Targeting Motor End Plates for Delivery of Adenoviruses: An Approach to Maximize Uptake and Transduction of Spinal Cord Motor Neurons

Gene therapy can take advantage of the skeletal muscles/motor neurons anatomical relationship to restrict gene expression to the spinal cord ventral horn. Furthermore, recombinant adenoviruses are attractive viral-vectors as they permit spatial and temporal modulation of transgene expression. In the literature, however, several inconsistencies exist with regard to the intramuscular delivery parameters of adenoviruses. The present study is an evaluation of the optimal injection sites on skeletal muscle, time course of expression and mice’s age for maximum transgene expression in motor neurons. Targeting motor end plates yielded a 2.5-fold increase in the number of transduced motor neurons compared to injections performed away from this region. Peak adenoviral transgene expression in motor neurons was detected after seven days. Further, greater numbers of transduced motor neurons were found in juvenile (3–7 week old) mice as compared with adults (8+ weeks old). Adenoviral injections produced robust transgene expression in motor neurons and skeletal myofibres. In addition, dendrites of transduced motor neurons were shown to extend well into the white matter where the descending motor pathways are located. These results also provide evidence that intramuscular delivery of adenovirus can be a suitable gene therapy approach to treat spinal cord injury.


Results
Intramuscular Injections. Ad.eGFP intramuscularly delivered to the motor end plate (MEP) region of triceps brachii was internalized by the motor axon terminals and retrogradely transported to the ventral horn of the spinal cord. The transgene eGFP was translated and distributed throughout motor neuron somata and their axonal/dendritic processes. Figure 1A,B reveals the intricate network of transduced motor neurons and their afferent/efferent processes. Figure 1C,D displays the morphology of single motor neurons with eGFP expression throughout its soma and branches. In this figure, up to eight branches from a single soma expressing eGFP can be observed (Fig. 1C,D).

Motor End Plate Targeting Analysis.
In order to determine if targeting the MEP region with Ad.eGFP maximizes the number of transduced spinal cord motor neurons, intramuscular injections with Ad.eGFP were also performed in areas of triceps brachii that are away from the motor end plate region. Figure 2A shows that targeting the MEPs with Ad.eGFP produced a 2.5 fold increase in the number of transduced motor neurons when compared with the non-MEP group. An unpaired t-test revealed a statistically significant difference between the non-MEP group (n = 7, SEM = 8.35) and the MEP group (n = 7, SEM = 5.01) with a two-tailed p value = 0.0007 (t = 4.52, df = 12).
Time Course of Expression. The presence of eGFP-positive motor neurons was observed in motor neurons after three, five, seven, nine and eleven days post intramuscular delivery. No eGFP expression was observed in motor neurons after fourteen, twenty-eight or forty-two days. The time course of expression for eGFP in spinal cord motor neurons exhibited a Gaussian curve-like distribution with peak expression at seven days (Fig. 2B). Statistical significance was determined between the different time points by using a one-way ANOVA (F [4,18] = 8.98, p = 0.004). Post-hoc analysis for multiple comparisons using Bonferroni's correction indicated that the seven-day group exhibited a greater number of transduced motor neurons than the three-day group (t = 5.25, p = 0.0002), the nine-day group (t = 3.22, p = 0.019) and the eleven-day group (t = 4.39, p = 0.0014).
