Summary of Experimental Design
hESCs were differentiated into high purity populations of hMNPs. The phenotype of the resulting cell population was confirmed by analyses of morphology, immunocytochemical marker expression, kainate-stimulated uptake of Co2+ ions by Ca2+ permeable AMPA channels, electrophysiology, glutamate-mediated excitotoxicity, and hMNP-muscle co-cultures. To investigate the neurotrophic potential of hMNP secretions in vitro, we performed growth factor expression analyses, followed by functional assays of hMNP secretions on cortical neurons to assess neurite branching, axonal regeneration and survival of cortical neurons in a neurotoxic environment. The neurotrophic activity of hMNP secretions was confirmed by exposing cortical neurons to hMNP-CM pre-incubated with function blocking antibodies to hMNP-secreted growth factors. The survival, differentiation, and site-specific integration of hMNPs transplanted in the ventral horns cranial and caudal to bilateral cervical contusion injuries was investigated at distances from the injury sites, to investigate the reciprocal interaction between the injured spinal cord and a neuronal transplant population. A rodent model of motor neuron loss was also transplanted with hMNPs, to ensure that the differentiation profile in spinal cord injured animals was a result of the injured spinal cord rather than an inherent property of the transplant population. To investigate the neurotrophic potential of hMNP secretions in vivo, we assayed transplanted animals for sprouting of endogenous fiber tracts, sparing of endogenous neurons, gross tissue sparing, suppression of intracellular signaling pathways associated with SCI pathogenesis, and functional ability.
At day 7 of the 28 day differentiation protocol, 47.8+/−18.2% of cells expressed the undifferentiated hESC marker OCT4, 42.5+/−13% of cells expressed the early neural progenitor marker Pax6, 38.3+/−12.1% of cells expressed the early neural progenitor marker nestin, 4.8+/−2% of cells expressed the mesodermal marker SMA, and 0.5+/−1.3% of cells expressed the mesodermal marker aFP. Cultures consisted of solid core 100–600 µm spheres (), and single cells were eliminated by daily feeding. After plating at day 21, hMNPs migrated radially from neurospheres.
Morphological and immunocytochemical characterization of hMNPs at different stages of differentiation.
At day 25, cells exhibited processes () and 99. 2+/−0.6% of cells expressed the MNP marker Olig1/2 (), 99.2+/−0.44% of cells expressed the MNP marker Isl-1, 98.8+/−1.3% of cells expressed the neuronal marker Tuj1 (, green). Less than 1% of cells expressed the astrocyte marker GFAP (, red), 5.5%+/−3.6% of cells expressed the MN marker HB9, and no cells expressed the mature MN marker ChAT. Less than 0.1% remained unlabeled.
At day 28, the day of transplantation, 96.7+/−2.62% of cells expressed the MN marker HB9 (). Similar to the day 21 immunocytochemical profile, less than 1% of cells expressed the astrocyte marker GFAP and no cells expressed the mature MN marker ChAT. No Oct4 positive stem cells could be identified on or after day 28.
To assess maturation in vitro, hMNPs were plated on astrocytes and allowed to mature for 3 weeks. Matured hMNPs displayed a mature, branched morphology (), 97.7+/−1.53% of cells expressed the mature neurofilament marker SMI-32 (), and 96.6+/−4.4% of cells expressed the MN marker ChAT (), while Olig 1/2 and HB9 expression decreased significantly (p<0.001; 2+/−1.3% and 6+/−4.4%, respectively). SMI-32 positive cells exhibited distinct morphological features of mature MNs, including a large cell body size (>25 µm), a well-developed dendritic tree, and a distinct axon. MN maturation was further demonstrated by kainate-stimulated uptake of Co2+ ions by Ca2+ permeable AMPA channels. Less than 1% of anti-human nuclear positive cells displayed GFAP+ immunoreactivity, and no SMI-32, or ChAT immunoreactive cells were GFAP+. Less than 0.1% remained unlabeled.
The functionality of hMNPs was assessed by electrophysiology, glutamate-mediated excitotoxicity, and hMNP-muscle co-culture experiments. hMNPs matured for 8 weeks consistently displayed resting potentials of −40 to −60 mV. Injection of 20pA to MNs current clamped in the whole cell configuration () elicited trains of action potentials, typical of mature MNs (). The presence of glutamate receptors was evidenced using symmetrical solutions such that at 0 mV, the addition of 100 µM glutamate mediated a small outward current, likely due to the presence of KOH in the internal solution (), and at −70 mV the addition of glutamate mediated a large inward current, as K+ and Ca2+ ions flowed through open glutamate receptors ().
Electrophysiological profile of hESC-derived MNs.
