of PrPc against a noxious insult mediated by the pro-inflammatory cytokine tumor necrosis factor-α (TNFα) has recently been demonstrated . The α-secretase activity mediated by the TNFα-converting enzyme (TACE) was impaired at the surface of Fukuoka and 22L scrapie prion-infected neurons. Furthermore, the activity of 3-phosphoinositide-dependent kinase-1 (PDK1) which inactivates phosphorylation and caveolin-1-mediated internalization of TACE is increased in scrapie-infected neurons. PDK1 was shown to be controlled by RhoA-associated coiled-coil containing kinases (ROCK) which favored the PrPTSE production. In these neurons, exacerbated ROCK activity overstimulated PDK1 activity which canceled the neuroprotective α-cleavage of PrPc by TACE α-secretase, physiologically precluding PrPTSE production. Inhibition of ROCK lowered PrPTSE in prion-infected cells as well as in the brain of prion-diseased mice which had extended lifespans . Indeed, the dysregulation of TACE resulted in PrPTSE accumulation and reduced the shedding of TNFα receptor type 1 (TNFR1) from the neuronal plasma membrane. Inversely, inhibition of PDK1 in vitro promoted TACE localization at the plasma membrane, restoring TACE-dependent α-secretase activity and shedding of PrPc and TNFR1, thereby attenuating PrPTSE-induced neurotoxicity. Similarly, inhibition or siRNA-mediated silencing of PDK1 extended survival and reduced motor impairment of scrapie-diseased mice . Mechanistically, PrPc coupling to the NADPH oxidase-TACE α-secretase signaling pathway limits the sensitivity of recipient cells to TNFα by promoting TACE-mediated cleavage of TNFα receptors (TNFRs) and the release of soluble TNFRs. PrPc expression was further shown to be necessary for maintaining TACE α-secretase at the plasma membrane and its TNFR shedding activity. The loss of PrPc provoked TACE internalization, canceling TACE-mediated cleavage of TNFR. This rendered PrPc-depleted cells and Prnp-knockout mice highly vulnerable to pro-inflammatory TNFα insult. Thus, abnormal trafficking and activity of TACE in prion diseases likely originates from a loss of PrPc cytoprotective function .
Synaptolysis is believed to initiate the neurodegeneration arising after a decrease in depolarization-induced calcium transients that progressively impairs glutamate release . However, although cytoskeletal disruption in dendritic spines plays a major role in neuronal dysfunction, neither changes in postsynaptic densities and presynaptic compartment nor disruption of afferent innervation have been systematically observed, suggesting that even at terminal stages of the disease neuronal loss may not result from deafferentation as previously proposed in the hippocampus and cerebellum of scrapie-infected mice [33, 58, 59]. Thus, neuronal vulnerability to pathological protein misfolding appears to be more strongly dependent than previously thought, on the structure and function of target neurons.
Recent investigations of scrapie pathogenesis in the mouse cerebellum revealed an early upregulation of tumor necrosis factor-α receptor type 1 (TNFR1), a key mediator of neuroinflammation at the membrane of astrocytes enveloping Purkinje cell (PC) excitatory synapses already at the preclinical stage of the disease before PrP22L precipitation, GFAP astrogliosis, and PC death . The contribution of perisynaptic astrocytes to prion pathogenesis through TNFR1 upregulation remains to be clarified and, although the cell types responsible for PrP22L production in the cerebellum are still uncertain, these data suggest a critical role for astrocytes in prion pathogenesis.
2. Mechanisms of neuronal death in prion diseases
Despite the overall advances made in this field during the last decades, the sequence of cellular and molecular events leading to neuronal cell demise in TSEs remains obscure. At present, neuronal cell death can be envisioned as resulting from several parallel, interacting, or sequential pathways involving protein processing and proteasome dysfunction , oxidative stress , inflammation  apoptosis, and autophagy . The repertoire of pathways that lead to neuronal death is however limited . In TSEs, apoptosis is the most popular theory of cell death but is not convincingly documented. In all cases, the probable disruption of both neuronal metabolism and circuits generates a pro-apoptotic signal for neurons. In addition to disruption of cellular proteostasis, UPS dysfunction may lead to neurotoxicity by activating pro-apoptotic pathways. PrPTSE aggresomes can associate with pro-apoptotic factors such as vimentin and caspases . On the other hand, autophagy has been reported in TSEs, but its role in prion disease pathology is not well established . However, the extensive synaptic autophagy observed in prion diseases  has been proposed to contribute to overall synaptic degeneration, a major precocious pathological feature leading to neuronal death in TSEs. This chapter reports recent biochemical and cytopathological studies investigating the involvement of apoptosis and autophagy in neuronal loss induced by infectious prions as well as by PrPc-deficiency in the mouse cerebellum.
