Peroxisomal Disorders: A Review on Cerebellar Pathologies

Abstract

Peroxisomes are organelles with diverse metabolic tasks including essential roles in lipid metabolism. They are of utmost importance for the normal functioning of the nervous system as most peroxisomal disorders are accompanied with neurological symptoms. Remarkably, the cerebellum exquisitely depends on intact peroxisomal function both during development and adulthood. In this review, we cover all aspects of cerebellar pathology that were reported in peroxisome biogenesis disorders and in diseases caused by dysfunction of the peroxisomal α‐oxidation, β‐oxidation or ether lipid synthesis pathways. We also discuss the phenotypes of mouse models in which cerebellar pathologies were recapitulated and search for connections with the metabolic abnormalities. It becomes increasingly clear that besides the most severe forms of peroxisome dysfunction that are associated with developmental cerebellar defects, milder impairments can give rise to ataxia later in life.

Introduction

Peroxisomal function and dysfunction

Peroxisomes are omnipresent in mammalian cells but they have a flexible abundance and content depending on the cell type, developmental and physiological state. Beside peroxide‐generating oxidases and peroxide‐degrading enzymes such as catalase, these organelles harbor several pathways of lipid metabolism 139, 143, 144. Via α‐oxidation they chain shorten the branched‐chain fatty acid phytanic acid taken up via the diet and hydroxy‐fatty acids that are enriched in myelin. Peroxisomal β‐oxidation is not only essential for the degradation of very long‐chain fatty acids (VLCFA) and for the further breakdown of branched‐chain fatty acids, but also for the synthesis of polyunsaturated fatty acids (PUFA) such as docosahexaenoic acid (DHA, C22:6ω‐3) and the conversion of cholesterol into bile acids. Moreover, peroxisomes are required for the synthesis of ether lipids, including plasmalogens that are abundantly present in myelin. In addition, they take part in purine, polyamine, glyoxylate and amino acid metabolism. The ongoing determination of the peroxisomal proteome 58 is unveiling new peroxisomal proteins, which might lead to the identification of new peroxisomal disorders.

Peroxisomal diseases can be categorized in biogenesis disorders (PBD) and single enzyme or transporter defects. PBDs arise when one of 14 PEX genes encoding so‐called peroxins are defective 16. All these proteins are involved in a chain of events that is necessary to import cytosolically synthesized proteins into the peroxisomal membrane or lumen, or for peroxisome fission 150. Peroxisomal proteins that are mislocalized to the cytosol are mostly instable or inactive. Depending on the severity of the PEX mutation, this causes an array of disorders with developmental and/or degenerative pathology currently known as the Zellweger spectrum 16, 124.

Other peroxisomal disorders with neurological pathologies are due to mutations in peroxisomal enzymes or membrane transporters of the above mentioned lipid pathways 145. Although their metabolic deficits are less wide‐ranging, the pathologies also vary from lethality in the postnatal period to neurological demise in adulthood, depending on the affected gene and residual protein function.

We first discuss the cerebellar pathologies in single enzyme diseases as these can be better correlated with the metabolic dysfunction before elaborating on the PBDs. The findings in patients and in the corresponding mouse models are dealt with in parallel.

Cerebellar development and architecture

The cerebellum is involved in the coordination of smooth, purposeful motor activities and in learning new motor skills. Also, a growing body of evidence infers additional cerebellar functions such as cognitive processing (eg, language) and emotional reactivity (eg, fear) 114.

In humans, cerebellar development is a slow process, starting in the embryonic stage and finishing in the first postnatal years, whereas in mice, it occurs postnatally over a 3‐week period. In newborn pups, Purkinje cells (PCs) extend multiple, non‐organized dendrites evolving in a monolayer with primary dendrites branching in a single, parasagittal plane by postnatal day 21. By providing the only projections to the deep cerebellar nuclei and vestibular nuclei, PCs are the sole output neurons of the cerebellar cortex (Figure 1). During the same timeframe, granule cells migrate from the external granule cell layer to the internal granule cell layer and extend their parallel fibers (PF). This process of cell migration is carefully guided by Bergmann glia fibers projecting into the molecular layer. After arrival in the internal granule cell layer, granule cells mature and synaptic contacts are established.

Figure 1.

Cerebellar architecture. The cerebellar cortex harbors five cells types. The PCs are the most obtrusive, appearing as an ordered monolayer located in between the inner granular and the outer molecular cell layer. By providing the only projections to the deep cerebellar and vestibular nuclei, these cells are the sole cerebellar output neurons. Mossy fibers arising from the pontine nuclei, spinal cord and vestibular system project into the granule cell layer to establish synaptic contacts on the granule cells. In turn, PFs arising from granule cells project into the molecular layer to make excitatory synapses on the elaborated dendritic tree of the PC. CFs arising from the inferior olive provide the second PC excitatory innervation. The deep cerebellar nuclei are stimulated by mossy fibers and efferents originating from the inferior olive, while PCs provide inhibitory input.

Mature PCs are unique neurons with an elaborate dendritic tree harboring many spines, allowing a multitude of synaptic contacts. It has been shown that extrinsic signals such as normal synaptic input, growth factors, hormones and neurotrophins are necessary for dendritic outgrowth and refinement. Initially, each PC receives excitatory input from multiple climbing fibers (CF), originating from the contralateral inferior olivary nucleus of the medulla. By postnatal day 20, surplus CF are eliminated, resulting in mono innervation. This CF–PC connection is one of the strongest in the central nervous system (CNS). Indirect PC innervation originates from axons arising from the spinal cord and brainstem. These so‐called “mossy fibers” project to the granule cells, which in turn send PF making synaptic contacts with dendritic spines of multiple PCs. Excitatory inputs on PCs are counteracted by basket‐ and stellate cells, both inhibitory interneurons present in the molecular layer. The circuit is closed by inhibitory feedback connections from the deep cerebellar nuclei to the inferior olivary nucleus and an excitatory feedback projection between the cerebellar nuclei and the granule cells (Figure 1). Moreover, neuronal connections between the cerebellum and diverse brain regions such as the basal ganglia, hippocampus, thalamus and hypothalamus should not be overlooked. Given this extensive cerebellar wiring, the aforementioned additional roles in higher order brain functions are not surprising.

Pathologies causing abnormalities in PC development, cerebellar malfunctioning or PC loss generally cause a range of motor disturbances (eg, tremor, dysmetria and hypotonia) collectively referred to as “cerebellar ataxia.” Currently, more than 60 different types of cerebellum‐based ataxias have been identified. In what follows, cerebellar anomalies caused by peroxisomal dysfunction during development and adulthood will be discussed in men and mice.

