Contrasting walking styles map to discrete neural substrates in the mouse brainstem
Curation statements for this article:-
Curated by eLife
eLife assessment
This is a valuable survey of movements and locomotor patterns produced by circuits in the medial reticular formation (MRF) of the brainstem. The authors provide solid evidence that activation of GABAergic MRF neurons slowed down walking, activation of glutamatergic neurons induced a specific "shuffle" limb trajectory, and the activation of serotonergic neurons increased locomotor speed without affecting walking signature. This study adds to the growing body of knowledge about the effects of brainstem circuits on specific aspects of locomotor function.
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (eLife)
Abstract
Walking is a slow gait which is particularly adaptable to meet internal or external needs and is prone to maladaptive alterations that lead to gait disorders. Alterations can affect speed, but also style (the way one walks). While slowed speed may signify the presence of a problem, style represents the hallmark essential for clinical classification of gait disorders. However, it has been challenging to objectively capture key stylistic features while uncovering neural substrates driving these features. Here we revealed brainstem hotspots that drive strikingly different walking styles by employing an unbiased mapping assay that combines quantitative walking signatures with focal, cell type specific activation. We found that activation of inhibitory neurons that mapped to the ventromedial caudal pons induced slow motion-like style. Activation of excitatory neurons that mapped to the ventromedial upper medulla induced shuffle-like style. Contrasting shifts in walking signatures distinguished these styles. Activation of inhibitory and excitatory neurons outside these territories or of serotonergic neurons modulated walking speed, but without walking signature shifts. Consistent with their contrasting modulatory actions, hotspots for slow-motion and shuffle-like gaits preferentially innervated different substrates. These findings lay the basis for new avenues to study mechanisms underlying (mal)adaptive walking styles and gait disorders.
Article activity feed
-
eLife assessment
This is a valuable survey of movements and locomotor patterns produced by circuits in the medial reticular formation (MRF) of the brainstem. The authors provide solid evidence that activation of GABAergic MRF neurons slowed down walking, activation of glutamatergic neurons induced a specific "shuffle" limb trajectory, and the activation of serotonergic neurons increased locomotor speed without affecting walking signature. This study adds to the growing body of knowledge about the effects of brainstem circuits on specific aspects of locomotor function.
-
Reviewer #1 (Public Review):
The medial reticular formation (MRF) in the brainstem has long been implicated in the regulation of locomotion. One common - albeit very simple - model often presents the MRF as a major relay station receiving inputs from MLR circuits, among other brain regions, that together convey locomotor signals through efferent projections targeting the caudal brainstem and the spinal cord. Yet, the MRF is a particularly large brain area whose cellular complexity is far from understood. How molecularly distinct MRF ensembles contribute to the regulation of locomotor behaviors is largely unknown. Here, the authors apply focal activation of either glutamatergic, GABAergic, or serotonergic neurons throughout the MRF using a chemogenetic gain-of-function approach to uncover the putative modulatory properties of these …
Reviewer #1 (Public Review):
The medial reticular formation (MRF) in the brainstem has long been implicated in the regulation of locomotion. One common - albeit very simple - model often presents the MRF as a major relay station receiving inputs from MLR circuits, among other brain regions, that together convey locomotor signals through efferent projections targeting the caudal brainstem and the spinal cord. Yet, the MRF is a particularly large brain area whose cellular complexity is far from understood. How molecularly distinct MRF ensembles contribute to the regulation of locomotor behaviors is largely unknown. Here, the authors apply focal activation of either glutamatergic, GABAergic, or serotonergic neurons throughout the MRF using a chemogenetic gain-of-function approach to uncover the putative modulatory properties of these neuronal ensembles during walking. Using kinematic analysis of mice limbs during self-paced over-ground walkway locomotion, the authors find that activation of GABAergic MRF neurons can selectively slow down walking, whereas activation of glutamatergic neurons can induce a specific "shuffle" limb trajectory, altogether revealing that distinct MRF populations may retain the capability to engage divergent walking signatures, whose behavioral relevance are not yet clear. In contrast, the activation of serotonergic neurons did not affect walking signatures as described for the other two subgroups but led to an increase of locomotor speed. Interestingly, MRF neurons in each regional activation "hotspots" appear to target different domains in the lumbar spinal cord, suggesting that distinct circuit mechanisms are at play for the slowmo vs shuffle effects.