Age Matching. Data sets were divided into juvenile and adult groups to see if the age of the mice at the time of surgery could influence the number of eGFP-expressing motor neurons. Animals aged between 5-7 weeks (n = 14) at the time of Ad.eGFP intramuscular injections were assigned into the juvenile group and can be further categorised into the following age brackets: 5 weeks old (n = 4), 6 weeks old (n = 2) and 7 weeks old (n = 8). The adult cohort were those aged 8 weeks and above (n = 24) and can be further categorised into the following age brackets: 8 weeks old (n = 4), 10 weeks old (n = 2), 12 weeks old (n = 4), 13 weeks old (n = 2), 15 weeks old (n = 4), 16 weeks old (n = 4) and 20 weeks old (n = 4). Unpaired t-test analysis revealed that the numbers of Ad.eGFP transduced motor neurons were significantly greater in the juvenile group when compared with the adult group (t = 3.07; df = 36, p = 0.0041) (Fig. 2C). In addition, comparisons between the numbers of transduced motor neurons after targeting the triceps MEP vs non-MEP regions were performed individually, for both juvenile and adult mice. For the juvenile mice, an unpaired t-test revealed that Ad.eGFP transduced more motor neurons when targeting the MEPs (n = 4) than targeting non-MEP (n = 3) regions (t = 2.040, df = 5, p = 0.0485) (Fig. 2D). For the adult mice, an unpaired t-test revealed more transduced motor neurons when targeting the MEPs (n = 3) than targeting non-MEP (n = 4) regions (t = 7.570, df = 5, p = 0.003) (Fig. 2E).
Cocktail Injections of Ad.eGFP and Tetramethylrhodamine (Mini-Ruby). Cocktail injections of Ad.eGFP and Mini-Ruby in triceps brachii resulted in a combination of single eGFP-expressing, single Mini-Ruby-labelled as well as eGFP-expressing/Mini-Ruby-labelled motor neurons (Fig. 3). Columns of eGFP-expressing-or Mini-Ruby-labelled motor neurons were found to span a similar number of spinal cord segments compared to those obtained after Ad.eGFP or Mini-Ruby injections alone (i.e., C5-T1). Moreover, the number of transduced motor neurons from the cocktail injections was similar to that of Ad.eGFP only injections. These data indicate that intramuscular injections of Ad.eGFP and Mini-Ruby does not appear to affect the expression or labelling patterns of each of the individual constituents of this cocktail. Qualitative analysis showed Scientific RepoRts | 6:33058 | DOI: 10.1038/srep33058 that there were more Mini-Ruby labelled motor neurons than eGFP-expressing motor neurons ( Fig. 3A-C). Figure 3D-F illustrates the typical punctate appearance of Mini-Ruby distributed throughout the cytoplasm and dendritic processes of the motor neurons, a labelling profile that is in sharp contrast with the smooth ubiquitous  29 100 μ l of 5.0 × 10 9 pfu/ml SD Rats 8-10 weeks 1-6 weeks CAG BDNF, LacZ Middle belly of the superficial layer of the Sternomastoid muscle

Motor Neuron Axonal and Dendritic Pattern of eGFP Expression.
Targeting the motor end plate region of triceps brachii with Ad.eGFP resulted in widespread expression of the transgene. Figure 4 is a series of four consecutive 50 μ m-thick sections through the ventral horn of the spinal cord displaying axonal and dendritic branching from one transduced motor neuron. Figure 5A shows a longitudinal section through the ventral horn of the spinal cord with eGFP-expressing motor axons exiting the ventral horn via the ventrolateral funiculus and entering the ventral roots. Locally in the ventral horn grey matter, eGFP-expressing dendritic processes with numerous possible spines were observed (Fig. 5B). Figure 5C shows eGFP-expression in processes also located in the ventrolateral funiculus after bilateral Ad.eGFP injections into triceps brachii. In this section of tissue, eGFP-expressing processes extend, at least, through two spinal cord segments. Furthermore, eGFP-expression was also found in processes extending medially into the ventral funiculus ( Fig. 5D) as well as laterally (Fig. 5E) into the ventrolateral funiculus as well as, in the dorsal funiculus ( Fig. 5F) and dorsolateral funiculus (Fig. 5G). Figure 6 summarizes the course taken by the eGFP-transduced axonal and dendritic processes extending from the grey matter toward and into the different motor tracts running within the white matter of the spinal cord. In addition to eGFP-expression found in the spinal cord, Ad.eGFP-targeted muscles exhibited robust expression of eGFP in the majority of muscle fibres (Fig. 7). These eGFP-expressing muscle fibres were located both proximal and distal to the injection sites (as shown in Fig. 7i).