Exposure of hMNPs to glutamic acid at increasing concentrations resulted in increasing amounts of hMNP death. The percentage of propidium iodide (PI) positive cells was 11.36+/−1.25% for 0 µM glutamic acid, 9.68+/−2.2% for 500 µM glutamic acid, and 17.47+/−1.3% for 2000 µM glutamic acid. The percentage of Annexin V positive cells was not significantly different (p>0.01) than the percentage of PI positive cells.
Separation of MN and muscle cell bodies in microfluidic culture platforms connected by 5–20 µm micro-channels 
was required for stable functional innervation of myotubes. hMNPs progressively matured, as evidenced by an enlarged cell body size, an increase in dendritic arborization, and extension of neurites through the micro-channels into the other chamber. Approximately 40% of the myoblasts fused to form myotubes, which displayed spontaneous contractile activity upon innervation with MN neurites. After innervation, α-bungarotoxin, synaptophysin and ChAT staining was significantly (p<0.001) increased over myotube cultures that lacked innervation. No spontaneous contractile activity, α-bungarotoxin, synaptophysin and ChAT staining could be detected in this system when no hMNPs were seeded.
hMNPs Secrete Physiologically Active Growth Factors
To investigate the neurotrophic potential of hMNP secretions, we performed growth factor expression analyses, followed by functional assays of hMNP secretions on cortical neurons in vitro
. PCR () and Western blot analyses (Fig. S1b
) indicated that hMNPs both express and secrete neurotrophin-3 (NT-3), neurotrophin-4 (NT-4), nerve growth factor (NGF), and vascular endothelial-derived growth factor (VEGF). Further analysis using real-time PCR revealed a 10-fold increase in NT-3, a 23-fold increase in NT-4, an 18-fold increase in VEGF, a 3.5-fold increase in CNTF (ciliary neurotrophic factor), a 6-fold increase in IL10, a 121-fold increase in NRG4 (neuregulin 4), and a 275-fold increase in NPY (neuropeptide Y) when compared to hFib controls. Interestingly, there was a 5-fold decrease in NGF expression in hMNP when compared to hFibs (Fig. S1a
). PCR analyses indicated that TGFα, BDNF and GDNF were not produced by the hMNPs.
hMNPs secreted physiologically active growth factors.
Neurite length was 58% longer () in cortical neuron cultures exposed to hMNP-CM for 7 days () as compared to cortical neuron cultures exposed to MN differentiation media (; 182+/−15 µm and 115+/−11 µm, respectively; p<0.001). The neurofilament optical density was 45% greater () in the axonal chamber of microfluidic culture platforms in axotomized cortical neuron cultures exposed to hMNP-CM for 7 days () as compared to axotomized cortical neuron cultures exposed to MN differentiation media (; 15.5+/−2.2 µm and 10.7+/−1.8 µm, respectively; p<0.001). MAP-2 immunostaining of the two chamber microfluidic culture platforms confirmed that no cells migrated to the axonal chamber.
The neurotrophic activity of hMNP secretions was confirmed by exposing cortical neurons to hMNP-CM pre-incubated with function blocking antibodies to hMNP-secreted growth factors. Neurite length was significantly attenuated in cortical neuron cultures exposed to hMNP-CM that contained function-blocking antibodies to hMNP-secreted growth factors (); average neurite length decreased from 185+/−10 µm to 135+/−9 µm in the presence of anti-NT3 (p<0.01), 138+/−8 µm in the presence of anti-NT4 (p<0.01), 149+/−8 µm in the presence of anti-NGF (p<0.01), and 119+/−6 µm in the presence of anti-VEGF (p<0.01). There was no significant difference (p>0.01) in neurite length of cortical neurons cultured in the presence of hMNP-CM containing function-blocking antibodies to GDNF, which is not detected in hMNPs (156+/−10 µm). In addition, there was no significant difference (p>0.01) in average neurite length of cortical neurons cultured in the presence of MN differentiation media (115+/−11 µm) and cortical neurons cultured in the presence of MN differentiation media containing function-blocking antibodies to NT3 (99+/−7 µm), NT4 (101+/−9 µm), NGF (133+/−17 µm), VEGF (102+/−6 µm), and GDNF (142+/−9 µm).
To investigate the effects of hMNP-CM on neuronal survival in a neurotoxic environment, cortical neurons were exposed to microglial conditioned media (MG CM) or LPS-activated microglial conditioned media (LPS MG-CM), in the presence of hMNP-CM or motor neuron differentiation media. Immunofluorescent staining for MAP-2 in cultures exposed to control media () or hMNP-CM () in the presence of MG-CM, indicated that the number of MAP-2 positive neurons in cultures without LPS activation was not significantly different (p<0.05). In cultures exposed to LPS MG-CM, however, the number of neurons was significantly higher (p<0.05) in the presence of hMNP-CM as compared to control media (). These findings demonstrate that hMNPs secrete physiologically active growth factors which promote neuronal survival in an inflammatory environment and enhance axonal sprouting and regeneration in vitro.