Among TSEs, scrapie is a natural ovine prion disease widely studied in mouse models using murine-adapted prion strains (22L, ME7) that, akin to natural prion strains, differ in their rate of disease progression (i.e., duration of the incubation period), as well as the extent and regional pattern of brain histopathology [66, 67]. For example, the characteristic of a prion strain mostly relies on specific biochemical properties related to PrPTSE misfolding. The variable susceptibility of neuronal types to prion infection also emerges as another critical parameter that underlies the complex mechanisms of prion pathogenesis [54, 68, 69] and affects PrPTSE progression along defined anatomical routes . The cellular and molecular mechanisms involved in targeting PrPTSE to specific neuronal populations [33, 71, 72] and neuron-to-neuron spreading of prions in the CNS remain elusive .
In several prion diseases, the cerebellum is a preferential prion target for scrapie [74, 75, 76, 77, 78], also observed in Creutzfeldt-Jakob disease (CJD) cases [79, 80, 81, 82, 83, 84, 85, 86, 87]. Cerebellar circuits are exquisitely patterned and the expression patterns of zebrins in PCs define a topographical map of genetically determined zones controlling sensory-motor behavior [88, 89]. Subsets of PCs expressing zebrins alternate with subsets of zebrin-free PCs, thus forming complementary stripes of biochemically distinct PCs . The most comprehensively studied zonal marker is zebrin II/aldolase C (ZII/AldC) . The expression of ZII/AldC by itself, however, is not sufficient to recapitulate the full complexity of the cerebellar cortex because of the many other PC subtypes [91, 92].
In a recent study , the parasagittal compartmentation of the cerebellar cortex restricted 22L scrapie pathogenesis, including PrP22L accumulation, PC neurodegeneration, and gliosis. Indeed, PCs displayed a differential, subtype-specific vulnerability to 22L prions with zebrin-expressing PCs being more resistant to prion toxicity, whereas in stripes where PrP22L accumulated most zebrin-deficient PCs were lost and spongiosis was accentuated ( Figure 1 ). Although this banding pattern of PrP22L accumulation is most likely delineated by structural constraints of compartmentation, different biochemical properties of PC subpopulations may well determine their differential resistance to scrapie prions.
2.1 Prion-induced apoptosis
2.1.1 Apoptotic pathways in prion-infected neurons
The mechanism of prion neurotoxicity requires neuronal expression of PrPc and is based on the subversion of its normal function triggered by an interaction with PrPTSE at the cell surface, thereby transducing a toxic signal into the cell. Nevertheless, this has been challenged by the discovery of a monomeric, highly α-helical form of PrPc with strong in vitro and in vivo neurotoxicity that elicits autophagy and apoptosis with a molecular signature similar to that observed in prion-infected animal brains . This toxic PrP (TPrP) killed PrP-deficient neurons in vitro suggesting that a PrP-derived toxic signal can be generated within neurons independently of endogenous membrane-bound PrPc. Indeed, postnatal ablation of PrPc expression in neurons reversed neurodegeneration and affected disease progression in mice even though glial replication was maintained and PrPTSE accumulated . Thus, prion pathogenesis is governed by both cell-autonomous mechanisms responsible for cellular dysfunction and neurodegeneration and noncellular-autonomous mechanisms propagating prion spread .
Endoplasmic-reticulum stress has recently been implicated in an apoptotic regulatory pathway activated by changes in Ca2+ homeostasis or accumulation of aggregated proteins. In both these situations, Ca2+ is released and caspase-12 is activated . ER stress and caspase-12 activation have been identified in prion-infected N2a cells as well as in the brains of prion-diseased mice and CJD patients . The synaptic dysfunction and neuronal death caused by PrPTSE accumulation via dysregulation of the Ca2+-sensitive phosphatase calcineurin (CaN) provides further evidence of the role of ER stress and Ca2+ homeostasis in prion-induced neurodegeneration . The increase in Ca2+ cytosolic levels following hyperactivation of CaN dysregulates the pro-apoptotic Bcl-2-associated death promoter (Bad), and the transcription factor cAMP response element-binding (CREB). Dephosphorylated Bad interacts with Bax causing mitochondrial stress and apoptosis while dephosphorylated CREB cannot translocate into the nucleus to regulate the transcription of synaptic proteins, resulting in synaptic loss .
