Descending motor circuitry required for NT-3 mediated locomotor recovery after spinal cord injury in mice

Locomotor function, mediated by lumbar neural circuitry, is modulated by descending spinal pathways. Spinal cord injury (SCI) interrupts descending projections and denervates lumbar motor neurons (MNs). We previously reported that retrogradely transported neurotrophin-3 (NT-3) to lumbar MNs attenuated SCI-induced lumbar MN dendritic atrophy and enabled functional recovery after a rostral thoracic contusion. Here we functionally dissected the role of descending neural pathways in response to NT-3-mediated recovery after a T9 contusive SCI in mice. We find that residual projections to lumbar MNs are required to produce leg movements after SCI. Next, we show that the spared descending propriospinal pathway, rather than other pathways (including the corticospinal, rubrospinal, serotonergic, and dopaminergic pathways), accounts for NT-3-enhanced recovery. Lastly, we show that NT-3 induced propriospino-MN circuit reorganization after the T9 contusion via promotion of dendritic regrowth rather than prevention of dendritic atrophy.

panel c: this panel is supposed to show close appositions between residual propriospinal fibers and motoneurons. What the images show is a higher density of propriospinal terminals after Nt-3 treatment but no appositions. While it could be expected that there is more appositions the images do not allow appreciating this point as the lack of quantification. Investigating appositions would require higher magnification images and quantifications of these appositions in single planes and 3D views. This should be presented in the paper and detailed in the material and methods as this is an important and interesting point. Figure 4 panel a: What is the rational to now investigate thoracic propriospinal neurons instead of the cervical ones investigated before? It is not clear why the experiment has not followed logically with using the cervical propriospinal neurons. Please comment. Figure 5 a1: what is the hole at T9? There is no contusion here. Figure 5 a3: here again there is no obvious colocalisation of puncta as could be defined in the image. Higher magnification images would be needed to carry on such analysis as the use of single plans and 3D images. The detail of such analysis would also need to be added to the material and methods section. Figure 6 c: was this analysis corrected for motoneuron size? Couldn't find the information on the material and method. This is quite important to interpret the images in b which do not provide the information themselves.
Reviewer #3: Remarks to the Author: It has been difficult to obtain direct evidence of specific axonal pathways involved in a return of function after spinal cord injury, with most previous reports ending up being correlational at best. The present study provides considerable tract tracing, pathway silencing, and electrophysiological data that propriospinal neurons rostral to the injury have a significant role in improved locomotor function after a lower thoracic contusion injury of adult mice, to the exclusion of coticospinal, rubrospinal serotonergic and dopaminergic axons. Part of the enhanced recovery is attributed to NT-3 mediated modulation of lumbar motoneuron circuitry and propriospinal neuron sprouting, and yet while there is observed regrowth of the motoneuron dendritic field there is no evidence that this has a role in return of function.
There are three major conclusions but the second one about identifying the important role of the propriospinal-motoneuron circuitry for recovery of motor control is the strongest and most convincingly argued by the data. The first conclusion, that spared axons are required for recovery of motor function (i.e. axons do not recover in a complete transection model) is not surprising. The third conclusion is that motoneuron dendrites atrophy after SCI but regrow if motoneurons are transfected with AAV-NT3. This is interesting but does not seem to fit in with the observation of propriospinal-motoneuron circuitry and locomotor function after SCI. At least there is no data provided to link dendrite restructuring to a functional role. It is not clear that dendritic regrowth is indispensable for locomotor recovery (line 391).
Overall the experimental design is outstanding, with precise tracing of pathways, clever use of a dual neuron silencing approach and incorporation of electrophysiological stimulation to determine whether cortical or red nucleus neuron stimulation was involved in direct stimulation of lumbar motoneurons after SCI. This is a strong body of work will be of use to the SCI regenerationplasticity field.
A point of concern is the absence of discussion of reticulospinal and vestibulospinal pathways and their role in rodent locomotion. It is not clear that the impact of these pathways can be discounted without being tested. It is stated in Abstract and Discussion that the spared descending propriospinal pathway and not other pathways account for recovery. Also it is stated (line 289) that the propriospinal-MN circuit reorganization is functionally required for recovery, but this study does not indicate that it is sufficient for recovery. Other pathways may be involved. It would appear that both ReST and VST pathways are minimally damaged by the contusion injury (Suppl Figure 3) but there may be some response to local increase in NT-3 as seen with other pathways. This should be discussed at least. Minor issues: Line 494 -change pre-myelinated to pre-demyelinated Line 618 -change trail to trial 1