Major points:
1. While the experiments are carefully done and the results are well analyzed and clearly presented in a series of beautiful figures, several aspects of the methodology remain very confusing. In particular, the initial choice for the injection coordinates is not justified and the authors don't leverage the mapping of spinal projection neurons to drive their chemogenetic screen. Similarly, the authors group very different injection schemes (unilateral or bilateral targeting of MRF neurons), that should be analyzed separately. The choice of Z score cutoff that dictates the in-depth analysis of the chemogenetic phenotypes appears arbitrary and is not grounded in a set of objective criteria.
2. One issue that arise from the work presented here is that we don't know if these MRF neurons are active during locomotion in normal, unperturbed conditions. Knowing the recruitment profile of these MRF neurons would clarify whether the chemogenetic activation boosts the firing of neurons that are already active during walking, or activate neurons that are otherwise silent. Disentangling between these possibilities may have a profound impact on the overall interpretation of the results.
3. The results should be discussed in the broader context of historic stimulation experiments, notably in cats and other species, as well as more recent circuit mapping approaches in rodents. For instance, the notion that focal stimulation of distinct area within the MRF can elicit or modify the pattern of locomotion is not really new, so is the notion that some of these modulations are phase-specific and can influence the duration of single muscle activation during stance or swing phases. This last point has for instance already been assessed through individual muscle recordings paired with MRF stimulation in cats. Perhaps better introducing these key studies and a thorough discussion of what the results presented in this manuscript bring in terms of novelty will help readers ground this work into a more comprehensive and larger body of work.
-
Reviewer #2 (Public Review):
This paper is an interesting conceptual work where certain hotspot areas were found to induce unique gait patterns. These patterns differed from a classic change in speed or gait pattern from a walk to a gallop. From this, a hypothesis was formed that these areas could be important for possible alternative walking patterns seen, for example, during pathologies such as Parkinson's disease or perhaps related to stalking behaviors.
While I liked the work and found it interesting, it remains descriptive in that the actual behaviors observed can't be causally related to a particular behavior such as stalking or shuffling. If the necessity or sufficiency of this region was related to a specific hunting behavior, for example, its interest to the field would be greater.
Nevertheless, this paper does contribute to …
Reviewer #2 (Public Review):
This paper is an interesting conceptual work where certain hotspot areas were found to induce unique gait patterns. These patterns differed from a classic change in speed or gait pattern from a walk to a gallop. From this, a hypothesis was formed that these areas could be important for possible alternative walking patterns seen, for example, during pathologies such as Parkinson's disease or perhaps related to stalking behaviors.
While I liked the work and found it interesting, it remains descriptive in that the actual behaviors observed can't be causally related to a particular behavior such as stalking or shuffling. If the necessity or sufficiency of this region was related to a specific hunting behavior, for example, its interest to the field would be greater.
Nevertheless, this paper does contribute to growing evidence that specific behaviors can be triggered by specific neuronal populations within the brainstem.
-
Author response:
Reviewer #1 (Public Review):
The medial reticular formation (MRF) in the brainstem has long been implicated in the regulation of locomotion. One common - albeit very simple - model often presents the MRF as a major relay station receiving inputs from MLR circuits, among other brain regions, that together convey locomotor signals through efferent projections targeting the caudal brainstem and the spinal cord. Yet, the MRF is a particularly large brain area whose cellular complexity is far from understood. How molecularly distinct MRF ensembles contribute to the regulation of locomotor behaviors is largely unknown. Here, the authors apply focal activation of either glutamatergic, GABAergic, or serotonergic neurons throughout the MRF using a chemogenetic gain-of-function approach to uncover the putative modulatory …
Author response:
Reviewer #1 (Public Review):
The medial reticular formation (MRF) in the brainstem has long been implicated in the regulation of locomotion. One common - albeit very simple - model often presents the MRF as a major relay station receiving inputs from MLR circuits, among other brain regions, that together convey locomotor signals through efferent projections targeting the caudal brainstem and the spinal cord. Yet, the MRF is a particularly large brain area whose cellular complexity is far from understood. How molecularly distinct MRF ensembles contribute to the regulation of locomotor behaviors is largely unknown. Here, the authors apply focal activation of either glutamatergic, GABAergic, or serotonergic neurons throughout the MRF using a chemogenetic gain-of-function approach to uncover the putative modulatory properties of these neuronal ensembles during walking. Using kinematic analysis of mice limbs during self-paced over-ground walkway locomotion, the authors find that activation of GABAergic MRF neurons can selectively slow down walking, whereas activation of glutamatergic neurons can induce a specific "shuffle" limb trajectory, altogether revealing that distinct MRF populations may retain the capability to engage divergent walking signatures, whose behavioral relevance are not yet clear. In contrast, the activation of serotonergic neurons did not affect walking signatures as described for the other two subgroups but led to an increase of locomotor speed. Interestingly, MRF neurons in each regional activation "hotspots" appear to target different domains in the lumbar spinal cord, suggesting that distinct circuit mechanisms are at play for the slowmo vs shuffle effects.