2.1.2 Mitochondrial apoptosis in prion-infected cerebellar neurons
PrPc has recently been suggested to participate in anti-apoptotic and anti-oxidative processes by interacting with the stress inducible protein 1 (STI-1) to regulate superoxide dismutase (SOD) activation . The PrPc octapeptide repeat region contains a B-cell-lymphoma 2 (Bcl-2) homology domain 2 (BH2) of the family of apoptosis regulating Bcl-2 proteins involved in the anti-apoptotic function of Bcl-2. A direct interaction between PrPc and the C-terminus of anti-apoptotic Bcl-2 has also been found [101, 102]. In addition, the third helix of PrPc impaired the BAX conformation changes required for apoptosis activation suggesting that PrPc may assure the neuroprotective function of Bcl-2 . Along this line, Prnp 0/0 neurons were more susceptible to apoptotic stimuli such as serum deprivation than their wild-type counterparts, whereas they were rescued by PrPc or Bcl-2 expression [104, 105]. PrPc also protected primary neurons against BAX-dependent apoptosis. Furthermore, transgenic expression of Bax or Bax and Prnp indicated that Prnp impairs Bax-dependent neuronal death .
Activation of the mitochondrial apoptotic pathway was observed when primary neurons were exposed to aggregated neurotoxic peptides like PrP106-126 or recombinant mutant PrP [107, 108, 109]. Apoptotic neuronal death demonstrated by activation of several caspases and DNA fragmentation is evident in natural prion diseases as well as in experimental models of TSEs [76, 110, 111]. In the cerebellum, apoptotic features have been observed in granule cells in CJD patients [112, 113] as well as in mice experimentally infected with CJD  and scrapie strains 301V, 87V, 22A , 79A , M1000/Fukuoka-1 , 127S , 22L, 139A, and RML [116, 117]. More recently, activation of caspase-3 was found in PCs of 22L-infected mice . However, cerebral upregulation of the pro-apoptotic factor BAX has been reported in some cases of scrapie-infected rodents [116, 118], whereas no changes in clinical illness and neuropathology could be detected in the brain of Bax-deficient mice infected with 6PB1 mouse-adapted BSE prions . This suggested that BAX-mediated cell death is not involved in the pathological mechanism induced by BSE. Nevertheless, BAX is known to be involved in neuronal death in Tg(PG14)  and Ngsk PrnP0/0  murine models of PrP-deficiency-linked diseases. In these cases, neuronal death is restricted to cerebellar neurons that are known to undergo BAX and BCL2-dependent apoptosis in other abnormal conditions [122, 123]. This led us to further investigate the involvement of intrinsic mitochondrial apoptotic pathways in a cerebellotropic prion disease such as the 22L scrapie. For this purpose, the pathogenesis of 22L scrapie in the brain of Bax-KO (Bax −/−) mice  and in mice expressing a human Bcl-2 transgene  was analyzed. Clinical signs of 22L scrapie (mainly ataxia) were similar to those previously described for C57Bl/6 mice . Bax −/− and HuBcl-2 mice infected by either intraperitoneal (ip.) or intracerebellar (icb.) route displayed ataxia 10–15 days sooner than wild-type mice. Survival times however, were similar in all genotypes (i.e., 223 dpi ip. and 129 dpi icb.). whereas 22L induced more severe cerebellar spongiosis via the icb. route than the ip. route, similar lesion profiles  were induced by 22L ip. in the brain of Bax −/− and wild-type mice and lesion profiles were not different in the brain of Bax −/−, HuBcl-2 and wild-type mice infected with 22L icb. ( Figure 2 ). Anatomopathological analysis of the cerebral and cerebellar cortices of the 22L-diseased Bax −/− and HuBcl-2 mice did not reveal any modified patterns of vacuolation, astrogliosis, and PrP22L deposits irrespective of the inoculation route. Synaptophysin and calcium-binding protein (CaBP) immunohistochemistry also revealed severe synapse and PC loss in all cases ( Figure 3 ). Finally, quantitative analysis of the cerebellar granule cells immunolabeled for the nuclear marker NeuN revealed a significant loss of neurons in all genotypes infected by the icb. route ( Figure 4 ). Surprisingly, no significant difference could be detected between Bax −/− and wild-type mice infected by the icb. route, whereas HuBcl-2 mice whose granule cells are rescued from developmental cell death  lost more granule cells than wild-type and Bax−/− mice ( Figure 4 ). These data indicate that neither suppression of Bax nor overexpression of Bcl-2 protected cerebellar neurons from 22L scrapie-induced neurotoxicity. Thus, the granule cell and PC death induced by 22L scrapie does not seem to involve BAX and cannot be counteracted by overexpression of the anti-apoptotic factor BCL-2. However, cleaved caspase-3 and -9 were observed in the brains of Bax−/− mice, suggesting that apoptosis may occur through (an) alternative mechanism(s) in TSEs of infectious origin. Indeed, apoptotic features have been reported in the brain of wild-type mice infected with RML, in the absence of Bax upregulation , while other proteins involved in cell death including those associated with the mitochondrial inner membrane, the UPS and the endoplasmic-reticulum-associated protein degradation (ERAD) pathway  were upregulated.
