Absence of functional peroxisomes in the postnatal mouse CNS causes a neurodegenerative phenotype that ultimately leads to extensive axon loss and early death. This closely mimics pathologies in mild peroxisome biogenesis disorders and the cerebral form of X-ALD. We here investigated the involvement of micro- and astroglia activation, metabolic factors and oxidative stress in the onset and progression of myelin and axon loss triggered by peroxisome inactivity.
In all demyelinating areas, increased numbers of microglia were present. Remarkably, in cerebellum and cortex microglia were never very abundant, whereas they were strongly represented in brain stem starting in the juvenile period and they massively invaded the demyelinating corpus callosum of 12-week-old mice. Also of note is that astroglial activation lagged behind microglia proliferation, as described before in other neurodegenerative diseases . Many microglia displayed features of phagocytosing cells including a swollen appearance, sometimes containing MBP-positive myelin debris and surrounding a demyelinating fiber. This was also confirmed by increased transcripts for macrophage markers such as Mpeg1 and lysosomal enzymes.
According to recent insights, the number and shape of microglia does not predict the type of immunological reaction that is developing. Besides the classical pro-inflammatory microglia, macrophages occur that are involved in the resolution of inflammation which have been named “alternatively activated” or “deactivated” . In all brain regions, markers of a pro-inflammatory response, such as TNFα, IL6 and IL1β were highly up regulated . Although both markers of alternative microglial activation (Arginase 1, Fizz1) and of acquired deactivation (IL10, TGFβ) were up regulated in peroxisome deficient corpus callosum before the peak of microglial activation occurred, this could not halt the progression of inflammation.
It was remarkable that the induction of markers of the innate immune system occurred at a very early stage in the disease process, within 3 weeks after birth. Strikingly, it was recently reported that peroxisome deficiency in Pex1 mutant Drosophila larvae also caused up regulation of genes involved in innate immunity, further supporting that peroxisome deficiency modulates immunological responses . In addition, peroxisomes were shown to be involved in the generation of defense factors following a viral infection and thus participate in antiviral signaling . The precise role of peroxisomes in shaping the innate immune responses will need to be further elucidated.
With regard to neuronal morphology, we did not find much evidence for cell death  but there was progressive axonal loss throughout the CNS which was already very obvious at the age of 12 weeks based on neurofilament staining with antibody SMI31. At this and earlier ages, axonal swellings (detected by staining with anti-APP) and damage (detected by staining unphosphorylated neurofilaments with SMI32) could be visualized, but these stainings strongly underestimated ongoing axonal degeneration. By EM analysis, some axonal irregularities were already observed at the age of 3 weeks, illustrating the very early onset of the degenerative phenotype. The progressive axonal loss coincided with severe and progressive motor and cognitive decline, which we reported before . In this respect, it is interesting to mention that axonal injury is currently considered as the most important cause of clinical disabilities in MS [33, 34].
With the exception of a variable onset of lesions in the cortex, the development of brain pathology was consistent from mouse to mouse. It was also recapitulated, but delayed in time, in mice in which peroxisomes were deleted from the CNS in the juvenile period after completion of myelination. This ruled out the possibility that the observed anomalies were a late consequence of peroxisome ablation during development. Our present findings that demyelination in cerebellum precedes white matter abnormalities in cerebrum is in striking agreement with reports in mildly affected peroxisome biogenesis patients [35, 36] in which regressive changes predominate over developmental anomalies. By MRI analysis the first foci of white matter abnormalities were found in the central cerebellar area, additional lesions were often seen in the brain stem whereas the cerebral hemispheres were affected later. This distribution of white matter lesions is clearly different from the typical pattern in X-ALD in which the splenium of the corpus callosum is affected first. Remarkably, the leukoencephalopathy in PBD patients was not predictive for the clinical outcome  suggesting that additional pathology causes the psychomotor retardation. In addition, patients have been identified with milder PEX mutations, displaying normal mental capacities, but developing progressive ataxia, caused by cerebellar atrophy [36–38]. Why cerebellar neurons are selectively affected in these longer surviving patients is not clear.