Defects in α‐Oxidation of Phytanic Acid

Patients lacking the first enzyme of the peroxisomal α‐oxidation pathway, phytanoyl‐CoA hydroxylase (PHYH), suffer from Refsum disease, an inherited disorder resulting in the accumulation of phytanic acid in several tissues and body fluids 148, 152. Although the majority of Refsum patients carry PHYH gene mutations, the disease can also be caused by PEX7 mutations, impairing the import of PHYH into the peroxisome 63, 66, 67, 134 (Figure 2). Patients lacking other enzymes of the peroxisomal α‐oxidation pathway have not been identified. Phytanic acid is derived from chlorophyll by bacterial metabolism in the gut of ruminants and is primarily taken up through ruminant meat and dairy products. Other peroxisome‐related disorders characterized by increased phytanic acid levels are Zellweger spectrum disorders, multifunctional protein 2 (MFP2) deficiency, and rhizomelic chondrodysplasia punctata (RCDP) type 1.

Figure 2.

Schematic overview of cerebellar and brainstem pathologies in peroxisomal single enzyme and transporter defects in men and mice. Defects in either of the three major lipid metabolic pathways can cause developmental and/or degenerative pathologies. Ataxia is a common but not obligate clinical presentation. Defects in ether lipid synthesis occur through mutations in peroxisomal enzymes AGPS or GNPAT, or in FAR1 located at the outer peroxisomal membrane that produces fatty alcohols. AGPS needs to be imported through PEX7 (7) that binds the long form of PEX5 (5L) before docking at the peroxisomal membrane. A similar mechanism is involved in the import of PHYH, the first enzyme of peroxisomal α‐oxidation, which functions in the breakdown of the branched‐chain FA phytanic acid reaching the peroxisomal lumen via import through ABCD3. Peroxisomal β‐oxidation defects related to cerebellar dysfunction are caused by mutations in the ABCD1 transporter or in any of the enzymes shown involved in the degradation pathway. Import of these enzymes occurs through PEX5 (see Figure 3). ABCD2 (not depicted), which transports VLCFA and other substrates, was only inactivated in mice. Figure adapted from (150). ND = not documented; * after phytol treatment.

Refsum disease

Cerebellar and brainstem pathologies in Refsum patients

Refsum was the first to describe cerebellar signs as one of the main manifestations of the disease that was named after him, although it is now clear that cerebellar problems can be late in onset compared with retino‐ or neuropathy 147. The clinical presentation of Refsum disease is variable, and cerebellar signs do not always occur 3, 120. Surprisingly, only few post‐mortem histopathological investigations on Refsum patients were performed and findings in the cerebellum and brainstem are scarce, revealing atrophy of the cerebellar vermis and patchy cell degeneration in the dentate and inferior olivary nuclei 22, 55, 106. It should be noted that cerebellar pathology may not be the sole origin of ataxia as gait disturbances without cerebellar pathological changes have been reported 46, 112. The loss of proprioception caused by the demyelinating neuropathy will cause a sensory ataxia on top of the cerebellar ataxia 46, 88. As their gait disturbances ameliorated upon dietary phytanic acid restrictions, this strongly indicated that phytanic acid accumulation induces toxicity 33. Interestingly, phytanic acid concentrations are 3 to 4 times higher in the peripheral nerves of Refsum patients as compared with the brain, caused by impaired access through the blood‐brain barrier 128. Phytanic acid accumulations in the peripheral nervous system slow down nerve conduction velocity, impair reflexes and perturb sensation. The demyelinating polyneuropathy generally starts before the cerebellar dysfunction, making it sometimes clinically difficult to detect the cerebellar ataxia. However, signs like nystagmus cannot be explained by the polyneuropathy, and other signs of ataxia are out of proportion with the degree of sensory involvement 133.

The Refsum mouse model

The in vivo short‐term pathological processes of Refsum disease were studied by Ferdinandusse et al 42, using PHYH‐deficient mutant mice on chow supplemented with phytol. Phytanic acid is derived from dietary sources only and the concentration of branched‐chain fatty acids or their precursors in standard rodent food is low. A moderate increase in the plasma concentration of phytanic acid was obtained when Phyh ^−/− mice were supplemented with 0.1% phytol for 6 weeks, whereas 3 weeks of 0.25% phytol‐supplementation yielded plasma levels similar to those reported in Refsum disease patients (>1 mmol/L). Phytanic acid markedly accumulated in the cerebellum, although levels were 3–10‐fold lower compared with testis, kidney and liver, in line with the relative impermeability of the blood–brain barrier 128. The phytol diet induced prominent gait disturbances in the mutant mice. Immunohistochemistry revealed cerebellar modifications ranging from focal PC loss in Phyh^−/− mice fed 0.1% phytol to severe cell loss and morphological changes of the remaining cells in Phyh^−/− mutants on the 0.25% phytol‐supplemented diet. As in patients, also in the phytol‐supplemented Phyh^−/− mice, the peripheral nervous system contributed to the symptoms as motor nerve conduction velocity was decreased, indicating peripheral neuropathy.

Molecular aspects of phytanic acid accumulation

The improvement of ataxia following reduced phytanic acid intake in Refsum patients, as well as the dosage‐dependent severity of PC degeneration in the phytol‐treated Refsum mouse model, strongly suggest a detrimental role of high phytanic acid concentrations on the cerebellum.

During the past decade, several in vitro studies were performed to elucidate the mechanisms underlying the toxicity caused by phytanic acid. These include experiments with isolated cerebellar mitochondria, cerebellar homogenates, cultured neural cells or fibroblasts incubated with phytanic acid concentrations in the range of those found in plasma of Refsum disease patients (1–500 μM). Impairments of mitochondrial function and morphology were commonly found, including a reduced matrix NAD(P)H content, alterations in energy production, loss of membrane potential and opening of the transition pore 19, 68, 74, 108, 110, 115. Phytanic acid also induced oxidative stress in some in vitro preparations 108.