2.1.3 Prion-induced neuronal death in cerebellar organotypic slice cultures (COCS)
In the recently developed prion cerebellar organotypic slice culture (COCS) assay, progressive spongiform neurodegeneration that closely reproduce features of prion disease can be induced ex vivo [117, 129]. Infecting COCS with three different scrapie strains (RML, 22L, 139A) produced three distinct patterns of prion protein deposition accompanied by salient features of prion disease pathogenesis such as severe neuronal loss, a pro-inflammatory response, and typical neuropathological changes (spongiform vacuolation, tubulovesicular structures, neuronal dystrophy, and gliosis). Neurodegeneration did not occur when PrP was genetically removed from neurons and was abrogated by compounds known to antagonize prion replication. Also, calpain inhibitors, but not caspase inhibitors, prevented neurotoxicity and fodrin cleavage; whereas, prion replication was unimpeded indicating that inhibiting calpain uncouples prion replication and neurotoxicity. These data validate the COCS as a powerful model system that faithfully reproduces many morphological hallmarks of prion infections and shows that prion neurotoxicity in cerebellar granule cells is calpain-dependent but caspase-independent.
Furthermore, significant spine loss and altered dendritic morphology, analogous to that seen in vivo were induced by RML scrapie in COCS , while the deposition pattern and subcellular distribution of PrP22L (i.e., granular deposits associated with neurons, astrocytes, and microglia but not PCs in the neuropil of the PC and molecular layers ), closely resembled that observed in vivo .
Following infection of COCS from C57Bl6/J, ZH-I Prnp0/0, and Tga20 PrP-over-expressing mice with brain homogenate from C57Bl6/J infected intracerebrally (ic.) with either 22L or 139A scrapie prions, PrP22L and PrP139A accumulation could be detected on histoblots from wild-type and Tga20 COCS, respectively, 30 and 20 days post infection (dpi), but not on histoblots from ZH-I mice ( Figure 5 ). Furthermore, quantitative analysis of PCs in these COCs indicated that a severe loss of neurons was induced by 22L prions in wild-type slices at 30 dpi (22 ± 2 surviving PCs/slice) and in Tga20 slices at 20 dpi (293 ± 68 surviving PCs) as well as by 139A prions in wild-type slices at 30 dpi (145 ± 63 surviving PCs/slice) and in Tga20 slices at 20 dpi (191 ± 31 surviving PCs/slice) compared to noninfected control COCS (220 ± 27 surviving PCs/slice in wild-type slices and 357 ± 71 surviving PCs/slice in Tga20 slices) ( Figure 6 ). At 30 dpi, the trilaminar organization of the cerebellar cortex was evident in noninfected COCs, which did not exhibit any clear ultrastructural modifications ( Figure 7 ). Nevertheless, numerous vacuoles, autophagosomes, and lysosomes had formed in granule cells infected by 22L and 139A ( Figure 8 ). In diseased PCs, autophagosomes with double membranes and rough endoplasmic reticulum (Nissl bodies) formed compartmented organelles of various sizes (1–10 compartments) resembling different stages leading to multivesicular vacuoles ( Figure 9 ). Although further investigations are necessary, these ultrastructural alterations were not observed in noninfected slices suggesting that a specific effect of prions links prion-induced ER stress to this morphological ER modification.