Discussion
The present study is a systematic evaluation of the parameters for the maximum transgene expression in spinal cord motor neurons after intramuscular delivery. These parameters are 1) the location of the injection sites within skeletal muscles, 2) the time course of expression of the transgene in spinal cord motor neurons and 3) the influence of the age of the animals on motor neuron transduction levels. Targeting the motor end plates (MEPs) in triceps brachii with Ad.eGFP yielded a 2.5-fold increase in the number of transduced motor neurons when compared with the delivery of the same volumes of adenoviral vectors in regions of the muscle that are away from the MEP region. Peak expression of the transgene was detected seven days after the injections were performed. In addition, more motor neurons were transduced in juvenile (i.e., 5-7 weeks old) than in adult mice (i.e., 8+ weeks old). In addition, eGFP-expressing dendrites from transduced motor neurons were observed to extend well into white matter regions where the different descending motor pathways run. This analysis also showed that targeting   35 . The green crosses represent the five areas targeted for full-length MEP injections and the red crosses represent the areas away from the MEP region (Non-MEP). ii) Quantification of the numbers of eGFP-expressing motor neurons from both age groups. An unpaired t-test revealed a statistically significant difference between the non-MEP group and the MEP group with a two-tailed p = 0.0007 (***). (B) Number of Ad.eGFP-positive neurons after intramuscular injection of 40 μ ls of Ad.eGFP into the MEP region in triceps brachii observed after 3, 5, 7, 9 and 11 days post-injections. Unpaired t-test analyses indicated that the number of transduced motor neurons from the seven-day group was significantly larger than the three-day group (***p = 0.0002), the nine-day group (*p = 0.019) and the eleven-day group (**p = 0.0014). (C) Comparison of the number of transduced motor neurons after intramuscular injections of Ad.eGFP between juvenile and adult animals. T-test analysis revealed a statistical difference between the two groups (t = 3.07; df = 36; p = 0.0041). The error bars in all panels reflect the mean with SEM. (D) Comparison of the number of transduced motor neurons in juvenile mice after intramuscular targeting of the MEP vs non-MEP areas in triceps brachii. An unpaired t-test revealed a statistically significant difference between the age-matched groups with a one-tailed p value = 0.0485 (*). (E) Comparison of the number of transduced motor neurons in adult mice after intramuscular targeting of the MEP vs non-MEP areas in triceps brachii. An unpaired t-test revealed a statistically significant difference between the age-matched groups with a one-tailed p value = 0.0003 (***). The mice in which the data was obtained for (D,E) received 40 μ l of Ad.eGFP distributed into the locations depicted in (A) and tissue was obtained seven days later.
the MEPs with a cocktail of Ad.eGFP and the retrograde tracer Mini-Ruby was a practical option to spatially determine the location of transduced motor neurons after adenoviral-mediated transgene expression had faded.

Targeting the Motor End Plate Region for Maximum Transgene Expression.
Once considered a promising technique to restrict transgene expression into ventral horn motor neurons, the popularity of intramuscular injections of adenoviral vectors have diminished over the last decade, mainly because it has resulted in suboptimal levels of transduction. In the literature, the exact location of the intramuscular injections of recombinant adenoviral vectors is not always disclosed in great detail, however most intramuscular injections of adenoviral vectors seem to be performed in the 'belly' of the muscle 25,29,[31][32][33] (Table 1). Over the last years, we have mapped the location of the motor end plates of the mouse and the rat forelimb and hindlimb and found that to maximise the delivery of retrograde tracers, the entire MEP region needs to be targeted [34][35][36][37][38] . Here, we demonstrated that this principle also applies to recombinant adenoviral vectors. Indeed, the present analysis showed that targeting the entire MEP region of triceps brachii with Ad.eGFP lead to approximately a 2.5-fold increase in the number of transduced motor neurons in comparison with injections that were directed to areas of the muscles distal to the MEPs. This was also the case for when the analysis was performed separately, both juvenile and adult mice whereby targeting the MEPs resulted in significantly more transduced motor neurons. Hence, regardless of age it is clear that targeting the MEPs results in greater transduction efficiency of spinal cord motor neurons.