As to the precise molecular impact, opposing mechanisms were proposed, probably because of the varying experimental setups. In view of the low permeability of the blood–brain barrier to phytanic acid 128, it can be questioned whether exposure of neural cells, and in particular isolated mitochondria, to phytanic acid concentrations occurring in Refsum plasma reflects the in vivo situation. In some experiments, 80% of cultured glial cells died after a 4‐h incubation with 100 μM phytanic acid 110 whereas, with exception of PCs, no overt cell death was reported in the brain of Refsum mice treated with a high dose of phytol 42. In addition, it is unfortunate that in several studies control incubations with straight‐chain fatty acids were not included. In a well‐controlled study using permeabilized and intact human fibroblasts 74, it was shown that phytanic acid acts as a protonophore leading to mitochondrial membrane depolarization and reduced ATP production. It was further proven that this activity depended on the branched‐chain and carboxyl function as it was mimicked by pristanic acid, the α‐oxidized shortened product of phytanic acid, but not by phytol or the straight‐chain palmitic acid with the same number of carbons in the backbone. In addition to mitochondrial dysfunction, it was reported that phytanic acid impairs Ca²⁺ homeostasis 68, 110 and deregulates synaptic Na⁺, K⁺‐ATPase in the cerebellum 20.

Taken together, phytanic acid seems to exert adverse effects on cerebellar PCs, which contribute to an unstable gait in men and mice. Nevertheless, the potential mechanisms deduced from in vitro experiments need to be confirmed in the in vivo situation, which is possible using the Refsum mouse model.

Peroxisomal β‐Oxidation Disorders

Currently, six divergent genetic diseases are known in which peroxisomal fatty acid β‐oxidation is deficient. In order of decreasing incidence the following pathologies are distinguished: (i) X‐linked adrenoleukodystrophy (X‐ALD), (ii) multifunctional protein‐2 (MFP2) deficiency, (iii) acyl‐CoA oxidase 1 (ACOX1) deficiency, (iv) 2‐methylacyl‐CoA racemase (AMACR) deficiency, (v) sterol carrier protein X (SCPx) deficiency and (vi) ABCD3 deficiency (Figure 2).

X‐linked adrenoleukodystrophy

Adrenoleukodystrophy protein, the peroxisomal ABC transporter encoded by the ATP binding cassette D1 (ABCD1) gene, transports VLCFA either as free acids or as CoA esters into peroxisomes prior to β‐oxidation 27, 153. As a consequence, adrenoleukodystrophy protein deficiency is biochemically characterized by C_24:0 and C_26:0 accumulations in plasma and tissues such as the CNS, adrenal cortex and gonads.

Cerebellar and brainstem pathologies in X‐ALD patients

In X‐ALD patients, the clinical spectrum ranges from asymptomatic and Addison‐only (isolated adrenocortical insufficiency) to progressive neurological dysfunction caused by cerebral demyelination in childhood, adolescence or adulthood 70.

The majority of X‐ALD males present either with childhood cerebral ALD (CCALD), characterized by major inflammatory demyelination of the cerebral white matter, or with adrenomyeloneuropathy (AMN), a progressive disorder involving the spinal cord and peripheral nerves (with more frequent axonal than demyelinating polyneuropathy) in adulthood. In both patient groups, adrenocortical failure, which also occurs as the sole clinical involvement, is frequent. About half of AMN patients develop cerebral white matter lesions, typically with a less aggressive evolution than the lesions in CCALD. These different presentations can occur in the same family (with the same ABCD1 mutation), and the factors determining who will develop which clinical presentation remain largely unknown. Moreover, recent evidence indicates that bone marrow transplantation, the therapy for CCALD, is not effective to prevent AMN, again indicating the differences in pathogenesis of these presentations 136. In about 1% of the patients 85, white matter damage in CCALD and AMN starts in the cerebellar white matter, resulting in cerebellar ataxia as a neurological presentation. Also, a form of X‐ALD presenting with primary cerebellar (neuronal) degeneration has been described 79. Almost all of these latter cases were reported before genetic confirmation of X‐ALD was part of clinical routine, and before the mild presentations of PBDs 136 (among which ataxia) were known. As the finding of elevated VLCFA was the only biochemical proof of the diagnosis, several of these cases may in fact have been patients with a mild PBD instead of X‐ALD. At least one report describes profound vermian atrophy in a genetically confirmed X‐ALD patient, even without white matter changes. The family history was suggestive of a cerebellar presentation in other affected family members, and the authors proposed that this presentation was linked to this particular mutation, delC1321in exon 2 31. Reports of more confirmed patients with this presentation would help to establish its real frequency.

Insights from mouse models

Three Abcd1 knockout mouse models were generated independently 48, 70, 72, 87, all showing VLCFA accumulation in the brain. A late‐onset phenotype affecting motor performance developed in 20‐month‐old Abcd1^−/− mice and was accompanied by axonal degeneration in the spinal cord, resembling AMN pathology. Remarkably, inflammatory demyelination as seen in CCALD patients did not develop in the CNS 104. However, PC death was reported in the cerebellum of aged ABCD1‐deficient mice 45, corresponding with the rare X‐ALD patients with primary cerebellar atrophy.

ABCD2 deficiency, so far not described in men, was created in mice. This transporter partially overlaps in function with ABCD1 92. Similar to Abcd1 mutants, ABCD2‐deficient mice developed cerebellar problems including PC atrophy and loss, concomitant with whole‐body tremor and cerebellar ataxia. Not surprisingly, the pathological signs occurred earlier and were more severe in Abcd1/Abcd2 double mutants 45.

The toxicity of VLCFA

The only metabolic abnormality in X‐ALD is the accumulation of VLCFA. Because of their exceptional length, increased levels of VLCFA are thought to be detrimental when incorporated into membrane phospholipids 61. The toxicity of C_26:0 was shown in several in vitro studies using diverse cell types including neural cells. In some experimental designs, tissues or cells with Abcd1 gene knockout or knockdown were used giving rise to elevated VLCFA levels from endogenous origin 4, 53. In other studies, cells were incubated with exogenously added C_26:0 up to 50 μM 60, 86, 154. It should however be noted that this exceeds the pathological concentrations of C_26:0 in X‐ALD patients which were estimated to be 1–5 μM 5, 28, 129, 137. Oxidative stress and mitochondrial impairments were recurrent cellular dysfunctions triggered by the VLCFA and were assumed to cause axonal degeneration in the spinal cord 52 and, together with an inflammatory component, the cerebral phenotype 119. Whether these mechanisms also underlie the demise of PCs is however unresolved.