2.2 Prion-induced autophagy
Autophagy and apoptosis are activated in many neurodegenerative diseases featured by ubiquitinated misfolded proteins. In neurons, the degradation of abnormal proteins such as α-synuclein in Parkinson’s disease, β-amyloid peptide in Alzheimer’s disease (AD), or PrP in TSEs occurs by autophagy [14, 132, 133, 134, 135]. These cardinal proteins contribute to synaptic dysregulation and altered organelles leading to apoptosis. The neurodegenerating neurons exhibit robust accumulation of cytosolic autophagosomes (see  for review, Figure 10 ) suggesting a dysregulation of the autophagic flux resulting from autophagic stress, due to an imbalance between protein synthesis and degradation . Autophagy reduces intraneuronal aggregates and slows down the progression of clinical disease in experimental models of AD [137, 138, 139] and prion diseases [140, 141]. Thus, dysregulation of the autophagic flux impairs the elimination of misfolded proteins and damaged organelles which then accumulate in the cytoplasm and contribute to cell dysfunction and death .
Together, spongiform vacuolation of the neuropil, synaptolysis, accompanied by neuronal cell loss and gliosis constitute the classical neuropathological quartet of TSEs. The typical “spongiform vacuoles” are believed to result from autophagy and develop within neuronal elements, myelinated axons, and myelin sheaths [143, 144]. Autophagic vacuoles are increased in prion-diseased neurons [64, 65, 145], and the scrapie responsive gene 1 (SRG1) protein is overexpressed and bound to neuronal autophagosomes in the brain of scrapie- and BSE-infected animals and CJD-diseased humans [146, 147]. In addition, LC3-II, a marker of autophagosomes is increased in the cytosol of neurons in scrapie-infected hamsters and CJD- and FFI-diseased patients.
Recent evidence indicated that PrPc, but not truncated PrP devoid of the N-terminal octapeptide repeat region, exerts a negative control on the induction of autophagy . Thus, the loss or subversion of PrPc function resulting from prion infection may upregulate autophagy in diseased neurons . While autophagy-inducing agents increased cellular clearance of PrPTSE [149, 150, 151], blocking the fusion of autophagosomes with lysosomes allowed visualization of PrPTSE in the autophagosomes suggesting that degradation of endosomal PrPTSE is by autophagy . However, saturation of the autolysosomal degradation process can release PrPTSE aggregates and degradation enzymes into the neuroplasm contributing to autophagy upregulation and neuronal death . Nevertheless, although autophagy-inducing agents delayed disease onset and PrPTSE accumulation in the CNS of mice , survival time was not modified . Along this line, neither autophagy-inducing nor -inhibiting treatments altered the time course or amplitude of prion-induced neuronal death, strongly suggesting that autophagy in protein misfolding diseases is a secondary mechanism in the neurodegenerative process [141, 154].
3. Neuronal death in prion protein-deficient mice
3.1 Impaired autophagy in Zrch-1 prion protein-deficient mice
With the exception of the Prnp-knockout models in which ectopic expression of Doppel (Dpl) in the CNS leads to PC death, most other Prnp-knockout mouse models do not show gross abnormalities indicating that PrPc may be dispensable for embryonic development and adulthood. Nevertheless, PrP-deficient mice exhibit an increased predilection for seizures, motor and cognitive disabilities, reduced synaptic inhibition, and long term potentiation in the hippocampus. Also, altered development of the granule cell layer, dysregulation of the cerebellar network and age-dependent spongiform changes with reactive astrogliosis have been observed [155, 156]. In cultures of PrP-deficient hippocampal neurons, autophagy is upregulated in the absence of serum or by hydrogen peroxide-induced oxidative stress [148, 157] suggesting that suppression of the protective effects of PrPc could impair the autophagic flux in PrP-deficient neurons in vivo. Indeed, ultrastructural examination of hippocampus and cerebral cortex of ZH-I Prnp 0/0 mice revealed an accumulation of autophagosomes containing incompletely digested material increasing from 3 to 12 months of age . In addition, an ultrastructural examination of PCs in the cerebellum of ZH-I Prnp 0/0 mice revealed significant autophagic accumulation in the somato-dendritic compartment of these neurons from 6 to 14.5 months of age ( Figure 11 ). Since autophagic cell death is known to induce neurodegeneration [136, 159, 160], these signs of autophagy blockade could reflect a sustained, progressive autophagic neuronal loss in the CNS of the ZH-I Prnp 0/0 mice.
