Neuron

Previews

shibirets mutant flies does not affect auto- about the actual substrates of neuronal Milosevic, I., Giovedi, S., Lou, X., Raimondi, A.,
Collesi, C., Shen, H., Paradise, S., O’Toole, E., Fer-
phagosome formation, in spite of the and synaptic autophagy, and we do not guson, S., Cremona, O., and De Camilli, P. (2011).
complete block of endocytosis (Soukup know the precise fate of synaptic auto- Neuron 72, 587–601.
et al., 2016). Third, loss of endophilin-A phagosomes that undergo retrograde
Murdoch, J.D., Rostosky, C.M., Gowrisankaran,
in mice has no overt effect on the levels transport to the neuronal soma (Maday S., Arora, A.S., Soukup, S.F., Vidal, R., Capece,
of ATG9 or ATG16L proteins (Murdoch and Holzbaur, 2014). Future studies will V., Freytag, S., Fischer, A., Verstreken, P., et al.
et al., 2016). Collectively, these results need to provide answers to these exciting (2016). Cell Rep. 17, 1071–1086.
support a direct function of endophilin-A questions. Rubinsztein, D.C., Shpilka, T., and Elazar, Z.
in autophagosome formation indepen- (2012). Curr. Biol. 22, R29–R34.
dent of its established role in endocytosis REFERENCES Soukup, S.F., Kuenen, S., Vanhauwaert, R.,
(Figure 1). As with all exciting new works, Manetsberger, J., Hernández-Dı́az, S., Swerts,
a number of open questions remain: How Cajigas, I.J., Will, T., and Schuman, E.M. (2010). J., Schoovaerts, N., Vilain, S., Gounko, N.V.,
EMBO J. 29, 2746–2752. Vints, K., et al. (2016). Neuron 92, this issue,
exactly is the function of endophilin-A
829–844.
regulated, and how do LRRK2 and other Cao, M., Milosevic, I., Giovedi, S., and De Camilli,
factors such as FBXO32/ FOXO3A sense P. (2014). J. Neurosci. 34, 16544–16549. Tian, Y., Chang, J.C., Fan, E.Y., Flajolet, M., and
Greengard, P. (2013). Proc. Natl. Acad. Sci. USA
neuronal activity? What physiological 110, 17071–17076.
Kjaerulff, O., Brodin, L., and Jung, A. (2011). Cell
mechanisms define the switch between Biochem. Biophys. 60, 137–154.
the functions of endophilin-A in endocy- Uytterhoeven, V., Lauwers, E., Maes, I., Miskie-
Maday, S., and Holzbaur, E.L. (2014). Dev. Cell 30, wicz, K., Melo, M.N., Swerts, J., Kuenen, S., Wit-
tosis and in autophagy? Which partner 71–85. tocx, R., Corthout, N., Marrink, S.J., et al. (2015).
proteins associate with endophilin-A in Neuron 88, 735–748.
autophagy, and which membranes are Matta, S., Van Kolen, K., da Cunha, R., van den Bo-
gaart, G., Mandemakers, W., Miskiewicz, K., De Verstreken, P., Kjaerulff, O., Lloyd, T.E., Atkinson,
the substrates for autophagosome forma- Bock, P.J., Morais, V.A., Vilain, S., Haddad, D., R., Zhou, Y., Meinertzhagen, I.A., and Bellen, H.J.
tion at synapses? Finally, we know little et al. (2012). Neuron 75, 1008–1021. (2002). Cell 109, 101–112.

Cut Your Losses:

Spastin Mediates Branch-Specific Axon Loss
Hagar Meltzer1 and Oren Schuldiner1,*
1Department of Molecular Cell Biology, Weizmann Institute of Sciences, Rehovot 7610001, Israel

*Correspondence: oren.schuldiner@weizmann.ac.il
http://dx.doi.org/10.1016/j.neuron.2016.11.004

In this issue of Neuron, Brill et al. (2016) demonstrate that, during synapse elimination in the developing
neuromuscular junction, branch-specific microtubule destabilization results in arrested axonal transport
and induces axon branch loss. This process is mediated in part by the neurodegeneration-associated, micro-
tubule-severing protein spastin.