Acyl‐CoA oxidase‐1 deficiency

ACOX1 catalyzes the first step in peroxisomal β‐oxidation. Its enzymatic function is restricted to the oxidation of VLCFA, long‐chain dicarboxylic acids and PUFA. ACOX1 deficiency is often referred to as pseudo‐neonatal ALD because of the similarities in clinical presentation with the PBD neonatal adrenoleukodystrophy (see below) 100. Patients have increased levels of VLCFA 41 although some ACOX1‐deficient patients with normal plasma VLCFA concentrations have also been described 111.

Currently, the pathogenic knowledge of ACOX1 deficiency is restricted to what has been observed in a limited number of patients worldwide. Delays in motor development such as head control, crawling and non‐assisted walking were described. In addition to a delay in attaining early developmental milestones, almost all ACOX1‐deficient children aged 24 to 48 months showed regression of motor achievements. Brain MRI studies demonstrated myelin loss starting in the white matter of the cerebellum, cerebellar peduncles and brainstem tracts, progressively extending to the midbrain and the cortical white matter 21, 80, 100, 111, 125, 146, 149.

In two ACOX1‐deficient siblings with a mild disease course, developmental milestones were normal but clumsiness appeared at an age of 8–10 years evolving in progressive unsteadiness and severe ataxia. Interestingly, the cerebellum, middle cerebellar peduncles and brainstem showed marked atrophy whereas the cerebrum was only mildly affected 43. These patients survived into their fifties. This moderate clinical phenotype was associated with only borderline increased levels of VLCFA, likely caused by a mutation outside of the catalytically active parts of the enzyme 51, 125.

Acox1^−/− mice present with a severe hepatic phenotype and, to our knowledge, the neuropathology has not been thoroughly investigated. Given that VLCFA accumulate in blood and ACOX1 expression was confirmed in neural and glial cells of the cerebellum and other motor‐related regions such as the pontine nuclei and inferior olive 35, 49 it is surprising that no neurological signs were reported.

Multifunctional protein‐2 deficiency

MFP2 deficiency occurs with an incidence of approximately 1:100 000 39. MFP2, alternately known as D‐bifunctional protein, is encoded by the hydroxy‐steroid dehydrogenase type 4 (HSD17B4) gene. MFP2 catalyzes the oxidation of a broad spectrum of substrates including VLCFA, 2‐methyl‐branched‐chain fatty acids (eg, pristanic acid), bile acid intermediates [dihydroxycholestanoic acid (DHCA) and trihydroxycholestanoic acid (THCA)], leukotrienes and long‐chain dicarboxylic acids. In addition, MFP2 is involved in the biosynthesis of DHA, the most common PUFA in the mammalian brain and retina 39.

Cerebellar and brainstem pathologies in MFP2‐deficient patients

The importance of MFP2 in the developing CNS is highlighted by the fact that the majority of MFP2‐deficient patients display severe abnormalities at birth, presenting with hypotonia and brain malformations, and die within the first years of life 39, 145. This phenotype is indistinguishable from Zellweger disease (see below). Concerning the cerebellum, imaging experiments uncovered hypoplasia, demyelination and atrophy. Brain autopsy studies revealed ectopic or degenerating PCs, gliosis and defects in the migration and maturation of granule neurons. Brainstem pathologies often include demyelination and malformations of the inferior olivary and dentate nuclei 39, 69, 71, 95, 98, 138.

Depending on the nature of the HSD17B4 mutation, MFP2‐deficient patients may achieve some developmental milestones followed by regression. Similar to mild PBD patients (see below), central cerebellar demyelination is then a common feature 39, 122. Importantly, several patients who presented with progressive cerebellar ataxia during childhood, adolescence or adulthood combined with peripheral neuropathy and hearing loss were identified in recent years as HSD17B4 mutants by next generation sequencing 82, 84, 89. The residual enzyme activity led to normal plasma levels of VLCFA and branched‐chain fatty acids and they were proposed to represent a novel subtype of MFP2 deficiency based on their clinical, genetic and biochemical features.

Of interest, several syndromes encompassing cerebellar manifestations could also be ascribed to MFP2 deficiency. For example, HSD17B4 mutations are the first recognized genetic cause of Perrault syndrome 99, characterized by ovarian dysgenesis and sensorineural hearing loss but often also associated with neurological problems. Shrinkage of the cerebellum and motor problems in Perrault patients are similar to the clinical profile of mild MFP2‐deficient patients 47, 57, 96, 99. However, mutations in several other genes have been identified to cause Perrault syndrome. Furthermore, the importance of MFP2 in ataxic phenotypes can be deduced from patients with Stiff‐man syndrome, a rare disorder of the CNS characterized by the production of auto‐antibodies against glutamic acid decarboxylase, resulting in muscle rigidity and spasms. MFP2 is an autoimmune target in a subset of Stiff‐man patients 26, 30, 62 and gait ataxia is a facultative neurological symptom 75. A correlation between the concentration of MFP2 auto‐antibodies and the severity of cerebellar ataxia has not been established so far, but the expression of both GAD and MFP2 97 in cerebellar PCs is in favor of PC dysfunction underlying the motor problems in these patients. Another post‐developmental disease linked to MFP2 deficiency is optico‐cochleo‐dentate degeneration characterized by cerebellar symptoms, regression of motor performance and degeneration of the optic nerve, the dentate nucleus and the cochlear nerve 116. Clinically, the distinction between mild MFP2 deficiency, Perrault syndrome caused by MFP2 deficiency, and optico‐cochleo‐dentate degeneration appears artificial, and all these syndromes could be gathered under “MFP2 ataxia syndrome.”

Cerebellar and brainstem abnormalities in mouse models with MFP2 deficiency

In contrast to patients with a total ablation of MFP2 function, cerebral development was normal and cerebellar formation was only mildly affected in Mfp2^−/− mice 6, 65. A delay in cerebellar foliation was observed in 7 days old Mfp2^−/− mice, but cerebellar architecture appeared normal in the post weaning period 77. However, progressive motor problems such as clasping of hind legs and poor performance on the rotarod already presented in four weeks old Mfp2^−/− mice and evolved in immobility and death by the age of 6 months 65, 140. Peripheral sensorimotor abnormalities were absent in Mfp2^−/− mice, pointing to MFP2 deficiency in the CNS as the origin of the motor problems 65. Studies in Nestin‐Mfp2^−/− mice, a model in which MFP2 is deleted from neural cells, showed a similar motor phenotype as the Mfp2^−/− mice 140.