3.2 Neuronal loss in Dpl-expressing Ngsk prion protein-deficient mice
Nagasaki (Ngsk) PrP-deficient mice which have a deletion of the entire Prnp gene [161, 162, 163] develop progressive cerebellar ataxia, which was later discovered to result from the absence of a splice acceptor site in exon 3 of Prnp . This leads to the aberrant overexpression of the Prnd gene encoding the PrPc paralogue Dpl [165, 166] that causes selective degeneration of cerebellar PCs. Notably, the reintroduction of Prnp in mice overexpressing Prnd in the brain rescued the phenotype, suggesting a functional link between the two proteins . Dpl has been shown to have intrinsic neurotoxic properties in cerebellar neurons  and has been proposed to interfere with PrPc and affect cell survival . According to this hypothesis, PrPc and Dpl bind a common ligand LPrP, where PrPc binding induces a cell survival signal while Dpl binding activates a death signaling cascade. In PrPc-deficient Prnp-knockout mice that do not express Dpl, the existence of a protein π has been proposed to induce a cell survival signal when bound to LPrP . For the moment, LPrP and π remain to be identified, as well as the neuronal death pathways involved in Dpl-induced PC loss.
Because Dpl neurotoxicity depends on PrPc-deficiency in PCs, investigating the underlying neurotoxic mechanism may provide important insight into the neuroprotective function of PrPc. The resistance of the PC population to neurotoxicity increased in the cerebellum of Ngsk mice, which were either deficient for the pro-apoptotic factor Bax  or over-express the anti-apoptotic factor Bcl-2 . Although this suggests that an intrinsic apoptotic process is involved in the death of the Ngsk Prnp 0/0 PCs, a significant PC loss still occurred in both (Bax−/−; Ngsk Prnp 0/0) and (HuBcl-2; Ngsk Prnp 0/0) double mutants. Thus, the Ngsk condition, i.e., Dpl neurotoxicity and PrP-deficiency, could activate BAX-independent mechanisms in the Ngsk Prnp 0/0 PCs. These neurons exhibited robust autophagy well before significant neuronal death in the cerebellar cortex of the Ngsk Prnp 0/0 mice [135, 171] suggesting that either “reactive” autophagy is initially induced as a neuroprotective response to Dpl neurotoxicity or impaired autophagy results from PrP-deficiency as in ZH-I Prnp 0/0 mice (see above and ). Indeed, the increased expression of the autophagic markers SCRG1, LC3-II, and P62 proteins without any changes in mRNA levels, indicates that the ultimate steps of autophagic degradation are impaired in Ngsk Prnp 0/0 PCs . Probably due to this impairment of autophagic proteolysis, LC3-II-, and Lamp-1-labeled autophagosomes and autolysosomes  accumulate in the Ngsk Prnp 0/0 PCs. How apoptosis and autophagy are involved in Ngsk Prnp 0/0 PC death remains to be determined.
To further investigate the role of autophagy in the death of Ngsk Prnp 0/0 PCs, a quantitative analysis of autophagic PCs was performed at the ultrastructural level in the cerebellum of Ngsk Prnp 0/0, Bax −/−; Ngsk Prnp 0/0, ZCH-I Prnp 0/0 and control Bax +/+;Prnp +/+ mice ( Figure 12 ). At 4.5 months of age, equivalent amounts of autophagic somato-dendritic compartments and axons of PCs were found in the cerebella of Ngsk Prnp 0/0 and Bax −/−; Ngsk Prnp 0/0 mutants and were significantly more than those in ZCH-I Prnp 0/0 and control Bax +/+;Prnp+/+ cerebella. Interestingly, the amounts of autophagic axons and somato-dendritic compartments of PCs in the ZCH-I Prnp 0/0 and control Bax +/+;Prnp +/+ cerebella were not different. These data suggest that while autophagy induction is already visible in PCs with the Ngsk condition, it is not induced in control Bax +/+;Prnp +/+ PCs, nor in the absence of PrPc in the ZCH-I Prnp 0/0 PCs. Thus, autophagy seems to be induced by Dpl neurotoxicity in the Ngsk condition whether BAX is present or not; whereas PrP-deficiency alone has no autophagy-inducing effect at this age ( Figure 13 ).
