A common theme for axonopathies? The dependency cycle of local axon homeostasis

The number of acquired or inherited conditions leading to axon degeneration (from now on referred to as axonopathies) is vast. To diagnose patients, clinicians use a range of indicators including physiology, morphology, family and patient history, as well as genetics, with the specific location of the lesion within the nervous system being a prominent feature. For the neurobiologist, these criteria are often unsatisfactory, and key questions remain unanswered. For example, does it make sense that different axonopathies affect distinct neuron groups through distinct mechanisms? Would it not be more likely that there are common routes to axon degeneration? In this opinion piece, I shall pose this fundamental question and try to find answers that are hopefully thought‐provoking and trigger some conceptual rethinking in the field. I will conclude by describing the ‘dependency cycle of axon homeostasis’ as a new approach to make sense of the intricate connections of axon biology and physiology, also suggesting that different axonopathies might share common paths to axon degeneration.

For an ever-increasing number of axonopathy-associated disorders, genetic links are being identified (Table S1), providing us with concrete molecular tools to study the mechanistic basis of axon decay.
However, even with this knowledge, it remains a challenge to explain how identified gene functions cause specific disease symptoms. When comparing functions of genes linked to these disorders (Table S1), several observations can be made: First, apart from spinal muscular atrophies (SMA), which is predominantly linked to the small ribonucleoprotein regulators survival of motor neuron 1 and 2 (SMN1 and 2) and was recently shown to also display a neurodevelopmental component (Kong et al., 2021), the other disease classes are usually caused by mutations in a variety of genes associated with a wide range of cellular functions (see colour code in Table S1A-C).
Second, different disorder classes seem to have prevalence for genes that relate to specific aspects of cell function; for example, F I G U R E 1 Schematic representation of motor and sensory neurons of the trunk. Upper motor neurons (UM; dark blue) project from the primary motor cortex into the spinal cord, terminating in the ventral horn of their target segments; here they form synapses with lower motor neurons (LM, red) that project through the peripheral nerves to form neuromuscular junctions with their target muscles or glands. Sensory neurons (SN, green) are pseudo-unipolar: they have their cell bodies in the dorsal root ganglia detached from their main axon via a short stem axon; the sensory neuron shown here is a myelinated Aβ-type low-threshold mechanoreceptor forming the longest fibres in our bodies; they project from the skin of our toes (lower arrow head) into the dorsal horn of the spinal cord where they form a branch that projects through the dorsal column white matter tract up to the brain stem (upper arrow head; Crawford & Caterina, 2019). Under each neuron type those axonopathies are listed that have their main lesion in this part of the nervous system (as listed in Box 2): HSP/hereditary spastic paraplegias; ALS/amyotrophic lateral scleroses, AMN/adrenomyeloneuropathy, SMA/spinal muscular atrophies, CMT/Charcot-Marie-Tooth diseases, dHMN/distal hereditary motor neuropathies, GAN/giant axonal neuropathy, HSAN/hereditary sensory and autonomic neuropathies [Color figure can be viewed at wileyonlinelibrary.com] BOX 1 Examples of acquired axonopathies (a) Physiological ageing; dying back pathology, axon swellings (Calkins, 2013;Chung et al., 2017;Marner et al., 2003;Salvadores et al., 2017) (b) Trauma; Wallerian degeneration (Bradke, Fawcett, & Spira, 2012;Curcio & Bradke, 2018) (c) Metabolic neuropathies (Minagar, 2010) • Vitamin E deficiency; dying back pathology, axonal swellings (Kohlschütter, 2009;Lampert, 1967) • Diabetic peripheral neuropathy/DPN; stocking-glove pathology (Juster-Switlyk & Smith, 2016) (d) Toxic neuropathies (Brandner, 2014) • Chemotherapy-induced peripheral neuropathies/CIPN; dying back pathology (Fukuda, Li, & Segal, 2017 • Neurotoxic hexacarbon-induced; dying back pathology, axonal swellings (Spencer & Schaumburg, 1977a, 1977b • Heavy metal-induced; stocking-glove pathology (Jang & Hoffman, 2011) (e) Upon inflammation or infectious diseases • Multiple sclerosis; dying back pathology (Cotsapas, Mitrovic, & Hafler, 2018;Dutta & Trapp, 2011;van den Berg, Hoogenraad, & Hintzen, 2017) • Carcinoma-associated paraneoplastic peripheral neuropathy; proximal-distal pathology (Dalmau, 1999) hereditary spastic paraplegias (HSPs), which primarily affect upper motorneurons, link to a disproportionately high number of genes associated with the formation or function of endomembrane organelles, preferentially the endoplasmic reticulum (Table S1B; Blackstone, 2018); this is surprising when considering that these organelles are essential in all cells of the brain and beyond. A similar statement can be made for Charcot-Marie-Tooth (CMT) diseases which affect primarily axons in peripheral nerves and show bias for genes associated with myelin formation or tRNA regulation (  (Bird, 2015;Pisciotta & Shy, 2018).