Histologically, several lesions were found in the cerebellum of both mouse models. Prominent lipid droplets containing neutral lipids progressively accumulated in Bergmann glial cells, whereas smaller inclusions were occasionally observed in cerebellar PCs. Interestingly, astrogliosis and up‐regulation of the antioxidant enzyme catalase were also found in the molecular layer of the Mfp2^−/− cerebellum 65, 140. These observations raised the question whether astroglial lesions underlie the neuromotor problems in MFP2 deficiency. It is however unlikely that the lipid droplets and catalase overexpression are detrimental as they also occurred in the brain of Gfap‐Pex5^−/− mice (see also below) that do not develop neurological problems and have a normal life span. Additional microscopic analyses in both Mfp2^−/− and Nestin‐Mfp2^−/− mice uncovered swellings on myelinated PC axons, followed by cerebellar microgliosis and profound astrogliosis in the deep cerebellar nuclei. Myelin loss started in the cerebellar lobules and expanded to the central white matter. In 12 months old Nestin‐Mfp2^−/− mice, a time point at which Mfp2^−/− mice already passed away, degeneration of the PC axon and dendritic tree culminated in PC loss and cerebellar atrophy. The fact that axonal abnormalities preceded neurodegeneration was compatible with a dying back neuropathy 140. It was investigated whether the absence of oligodendroglial MFP2 plays a causal role in the cerebellar demyelination by using Cnp‐Mfp2^−/− mice, an oligodendrocyte selective knockout model. However, these mice showed normal motor skills and their cerebellum was unaffected until the age of 12 months ruling out a primary role for peroxisomal β‐oxidation in oligodendrocytes for the maintenance of the cerebellum 140.

The early‐onset ataxic phenotype of both Mfp2^−/− and Nestin‐Mfp2^−/− mice clearly preceded cerebellar histopathological changes 91, 140. As in other neurodegenerative diseases such as Huntington's disease 132 and spinocerebellar ataxia type 1 64 early cell dysfunction rather than cell death might be involved in the initial disease stages. Currently, the cellular mechanisms underlying loss of PC integrity and cerebellar demise in MFP2 knockout mice are still unsolved. One of the outstanding questions is whether the PC degeneration is inherent to peroxisomal β‐oxidation loss within these cells, which is supported by their elevated MFP2 expression as compared with other neural cells 91, or by non‐cell autonomous mechanisms.

Candidate metabolites causing cerebellar defects in MFP2 deficiency

Given the broad substrate spectrum of MFP2, several metabolic disturbances could account for cerebellar degeneration in MFP2‐deficient patients and mice. However, it remains unresolved how defective peroxisomal β‐oxidation impacts on cerebellar integrity. First, as in ABCD1 deficiency, VLCFA accumulate in the nervous system after loss of MFP2. However, when comparing the mouse models, it is striking that cerebellar degeneration occurs much earlier and is more drastic in Mfp2^−/− than in double Abcd1/Abcd2^−/− mutants. Furthermore, the absence of a clinical phenotype in Cnp‐Mfp2^−/− mice, despite elevated C_26:0 levels in the cerebellum, does not support a crucial role for VLCFA in cerebellar degeneration 140.

MFP2 is also necessary for the breakdown of pristanic acid, the branched‐chain fatty acid formed by α‐oxidation of phytanic acid, both presumably having a similar toxicity profile when accumulating. As already mentioned, mice are exposed to very low levels of these fatty acids when fed a normal diet. Whereas in the Refsum mouse model, the cerebellum only degenerated after supplementation of phytol, this occurred in MFP2‐deficient mice on a standard diet. In addition, the cerebellum of Mfp2^−/− and Nestin‐Mfp2^−/− mice is free of phytanic or pristanic acid accumulation and the ataxic behavior of Mfp2^−/− mice fed with pellets enriched in the branched‐chain fatty acid precursor phytol does not differ from MFP2 knockout mice on a normal diet 65. As already mentioned, it is striking that plasma levels of VLCFA and the branched‐chain fatty acids are normal or near‐normal in MFP2 patients with post‐developmental cerebellar degeneration (MFP2 ataxia syndrome) 43, 82, 84, 89, 99, 122.

Another potential metabolic candidate is a lack of DHA. It is well established that this PUFA requires one peroxisomal β‐oxidation cycle for its synthesis and in MFP2‐deficient patients, a shortage of this important PUFA was shown 38. Quite surprisingly, and for unclear reasons, DHA levels are normal in cerebellum and other brain regions of MFP2 knockout mice 65. Unless cell type or regional differences in DHA concentration would exist, a lack of PUFA cannot be claimed as the origin of cerebellar pathology.

In summary, the cerebellar pathology in Mfp2^−/− and Nestin‐Mfp2^−/− mice cannot be correlated with levels of known MFP2 substrates nor by myelin loss. It appears therefore that MFP2 could have other, unknown functions.

Alpha‐methylacyl‐CoA racemase, sterol carrier protein X and ABCD3 gene mutations

AMACR, SCPX and PMP70 are involved in the metabolism of branched‐chain substrates by peroxisomal β‐oxidation. AMACR converts 2R‐pristanic acid, 25R‐DHCA and 25R‐THCA to their S‐configuration, prior to their passage through the peroxisomal β‐oxidation pathway. Most AMACR‐deficient patients are asymptomatic until adolescence where after they start showing symptoms similar to those observed in Refsum disease patients 29, 59, 131. Cerebellar dysfunction was only seen in two out of ten patients 24, 29, 59. An MRI study on these two siblings revealed chronic degenerative lesions of efferent cerebellar pathways 59. Altered signal intensities in the dentate nucleus, superior cerebellar peduncle, thalamus and red nucleus indicated degeneration of the cerebellothalamic, dentatothalamic and dentatorubral tracts. In addition, pontine atrophy and signal changes of the transverse pontine fibers also revealed involvement of cerebellar afferents. No inferior olivary lesions were observed 59. The sensory (‐motor) neuropathy observed in most patients likely contributes to their motor abnormalities 59.

Pristanic acid also accumulates in patients with mutations in the SCPx gene. Once SCPx is imported into the peroxisome, it is split into two functional domains: a thiolase domain that catalyzes the final step of the peroxisomal β‐oxidation, and a sterol carrier protein 2 (SCP2) domain. Gait disturbances, tremor and impairment of balance in an adult SCPx‐deficient patient was consistent with cerebellar ataxia and slowed saccades indicated brainstem malfunction 40. As in other peroxisomal patients with defective branched‐chain fatty acid catabolism, the motor and sensory neuropathy should not be overlooked as an additional source of motor problems. Similar to Refsum disease patients, restriction of phytanic acid intake halted symptom progression in the SCPx patient 40, 43, while in AMACR‐deficient patients, the benefit of a phytanic acid restricted diet is inconclusive 121. Scpx knockout mice challenged with a phytol diet developed ataxia and peripheral neuropathy with an unsteady gait 117 but this was not the case in Amacr knockouts 113. Unfortunately, the cerebellum was not investigated in these mouse models.