Point-by-point Responses to the Reviewers Comments
Reviewer #1 (Remarks to the Author)

Overall comments:
As far as I can tell, the work appears to be have been rigorously conducted and is of high quality. The experiments are well designed and properly controlled and have sufficient replicates to have statistical power. The data presented look good and are convincing. The findings are important on several levels and should be of interest to a broad audience interested in restoring function after spinal cord injury and more generally in understanding the mechanisms that can promote the regeneration of functional neural circuits. I have no concerns regarding technical aspects of the study. It seems very thorough. I have only a few suggestions for edits regarding the text.
Response: We thank the reviewer for his/her strong enthusiasm for this manuscript.

Specific comments:
1) About half-way through the results, around page 7 lines 176 onward, the experimental model switches from evaluations long projecting propriospinal neurons in the cervical area injury for anterograde tracing studies to evaluating short projecting propriospinal neurons in the thoracic area around T6 just above by retrograde tracing studies after contusion injuries which it seems were always at T9 for both anterograde and retrograde tracing. Studying both sets of neurons can be valid in their own right, but the switch is never explained. Why use one set of neurons for anterograde another for retrograde analyses? Some explanation should be provided and perhaps a bit of discussion on differences between long cervical and short thoracic propriospinal neurons.
Response: The reviewer raised an important question which was also shared by reviewer #2. In an early study, we found that retrograde transport of AAV-NT-3 to lumbar motorneurons (MNs) stimulated the sprouting of spared cervical dPST terminals in lumbar MN pools 1 . The increased dPST fibers were in close apposition with CTB-labeled MNs, indicating that NT-3 treatment reinforced residual propriospino-MN circuit reorganization. To confirm the NT-3-modulated propriospino-MN connections, we took advantage of a time-dependent, muti-transsynaptic virus, PRV, to retrogradely label the MN-related descending propriospinal neurons (dPNs). At 72 h, we found that, despite the apparent disruption of propriospino-MN connections, there was still a significant increase in PRV-labelled dPNs at the T7 level in SCI mice after the AAV-NT-3 treatment, as compared to SCI control mice with AAV-GFP treatment ( Supplementary Fig. 5c,  d). Notably, these dPNs above the lesion relayed descending commands, such as CST, RST, and cervical dPST, down to the lumbar motor circuit (Supplementary Fig. 6). In contrast to dozens of PRV transsynaptically-labeled neurons at the thoracic level, few, if any, labeled dPNs were found in the cervical level in contusive mice with either AAV-GFP or AAV-NT-3 treatment at 72 h post-inoculation when the time is sufficient for viral propagation to label plenty of neurons in cervical segment in uninjured sham mice (Response Fig.1, inserted below, also revised Supplementary Fig. 5c). This result suggests that more direct synaptic connections between the thoracic dPN-MN circuit might exist as compared to the cervical dPN-MN circuit. With consideration of the feasibility and efficacy of doxycycline-induced dual-viral system, the finding of more PRV-labeled, MN-related dPNs in the thoracic level than the cervical level, is the primary reason that prompted us to use thoracic dPNs for functionally dissecting the role of descending propriospino-MN circuit in NT-3-mediated recovry. Secondly, although the T9 contusion can result in large-scale alterations of spared dPST that relay the supraspinal commands downstream, the considerable reorganization may occur around or near the lesion site. Therefore, in comparison to the cervical dPST, the thoracic spared dPST axons may contribute more directly to the recovery of function after a thoracic contusion. In the dPN selective silencing experiment, we choose to inject AAV/Tet-On into T5-T7, but not T8, due to the concern of its diffusion across the lesion site.