Adenoviruses enter cells through receptor-mediated endocytosis where they bind with high affinity to receptors such as coxsackie and adenovirus receptors (CAR) and members of the integrin receptor family, amongst others [39][40][41] . In skeletal muscle, CARs are confined to the neuromuscular junction in adult animals 40,42 and after internalization, adenoviral particles are sorted and trafficked by cytoplasmic dynein along the microtubule network to the cell soma where they gain entry into the nucleus to induce gene expression 39,41 . This, at least partially, explains the success of the MEP targeting. The present study, therefore, suggests that the sub-optimal levels of Time Course of Expression of Ad.eGFP. In line with the literature, the present investigation showed that expression of the eGFP transgene could be observed in motor neurons as soon as three days post delivery and up to eleven days, with expression levels peaking at seven days 21,22,29,31,32,[43][44][45] . These data represent both age groups, as age-dependent time-course of expression experiments were not performed. There have been reports, however, that transgene expression after intramuscular injections can last for much longer (i.e., up to four months), specifically in new-born animals 23,27,46 or when measured by RT-PCR techniques 27 . This temporal regulation of adenoviral-mediated gene expression is advantageous where short-term up-regulation of the transgene is advantageous. In particular, transient gene expression can be beneficial for spinal cord regeneration whereby tightly regulated control of synaptogenesis is required between descending motor axons and their targets, motor neurons and interneurons 6,[47][48][49] . In this context, short-term neurotrophic factor treatment facilitates synaptic plasticity 9,16 whereas chronic treatment has resulted in spasticity and hyper-excitability 7,8 , due to the probable formation of novel, aberrant connections [6][7][8]30 . Further, short-term expression of neurotrophic factors fits the growth rate of regenerating axons, which is estimated to be 1 mm per day 50 . Therefore, the auto-regulatory temporal gene-expression, as mediated by the transient nature of adenovirus, is ideal for short-term expression of therapeutic genes.
Age Matching. In our hands, the total numbers of transduced motor neurons arising from intramuscular injections of Ad.eGFP was greater in juvenile mice. A caveat, however, is that this statistical analysis was conducted post-hoc on all data sets, (i.e., time-course and optimal injection sites). Despite this limitation, we are confident that the success of intramuscular delivery and retrograde transport of adenovirus is age-dependent. One explanation for the greater numbers of transduced motor neurons observed in the juvenile group could be, at least in part, attributable to the fact that they received the same volume of Ad.eGFP as the adult cohort, despite that they would have smaller triceps. We do not think, however, that the difference in muscle mass between the juvenile and the adult cohorts is the main factor that contributes to the difference in the numbers of transduced motor neurons between the two groups of mice. Rather, coxsackie and adenovirus receptor (CAR) expression would be the key factor accounting for this disparity. Indeed, other groups of researchers have established that the age of the animals directly influences vector-spread, cellular tropism and, ultimately, transgene expression 51,52 . It has also  been established that the abundance of myoblasts and myotubes as well as the relative abundance of CARs found in young mice contributes to greater transduction efficiency 42,53,54 . Further, the overexpression of CARs in muscle from adult transgenic mice has been shown to produce a significant increase in the transduction of skeletal muscle 55,56 . Taken together, these findings support the view that the reduced numbers of transduced motor neurons in adult mice is more than likely a consequence of the down-regulation of CARs observed throughout age.
