Non-Cell-Autonomous Regulation of Retrograde Motoneuronal Axonal Transport in an SBMA Mouse Model

Discussion

Here we demonstrate that retrograde axonal transport in motoneurons is impaired via non-cell-autonomous AR action in muscle. Myogenic contributions to SBMA were first recognized in two different mouse models (Yu et al., 2006; Monks et al., 2007), where myopathic changes were discovered in a knock-in model well before neuropathy and transgenically overexpressed WT-AR specifically in muscle cells triggered SBMA-like symptoms only in myogenic TG males, including a loss of ventral root axons. These unexpected findings suggested that non-cell-autonomous mechanisms originating in muscle could drive pathogenesis in motoneurons in SBMA mice. We now find that AR in muscle is sufficient to disrupt retrograde transport in the motoneurons that innervate diseased muscle. Previous findings indicated that myogenic TG mice exhibit perturbed axonal transport compared with WT littermates (Kemp et al., 2011). However, such TG mice also had AR in other cell types due to expression of the endogenous gene, leaving open the possibility that the transport deficits found in this model also depended on endogenous AR. Thus, we directly addressed this question by assessing axonal transport in myogenic TG males that also have the Tfm allele of the endogenous AR gene (Tfm/TG males), thus eliminating all functional AR from the endogenous gene. Consequently, such Tfm/TG males had functional AR only in skeletal muscle fibers due to the expression of the AR transgene. We found that diseased Tfm/TG males showed comparable deficits in the number of retrogradely labeled motoneurons and in endosomal flux as previously reported for diseased TG males on a WT AR background (Kemp et al., 2011), indicating that AR acts solely in muscle fibers to perturb retrograde transport in motoneurons.

As expected, Tfm/TG males exhibited androgen-dependent motor dysfunction, replicating a previous finding from this same mouse model showing that such dysfunction is fully reversible when androgen treatment ceases (Monks et al., 2007; Johansen et al., 2009, 2011; Fig. 2A) and is consistent with the phenotype of SBMA mouse models expressing a CAG-expanded AR allele (Katsuno et al., 2002; Chevalier-Larsen et al., 2004; Sopher et al., 2004; Yu et al., 2006; Renier et al., 2014). Such mice also showed a striking deficit in the number of retrogradely labeled motoneurons compared to nondiseased Tfm male controls (Fig. 2B), correlating with the deficit in motor function (Fig. 2A). Interestingly, the magnitude (∼50%) of the labeling deficit in Tfm/TG males is comparable to that previously reported for both chronically diseased myogenic TG males and acutely diseased myogenic TG females (Kemp et al., 2011), indicating that TG AR in muscle, when activated by androgens, rapidly perturbs the uptake/retrograde transport of CT-HRP by motoneuronal axons independent of AR expressed by the native gene, including that in motoneurons and/or other cell types.

The deficit in axonal transport exhibited by androgen-treated Tfm/TG males is largely comparable to that previously observed in TG-only males and females. Notably, deficits in CT-HRP staining and grip strength are apparent after only 24 h of T treatment in TG-only females (Kemp et al., 2011), the time point at which Tfm/TG males, given the same T treatment, show deficits in grip strength. Moreover, CT-HRP transport recovers fully in chronically symptomatic TG males that have been castrated, as does motor function in symptomatic Tfm/TG males following T removal (Johansen et al., 2011; Kemp et al., 2011). It is clear that the transgene alone without androgen stimulation is not sufficient to induce transport dysfunction, in parallel with the effect of the transgene on motor function, since castrated TG males and untreated TG females show deficits in neither motor function nor axonal transport (Kemp et al., 2011). Unfortunately, any direct comparison between TG and Tfm/TG males in disease traits would be confounded by long-term versus short-term effects of disease. TG-only males are exposed throughout life to endogenous male levels of androgens, causing secondary effects of disease due to its chronic nature in TG-only males. On the other hand, Tfm/TG males, like TG-only females, lack endogenous male levels of androgens postnatally, and only express an acute disease phenotype when provided with exogenous androgens. In sum, the defects in transport and motor dysfunction induced by T treatment of Tfm/TG males, who, like TG-only males and females, express the disease-causing AR allele only in muscle fibers, is comparable to what is seen in TG-only males and females in which both motor function and axonal transport are disrupted only in the presence of male levels of androgens.