| IS THERE A COMMON GENETIC EXPLANATION FOR AXONOPATHIES?
The last notion might indicate that the different classes of axonopathies may, to a degree, root in a common fundamental cell biology upheld by comparable genetic networks (see also Züchner & Vance, 2005). Therefore, the observed bias of specific axonopathies linking to certain gene functions, may have to be interpreted within the specific context of the neurons affected in each case, as will be briefly discussed in the following.
Thus, different neuron types have different developmental histories, and their genetic programs specified during early neurogenesis, trigger distinct gene expression profiles translating into distinct gene expression patterns across nervous tissues (Jessell, 2000). ing myelination in the peripheral nerves but not CNS because they are specifically expressed only in peripheral myelin (Stassart, Möbius, Nave, & Edgar, 2018). Based on this thinking, one might be able to predict novel gene linkages: for example, microtubule crosslinking factor 1 (MTCL1) is an organiser of the axon initial segment specifically expressed in Purkinje cells (Satake et al., 2017) and might therefore turn out to be linked to HCAs in the future.
Furthermore, different neuron types display distinct properties that make them differentially dependent on specific aspects of their cell biology. Axons of different neuron classes display significant differences in lengths, diameters, neurofilament contents and arborisations, and they can be non-myelinated, myelinated or grouped in Remak bundles (Gardiner, 2011;Prokop, 2020). For example, the very specific dendritic morphology and physiology of cerebellar Purkinje cells makes them prime targets for specific genetic links to cerebellar ataxias (Hoxha, Balbo, Miniaci, & Tempia, 2018 Finally, we need to consider the complications of human genetics. It is the most refined genetics we have, but it also raises two funda-  (Bamburg & Bloom, 2009) or actin-rich rods (Fulga et al., 2007). Pinpointing which of the loss-or gain-of-function mechanisms, or combination thereof, causes the axonal decay can be a challenging task and may involve aspects of cellular function that are not related to endogenous roles of the affected gene locus (Kim et al., 2020).
Second, the disease-associated mutations listed in Table S1 usually allow individuals to survive for long enough to be identified and sequenced by clinicians; they are selected for restricted functional impacts, and this could be one potential explanation as to why they often cause symptoms restricted to specific neuron types. Complete loss-of-function conditions (also eliminating genetic redundancy) or stronger gain-of-function conditions are more likely to give us profound genetic insights into axonopathies. However, such conditions likely have a strong bias to cause embryonic or early postnatal lethality. These would therefore be discovered primarily through foetal sequencing in missed abortions (Fu et al., 2018) or through work in model organisms, which provide powerful strategies to complement human genetics and gain more profound genetic, conceptual and mechanistic understanding of axonopathies.

| IS THERE A COMMON PATHOLOGY?
Apart from genetic considerations, we need better knowledge of the cell biology of axon pathology as an important further source of information to clarify commonalities and differences between axonopathies. However, as stated by Vallat and colleagues: 'The characteristic microscopic lesions can only be identified by ultrastructural analysis of a nerve biopsy, which is nowadays only carried out in rare cases' (Vallat et al., 2016). Even if more biopsies were available, their interpretation remains a challenging task. Axon degeneration occurs over an extended time period, and key traits of pathology are likely to change from early to late stages of an axon's degenerative process; any sample taken will only provide us with a narrow time window of a long process. Furthermore, axonopathies are unlikely to synchronously affect all axons in a nerve, and they may initiate in varying proximo-distal positions, thus making it difficult to stage disease progression of individual axons in biopsies. To really understand the origin and progression of decay, we would need pathological timelines of events-but such systematic descriptions are rare (e.g., Spencer & Schaumburg, 1977a, 1977b. probably as manifestations of a major pathway of CNS axonal death' (Coleman, 2005). Usually, these swellings are characterised by the accumulation of organelles and disorganisation or even disappearance of the cytoskeleton (Figure 2b, top; e.g., Berard-Badier, Gambarelli, Pinsard, Hassoun, & Toga, 1971;Bridge et al., 2009;Fassier et al., 2013;Fiala, Feinberg, Peters, & Barbas, 2007;Havlicek et al., 2014;Jellinger & Jirasek, 1971;Probst et al., 2000;Seitelberger, 1971;Tarrade et al., 2006;Yang et al., 1999); they are known to interfere with action potential propagation (Gu, 2021).
Finally, there is little indication that chronic damage to axons has prominent links to Wallerian degeneration-type mechanisms, which seem restricted primarily to acute damage of axons instead (M. P. Coleman & Höke, 2020;Llobet Rosell & Neukomm, 2019).
Taken together, there might be a substantial degree of commonality between different forms of axonopathy; likely the seven pillars of axon maintenance decaying during ageing (epigenetics, metabolism, proteostasis, adaption to stress, macromolecular damage, inflammation, regeneration; Salvadores et al., 2017) may all apply also during processes of axonopathy. In the following I will discuss conceptual frameworks that may explain common features of axonopathies, and why it is so difficult to decipher causes and consequences. This scenario where MT bundle-damage due to life-sustaining axonal transport is counterbalanced by bundle-maintaining actions of MTBPs and the cortex, could be an essential pillar of axon maintenance and has originally been formulated in our model of 'local axon homeostasis' .  Table S1). This might be a co-incidence, or it might hint at specific roles of these particular motor proteins during axon biology.