Localization and Differentiation of hMNPs in a SCI Model
To investigate the reciprocal interaction between the injured spinal cord and a neuronal transplant population, the survival, differentiation, and site-specific integration of hMNPs was investigated at distances from SCI sites, following transplantation of hMNPs cranial and caudal to bilateral contusion injuries. Human nuclear antigen-positive cells were detected in all transplanted animals, and did not migrate from the transplant sites cranial and caudal to the injury epicenter. Human cells () within the ventral horns double stained with Isl-1 (), p75 (; p75 in red, human nuclei in green), neurofilament or ChAT (; ChAT in blue, human nuclei in brown), consistent with a MN lineage of mixed maturation state. Some human cells in the ventral horns were surrounded by synaptophysin positive processes, suggesting integration with host tissue (; synaptophysin in red, human nuclei in green). Many human nuclear antigen-positive cells extended Tuj1 positive processes, and in some animals, ectopic motor tracts were present in the dorsal and ventral white matter (; Tuj1 in red, human nuclei in green). Tuj1 positive structures had an irregular trajectory and included varicosities along their length, characteristic of new axon growth 
. No-primary and no-secondary antibody controls yielded no Tuj1 tissue staining. None of the CTB+ MNs retrogradely labeled following CTB injection into peripheral muscle contained a human nucleus, demonstrating that transplanted cells did not extend axons into the periphery or form neuromuscular junctions with host tissue, as expected.
Transplanted hMNPs differentiated following transplantation.
Although most transplanted cells were located in the parenchyma cranial and caudal to the injury epicenter where they were placed, transplanted cells were found within the injury epicenter in 5 out of 15 animals. These human cells outside of the ventral horns were predominantly nestin positive or double cortin positive, suggesting back-differentiation to an early neuronal phenotype, and some were GFAP positive suggesting amplification of the minor astrocyte population within the transplant. No human nuclear antigen-positive cells expressed markers for oligodendrocytes.
Localization and Differentiation of hMNPs in a Motor Neuron Loss Model
The Smn−/−;SMNΔ7+/+;SMN2+/+ model of motor neuron loss was also transplanted with hMNPs, to ensure that the differentiation profile in spinal cord injured animals was a result of the injured spinal cord rather than an inherent property of the transplant population. As these mutants have a shortened lifespan, animals were sacrificed within 13 days of transplantation; the lifespan of this animal model limits the time available for differentiation of transplanted cells. Human cells were detected in all Smn−/−;SMNΔ7+/+;SMN2+/+ mice in the ventral horns of the spinal cord. All human nuclear antigen-positive cells double stained with Isl-1, confirming the MN differentiation potential of the transplant population in a model of MN loss that lacks a spinal cord injury. Isl-1 staining was absent in non-transplanted animals, consistent with their MN pathology. Human nuclear antigen-positive cells did not double label with markers for the mature motor neuron markers ChAT or SMI-32, indicating that 13 days of survival in vivo was insufficient for differentiation of transplanted hMNPs. Importantly, very few of the human-positive cells were nestin, double-cortin, or GFAP positive, indicating that the abundance of these cells in SCI sites was a result of the SCI environment rather than the default differentiation profile of the transplant population.
Transplantation of hMNPs Caused Histological Benefit
To investigate the neurotrophic potential of hMNP secretions in vivo, we assayed transplanted animals for sprouting of endogenous fiber tracts, sparing of endogenous neurons, gross tissue sparing, and suppression of intracellular signaling pathways associated with SCI pathogenesis.
hMNP transplantation enhanced sprouting of endogenous serotonergic (5-HT) projections (). In vehicle-injected animals, the patterning of serotonergic projections was consistent with that documented in the literature; there was high innervation of laminae I and II, some diffuse labeling in lamina V, and highly distributed innervation throughout the ventral horn (). The amount of 5-HT immunoreactivity was lowest at 3 mm cranial to the injury epicenter (9±1) but progressively increased up to 1 mm cranial to the injury epicenter (14±3, ). Immediately caudal to the injury epicenter, the labeling of serotonergic projections decreased again (9±1) followed by a slight increase up to 3 mm caudal to the injury epicenter (12±2, ). In contrast, hMNP-transplanted animals consistently contained a higher degree of 5-HT immunoreactivity in regions corresponding to the location of transplanted cells. hMNP-transplanted animals consistently contained aberrant projections throughout the deeper laminae of the dorsal horn and dense innervation of the ventral horns (, arrows). At 2 mm and 3 mm cranial to the injury epicenter, and 1 mm caudal to the injury epicenter, 5-HT immunoreactivity was significantly greater than that observed in vehicle-injected animals (17±2 vs. 12±1, p<0.05; 18±1 vs. 9±1, p<0.001; 14±1 vs. 9±1, p<0.05, respectively). At 1 mm cranial to the injury epicenter, and 2 mm and 3 mm caudal to the injury epicenter, 5-HT immunoreactivity was not significantly different than in vehicle-injected animals (p>0.5).