Developmental neuronal remodeling is tral and peripheral nervous systems of Purves and Lichtman, 1980). When the
crucial for sculpting the mature nervous both vertebrates and invertebrates. In mammalian NMJ initially forms, several
system. In mammals, neuronal remodel- addition to its developmental signifi- axon branches originating from different
ing is largely timed to postnatal develop- cance, understanding the molecular basis motor neurons converge to a single syn-
ment and serves to refine neural circuits of axon pruning may shed light on the aptic site on the muscle, which is at this
that were formed during the embryonic resembling process of axon degeneration stage poly-innervated. However, during
period, often by eliminating exuberant during certain ‘‘dying-back’’ neurodegen- the early postnatal period, extensive
connections. This may include pruning of erative diseases (Luo and O’Leary, 2005; axon pruning takes place until only a sin-
individual axonal projections while the Yaron and Schuldiner, 2016). gle axon branch remains. This synaptic
cell body, and in some cases sister axonal One of the first neural structures in elimination is based on ongoing competi-
projections, remains intact. Axon pruning which axon pruning was demonstrated tion between axon branches, in which the
is a widespread phenomenon and has to be essential for establishing proper wir- ‘‘winning’’ branch gradually expands its
been demonstrated throughout the cen- ing is the neuromuscular junction (NMJ; territory at the expanse of the ‘‘losing’’

Neuron 92, November 23, 2016 ª 2016 Elsevier Inc. 677

 Neuron

Previews

Figure 1. A Schematic Illustration of the Developing Neuromuscular Junction

Spastin mediates microtubule destabilization and axon branch dismantling.

branch, eventually leading to complete nerve-muscle explants at extremely high et al. (2016), however, demonstrated
retraction of the latter (Walsh and Licht- resolution. The sequential photo-bleach- transport arrest in the dismantling, losing
man, 2003). It is crucial to note that the ing method, which they have previously axon branches. By tracing labeled mito-
outcome of each such ‘‘battle’’ is branch optimized, permits non-invasive time- chondria and peroxisomes, they estab-
specific, as various axon branches origi- lapse visualization of a single axon branch lished absence of both anterograde and
nating from a single motor neuron can at its neuromuscular synaptic site. This retrograde transport in retreating axon
simultaneously expand or retract at allows determination of the fraction of branches (which they defined as occu-
different NMJs along the muscle. In other the postsynaptic territory occupied by pying 40% or less of the synaptic terri-
words, at a single time point, all stages of each axon branch—a reliable predictor tory), as opposed to stable bi-directional
the competition may coexist within one of the synaptic competition outcome trafficking in the growing or competing
motor unit. Despite our thorough under- (i.e., a branch that occupies the majority (i.e., occupying between 40%–60% of
standing of the temporal and spatial order of the postsynaptic site at a given time the synaptic territory) branches. Impor-
of events during synapse elimination in point will most likely be the ‘‘winner’’; tantly, this argues against the ‘‘evacua-
the remodeling NMJ, little is known about Walsh and Lichtman, 2003). Moreover, tion’’ model, but rather supports the ‘‘axo-
the molecular mechanisms that govern using time-lapse imaging, the system en- some shedding’’ model, in which axon
its regulation and execution, especially ables tracing of microtubule dynamics as branches are locally dismantled along
regarding the manner by which branch well as evaluation of the directionality and with their content, thus shedding mem-
specificity is achieved. The elegantly de- rate of axonal transport. brane-bound remnants that contain a
signed study by Brill et al. (2016), pub- In this study, Brill et al. (2016) begin by high density of axonal organelles, which
lished in this issue of Neuron, takes an testing the hypothesis that subcellular are later engulfed and degraded by neigh-
important step toward delineating the structures are evacuated from axon boring Schwann cells (Bishop et al.,
molecular basis of NMJ remodeling by branches prior to pruning. This theory 2004).
identifying a key role for spastin-medi- was suggested in the original model of Naturally, the arrest in axonal transport
ated, branch-specific microtubule desta- NMJ synapse elimination (Riley, 1981) directed Brill et al. (2016) toward the po-
bilization during synapse elimination and was reinforced by a more recent pa- tential role of the microtubular cytoskel-
(Figure 1). per that demonstrated retrograde trans- eton in the process, as it comprises the
Brill et al. (2016) developed a unique port of pre-synaptic vesicles along axonal tracts on which axonal transport takes
system that allows studying of the devel- microtubular tracts during remodeling of place. The involvement of microtubules
oping mammalian NMJ in postnatal the Drosophila NMJ (Liu et al., 2010). Brill in synapse competition and elimination