At 6.5–7 months of age, the amount of autophagic somato-dendritic compartments and axons of PC were significantly decreased in Bax −/−; Ngsk Prnp 0/0 cerebella compared with Ngsk Prnp 0/0 cerebella. Consequently, the amount of autophagic PC profiles in the Bax −/−; Ngsk Prnp 0/0 and ZH-I Prnp 0/0 cerebella was equivalent, yet more than in the control Bax +/+;Prnp +/+ cerebella. Furthermore, autophagic PC somato-dendritic compartments and axons did not change from 4.5 to 6.5–7 months of age in the Bax −/−; Ngsk Prnp 0/0, whereas many more PCs were autophagic in the 6.5–7 month-old compared to the 4.5 month-old Ngsk Prnp 0/0 cerebella. This increase was also observed in ZH-I Prnp 0/0 cerebella, while no autophagic PCs were found in 6.5–7 month-old control Bax +/+;Prnp +/+ cerebella ( Figure 13 ).
This suggests that BAX-deficiency modulates autophagy in Ngsk Prnp 0/0 PCs after 4.5 months of age. Autophagy in the ZH-I Prnp 0/0 PCs had increased to the same level as observed in the Bax −/−;Ngsk Prnp 0/0 cerebella. Thus, the persistent autophagy in the PCs of the Bax −/−;Ngsk Prnp 0/0 double mutants is likely related to PrP-deficiency. Also, autophagy- and Bax-dependent apoptosis are likely to occur in the same PCs that are rescued by Bax deletion.
At 12 months of age, the amount of autophagic somato-dendritic compartments and axons of PCs in Bax −/−;Ngsk Prnp 0/0 cerebella was equivalent to that found in 4.5 month-old cerebella suggesting that autophagy remains stable in this PC population, at a level similar to that maintained in the ZH-I Prnp 0/0, and this likely results from PrP-deficiency. Indeed, many more autophagic PC somato-dendritic compartments and axons were observed in ZH-I Prnp 0/0 cerebella than in the cerebella of 12 month-old control Bax +/+;Prnp +/+ mice which did not contain autophagic PCs. However, the autophagic PC somato-dendritic compartments were still more in Ngsk Prnp 0/0 cerebella (16.36 ± 7.9) compared with Bax −/−;Ngsk Prnp 0/0 (5.08 ± 5) cerebella, and there was a significant increase from 6.5–7 (14.38 ± 7.8) to 12 months of age. The increased amount of autophagic PC axons in Bax −/−;Ngsk Prnp 0/0 cerebella was stable during this same period (3.9 ± 5.6 at 6.5–7 months; 5.08 ± 5.1 at 12 months), suggesting that the initiation of axonal autophagy peaks at 6.5–7 months of age ( Figure 12 ) [135, 171, 173, 174, 175, 176]. In agreement, an examination of autophagy in the presynaptic terminals of PCs impinging on the somato-dendritic compartments of the deep nuclear neurons in the fastigial, interposed and dentate deep cerebellar nuclei, revealed a significantly greater amount of autophagic PC presynaptic boutons in the deep nuclei of all mutants compared to control Bax +/+;Prnp +/+ mice ( Figure 14 ).
The absence of BAX not only protected some PCs from neurotoxicity in the cerebellum of the Ngsk Prnp 0/0 mice , but also decreased the number of autophagic neurons suggesting that the PCs rescued by Bax deficiency do not display activated autophagy, whereas the autophagic PCs in the Bax −/−;Ngsk Prnp 0/0 cerebellum are likely to result from PrP-deficiency as in the ZH-I Prnp 0/0 cerebellum. Nevertheless, the persistent loss of Bax−/−;Ngsk Prnp 0/0 PCs could result from an increased sensitivity of these PCs to the Ngsk condition compared to ZH-I Prnp 0/0 PCs.
The complex pattern of neuronal death observed in neurodegenerative diseases is believed to involve an extensive interplay between the major cell death pathways [177, 178]. This is likely the case in prion-infected, as well as PrP-deficient neurons such as PCs. We further investigated PC death in Ngsk Prnp 0/0 and ZH-I Prnp 0/0 COCS by measuring PC survival and development using morphometric methods  in COCS from these PrP-deficient mice. Similar timing and amplitude of PC growth impairment and death were observed in all PrP-deficient genotypes. Indeed, PC surface, perimeter, and dendritic extension increased between 7 and 21 DIV in the wild-type COCS, while no significant variation of surface and perimeter could be measured in the PrP-deficient mutant COCS during this period ( Figure 15 ). Similarly, wild-type and PrP-deficient PCs displayed equivalent maximal dendritic extension after 7 days ex vivo, but wild-type PCs continued to increase their maximal dendritic length until 21 DIV, while the dendrites of PrP-deficient PCs did not grow during this period [14, 180]. Thus, PrP-deficient PCs exhibit a similar developmental deficit which seems to be independent of Dpl expression in COCS.