Although thoracic dPNs above the lesion site are considered to be more functionally relevant to the contribution of NT-3-mediated motor recovery, we can not exclude the engagement of the cervical dPNs. The approach of PRV-labeled dPNs is in a time-dependent manner. The contusive animals may require more time to label MN-related dPNs in the cervical level than sham mice due to the lesion that may lead to severe disruptions of the direct dPST-MN circuit. Therefore, future work determining whether cervical propriospinal neurons also play a role in NT-3mediated locomotor recovery would be interesting; if they do, it would also be interesting to know which dPNs play a more significant role, and what will be the consequence if both dPNs are silenced at the same time.
We have added data analysis to Supplementary Fig. 5c and incorporated such discussion on page 10 and 18.
2) NT3 is a pleiotropic growth factor with many targets and many functions and effects. The authors should provide a bit of background and discussion with literature citations on why they think that providing NT3 as a retrograde factor to motor neurons would induce motor neurons to attract more connections by propriospinal neurons. Why did they try this strategy, and what background evidence (literature) is there about how it might work? Is this a mechanism that is documented to function during development?
Response: This is also an excellent comment made by the reviewer. We agree that introducing more background and discussion of NT-3 will provide a more detailed rationale behind why we chose NT-3 as a retrogradely transported trophic factor to reverse MN atrophy and restore local motor circuits after a rostral SCI.
The lumbar MNs are the final common pathway for the motor output of the hindlimbs. They can be impaired by a direct injury to the lumbar cord or by an indirect injury that occurred at levels above the lumbar cord. When an SCI occurs at the above lumbar levels (namely above-level injury), the lumbar MNs are not directly injured by the trauma, but they undergo profound 3 dendritic atrophy and synaptic stripping from denervated supraspinal and propriospinal axons. Such altered lumbar MN morphological and synaptic changes could result in impaired motor outputs to hindlimb muscles and therefore impaired locomotor functions. Whereas most SCI studies have been focused on the regeneration or protection of injured spinal cord at the site of injury, few studies have explored how modulation of lumbar MN circuitry would affect pathological and functional consequences after an above-level SCI. Our previous study showed that NT-3 could be a promising restorative treatment to recover MN atrophy and locomotor deficits after an above-level SCI. In the present study, we aim to investigate the anatomical and functional mechanism of NT-3-enhanced locomotor recovery.
Neurotrophins are a family of proteins that regulate neuronal survival, neurite outgrowth, synaptic plasticity, and neurotransmission. Among them, NT-3 mRNA is highly expressed in the developing spinal cord in motor neurons but decreased in the adult spinal cord 2,3 . NT-3 is essential for MN survival, target finding, innervation, and synapse formation, mainly during development and early postnatal maturation 2,4 . Therefore, we hypothesize that vector-induced NT-3 expression in MNs may allow us to mimic some of the observations seen during development, which can either protect MNs from atrophy or promote dendritic regrowth of these MNs after SCI.
It should be noted that the expression of TrkC receptor that NT-3 binds with the highest affinity was not limited to motoneurons, but was seen on the vast majority of neurons throughout the grey matter of spinal cord 2,5,6 , suggesting that various other populations of neurons also respond to the NT-3 expression. A body of studies has demonstrated that local, sustained expression of NT-3 supports the plasticity of corticospinal tract 7-9 , spinal neuronal survival, and regeneration 10,11 . More importantly, NT-3 can also serve as chemotropic guidance for regenerated axons in establishing their projection, selecting their targets and reforming appropriate connections 10,12,13 . Together, the rationales we mentioned above channel the hypothesis that retrogradely-transported NT-3 to lumbar MNs may be a beneficial strategy that is able to attenuate SCI-induced lumbar MN dendritic atrophy, attract more connections from spared descending pathways, and, therefore, promote locomotor functional recovery after a thoracic contusion. We have added a new discussion with new literature citations on why NT-3 was chosen as a treatment for SCI (page 18-19).