These data demonstrate that the retrograde machinery of motor neurons for adenoviral transport is functional from 5 weeks through to 20 weeks of age, in which period of time the ability for uptake from the MEPs into pre-synaptic axon terminals and subsequent delivery is preserved. Indeed, for the wild-type mouse, the formation of the neuromuscular junction begins during embryonic development (for a recent review see ref. 57), however, the pre-synaptic axonal terminals are in an adult-like state after one month 58 . Furthermore, as there are no differences in the signalling endosome dynamics of retrograde axonal transport of wild-type mice aged from 1 month up to 13 months old 59 the motor neuron transduction differences cannot be due to delays in retrograde axonal transport. In summary, although it has been previously reported that gene transfer is greater in younger animals, the present study is the first to report that adenoviral delivery in skeletal muscles leads to significantly greater levels of transduction in spinal cord motor neurons of juvenile mice.  Putative Mechanism(s) Underlying Adenoviral Vectors Transient Expression. The most common explanation for the transient expression of adenovirus is the silencing of the episomally located expression cassette, although some authors have suggested that this phenomenon is caused by an immune response triggered toward the virus from the host tissue. For instance, Zirger et al. 60 have demonstrated that direct brain injections result in CD4+ and CD8+ T cell-mediated elimination of adenoviral transgene expression through both cytotoxic and non-cytotoxic mechanisms 61 . On the other hand, intramuscular injections of adenoviral particles in the diaphragm has been reported to initiate an immune response in the targeted muscle, however the central nervous system (CNS) was not scrutinized in this study to see whether the immune response has spread centrally 62 . In another study, no significant loss of motor neurons was reported after peripheral injections of adenoviral-vector in the sciatic nerve 63 , supporting the notion that the immune system is activated after central but not after peripheral nerve or muscle injections.

Cocktail Injections of Adenoviral Vectors and Retrograde Neuronal Tracers.
As adenoviral expression is transient, intramuscular injections of a cocktail of adenovirus and permanent neuronal retrograde tracer enables to decipher the exact rostro-caudal location, within the spinal cord, of transduced motor neurons after the peak of the protein marker has faded 28 . In our hands, cocktail injections lead to approximately twice as many Mini-Ruby-labelled motor neurons than eGFP-expressing motor neurons. In line with Martinov et al. 28 , these cocktail injections resulted in both single Mini-Ruby-labelled and eGFP-expressing motor neurons as well as double-labelled motor neurons. It is worth noting that, unlike that of adenovirus, the mechanism of uptake of dextran conjugates such as Mini-Ruby is not receptor-mediated endocytosis, perhaps explaining why the tracer and the adenoviral vector are not always co-localized inside the same motor neurons. Despite a difference in their expression patterns, both Ad.eGFP-and/or Mini-Ruby-positive motor neurons span the same number of spinal cord segments and as such, cocktail injections are a viable approach to determine the spatial distribution of transduced motor neurons after transgene expression has faded.

Axonal and Dendritic Patterns of eGFP Expression. After the retrograde transport of recombinant
Ad.eGFP, translated eGFP does not require active transport and freely diffuses throughout the transduced neurons 64 . As expected, eGFP expression was ubiquitously distributed throughout the cytoplasm, nucleus and axonal and dendritic processes of the transduced motor neurons, giving them a Golgi stain-like appearance. Interestingly, eGFP-positive dendrites were observed to extend into several compartment of the white matter (as summarized in Fig. 6). The eGFP-positive processes extending in the ventral roots are presumably axons originating from transduced motor neurons and travel within the peripheral nerve toward the targeted muscles (Fig. 5A). On the other hand, dendrites expressing eGFP were observed extending away from the neuron somata to reach into various regions of the spinal cord white matter, i.e., in close proximity to the major descending motor tracts (Fig. 5B-G). For instance, eGFP-positive dendrites were observed to extend into the dorsal funiculus where the dorsal corticospinal tract is located, the dorsolateral funiculus where both the rubrospinal and the lateral corticospinal tracts travel as well as into the ventral funiculus where the ventral corticospinal tract and the reticulospinal tracts run 65,66 . The extensive dendritic arborisation pattern of motor neurons suggests that many synaptic contacts between the main descending motor tracts and spinal cord motor neurons would be axo-dendritic in nature. Together these data show that motor neuron dendrites can extend a relatively great distance away from their somata to reach for synaptic contact 67,68 . Implications of the Current Work and Future Direction. This work provides a consistent adenoviralmediated gene delivery approach for gene therapy scenarios that aim to overcome the deleterious effects of SCI and other neuromuscular dysfunction. Intramuscular injections of viral vectors are a minimally invasive and therefore constitute a clinically relevant way to deliver therapeutic genes to motor neurons 29 . Intramuscular injections of recombinant adenovirus transduce both the skeletal muscle fibres and the motor neurons that supply them, thus delivering therapeutic genes to both the central and peripheral aspects of the motor units is of interest for amyotrophic lateral sclerosis.