Evidence strongly argues that the deficit found in the number of retrogradely labeled motoneurons is not caused by a net loss of motoneurons. Cell counts of motoneuronal cell bodies in Nissl-stained sections of the spinal cord show no evidence of motoneuronal loss in the myogenic TG model (Monks et al., 2007; Johansen et al., 2009). Moreover, extending CT transport time to 24 h rather than the 12 h used here eliminates the deficit in retrograde labeling of motoneurons (Kemp et al., 2011). Notably, the defects in axonal transport observed here are not likely due to secondary effects of disease since deficits in CT-HRP transport occur just 24 h after the commencement of T treatment in female TG mice (Kemp et al., 2011) who display the same time course of deterioration as the inducible Tfm/TG model used here. Moreover, no muscle atrophy is apparent in TG females treated for 5 d, despite their displaying severe motor impairment and a loss of muscle force that does not involve a loss of muscle mass (Oki et al., 2013). Additionally, no dying back of the motoneuron occurs in even chronically diseased TG males, as all endplates are contacted by a motor nerve terminal (Kemp et al., 2011). Thus, deficits in axonal transport but not in neuronal loss or muscle atrophy correlate with motor dysfunction, suggesting that axonal transport dysfunction may contribute to early losses in motor function and ultimately to motoneuron death in SBMA.

This deficit in the retrograde labeling of motoneurons may be due to a number of factors, including dysfunction originating in the distal most aspect of the axon, the synaptic nerve terminal, where material is first taken up, sorted, and packaged into vesicles for either local recycling or retrograde transport to the motoneuronal cell bodies. Deficits could also originate in more proximal aspects of the axon. Transport of cargo in these two axonal compartments depends on different transport machinery, with local transport of cargo in the synaptic nerve terminal moved along actin by specialized myosin, whereas long-distance transport of cargo toward motoneuronal cell bodies is moved along microtubules by dynein and the motor-associated protein complex dynactin (Chevalier-Larsen and Holzbaur, 2006). Our first-level analysis of endosomal transport in living axons of diseased Tfm/TG mice indicated that endosomal flux, but not net velocity, was perturbed in Tfm/TG males, as found previously in chronically diseased myogenic males and acutely diseased myogenic females (Kemp et al., 2011). Thus, fewer labeled endosomes than normal move along the diseased axon per unit of time, offering an explanation for why fewer motoneurons in diseased Tfm/TG mice are retrogradely labeled after 12 h of CT transport. Such a deficit in flux suggests that the early endocytotic pathway in the nerve terminal is susceptible to non-cell-autonomous regulation, with disease signals originating in muscle perturbing some aspect of this presynaptic pathway, including possibly the uptake, local sorting, packaging/fusion, and/or movement of CT-containing cargo within the synaptic nerve terminal. Moreover, the nerve terminal exhibits the accumulation of neurofilament in SBMA patients, further linking the distal axon to defects in transport (Katsuno et al., 2006). Nevertheless, which of these processes is affected is not clear and would require direct monitoring of endocytosis in synaptic nerve endings to determine.