Overall comments:
The article by Han et al is interested in dissecting the effects of NT-3 and follows on a recent report demonstrating that retrograde delivery of NT-3 to motoneurons attenuate SCI-induced lumbar motoneurons dendritic atrophy and improved functional recovery. In this report the authors go further in their investigations to demonstrate that (i) residual projections to lumbar motoneurons are necessary to produce leg movements, (ii) that spared propriospinal connections are more important that supraspinal motor tracts to mediate NT-3 induced recovery and (iii) that NT-3 treatment promotes motoneuron dendritic regrowth. The paper is well written, very well illustrated and overall provides new insights on NT-3-induced spinal remodeling of axonal connections. The material and method section relative to the quantifications should be more detailed to allow a correct interpretation of the data.
Response: We thank the reviewer #2 for his/her positive comment on our manuscript. We have provided more detailed information on the material and method section relative to the quantifications to allow a correct interpretation of the data.

panel b: no scale bar is provided.
Response: Scale bar has been added to the panel b in Figure 3.

panel c: this panel is supposed to show close appositions between residual propriospinal fibers and motoneurons. What the images show is a higher density of propriospinal terminals after Nt-3 treatment but no appositions. While it could be expected that there is more appositions the images do not allow appreciating this point as the lack of quantification. Investigating appositions would require higher magnification images and quantifications of these appositions in single planes and 3D views. This should be presented in the paper and detailed in the material and methods as this is an important and interesting point.
Response: We appreciate the reviewer's comment. We found that NT-3 treatment stimulated a higher density of propriospinal terminals in lumbar MN pools, but we did not directly quantify the close appositions between propriospinal terminals and lumbar MNs. We agree with the reviewer's suggestion to quantify these appositions in single planes and 3D views via higher magnification images. Sections of the lumbar spinal cords (L2-L4) with both CTB-labeled MNs and BDA-labeled dPST axons were imaged every fourth 40-μm cross-sections. All the apposition data described were acquired in the ventral horns (MN pools) of these sections using a 60x oil objective confocal fluoview microscope (Olympus, Japan). To determine the appositions, we studied the co-labeling of MNs with propriospinal axons in a 3D view. The criteria for quantification are that only the apposition observed in all three dimensions is considered a valid candidate to be counted (indicated as arrowhead in Supplementary Fig 4e), with the intention of eliminating other appositions just in one or two dimensions (indicated as arrow in Supplementary Fig 4e). We counted all valid appositions of BDA-labeled dPST axon terminals with 2-4 CTB-labeled MNs within an unbiased virtual courting space in each section. In each animal, we counted 4-6 sections. We have incorporated this quantification in Supplementary  Fig 4 e and 4f. With this data, we confirmed that retrograde transport of AAV-NT-3 to lumbar MNs increased the terminal BDA-labeled dPST fiber density in lumbar MN pools, as compared to AAV-GFP treatment. Importantly, the increased dPST terminals in the ventral horn intermingled and apposed with CTB-labeled MNs (Supplementary Fig 4e, f). This result suggests that retrograde transport of AAV-NT-3 to lumbar MNs stimulates the reorganization of spared dPST connections with lumbar MN pools following contusive SCI. As the reviewer suggested, we have presented and highlighted the detailed methods in the materials and methods section (page 28-29).