Finally, the transport of the C₂₇‐bile acid intermediates and branched‐chain fatty acids across the peroxisomal membrane requires ABCD3 also denoted as PMP70. In a single recently identified ABCD3‐deficient patient and in the mouse model, no cerebellar defects were observed 44. However, the patient died at young age from liver disease and the mice were only supplemented with phytol for a short period, precluding potential effects of branched‐chain fatty acids on the central or peripheral nervous system.

Ether Phospholipid Deficiency

Ether phospholipids are a subclass of glycerophospholipids with a long‐chain fatty alcohol at the sn‐1 position. Of all ether lipid species, plasmalogens that have a vinyl‐ether bond are the most abundant in mammalian tissues 15, 81, 94. The synthesis of this ether bond requires the enzymatic machinery in peroxisomes, namely glyceronephosphate O‐acyltransferase (GNPAT) and alkylglycerone phosphate synthase (AGPS) 56 (Figure 2). Being an essential constituent of myelin and neurological membranes, plasmalogens are thought to have a pivotal role in signal transduction both in the central and peripheral nervous system. At the molecular level, they were suggested to act as antioxidants, as a reservoir of PUFA and as mediators of membrane dynamics 15, 83, 95.

Rhizomelic chondrodysplasia punctata

Rhizomelic Chondrodysplasia Punctata (RCDP) has an overall incidence of 1:100 000 10. RCDP type 1 is caused by mutations in the PTS2 receptor PEX7, impairing the import of three peroxisomal enzymes, that is, AGPS, phytanoyl‐CoA hydroxylase and 3‐ketoacyl thiolase. The majority of PEX7 deficient patients show a very severe to moderate depletion in plasmalogen levels together with an increase in plasma phytanic acid levels 10. In a few PEX7 patients 63, 134 mutations are so mild that plasmalogen levels are close to normal and, as already mentioned, their phenotype is similar to Refsum disease patients, caused by impaired activity of phytanoyl‐CoA hydroxylase 67. In all other cases, the phenotype is fully determined by the impaired ether‐phospholipid synthesis. This is deduced from the fact that the clinical signs are indistinguishable from patients with RCDP type 2 or 3 with an isolated defect in ether‐phospholipid synthesis caused by mutations in respectively GNPAT and AGPS.

Cerebellar and brainstem pathologies in RCDP patients

The clinical image of RCDP types 1, 2 and 3 is generally severe but varies according to the residual capacity in plasmalogen biosynthesis 9. The most severely affected patients with undetectable plasmalogens in red blood cells showed progressive cerebellar atrophy caused by PC and granule cell degeneration in the neonatal period. The onset of cerebellar deterioration was delayed by 1 year in patients with residual albeit very low plasmalogen levels 9, 10. Non‐inflammatory myelin abnormalities were frequently observed 2, 102, 126, 127 and dysplastic olives and cerebellar heterotaxias noted in post‐mortem studies may refer to developmental defects 101, 102, 123. The achievement of motor skills in patients with the severe phenotype is poor 9. Other typical features include skeletal anomalies, audiovisual problems and psychomotor retardation with a considerably reduced life expectancy.

Less is known about the disease process of chronic RCDP patients with a milder clinical and biochemical course. The majority shows a delay in motor development such as the inability to sit or walk independently 9, 10, 12. Furthermore, patients with mild ataxia, nystagmus and cerebellar atrophy with PC and granule cell depletion have also been reported 103. As in the severely affected children, abnormalities in cerebellar architecture or myelination were not always present in these mild RCDP patients 9, 12.

Insights from mouse models

Knowledge about the consequences of deficient ether lipid synthesis has greatly increased by the analysis of Pex7^−/− and Gnpat^−/− (also called Dhapat^−/−) mouse models 14, 17, 83, 109. Cerebellar development was mainly studied in Gnpat knockout mice revealing foliation abnormalities, a delay in granule cell migration and atrophy during the first postnatal weeks 77, 130. The PC showed multiple aberrations including axonal spheroids, defects in paranode organization and a hyperspiny appearance accompanied by altered CF and PF innervations 130. Dysmyelination occurred both in the cerebellum and the cerebrum 130 but cerebellar myelin loss was not progressive 13. Decreased rotarod and vertical pole performances in Gnpat^−/− mice suggested cerebellar ataxia 130. However, a recent study on both Pex7 and Gnpat knockout mice has shown that plasmalogen deficiency in the peripheral nervous system affects axonal sorting and the myelination process by influencing Schwann cell differentiation and maturation, thereby causing peripheral neuropathy. Impairment of the AKT–GSK3β pathway induced by plasmalogen deficiency in Schwann cell membranes was proposed as the underlying mechanism 25. Therefore, the ataxia of Gnpat^−/− mice may have a dual origin.

Fatty acyl‐CoA reductase 1 deficiency

The importance of plasmalogens for the maintenance of cerebellar integrity is underscored by the recent identification of a few patients with Fatty Acyl‐CoA Reductase 1 (FAR1) deficiency. FAR1 plays a critical role in the plasmalogen biosynthesis pathway by reducing saturated and unsaturated fatty acyl‐CoAs to their respective fatty alcohols. The enzyme is located at the cytosolic side of the peroxisomal membrane. Remarkably, a 19‐year‐old FAR1‐deficient patient presented with atrophy of the cerebellar vermis and hemispheres accompanied by cortical white matter lesions 18.

Peroxisome Biogenesis Defects

The import of peroxisomal matrix and membrane proteins requires the concerted action of at least 14 peroxins that have diverse functions including cytosolic receptors (PEX5 and PEX7) for enzymes carrying the peroxisome targeting signal (PTS) 1 or 2, docking proteins in the peroxisomal membrane and proteins that allow the membrane translocation and the recycling of receptors to the cytosol 150 (Figure 3). PEX7 is only needed for the import of a minority of proteins harboring a PTS2 signal (AGPS, phytanoyl‐CoA hydroxylase and 3‐ketoacyl thiolase). As already mentioned, severe PEX7 mutations cause RCDP, whereas very mild mutations cause Refsum pathologies because of, respectively, plasmalogen synthesis and α‐oxidation defects 134. Defects in all the other peroxins will impede all peroxisomal functions leading to an array of clinical presentations now designated as Zellweger spectrum disorders 16, 124. The phenotypes range from severe developmental disorders (Zellweger syndrome), that are fatal in the first year of life, to milder syndromes previously named neonatal adrenoleukodystrophy and Infantile Refsum disease that sometimes allow survival into adulthood.