One question remains to be answered: is a 2.5-fold increase in therapeutic genes sufficient to promote CNS plasticity and functional recovery? A number of studies have shown that the introduction of neurotrophic factor in the injured spinal cord encourages the elongation of injured axons toward the exogenous neurotrophic source [3][4][5][6]9,16,49 . As transduced dendrites are not confined within the limits of the grey matter but extend well into the different regions of the white matter, (i.e., where descending motor tracts run) has great relevance for spinal cord regeneration. For example, in a gene therapy scenario, transduced motor neurons expressing exogenous chemoattractant molecules would become 'therapeutic bait' for the transected axons and that the latter will not have to leave the white matter in order to make axo-dendritic contact with the motor neurons that were once their post-synaptic targets (see 6,50). These considerations do not determine whether intramuscular delivery of adenoviral vectors is sufficient to trigger functional regeneration. The current study, however, by improving the delivery parameters to achieve up-regulation of gene expression in spinal cord motor will contribute to making gene therapy a step closer to treat SCI and other neuromuscular disorders. Adenovirus. An expression cassette comprising the cytomegalovirus promoter and the cDNA encoding enhanced green fluorescent protein (eGFP) packaged into Ad serotype 5 was obtained from the University of Pennsylvania's Penn Vector Core (Ad.eGFP). There were two separate batches of adenovirus used in this study. The viral titre of the first batch was 4.84 × 10 12 pfu/ml whereas the viral titre of the second batch was 5.48 × 10 12 pfu/ml as quantified using real-time PCR primer-probe sets specific for the E2a gene (data not shown). Each experimental group received intramuscular injections from both adenoviral batches. For example, for the seven-day experimental group, four animals received adenovirus from the first batch (i.e., 4.84 × 10 12 pfu/ml) and three animals received adenovirus from the second batch (i.e., 5.48 × 10 12 pfu/ml).