Transplanted hMNPs promote histological recovery and alter intracellular signaling pathways.
hMNP transplantation enhanced survival of endogenous neurons (). Serial sections were labeled for human nuclei to ensure no human neurons were included in the NeuN cell counts. Cranial to the injury epicenter, vehicle-injected animals had an average of 865±80 endogenous neurons whereas hMNP-transplanted animals had an average of 1091±51 endogenous neurons (p<0.05). Caudal to the injury epicenter, vehicle-injected animals had an average of 871±51 endogenous neurons whereas hMNP-transplanted animals had an average of 1098±57 endogenous neurons (p<0.01).
hMNP transplantation enhanced gross tissue sparing. Morphometric analyses revealed a significantly greater amount of spared tissue in hMNP-transplanted animals compared to vehicle controls, both rostral (7.4±0.5 mm2
versus 5.44±0.58 mm2
) and caudal (8.13±0.6 mm2
versus 5.67±0.34 mm2
) to the injury epicenter. There was no significant difference in tissue sparing between hMNP-transplanted and hFib-transplanted animals. This latter finding indicates that increases in neuronal survival, serotonergic innervation and functional recovery (see below) observed in hMNP-transplanted animals were not due to tissue sparing, as they were absent in hFib-transplanted animals (). This finding is similar to previous reports comparing hESC-derivates and hFib transplant controls 
Transplanted hMNPs caused functional benefit.
hMNP transplantation altered intracellular signaling pathways (). 1 day following transplantation, we observed an 86.3% decrease in phosphorylation of stress-associated protein kinase (SAPK) in hMNP-transplanted animals (p<0.001). This decrease was maintained at 4 days post-transplant where hMNP-transplanted animals had 59.4% less phosphorylation of SAPK than controls (p<0.05). At 7 days post-transplant SAPK phosphorylation increased in hMNP-transplanted animals, however, no significant difference was observed between groups. Similarly, at 10 days post-transplant the two groups equilibrated and SAPK phosphorylation was not significantly different. As inflammatory molecules (eg. TNFα and CD40) and cellular stressors activate SAPK, it is possible that hMNP-secreted factors inhibited upstream activators of SAPK thus, attenuating downstream effects such as apoptosis, immune activation, and inflammation. PCR array analysis showed that hMNP express 15 fold less CD40 than hFib, in addition to other downregulated pro-apoptotic and inflammatory genes. Therefore, it is possible that transplantation of hMNP alters the immune response following SCI.
Transplantation of hMNPs Improved Functional Outcome
hMNP transplantation enhanced performance on a balance beam task, which assesses balance and coordination (). hMNP-transplanted animals performed significantly (p<0.01, repeated measures ANOVA) better than hFib-transplanted animals on the balance beam, demonstrating fewer foot faults (0.7±0.3 left limb and 0.6±0.4 right limb in hMNP transplants vs. 10±3.5 left limb and 9.2±3.1 right limb in hFib-transplanted), a greater average fault distance (131±1.7 left limb and 133.9±0.7 right limb in hMNP transplants vs. 91.2±8 left limb and 86.7±15.2 right limb in hFib-transplanted), and a greater minimum fault distance (107.6±11.5 left limb and 131.9±2.2 right limb in hMNP transplants vs. 57.3±17.5 left limb and 53.2±16.8 right limb in hFib-transplanted). hMNP-transplanted animals began to recover one week earlier than hFib-transplanted animals on all parameters of the task. Both groups began to plateau at about 6 weeks post-transplant.
hMNP transplantation enhanced performance on a Montoya Staircase apparatus, which assesses skilled forelimb reaching. hMNP-transplanted animals performed significantly (p<0.01) better than hFib-transplanted animals on number of pellets taken with the right forelimb at 5 weeks (6.5±2 vs. 2±1) and 7 weeks (8±2 vs. 3±1) post-transplant. No differences were observed in left forelimb function. Similarly, hMNP-transplanted animals performed significantly (p<0.01) better than hFib-transplanted animals on the success rate (pellets taken/pellets eaten) of the right forelimb at 5 weeks (15±6% vs. 0%) and 7 weeks (22±9% vs. 2±2%), and of the left forelimb at 7 weeks (27±7% vs. 5±3%) post-transplant. No differences were observed in left forelimb success rate at 5 weeks post-transplant.
hMNP transplantation did not enhanced performance on a grip strength task, which assesses distal limb strength. There was no significant difference (p>0.01) in grip strength of hMNP-transplanted animals compared to hFib-transplanted animals at any time point.