Figure 3.

Schematic overview of cerebellar and brainstem pathologies in PBD in men and mice. PBD are caused by mutations in proteins involved in peroxisomal matrix protein import and peroxisomal membrane protein (PMP) import. PMPs contain a C‐terminal (PTS1) or a N‐terminal (PTS2) peroxisomal targeting signal. PTS1 matrix proteins are recognized by a shorter variant of the peroxisomal import receptor PEX5 (5S). The PTS1/PEX5 complex binds to the docking complex at the peroxisomal membrane, which leads to import of the protein in the peroxisomal lumen. PTS2 proteins are imported in a similar way, but are first recognized by PEX7 that binds to the longer variant of PEX5. Peroxisomal membrane proteins are incorporated via a mechanism involving PEX19, PEX16 and PEX3.

Developmental cerebellar pathologies

Newborns with Zellweger syndrome, also known as the cerebro‐hepato‐renal syndrome, present with hypotonia and craniofacial dysmorphisms 16. MRI of the brain shows gyric abnormalities including polymicrogyria in the Sylvian fissure, frontoparietal pachygyria and heterotopias 11, 151, indicative of a neuronal migration defect. Brain atrophy is another common feature. The landmark histological studies by Volpe 141 and Evrard 34 showed that in Zellweger syndrome, besides the cytoarchitectonic abnormalities of the cortex, malformations of the cerebellum and the inferior olivary nucleus occur. In the postnatal cerebellum, a subset of PCs are heterotopically located in the white matter, or abnormally positioned relative to the granule cells as a result of hampered neuronal migration. The dysplasia of the olivary and dentate nuclei also encompass laminar discontinuities. By MRI, no abnormalities were however observed in the cerebellum within the age of 2 months 11.

Cerebellar malformation in PBDs was studied in detail using mouse models lacking PEX2, PEX5 or PEX13. Depending on the genetic background, the global knockout mice die in the perinatal period or survive only for a few weeks 7. Only in the longer surviving Pex2^−/− knockouts, cerebellar development could be studied as it occurs postnatally in mice. Alternatively, neural selective knockouts were generated for PEX5 and PEX13 circumventing severe liver pathology in the postnatal period and enabling to study cerebellar histogenesis 76, 93.

A thorough investigation in Pex2^−/− mice revealed multiple anomalies in the developing cerebellum in which granule cells, PC and CF were affected leading to persistent abnormalities in cerebellar foliation 36. The migration of granule cells from the external granule cell layer to the internal granule cell layer was delayed. As the Bergmann glia scaffold appeared normal, the migration impairment was due to an intrinsic problem of the granule cells, which also showed increased apoptotic death. The impaired granule cell migratory capacity was confirmed using in vitro setups 37. The heterotopia of PCs was less pronounced in the mice as compared with Zellweger patients as only a minor fraction of PCs were slightly delocalized to the internal granule cell layer. However, although the initial development of PC was indistinguishable from wild type until postnatal day 3, the subsequent maturation was impaired with less complex branching of the dendritic tree and aberrant formation and orientation of dendritic spines. The latter coincided with the abnormal development of CFs. Although these innervated the PC somata at P3 in a normal fashion, the translocation of the CF dendritic tree to the PC dendrite compartment was delayed 36. In addition, the axonal compartment of PCs was affected with the occurrence of spheroids, altogether indicative of ongoing degenerative processes.

These data were confirmed in mice with inactivation of Pex13 93 or Pex5 76 restricted to neural cells by using Nestin‐Cre and respective floxed mice. Their cerebellar development was abnormal including granule cell migration impairment, increased cell death and defects in PC positioning and arborization. In cultured cerebellar Pex13^−/− neurons, oxidative stress mediated by mitochondrial dysfunction was demonstrated 93. Surprisingly, no elevation of VLCFA was found in Nestin‐Pex13^−/− brains eliminating this as a potential mechanism.

Remarkably, also in mice with a liver selective depletion of functional peroxisomes, major abnormalities in cerebellar morphogenesis leading to impaired foliation were observed 76, indicating that brain extrinsic mechanisms also play a role. The mechanisms were not resolved but given that bile acid treatment can partially restore the cerebellar anomalies in Pex2 knockout mice, accumulation of immature bile acids may be involved 37.

Taken together, the conservation of cerebellar dysgenesis in peroxisome‐deficient mice and men, underscores the necessity of intact peroxisomal metabolism for the normal formation of the cerebellum. Both nervous system intrinsic and extrinsic factors may impact on the intricate interplay between granule cells, CFs and PCs.

Milder peroxisome biogenesis defects and their cerebellar pathologies

The milder Zellweger spectrum disorders have a variable onset and presentation. The age at diagnosis ranges from the neonatal period up to adulthood and survival varies from a few years to several decades. Impairment of vision and hearing, psychomotor retardation and liver disease are among the most frequent clinical signs. Other symptoms include hypotonia, leukodystrophy and ataxia. A major obstacle for the diagnosis is that the typical peroxisome‐related biochemical anomalies (increased levels of VLCFA, branched‐chain fatty acids, bile acid intermediates and plasmalogens) are often not or only borderline affected. Frequently, the genetic cause is a point mutation in PEX1, the gene accounting for more than 50% of Zellweger spectrum cases 150.

Leukodystrophies initiate in the hindbrain

Infants with the adrenoleukodystrophy phenotype of PBD reach some developmental milestones but deteriorate after the age of 1 year 135. The leukodystrophy affects both the cerebrum and the cerebellum and can be stable or progressive. In the absence of clear‐cut biochemical peroxisomal hallmarks, the differential diagnosis with other leukodystrophies is challenging. By performing sequential MRI, a typical pattern was revealed for the ZSS, whereby abnormalities start in the hilus of the dentate nucleus and the superior cerebellar peduncles before affecting the cerebellar white matter and the brainstem. The white matter degeneration in the forebrain appears thereafter 135. Other unique presentations were reported such as an infant with acute neurological deterioration accompanied by inflammatory demyelination in the brainstem 78. Remarkably, in the Nestin‐Pex5 mouse model lacking functional peroxisomes in neural cells, extensive inflammatory demyelination develops. This also initiated in the cerebellum and progressed to brainstem, cortex and corpus callosum mimicking the sequence in patients 13.