A direct comparison of the lesions in Nestin-Pex5
mice with those in Gnpat
−/− mice allowed to determine the contribution of hampered ether phospholipid synthesis to the observed pathologies. Ether phospholipids are quintessential products of peroxisomal metabolism that are very enriched in the CNS and myelin under the form of plasmalogens. Gnpat
−/− mice were previously shown to have myelin deficits in cerebellum and swellings on Purkinje cell axons . Although the pattern of hypomyelination, axonal swellings and axonal loss was very similar in cerebellum of juvenile Gnpat and Nestin-Pex5 knockout mice, with increasing age, Gnpat
−/− mice did not further lose myelin in cerebellum, nor in other brain areas. Also in rhizomelic chondrodysplasia punctata RCDP patients, abnormalities in white matter were reported which were however rather located in supratentorial areas . In sharp contrast with Nestin-Pex5 knockout mice no phagocytotic microglia nor reactive astrocytes were detected at the age of 6 months throughout the Gnpat
−/− brain which was further confirmed by the absence of pro-inflammatory markers. These findings are in accordance with the pathology observed in the Pex7
mouse model for RCDP type 1. At the age of 9 – 11 months the latter mice do not display a reactive response of microglia  but they show mild astrocytosis. Likewise, in RCDP patients non-inflammatory dysmyelination and occasionally astrocytosis, but no inflammatory demyelination has been reported . A mild degree of gliosis is observed in a case of GNPAT deficiency . In view of the inflammatory demyelination in patients with peroxisomal β-oxidation deficiency , the latter metabolic defect is likely facilitating the inflammatory response in brain lacking peroxisomes. Extensive gliosis was indeed observed in Mfp2 knockout mice although the pro- versus anti-inflammatory character was not investigated yet. A role for plasmalogen shortage as a factor synergizing with peroxisomal β-oxidation defects to induce inflammation can however not be excluded. Indeed, microglia activation was observed in the Pex7:Abcd1 double knockout mouse model but not in the single knockouts .
An important question is whether the axon injury culminating in axonal loss is due to myelin abnormalities, to neuroinflammation, to other factors, or a combination of these. We can indeed exclude that this is the consequence of peroxisome deficiency within neurons, as we previously showed that neuron selective inactivation of Pex5 did not cause a neurodegenerative phenotype . Based on fluorescence microscopy, and particularly obvious in cerebellum, axonal swellings and degeneration mostly appear after demyelination, indicating that the lack of myelin evokes loss of the denuded axon. On the other hand, as previously reported and now confirmed at younger ages, ultrastructural analysis showed degenerating axons that were still surrounded by a full myelin sheath, indicating that alternative mechanisms may also be operating. The milder phenotype of Gnpat knockout mice, in which an inflammatory reaction is absent, strongly indicates that the early and severe activation of the innate immune system contributes to the neurodegenerative phenotype in Nestin-Pex5
mice. The detrimental impact of neuroinflammation on axonal survival is well established. In MS patients, axonal loss is closely related to the degree of inflammation in the active lesions [34, 44]. In addition, it should be kept in mind that peroxisomes play an essential role in oligodendrocytes, because the selective loss of functional peroxisomes from these cells also results in inflammatory demyelination . In this respect the early expression of the complement component C1q on oligodendrocytes and neurons might be indicative of an endangered response which secondarily evokes inflammation. Indeed, although the complement system can be stimulated by both the innate and the adaptive immune system, it here seems to play a role in the early activation of innate immunity as the adaptive system is only involved much later with infiltrating T cells. Also in a mouse model for spinal cord injury, C1q is expressed on axons and oligodendrocytes . In Alzheimer disease, C1q causes activation of C3d, and via the complement pathway it can activate microglia with a damaging effect on myelin and axons . Although the relationship between peroxisomes and innate immunity needs further investigation, we speculate that metabolic abnormalities, likely related to peroxisomal β-oxidation defects, initiate an early activation of the innate immune system, which together with abnormalities in the formation and maintenance of myelin creates an environment which is detrimental for axons.
The absence of oxidative stress in young Nestin-Pex5
mice is in contrast with recent findings in Abcd1 knockout mice, a model for the AMN form of X-ALD. In these mice oxidative damage to proteins was detected in spinal cord but not in brain, preceding by several months axonal damage in spinal cord. The crucial role of this oxidative stress for the pathogenesis was further proven as anti-oxidative therapy could prevent the degenerative phenotype. Increased concentrations of the very long chain fatty acid, C26:0 is thought to be the metabolic factor initiating oxidative stress . It remains unclear why this occurs in spinal cord and not in brain with Abcd1 deficiency and likewise, why C26:0 accumulation in Nestin-Pex5 knockout mice  does not trigger an oxidative stress response. In fact, in patients with ABCD1 deficiency, lipid peroxidation in plasma was higher in the AMN phenotype than in cerebral ALD and in asymptomatic patients . On the other hand, in PBD patients oxidative stress markers were not increased in plasma or urine . Taken together, the causal relationship between oxidative stress and inflammatory demyelination is currently unclear and we have no evidence that degeneration in the peroxisome deficient brain is triggered or boosted by the development of oxidative stress.