| EXPANDING THE MODEL
Our further investigations dedicated to kinesin-1 or −3 deficient neurons revealed that the patho-mechanisms leading to MT curling involved oxidative stress in form of harmful reactive oxygen species (ROS; our unpublished data). How this harmful ROS is generated upon loss of these kinesins, and whether it involves aberrant organelle dynamics (e.g., Fransen, Lismont, & Walton, 2017;Pascual-Ahuir, Manzanares-Estreder, & Proft, 2017), remains to be explored; but our finding is consistent with previous reports that the actin cytoskeleton is modified by ROS (Wilson, Terman, Gonzalez-Billault, & Ahmed, 2016;Wioland et al., 2021), that oxidative stress leads to the oxidation and damage of MTs in cardiac myocytes (Goldblum et al., 2020 preprint), and that it explains axon swellings in models of Parkinson's disease and multiple sclerosis (Czaniecki et al., 2019;Niki c et al., 2011).
Importantly, our findings triggered new ways of thinking: loss of motor proteins might potentially relieve the MT bundles from mechanical damage, but such an effect appears to be outweighed by pathomechanisms triggered by loss of transport: transport deficiencies seem to trigger changes in axonal physiology that become severely harmful to MT bundles through alternative routes ('5' in Figure 2).
This idea is gaining momentum through various independent publications (Guo et al., 2020). The most striking case was the recent report depleting JNK-interacting protein 3 (JIP3), a kinesin-1 linker protein associated with the neurodevelopmental disease neurodevelopmental disorder with or without variable brain abnormalities (NEDBA; #618443). JIP3-depleted human iPSCs (inducible pluripotent stem cells) displayed lysosome-filled axonal swellings in which the actin cortex was disturbed and MTs showed the exact same curling as described in our work (Rafiq, Lyons, Gowrishankar, De Camilli, & Ferguson, 2020 preprint); upon JIP3 knock-down, TAU is less likely to associate with MTs as suggested by hyperphosphorylation at several assessed sites (T181, S202/T205, S396), and TAU depletion from MTs could be one cause for their curling (Hahn et al., 2020 preprint

| WHAT ARE THE IMPLICATIONS?
The notion that axonal physiology downstream of axon transport Obviously, the sequence of events will differ depending on the site of lesion. However, the model would predict that the outcome should be very similar. Unfortunately, the existing descriptions of axonal pathology are too scarce to validate our predictions, and substantial work would be required to compensate for this lack, ideally using standardised approaches in animal models and patient biopsies that cover a broader set of readouts. For example, as argued earlier , MTs are too often neglected in axon analyses, with more emphasis being given to the staining with less relevant neurofilaments; or in ultrastructural analyses resolutions are kept too low to properly visualise MTs.
Taken together, I strongly believe that all the arguments listed above, including genetics, pathological observations and the 'dependency cycle of axon homeostasis', build an increasingly strong case for a common concept of axonopathies. I predict that axonopathies will have their individual variations but also share a fundamental common pathway of auto-destruction. Clearly, we require more pathological analyses to validate or disprove this point. Whatever the outcome, such an effort will pay off through providing us with essential new understanding of the essential patho-mechanisms that lead to the tragic occurrence of axonopathies. no conflict of interests and no sharable data were generated.

DATA AVAILABILITY STATEMENT
No sharable data were generated.