Major points:
(1) While the experiments are carefully done and the results are well analyzed and clearly presented in a series of beautiful figures, several aspects of the methodology remain very confusing.
A) In particular, the initial choice for the injection coordinates is not justified and the authors don't leverage the mapping of spinal projection neurons to drive their chemogenetic screen.
Thank you for pointing this out. To clarify this, we now start the results with an extra paragraph and accompanying figures (Figure 2 and its supplementary figures) in which we define the region of interest (ROI) within the mRF. The ROI is based upon the distribution of reticulospinal neurons in the brainstem mRF that connect directly with the lumbosacral enlargement (whether or not this ROI projects to other CNS sites), which contains the main networks important for hindlimb control during locomotion, including walking gait. Reticulospinal neurons in the mRF in the caudal pons and medulla oblongata form longitudinal columns that together occupy up to more than half of the entire brainstem. While the morphology of the medulla and caudal pons varies little from level to level, in contrast to rapid changes at the midbrain level, this doesn’t necessarily mean that the neuronal populations, even within neurotransmitter classes, are homogeneous in connectivity and function. We have now clearly denoted the rostrocaudally extensive field with its dorsoventral and mediolateral dimensions that comprises the anatomical region of interest in the new figure. While this dataset is rather basic, it allows us to directly refer back to it and clarify additional queries that came up related to the anatomy (i.e. that the hotspots for slomo- and shuffle-like gaits only cover a small portion of the reticulospinal field).
We then included detailed anatomical mapping of the spinal projections for the identified hotspots for changes in walking quality (phenomenology), the central theme of the study, and immediately adjacent regions to highlight contrasting location-connectivity-functional properties between these adjacent sites. To better incorporate these mapping results we now present it directly following the walking function based transfection site mapping, but before delving into the details of the walking gait phenotypes. We did not systematically include mapping results from all sites in the mRF ROI into this manuscript as this was beyond the scope of this already very large functional-anatomical study.
B) Similarly, the authors group very different injection schemes (unilateral or bilateral targeting of MRF neurons), that should be analyzed separately.
We now clarify early in the results section how uni- and bilateral groups were composed and what the rationale was for this. As pilot data suggested that the slomo gait style was only seen following bilateral activation in VGaT-cre mice, but not in all bilateral cases, we designed the VGaT cohort to contain mainly bilateral injections, spread across the mRF region of interest, with a smaller group of unilateral injections to verify the pilot data.
For the shuffle gait style, pilot data suggested that both uni- and bilateral activation of VGluT2 neurons could elicit this style, but only in a subset of uni- and bilateral cases. Therefore we mainly included unilateral injections in this group with a smaller bilateral cohort for verification. This approach served the main goal of the study, which was to map the walking style changes to subregions in the mRF.
However, laterality is indeed very important when it comes to locomotor control. The effects of laterality on the walking gait styles generated from the hotspots were included in supplemental figures and accompanying Tables. We have now better highlighted these in the body of the text and we have added analyses of the motor tests for uni- or bilateral groups.
Furthermore, it should be noted that the uni- and bilateral groups are heterogeneous when it comes to rostrocaudal and dorsoventral placement within the mRF ROI. As such, we were not able to rigorously compare uni- versus bilateral activation effects while at the same time separating cases out by dorsoventral and rostrocaudal location (which would be needed to do justice to the functional anatomical organization of the mRF) as we do not have sufficient power in each of the subgroups (i.e. 3 rostrocaudal levels, with each a dorsal, intermediate and ventral region to target, which each would have to be injected unilaterally and bilaterally). This was beyond the scope of this already very large study. Further studies designed to balance ipsi- and contralateral groups will be necessary to map out the hotspots for mobility phenotypes that may be driven by the mRF beyond the slomo- and shuffle-hotspots or to systematically study the impact of laterality on mobility from the mRF.
To summarize, analyses of uni- vs bilateral stimulation demonstrate that bilateral inhibition within the slomo hotspot is necessary to create the slomo walking phenotype, and that unilateral inhibition within the shuffle hotspot is sufficient to create the shuffle walking phenotype (with bilateral stimulation not enhancing the phenotype further). Unilateral activation of the slomo hotspot did not induce asymmetries in gait or a reduction in motor performance, whereas unilateral activation of the shuffle hotspot induced an asymmetry in swing time but not stride length, with laterality affecting horizontal ladder but not other motor tests. Mice with transfection sites within the mRF region of interest but outside of the slomo and shuffle hotspots did not display these walking phenotypes but did display slowed walking without qualitative changes. The connectivity to spinal and other supraspinal substrates differed between these sites, providing clues for the substrates that mediate these differential functions.