Intramuscular Injections in Triceps
Brachii. Anesthesia was induced with isoflurane (Provet, Sydney, NSW, Australia; 1-2% in O 2 ). The fur covering the targeted forelimb was shaved and cleaned with 70% ethanol. A small incision was made directly in the skin to expose triceps brachii. Each animal received 40 μ l of Ad.eGFP delivered in five injections evenly distributed along the full length of the motor end plates (MEPs) with graded glass micropipettes with lumen size at the tip of the micropipettes of 0.5 μ m (DKSH, Zurich, Switzerland) as per Mohan et al. 37 . These 1 μ l-graded glass micropipettes have a maximum volume capacity of 10 μ l. The Ad.eGFP-filled micropipettes were slowly lowered in the muscle until a slight resistance is encountered that indicates that the tip of the micropipettes has reached the fasciae that cover the deepest part of the muscle. The adenovirus is then slowly delivered manually, 1 μ m at a time, while the tip of the micropipette is steadily lifted towards the exposed surface of the muscle. The motor end plate region for triceps brachii forms an upside down 'V' shape and traverses the entire width of the muscle 35 . Great care was taken to preserve the fasciae covering the muscle and those in the surrounding as well as to spare the blood vessels near the targeted area. After the injections, the muscle was wiped with gauze to remove any virus that may have inadvertently seeped out from the injected muscle, after which the skin was closed with surgical clips (Texas Scientific Instruments LLC, Boerne, TX, USA). In order to determine whether targeting the motor end plates would lead to maximum uptake of adenovirus and expression of the transgene, 40 μ l of Ad.eGFP was also injected in areas of triceps brachii that surround, but do not include, the MEP region (n = 7). Tissue Collection and Histological Processing. At the conclusion of these experiments, the mice received a lethal dose of Lethabarb (100 mg/ml; Virbac, Sydney, NSW, Australia) and were intracardially perfused with 0.1 M phosphate buffer (PB) followed by 4% paraformaldehyde (Sigma Aldrich, Castle Hill, NSW, Australia) in 0.1 M PB. Cervical spinal cord dissections were performed by making an incision into the skin covering the midline of the animal, from the base of the skull to the thoracic region (as per 35,37). The paravertebral muscles were then reflected, exposing the cervical part of the vertebral column and the large spinous process of C2 was identified and then removed. This process exposed the dorsal roots of C2, which were then colored with a permanent marker for easy identification. C3-T1 vertebrae were subsequently removed, one at a time, exposing their dorsal roots that were also coloured with permanent markers using alternative colours (e.g., C3, C5, and C7 were coloured with a green marker and C4, C6 and C8 were coloured with a blue marker). A fiducial mark was made in the white matter approximately half way between two adjacent dorsal roots to indicate segmental boundaries. After this process, the cervical spinal cord was cut transversely into two-segment blocks (i.e., C3-C4, C5-C6, C7-C8 blocks). The spinal cord tissue blocks were then removed from the carcass, post-fixed overnight in a solution containing 4% paraformaldehyde in 0.1 M PB and then cryoprotected in a 30% sucrose solution (Sigma-Aldrich, Castle Hill, NSW Australia) in distilled water for 2 days at 4 °C. The spinal cord blocks were then cut along the horizontal axis into 50 μ m-thick sections using a cryostat. The tissue sections were collected into 48 well plates containing 0.1 M PB and then mounted onto Superfrost TM Plus microscope slides (Thermo Scientific, Scoresby, VIC, Australia). Triceps brachii muscles were also dissected, post-fixed, cryoprotected and sectioned at 50 μ m. The muscles were oriented so to obtain tissue sections that are transverse to the myofibres direction. The tissue was mounted directly onto Superfrost TM Plus microscope slides (Thermo Scientific, Scoresby, VIC, Australia). The microscope slides were air-dried and then coverslipped with a mounting medium suitable for fluorescence microscopy (Dako, North Sydney, NSW, Australia).

Image
Analyses. An epifluorescence BX51 (Olympus, Notting Hill, VIC, Australia) and a Zeiss Z1 AxioExaminer NLO710 confocal microscopes (Carl Zeiss Pty Ltd, North Ryde, NSW, Australia) were used to capture images of the spinal cord and muscle tissue. All images were processed with Adobe Photoshop CS6 (Adobe Systems Incorporated, San Jose, CA, USA). Motor neurons were considered positively labelled when eGFP expression and/or Mini-Ruby labelling were present within both the soma and at least one axonal/dendritic process. The total numbers of eGFP-or Mini Ruby-labelled motor neurons were then counted.

Statistical analysis.
Comparisons between two groups of animals were analysed using unpaired, two-tailed t-tests. Comparisons between several groups of animals (e.g., time course of expression) were performed using a one-way analysis of variance (ANOVA) followed by post hoc Bonferroni's correction. Statistical significance was determined using Prism version 6 (GraphPad Software, La Jolla, CA, USA).