While net velocity was not affected by disease in Tfm/TG males, paradoxically, instantaneous velocity was, which is counter to previous findings reported for myogenic TG males (Kemp et al., 2011). A deficit in instantaneous velocity means that when endosomes are in motion, they move slower in diseased axons than in healthy axons, raising questions about whether AR in muscle can also cause defects in the retrograde motor dynein (Alper et al., 2013). A possible explanation for dynein dysfunction per se is suggested by a report by Ramzan et al. (2015). Using Cre-lox technology to drive expanded polyglutamine AR expression specifically in muscle fibers (MyoAR), they found reduced nitrotyrosine staining in motoneurons (Ramzan et al., 2015). These data suggest that a loss of oxidative homeostasis in motoneurons triggered by diseased muscle can, in turn, impair dynein. In fact, dynein heavy chain is reduced in the ventral spinal root of symptomatic AR97Q SBMA mice, but only in later stages of the disease (Katsuno et al., 2006).

The effect of disease on instantaneous velocity in Tfm/TG males is reminiscent of previous findings in a knock-in model of SBMA, also involving a deficit in instantaneous velocity (Kemp et al., 2011). While these previous data suggested that mutant AR expressed in motoneurons can cause defects in dynein/dynactin function in a cell-autonomous fashion, the current data suggest an additional hypothesis that endogenous (WT) AR in motoneurons of TG males may have protected against the toxic effects of muscle TG AR on instantaneous velocity. Knock-in males would not have this protection since every cell that expresses AR in this model expresses only the mutant AR allele. This suggestion is consistent with previous findings (Thomas et al., 2006) that endogenous WT AR can blunt the toxicity of a disease-causing TG AR in a different SBMA model. It is also possible that endogenous AR in muscle could have a protective role, but this scenario seems less parsimonious. Finally, while AR aggregates perturbed axonal transport in one SBMA cell model (Piccioni et al., 2002), this cannot explain the defects seen in the current TG model as AR aggregates do not occur in myogenic mice (Monks et al., 2007) or in axons of other SBMA models in which axonal transport defects have been found (Szebenyi et al., 2003; Katsuno et al., 2006).

We also found that retrogradely transporting endosomes in diseased axons had a higher overall incidence of trafficking perturbations, with more endosomes in the axons of Tfm/TG males showing reversals, stalls, and velocity changes than endosomes in the axons of healthy Tfm males (Fig. 3). Such “error” events reflect reduced processivity. An important regulator of processivity is the motor-associated protein dynactin (King and Schroer, 2000). In fact, dynactin is reduced in motoneurons of SBMA patients and in a mouse model of SBMA in which the disease allele is globally expressed and shows early defects in axonal transport (Katsuno et al., 2006), suggesting that it is likely to also be perturbed in the myogenic model. That transport of endosomes in axons of diseased Tfm/TG males is less efficient suggests that dynactin function is also perturbed in a non-cell-autonomous manner by diseased muscle. The run length of endosomes in axons of diseased Tfm/Tg males is also shorter than in axons of healthy Tfm axons, which is consistent with reduced processivity, though this difference is not significant, perhaps because of the increased error variance seen in healthy Tfm controls. Nonetheless, together these data suggest that signals from diseased muscle can instigate defects in the core transport machinery involving the retrograde motor and its associated proteins, although the effect of diseased muscle on retrograde transport seems to lie predominantly in the motor synapse where cargo is first taken up and packaged. While the current study represents a small step toward recognizing the potential role of non-cell-autonomous regulators of axonal transport, this possibility is sufficiently important to merit further investigation.

The fact that instantaneous velocity, but not net velocity, is perturbed in Tfm/TG diseased males suggests that compensation occurred. One possible explanation is that while reversals are more common in diseased axons, perhaps the distance traversed in the wrong direction is less in diseased axons than in healthy axons. Regardless of the explanation for the apparent discrepancy in instantaneous versus net velocity, the current findings raise the possibility that dynein and dynactin, which are essential in the retrograde transport of cargo, are each susceptible to non-cell-autonomous regulation by the muscle. While there is precedence that both dynein and dynactin are impaired in models of SBMA (Szebenyi et al., 2003; Katsuno et al., 2006), the current data are the first to suggest that such defects in motor proteins could be caused via a non-cell-autonomous mechanism.