panel a: What is the rationale to now investigate thoracic propriospinal neurons instead of the cervical ones investigated before? It is not clear why the experiment has not followed logically with using the cervical propriospinal neurons. Please comment.
Response: The reviewer raised an important question which was also raised by reviewer #1. Please refer to our response to comment #1 of reviewer #1 (page 10 and 18).
(4) Figure 5 a1: what is the hole at T9? There is no contusion here.
Response: Figure 5 a1is a representative image of an anti-GFAP stained horizontal spinal section showing the lesion borders of staggered hemisections (T7 and T12) as well as two BDA injection sites (T9/T10) for unilaterally labeling the thoracic dPST on the right side of the spinal cord. Therefore, the lesion area at T9, which is surrounded by a strong expression of reactive astrocytes, is the center of BDA injection sites. While we realize that BDA injections appear to induce a strong GFAP response which spreads over the T9/T10 segments. Two possibilities may cause such an occurrence. First, the puncture of the injector tip, as well as respiration and other movements from the animal during tracer injections can cause the deformation of spinal tissue, leading to additional contusion beyond the injection site. Second, for the thoracic dPST tracing, we sacrificed the mice 5-7 days post-injection. Such a time period is an acute stage when excessive reactive astrocytes are generated in response to CNS damages 14 . Together, these reasons probably explain why we observed that substantial GFAP response spread over the injection site.
(5) Figure 5 a3: here again there is no obvious colocalisation of puncta as could be defined in the image. Higher magnification images would be needed to carry on such analysis as the use of single plans and 3D images. The detail of such analysis would also need to be added to the material and methods section.
Response: This is a similar issue raised in comment #2 of reviewer #2 (see above). We agree with the reviewer that it would be more evident to identify the colocalized puncta in the images with a 3D view. Therefore, we replaced the representative images with images in a 3D view. All images used for quantification of appositions between BDA-labeled dPST fibers and CTBlabeled neurons are Z-stacks of confocal images acquired using a 60X oil-immersive objective in sequential mode to avoid crosstalk between channels. We also follow the same criteria as described before in the execution of apposition analysis. To be consistent with previous presentations, the data were expressed as appositions per MN per animal. As the reviewer suggested, we also added a detailed analysis in the materials and methods section (page 28-29).
(6) Figure 6 c: was this analysis corrected for motoneuron size? Couldn't find the information on the material and method. This is quite important to interpret the images in b which do not provide the information themselves.
Response: Thanks for this comment! The analysis in Figure 6c was not corrected for motoneuron size. We determine the propriospino-MN synaptic connections by quantification of the number of colocalized puncta stained for both BDA-labeled dPST axonal terminals, CTB-labeled MNs 6 and a presynaptic marker, synaptophysin. The synaptic numbers were determined by counting the number of triple appositions in the counting frame by defined tissue space. As we described in the previous response, we unbiasedly counted 2-4 MNs in each section and we counted 4-6 sections per each animal. The synapse numbers were finally expressed as the number of synapses per MN per section. We added a new detailed description of synapse quantification in the materials and methods section (page 29).

Reviewer #3 (Remarks to the Author)
Overall comments: It has been difficult to obtain direct evidence of specific axonal pathways involved in a return of function after spinal cord injury, with most previous reports ending up being correlational at best. The present study provides considerable tract tracing, pathway silencing, and electrophysiological data that propriospinal neurons rostral to the injury have a significant role in improved locomotor function after a lower thoracic contusion injury of adult mice, to the exclusion of corticospinal, rubrospinal serotonergic and dopaminergic axons. Part of the enhanced recovery is attributed to NT-3 mediated modulation of lumbar motoneuron circuitry and propriospinal neuron sprouting, and yet while there is observed regrowth of the motoneuron dendritic field there is no evidence that this has a role in return of function.
There are three major conclusions but the second one about identifying the important role of the propriospinal-motoneuron circuitry for recovery of motor control is the strongest and most convincingly argued by the data. The first conclusion, that spared axons are required for recovery of motor function (i.e. axons do not recover in a complete transection model) is not surprising. The third conclusion is that motoneuron dendrites atrophy after SCI but regrow if motoneurons are transfected with AAV-NT3. This is interesting but does not seem to fit in with the observation of propriospinal-motoneuron circuitry and locomotor function after SCI. At least there is no data provided to link dendrite restructuring to a functional role. It is not clear that dendritic regrowth is indispensable for locomotor recovery (line 391).
Overall the experimental design is outstanding, with precise tracing of pathways, clever use of a dual neuron silencing approach and incorporation of electrophysiological stimulation to determine whether cortical or red nucleus neuron stimulation was involved in direct stimulation of lumbar motoneurons after SCI. This is a strong body of work will be of use to the SCI regeneration-plasticity field.
Response: We thank Dr. Houle for his insightful comments on this manuscript.
A point of concern is the absence of discussion of reticulospinal and vestibulospinal pathways and their role in rodent locomotion. It is not clear that the impact of these pathways can be discounted without being tested. It is stated in Abstract and Discussion that the spared descending propriospinal pathway and not other pathways account for recovery. Also it is stated (line 289) that the propriospinal-MN circuit reorganization is functionally required for recovery, but this study does not indicate that it is sufficient for recovery. Other pathways may be involved. It would appear that both ReST and VST pathways are minimally damaged by the contusion REVIEWERS' COMMENTS:

Reviewer #1 (Remarks to the Author):
In this revised manuscript, the authors edited the paper well and have appropriately taken care of my comments and concerns. They have also responded to requests by other reviewers. I have no further comments or concerns and I continue to find the paper interesting and convincing.
Response: We truly appreciate this reviewer's constructive comments to improve the quality of our work.

