The mechanisms underlying the postnatal maturation of pre- and postsynaptic specializations are finely controlled to assure reliable spatial and temporal transmission of information in and from the nervous system. At the end of the synaptic maturation process optimal juxtaposition of postsynaptic receptors and presynaptic neurotransmitter release sites is achieved. From the presynaptic side, the key elements to regulate are the position and density of proteins and organelles within the terminal. How this process is controlled, however, remains obscure. Here, we report that SMN-deficient mice present presynaptic defects in SVs, mitochondria, AZs, NFs and microtubules, as well as in neurotransmission, with a pattern compatible with a deficiency in presynaptic differentiation.
In the TVA muscle of SMN-deficient mice SVs remained clustered at P7, contrary to what happened in control mice where SVs dispersed and occupied larger areas of the terminal. In addition, SMN-deficient motor nerve terminals stopped accumulating SVs at P7, resulting in a 50% decrease in vesicles in mutants compared to controls at two week of age (). This last finding was in agreement with the decrease in size of the synaptic vesicles pools found electrophysiologically at this age (). Strikingly, in the rostral band of the LAL these alterations appeared about a week later (). In the Tibialis anterior
(TA) and in the Extensor digitorum longus
(EDL) muscles two different ultrastructural studies, in the same SMA mouse model, also found a decrease in SVs density 
. Conversely, in the diaphragm no significant difference in vesicle number has been found 
. Moreover, the size of the pool of SVs ready to be released (RRP) has been reported to be comparable in SMNΔ7 and littermate controls in the EDL (Ling et al., 2010), a mildly SMA affected distal muscle. However, here we found that the RRP size was decreased in the TVA in mutant mice (), in agreement with the ultrastructural observation of a similar decrease in docked vesicles in nerve terminals from TA muscle 
. Therefore, there seems to exist a correlation between the degree of muscle pathology and the variability in SV content.
In the present experiments we also found a ~50% reduction of total active mitochondrial surface in mutant presynaptic terminals (), with no apparent alteration in their spatial organization close to SV clusters. Previous evidence also suggests that mitochondrial defects are present in SMA nerve terminals. For example, in the TA muscle mitochondrial density is reduced by half in mutants while their morphology is normal 
. In the diaphragm, however, presynaptic mitochondria are smaller in mutants than in wild-type littermates, while no differences are found at the postsynaptic site 
. In the TVA, the amount of Ca2+
-dependent asynchronous neurotransmitter release during prolonged stimulation is increased, which might suggest an altered regulation of bulk [Ca2+
] by the mitochondria 
. Mitochondrial dysfunction has also been reported when Smn is knocked down in cultured neuronal cells, a cell model of SMA 
. Mitochondrial defects have been demonstrated in other motor neuron diseases, including ALS 
and Spinal and Bulbar Muscular Atrophy 
Beside the reduction and clustering of SVs and mitochondria in SMA mutant terminals, we also found a reduction in the density of AZs (visualized with anti-bassoon antibodies), together with an alteration in their distribution (). This, together with the decrease in the RRP size (), may explain the drop in neurotransmitter release as evidenced by estimation of the quantal content 
. A smaller number of AZs might be caused by a deficiency of presynaptic P/Q- and N-type voltage-dependent calcium channels in the nerve terminal 
. In motoneurons in culture from SMN deficient embryos, N-type calcium channels have been described to be diminished at growth cones 
. It could be of interest, therefore, to explore this possibility at motor terminals of postnatal mutant mice in the future.
It is interesting to note the parallels between the decrease in SVs, mitochondria and AZs in SMN deficient nerve terminals (, & ). The association between mitochondria and SVs is relevant for ATP-dependent functions such as refilling of SVs with neurotransmitter. On the other hand, the close relationship between AZs and SVs is essential for the efficient refilling of the release sites. In addition, the proximity of mitochondria to the plasma membrane may also support SV cycling. Thus, in SMN deficient terminals, the reduction of these organelles could be partially responsible for the functional impairment of the synapse. It is difficult to discern, however, whether the decrease in mitochondrial density is a consequence of the decrease of the SV pools. Other possibilities are also feasible, for example, a malfunction of the control mechanisms that maintains an appropriate pool of mitochondria within the presynaptic terminal. The interactions of mitochondria and SVs with the cytoskeleton are crucial for localization and maintenance of these organelles at their sites of action. For example, the subcellular localization (docking process) of mitochondria is likely based on F-actin filaments 
, in turn regulated through the RhoA/formins pathway 
. Finally, a deficiency in the microtubule-based anterograde transport of mitochondria, SVs and AZ precursors, is also plausible. A fault in the microtubule network itself, or in the motor proteins responsible for those cargos, may cause this phenotype. In support of this possibility, it has recently been reported that there is a decrease in polymerized tubulin at distal axons of SMN deficient terminals and less amount of mitochondria in the motor neurons 
For all these reasons we also studied the cytoskeleton (NFs, actin filaments, and microtubules) in SMNΔ7 motor terminals. Previously, it has been reported that NFs accumulated in motor axons and in the terminals of SMA mouse models 
. The importance of NF accumulation in the SMA pathogenesis is difficult to determine. NF overpacking may impair axonal transport of vesicles to terminals. However, we found that SV content in nerve terminals of the TVA muscle was already reduced at one-week of age (), a time at which abnormal NF accumulation was not apparent (), suggesting that NF accumulation is a late event in the disease progression. In addition, we also found a marked tendency of intraterminal NFs to end in balls in SMA mutants at P14 (), a sign of immaturity that suggests impairment in NF assembly and turnover.
A defect in axonal transport in SMN-deficient animal models cannot, nevertheless, be discarded. SMN has been implicated in the axonal transport of mRNA of beta-actin and other cargoes necessary for motor neuron integrity and function 
. In nerve terminals, F-actin is known to play important roles in SV recycling 
and, probably, in the tethering of SVs at AZs 
. Therefore, even a small decrease in actin content in SMA motor terminals (), may affect one or more functions related to the transport or the stability of these organelles. Although this hypothesis is attractive, however, a newly generated motor neuron specific beta-actin conditional knock-out mouse does not present an altered phenotype at the NMJ 
Recently, it has been reported that microtubule polymerization is disrupted in Smn-deficient NSC34 cells in culture, that the amount of acetylated tubulin is about one third of controls in sciatic nerves of SMA mice and that the number of axonal microtubules per axon is reduced by 25% in mutant mice 
. We here found that the presynaptic motor terminals of SMNΔ7 mice show a reduction and an abnormal distribution of microtubules (), compatible with the arrest of postnatal maturation at the presynaptic terminal. The scattering of microtubules in mutant terminals might, in turn, contribute to the delayed organization of the synaptic organelles.
In summary, our data show that SMN is essential for postnatal maturation of SVs, AZs, mitochondria and the cytoskeleton at the motor nerve terminal. We also suggest that this disruption in the presynaptic architecture might limits synaptic transmission in most affected muscles. These results, together with data from others showing delay in the maturation of the postsynaptic terminal, failure of muscle fiber growth, and a decrease in synaptic inputs to spinal motor neurons 
, support a possible role of SMN in neuromuscular development. Future investigations in this direction, therefore, may help to better understand the pathophysiology of this disease.