How might the action of a toxic AR in muscle perturb axonal transport in the motoneurons? One attractive scenario is the loss of muscle-derived trophic factors. Skeletal muscle releases many kinds of trophic factors that can act on receptors in motor nerve terminals to promote healthy cell function. Diseased muscle from various SBMA mouse models exhibits deficits in multiple trophic factors, including brain-derived neurotrophic factor (BDNF), glial cell line-derived factor, insulin-like growth factor 1, neurotrophin (NT)-4, and vascular endothelial growth factor (VEGF) (Sopher et al., 2004; Yu et al., 2006; Monks et al., 2007; Mo et al., 2010; Halievski et al., 2015). Moreover, replenishing the supply of such neurotrophic factors can reverse disease-related deficits in axonal transport in models of motoneuron/neuromuscular disease. For example, delivery of ciliary neurotrophic factor, BDNF, or NT-3 to muscle improves retrograde transport in the progressive motor neuronopathy mouse model (Sagot et al., 1998). Additionally, treating diseased muscles of myogenic TG SBMA females with VEGF reverses a disease-related deficit in retrograde labeling of motoneurons and endosomal flux (Kemp et al., 2011). These data are compelling precedence for a cause–effect relationship between muscle-supplied neurotrophic factors and retrograde transport in motoneurons.

Loss of such trophic support from target musculature could lead to defective transport in at least two ways. First, lack of trophic factor signaling might mean a loss of survival signaling, leaving the cell deprived of needed resources for proper maintenance of axonal transport (Chowdary et al., 2012). Alternatively, local neurotrophic signaling may be needed for endocytosis per se, since VEGF can reverse a deficit in flux (Kemp et al., 2011). Internalization of the neurotrophin-bound Trk receptor requires PI3K signaling to recruit Rab5 (Christoforidis et al., 1999; for review, see Harrington and Ginty, 2013). Since internalization and sorting of the CT receptor GM1 also depends on Rab5 (Pelkmans et al., 2004), neurotrophic factors could enhance endocytosis through this pathway. Once internalized, Trk signaling also leads to the phosphorylation of dynein intermediate chains through the MAPK/ERK pathway to enhance the binding of dynein to Trk-activated signaling endosomes (Mitchell et al., 2012). Defects in this pathway could explain the deficit in flux as well as the reduced efficiency of transport in diseased axons. In short, muscle-derived neurotrophic factors are a strong candidate for how diseased muscle might disrupt the uptake and/or retrograde transport of cargo in motoneuronal axons of SBMA mice (Fig. 4, proposed model). While data are scant, the relevance of these data to SBMA in humans is likely (Katsuno et al., 2006).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Proposed model of the non-cell-autonomous influence of muscle on motoneuronal axonal transport. In the healthy neuromuscular system, neurotrophic factors produced and released from skeletal muscle support motoneuronal health and survival. SBMA skeletal muscle is deficient in neurotrophic factors, which may lead to reduced endocytosis in motoneurons, which may manifest as reduced retrograde transport of cholera toxin and reduced endosomal flux, as observed in the present study. Therapies aimed at increasing muscle-derived neurotrophic factors may remedy defective axonal transport and ameliorate disease symptoms.

While considerable attention has recently been redirected to muscle, recent work also supports the idea that mutant AR acting in motoneurons contributes to SBMA pathogenesis (Ramzan et al., 2015; Sahashi et al., 2015). Given that both motoneurons and muscle normally express AR, and that motoneurons and muscles are interdependent, it would not be surprising if the symptoms of SBMA reflect a complex interplay of mutant AR acting in both motoneurons and skeletal muscles, if not in other cell types too. The important task before us is to better understand where and how mutant AR acts to cause disease and to sort out the primary effects of AR from the secondary effects of disease. AR acting in muscle to impair axonal transport may be one such early pathogenic event leading to widespread neuromuscular dysfunction.