Reviewer #2 (Remarks to the Author):
The paper is now strengthened and suitable for publication.

Response:
We also appreciate the reviewer`s insightful comments, particularly on the difference between cervical and thoracical propriospinal neurons.

Reviewer #3 (Remarks to the Author):
Clearly the importance of propriospinal-motoneuron circuitry for recovery is presented in this study but it is important that the question raised about possible involvement of descending pathways in locomotor recovery not examined in this study was added in the Discussion.
Something that was raised but not asked for directly was clarification about the role of dendritic sprouting related to locomotor recovery. There appears to be a correlation but not direct evidence that sprouting has a role or is essential for locomotor recovery. It is not clear that the answer to the question asked on lines 88-90 is both. I don't see the data that NT-3 mediates locomotor recovery by promoting dendritic regrowth (lines 103-105 and 391-394). Is dendritic sprouting indispensable?
Response: We appreciate Dr. Houle's further insightful comments particularly on the role of MN dendritic sprouting in locomotor recovery. We agree that addressing this issue is important to clarify major principles and mechanisms underlying NT-3-mediated locomotor recovery.
Our moderate T9 contusion model showed that lumbar MNs were not directly affected by a rostral level injury, but they underwent a profound dendritic withdrawal and synaptic stripping due to supraspinal denervation. In this SCI model, although contusions often spared some descending pathways, including the descending propriospinal tract (dPST), the contusive mice only exhibited limited ability to perform consistent locomotion, suggesting that, without recovery of lumbar MNs, it is difficult to restore locomotion with some spared descending pathways. Our T9 transection model eliminated all descending projections to the lumbar spinal cord. In this condition, retrograde transport of AAV-NT-3 into the lumbar MNs reversed their dendritic atrophy by sprouting but failed to improve locomotor recovery, as compared to the AAV-GFP group. This suggests that, in the absence of spared descending innervation, the lumbar MN dendritic sprouting alone is insufficient to restore locomotion. Together, these results indicate that therapeutic strategies aimed at modulating the plasticity of both lumbar MN and spared descending pathways may enable better functional recovery. Indeed, our results showed that retrogradely transported NT-3 to lumbar MNs enhanced propriospino-MN circuit reorganization, as reflected with an increase in sprouting of both MN dendrites and dPST terminals as well as synaptic formations of neural circuits, which leads to functional improvement after SCI.
In the context of NT-3 gene therapy via the sciatic nerve-related route following SCI, we believe MN dendritic sprouting plays an indispensable role in propriospino-MN circuit reorganization which functionally accounts for NT-3-mediated locomotor recovery. In the absence of descending innervation (eg, in the T9 transection model), MN dendritic sprouting alone is insufficient and plays a limited role in locomotor improvement.
In lines 103-104, we previously concluded that NT-3 mediates locomotor recovery via promoting dendritic regrowth rather than by preventing dendritic atrophy. This conclusion, as Dr. Houle pointed out, is less accurate because we did not compare the behavioral changes between NT-3 pre-treatment and NT-3 post-treatment. We have now used "NT-3 mediates MN recovery" instead of "NT-3 mediates locomotor recovery" in our conclusion (lines 101-102).
In lines 391-394, we summarized that NT-3 released from lumbar MNs worked as a local modulatory factor on lumbar motor circuit remodeling, and that it did not expand over to the lesion site with an effect on axon sprouting or regeneration of damaged descending pathways.
In summary, our results demonstrate that NT-3-induced MN dendritic regrowth/sprouting may not be a correlation but is indispensable for propriospino-MN circuit reorganization which is required for NT-3-mediated functional recovery.