The neurotoxic effects of PrP-deficiency were quantitatively analyzed by counting PCs at 3, 5, 7, 12, and 21 days in COCS from wild-type, Ngsk Prnp +/0, Ngsk Prnp 0/0, and ZH-I Prnp 0/0. Whereas, wild-type PCs’ numbers remained stable during the whole period, severe PC loss (68–69%) had occurred at 7 DIV and slightly increased up to 21 DIV in all PrP-deficient mutant COCS. PC loss displayed similar kinetics and amplitude in Ngsk Prnp +/0, Ngsk Prnp 0/0, and ZH-I Prnp 0/0 COCS suggesting that despite detectable levels of 15–20 kDa glycosylated form of Dpl in the Ngsk Prnp 0/0 COCS ( Figure 16 ), it may be not implicated in PC death in ex vivo cultures.
Furthermore, at the ultrastructural level, whereas autophagic organelles were rare in wild-type PCs after 7 and 12 DIV, Ngsk Prnp 0/0 PCs contained numerous autophagosomes and autophagolysosomes at different maturation stages ( Figure 17 ). During the period of PC death in the Prnp 0/0 COCS (i.e., 3, 5, and 7 DIV) Western blotting of apoptotic and autophagic markers revealed a 4- to 5-fold increase in markers of autophagosomal formation such as LC3B-II (at 5 DIV), p62, and beclin-1 (at 3 and 5 DIV) in the ZH-I and Ngsk Prnp 0/0 COCS and the lysosomal receptor LAMP-1 in the Ngsk Prnp 0/0 COCS at 7 DIV ( Figure 18 ). Increased amounts of activated caspase-3 indicated the apoptosis in protein extracts of COCS from both Prnp 0/0 genotypes as early as 3DIV .
This morphometric and quantitative analysis of COCS suggests that PrP-deficiency, rather than Dpl neurotoxicity, is responsible for the neuronal growth deficit and loss ex vivo. Indeed, the neurotoxic properties of Dpl did not seem to contribute to Ngsk PC loss in the COCS, whereas Dpl-induced PC loss is detectable in 6-month-old Ngsk Prnp 0/0 mice. A possible explanation for this difference is that COCS are not mature enough to model 6-month-old cerebellar tissue. Nevertheless, in Ngsk Prnp 0/0 and ZH-I Prnp 0/0 COCs, activation of autophagy and apoptosis is contemporaneous with the atrophy and death of PCs during the first week of culture suggesting that PrP-deficiency is solely responsible for neuronal death in this ex vivo system and that PrPc is neuroprotective for cerebellar PCs. As ZH-I Prnp 0/0 PCs survive in vivo, PC death in ZH-I Prnp 0/0 and Ngsk Prnp 0/0 COCS could result from a noxious exacerbation of PrP-deficiency by ex vivo conditions.
Although the contribution of apoptosis to prion-induced death of central neurons including cerebellar ones is strongly supported, our studies of scrapie-infected PCs show that although caspase-3 is activated, the pro-apoptotic BAX/BCL-2-dependent mitochondrial pathway is not involved in the prion-induced death of these neurons. This is also the case for BSE-induced death of hippocampal and thalamic neurons , suggesting that prions exert neurotoxicity through BAX-independent activation of caspase-3. Ultrastructural evidence of ER stress and robust autophagy in the scrapie-infected cerebellar neurons both in vivo and ex vivo implicate them in these BAX-independent neurotoxic mechanisms. Furthermore, the autophagic blockade resulting from prion protein-deficiency in ZH-I and Ngsk Prnp 0/0 mice may contribute to neuronal death in infectious prion-diseased cerebellar neurons. In Ngsk Prnp 0/0 cerebellar neurons, Dpl neurotoxicity and PrP-deficiency contribute to neuronal death probably through an interplay between autophagic blockade and BAX-dependent apoptosis.