C) The choice of Z score cutoff that dictates the in-depth analysis of the chemogenetic phenotypes appears arbitrary and is not grounded in a set of objective criteria.
We are sorry that the Z score cutoff appeared arbitrary as that was not our intention.
The values to separate mice with and without a significant change were simply set at 2 standard deviations from the population mean in the control mice (i.e. Z=2). Two standard deviations from the population mean is widely used in all types of statistical analyses. We have now included the rationale for the cutoff of Z=2 in the text. Where group size allowed, to increase contrast between positive and negative groups in terms of gait characteristics, other behavioral assays and mapping, we used data from Z scores >3 (or < -3), but can assure that all moderately positive data (i.e. from mice with gait style Z scores between 2 and 3, and between -3 and -2) was reported as well in the statistical tables or supplementary figures. We have now included the links to theses supplementary tables and figures in the text, rather than only in the figure legends.
The Z scores for the different gait styles indeed appear to map to discrete sites, but the Z score cutoff was not informed by these sites or by anatomical data. Similarly, Z scores for changes in tonic muscle activity elicited by activation of inhibitory neurons also mapped to a hotspot in the same rostrocaudal column as the slomo gait style, but further caudally. This further demonstrates the strength of function-based mapping.
(2) One issue that arise from the work presented here is that we don't know if these MRF neurons are active during locomotion in normal, unperturbed conditions. Knowing the recruitment profile of these MRF neurons would clarify whether the chemogenetic activation boosts the firing of neurons that are already active during walking, or activate neurons that are otherwise silent. Disentangling between these possibilities may have a profound impact on the overall interpretation of the results.
We agree that this knowledge would improve our ability to interpret and apply the findings of the current study. It is indeed important to learn when these mRF sites are being recruited, whether part of normal modulatory strategies in order to navigate through a complex environment or as part of specialized behavioral modules or both. Another question is how loss of function in these sites impacts behavior and function. This concept has been added to the discussion and these questions can now be pursued in future experiments.
(3) The results should be discussed in the broader context of historic stimulation experiments, notably in cats and other species, as well as more recent circuit mapping approaches in rodents. For instance, the notion that focal stimulation of distinct area within the MRF can elicit or modify the pattern of locomotion is not really new, so is the notion that some of these modulations are phase-specific and can influence the duration of single muscle activation during stance or swing phases. This last point has for instance already been assessed through individual muscle recordings paired with MRF stimulation in cats. Perhaps better introducing these key studies and a thorough discussion of what the results presented in this manuscript bring in terms of novelty will help readers ground this work into a more comprehensive and larger body of work.
There is indeed a rich series of meticulous work done in cats, which included effects from stimulation of inhibitory and excitatory neurons on limb EMG, and rodent work focusing on excitatory mRF neurons. These studies show that distinct neurons or sites within the mRF drive distinct changes in motor readouts, albeit not described in terms of modulation of walking gait as we do here in terms of gait signatures. Despite this solid body of prior work, the notion of phase specificity and separate modulation of swing versus stance phase metrics has been underappreciated and therefore deserves to be emphasized. We have expanded the discussion to better highlight prior work and the interpretation of phase specificity has been enriched.
Reviewer #2 (Public Review):
This paper is an interesting conceptual work where certain hotspot areas were found to induce unique gait patterns. These patterns differed from a classic change in speed or gait pattern from a walk to a gallop. From this, a hypothesis was formed that these areas could be important for possible alternative walking patterns seen, for example, during pathologies such as Parkinson's disease or perhaps related to stalking behaviors.
While I liked the work and found it interesting, it remains descriptive in that the actual behaviors observed can't be causally related to a particular behavior such as stalking or shuffling. If the necessity or sufficiency of this region was related to a specific hunting behavior, for example, its interest to the field would be greater.
Nevertheless, this paper does contribute to growing evidence that specific behaviors can be triggered by specific neuronal populations within the brainstem.
We thank the reviewer for their thoughtful comments. We agree that more studies are necessary to understand how the slomo and shuffle hotspots serve behavioral repertoires (such as stalking or other internally driven activities) and adaptations (such as object avoidance or more subtle adjustments to terrain or internal cues). The experimental details of the present study leave ample leads for the research community to pursue these new directions.
-