PEX16 mutations as a cause of cerebellar ataxia

The PEX16 protein is essential for peroxisomal membrane assembly, which precedes the import of matrix proteins. In six PEX16 mutant patients, the most pronounced neurological defects were related to cerebellar dysfunction 32. They developed spastic paraplegia and ataxia in preschool years, whereas cognitive function was unaffected. Nevertheless, besides cerebellar atrophy, widespread changes in white matter in both sub‐ and supratentorial regions were detected and several patients were diagnosed with a demyelinating motor and sensory neuropathy. Plasma VLCFA were mildly increased in all cases, whereas branched‐chain fatty acids only in half of them.

The ring finger PEX disorders: PEX12, PEX10 and PEX2

An emerging subgroup of mild Zellweger spectrum patients present with ataxia as a primary clinical sign. Gootjes et al 54 reinvestigated a patient first diagnosed with trihydroxycholestanoic acidemia and found that PEX12 deficiency was the cause of the clinical and biochemical abnormalities. The patient presented around age 5 with mild intellectual disability, cerebellar ataxia, hypotonia and absent reflexes. Later, she developed retinopathy. Cerebellar imaging was not reported. Two patients with compound heterozygosity for PEX10 showing an analogous clinical presentation were reported by Régal et al 107. After a normal early development, they were referred to the clinic between 6 and 8 years because of worsening gait disturbances. Cerebellar abnormalities such as gait ataxia and dysarthria were confirmed by MRI, showing cerebellar atrophy. Both patients also showed an axonal motor neuropathy and posterior column dysfunction. Cognition was normal in both patients, and there was no retinopathy. Although VLCFA were not increased, a peroxisomal disorder was suspected because of increased phytanic and pristanic acid levels.

A pure cerebellar syndrome was seen in two brothers with a mutation in PEX2 118. Although both patients had the same mutation and showed similar clinical features, there was a striking difference in the onset of symptoms. One patient started to develop gait disturbances around 3.5 years and was diagnosed at 14 years with isolated progressive cerebellar ataxia. The other patient showed cerebellar signs such as impaired gait and dysmetria around the age of 18 years. Both patients showed pronounced cerebellar atrophy but no signs of myelin abnormalities or neuronal migration defects on MRI. Also in these patients, only pristanic acid and phytanic acid were moderately elevated, whereas VLCFA levels were normal. In a third PEX2 patient with a very similar clinical phenotype, biochemical analysis revealed a slight increase in VLCFA levels 90.

Although their numbers are still low, it is intriguing that these patients carry a mutation in PEX12, PEX10 or PEX2. These peroxins act as ubiquitin ligase (E3)‐like proteins and contain a RING finger domain. Mono‐ubiquitination allows PEX5 to be recycled, whereas poly‐ubiquitination targets PEX5 to the proteasome for degradation 50.

At present, it is unclear why the mutations in these genes with a common task in the peroxisome biogenesis process give rise to the cerebellar ataxic phenotype. Metabolically, these patients share mildly increased levels of branched‐chain fatty acids but minor alterations in VLCFA. Interestingly, not all mild PBDs cause ataxia. Indeed, the phenotype of mild PEX6 deficiency is characterized by retinopathy and deafness (105 and pers. comm.), signs absent in mild PEX10 and mild PEX2 deficiency, but no cerebellar degeneration. Our current knowledge of the function of these genes as necessary for global peroxisomal function does not explain why mild defects result in gene‐specific clinical syndromes. At least two possible explanations deserve further investigation. The reserve for normal peroxisomal function may be different for different PEX genes in different tissues. Another explanation would be unknown additional functions beside peroxisomal biogenesis.

Conclusion

It is striking that dysfunction of each of the three major peroxisomal lipid pathways gives rise to cerebellar defects. α‐Oxidation impairment causes a pure degenerative process that can be halted by dietary intervention. The diversity in β‐oxidation and ether lipid synthesis diseases is much wider, ranging from developmental malformations of the cerebellum and/or brainstem to degeneration later in life. It is remarkable that, during the last decade, besides the prototypical severe peroxisomal diseases such as Zellweger syndrome, MFP2 deficiency and CCALD, a number of patients have been diagnosed with mild deficiencies. Their identification is hampered by their apparent rarity, clinical signs that are not pathognomonic such as ataxia, neuropathy and psychomotor retardation in combination with (near) normal levels of peroxisomal metabolites in blood. There is no doubt that with the increasing application of genetic techniques such as exome sequencing, many more of these patients will be identified.

The mechanisms linking metabolic abnormalities to cerebellar pathology in peroxisomal disorders remain largely unresolved. Phytanic acid levels correlate with ataxia in Refsum disease, but the precise molecular impact (in PCs) needs to be further elucidated. Furthermore, it remains unclear whether and how elevated levels of VLCFA affect the cerebellum. It is most intriguing that in recently diagnosed adolescent or adult peroxisomal patients with progressive ataxia, most peroxisomal metabolites are (near) normal, contrasting with the high concentrations necessary to cause acute neurotoxicity in vitro. This raises the question whether unknown metabolic factors may be at play. To sort this out, unbiased approaches should be used to unravel whether mild peroxisome dysfunction may deregulate other metabolites. With regard to the cerebellar pathologies induced by the lack of ether lipids, the recent demonstration that AKT‐GSK3β signaling is impaired in the peripheral nervous system of an RCDP mouse model might shed new light on the pathogenesis in the brain 25.

Cerebellar PCs are among the most metabolically active of all neurons, thereby making them more vulnerable for derangement of cellular homeostasis. This not only relates to peroxisomal deficiencies but also to lysosomal 8, 142, mitochondrial 23 and autophagy 1, 73 processes. In this view, it cannot be excluded that a primary peroxisomal metabolic defect impairs mitochondrial function or autophagy processes. Whereas some peroxisomal pathologies may be PC autonomous, others are likely caused by aberrations in the circuitry innervating these pivotal cells. This can be addressed by creating mouse models with PC selective inactivation of peroxisomal proteins. It remains to be elucidated whether neurons in other brain areas develop similar pathologies at later time points as a consequence of peroxisome dysfunction.