678 Neuron 92, November 23, 2016

Neuron

Previews

was verified by a series of immunostain- destabilization in vivo arises. Microtu- microtubule polyglutamylation (Zempel
ings, which demonstrated a specific bules are known to be subjected to a vari- et al., 2013). Brill et al. (2016), however,
decrease in tubulin intensity that corre- ety of post-translational modifications, found decreased, rather than increased,
lated with reduction in synaptic territory some of which are shown to be associ- polyglutamylation in retreating axon
of the axon branches. In addition, retreat- ated with changes in stability (Janke and branches, in which spastin is supposedly
ing branches displayed a higher density of Bulinski, 2011). As expected, the retreat- more active, compared to the highly
EB3 ‘‘comets,’’ demarcating dynamic ing branches displayed a high glutamy- polyglutamylated winning branches. This
plus-end microtubule tips. Together with lated/tyrosinated a-tubulin ratio, typical potential lack of congruence could be
the selective reduction in microtubular of hyper-dynamic microtubules, accom- resolved if one assumes that poly-
mass, this finding is indicative of fragmen- panied by decreased polyglutamylation. glutamylated microtubules in the losing
tation and destabilization of the microtu- Due to the surprising lack of evidence for branch are severed by spastin. Indeed,
bular cytoskeleton and was evident in the action of deacetylases (as deacetyla- Brill et al. (2016) provide evidence to sup-
the retreating axon branches, but not in tion is the most consistent microtubule port this hypothesis by demonstrating
the growing ones. Furthermore, while modification linked to instability; Janke that, in spastin knockout mice, a promi-
EB3 comet density was equally high in and Bulinski, 2011), Brill et al. (2016) nent increase in polyglutamylation is
the proximal and distal areas of a given searched for an alternative mediator of observed, especially in the retreating
branch, it was significantly lower in the the local destabilization effects. They axon branches. Evidently, further work is
stem axon of that branch (in which it did turned to spastin, one of several micro- warranted in order to delineate the pre-
not differ from that of stem axons of win- tubule-severing enzymes and, as such, cise regulation of microtubule polygluta-
ning branches). This indicates that micro- predicted to induce an increase in the mylation in vivo. In this context, examining
tubule destabilization is a highly compart- number of microtubule ends as well as the potential involvement of other post-
mentalized event, which does not extend microtubule loss. In a new spastin translational modifications, some of
beyond the branch point of a retreating knockout mouse that they generated, Brill which were highlighted in this study
axon. This branch specificity is in accor- et al. (2016) demonstrated that while the (such a tyrosination), could provide addi-
dance with the coexistence of losing and initial development of the NMJ did not tional insight into the regulation of spastin.
winning branches within a single neuron, seem to be affected, synapse elimination Furthermore, spastin is not the only
as earlier described. Careful time-course was defective and resulted in persistence microtubule-severing protein, and it might
analysis established destabilization of of doubly innervated NMJs, similar to be interesting to investigate whether other
the microtubules in the losing branches, the effect observed following epothilone such proteins, namely katanin and fidge-
ruling out the alternative option in which treatment. The spastin mutants also dis- tin, also play a role in branch-specific
microtubules were being stabilized in all played a drastic decrease in the rate of microtubule destabilization.
branches excluding the losing one. The retraction bulb shortening. Lastly, Brill Taken together, Brill et al. (2016) pro-
profound loss of microtubular tracts in re- et al. (2016) generated a conditional vide important insight into unresolved
treating branches was further validated by knockout mouse model that selectively questions regarding the molecular mech-
reconstruction of a subset of the NMJs lacks spastin in cholinergic neurons. This anisms that underlie NMJ development
using correlated electron microscopy. mutant line displayed a similar delay in in particular and neuronal remodeling in
If microtubule destabilization induces synapse elimination, establishing that general. Furthermore, their identifica-
dismantling of the losing axon branch, spastin activity was cell autonomously tion of spastin as a mediator of branch-
then pharmacological stabilization of the required within the motor neurons. specific microtubule destabilization is
microtubular cytoskeleton is expected to The microtubule stabilization effects exciting, especially in light of its role
interfere with the normal progression of displayed by spastin mutants seemed to in hereditary spastic paraplegia and other
synapse elimination. Indeed, the use of manifest as a weaker version of those neurodegenerative diseases (Errico et al.,
the microtubule-stabilizing drug epothi- induced by epothilone treatment, with 2002). It is intriguing to mention that one of
lone resulted in a significant delay in syn- winning branches less affected than the most affected axon projections in
apse elimination, manifested by the losing ones. This finding might suggest hereditary spastic paraplegia is the corti-
persistence of doubly innervated NMJs that spastin activity is tightly regulated in cospinal tract, which is another classical
as late as postnatal day 21. Accordingly, a branch-specific manner. Indeed, one example for branch-specific neuronal re-
following epothilone treatment, both the of the most important questions that modeling (Luo and O’Leary, 2005).
losing and winning branches displayed arises from this study relates to the regu- It comes as no surprise that microtubule
increased tubulin content and reduced lation of spastin—namely, how is spastin fragmentation plays a role in axon dis-
EB3 comet density. Furthermore, the selectively directed to its target site, mantling, as microtubules constitute a
treatment seemed to restore organelle enabling the highly defined compartmen- major part of the axon cytoskeleton. In
transport in the retreating branches. talization of retreating branches. Is the fact, it is well appreciated that microtu-
Since branch-specific destabilization of regulation achieved due to a post-transla- bule disruption is one of the first and
the microtubular cytoskeleton seems to tional modification of spastin itself or essential steps during axon and dendrite
drive axon branch dismantling, the ques- rather by microtubule modifications or pruning in Drosophila, preceding mor-
tion of what is the molecular mechanism both? Previous studies demonstrated phological disruption of the neurites
that drives branch-specific microtubule recruitment of spastin to its target site by (Yu and Schuldiner, 2014). Nevertheless,

Neuron 92, November 23, 2016 679

 Neuron

Previews

identifying microtubules as the seemingly Minerva Foundation, and the European Research Liu, Z., Chen, Y., Wang, D., Wang, S., and Zhang,
Council (erc CoG). O.S. is an incumbent of the Y.Q. (2010). J. Neurosci. 30, 11624–11634.
causative force of this process is an
Rothstein Career Development Chair of Genetic
intriguing finding with potentially impor- Diseases. Luo, L., and O’Leary, D.D. (2005). Annu. Rev. Neu-
tant clinical implications. Much effort is rosci. 28, 127–156.
focused on the use of microtubule stabili-
zation compounds (such as epothilone) as REFERENCES Purves, D., and Lichtman, J.W. (1980). Science
therapeutic agents for a variety of neuro- 210, 153–157.
logical conditions. The findings demon- Bareyre, F.M., Kerschensteiner, M., Raineteau, O.,
Mettenleiter, T.C., Weinmann, O., and Schwab, Riley, D.A. (1981). J. Neurocytol. 10, 425–440.
strated by Brill et al. (2016) warrant careful, M.E. (2004). Nat. Neurosci. 7, 269–277.
context-dependent consideration of their
Bishop, D.L., Misgeld, T., Walsh, M.K., Gan, W.B., Walsh, M.K., and Lichtman, J.W. (2003). Neuron
use, as physiological microtubule destabi- and Lichtman, J.W. (2004). Neuron 44, 651–661. 37, 67–73.
lization might contribute to compensatory
plasticity, which might be important in Brill, M.S., Kleele, T., Ruschkies, L., Wang, M., Yaron, A., and Schuldiner, O. (2016). Curr. Biol. 26,
Marahori, N.A., Reuter, M.S., Hausrat, T.J., Wei- R628–R639.
neural repair following neurological injury gand, E., Fisher, M., Ahles, A., et al. (2016). Neuron
or during disease (Bareyre et al., 2004). 92, S0896-6273(16)30654-7.
Yu, F., and Schuldiner, O. (2014). Curr. Opin. Neu-
robiol. 27, 192–198.
Errico, A., Ballabio, A., and Rugarli, E.I. (2002).
ACKNOWLEDGMENTS
Hum. Mol. Genet. 11, 153–163.
Zempel, H., Luedtke, J., Kumar, Y., Biernat, J.,
Work in our laboratory is mainly funded by grants Janke, C., and Bulinski, J.C. (2011). Nat. Rev. Mol. Dawson, H., Mandelkow, E., and Mandelkow,
from the Israeli Science Foundation (ISF), the Cell Biol. 12, 773–786. E.M. (2013). EMBO J. 32, 2920–2937.

680 Neuron 92, November 23, 2016