Visibly ataxic mouse mutants exhibit varying patterns of neuropathology throughout the brain (Cendelin, 2014; Fortier et al., 1987; Goldowitz et al., 1997; Lalonde and Strazielle, 2007; Lalonde and Strazielle, 2019; Mullen et al., 1976; Walter et al., 2006). Although their motor coordination deficits are generally attributed to abnormal cell patterning within the cerebellum (Arshavsky et al., 1983; Orlovsky et al., 1999), these lines have distinct patterning defects within the cerebellum, varying degrees of extracerebellar involvement, and differences in age of onset (Cendelin, 2014; Lalonde and Strazielle, 2019). The nature of the motor deficits exhibited by these mice also varies, and can often be distinguished by trained observers (Berman, 2018; Brooks and Dunnett, 2009; Hoogland et al., 2015; Lalonde and Strazielle, 2019; Schiffmann et al., 1999; Stroobants et al., 2013; Vinueza Veloz et al., 2015). However, analysis of motor coordination is often limited to low dimensional descriptions of limited specificity that fail to distinguish between related behavioral phenotypes (Brooks and Dunnett, 2009; Lalonde and Strazielle, 2019). Analysis of locomotor kinematics can provide higher dimensional readouts of locomotor behavior (Cendelín et al., 2010; Gabriel et al., 2009; Zörner et al., 2010), but can still suffer from a lack of specificity due to an abundance of highly correlated measures that ultimately reflect non-specific features such as changes in walking speed or body size (Batka et al., 2014; Cendelín et al., 2010; Machado et al., 2015). A quantitative understanding of the specific nature of gait ataxia in mutants with well-described abnormalities in circuit architecture could provide important clues into neural mechanisms of motor coordination (Anderson and Perona, 2014; Bastian et al., 1996; Berman, 2018; Brown and de Bivort, 2018; Darmohray et al., 2019; Datta et al., 2019; Kiehn, 2016; Morton and Bastian, 2007; Powell et al., 2015; Sarnaik and Raman, 2018; Udo et al., 1980).

We previously used the LocoMouse system (Machado et al., 2015) to analyze the locomotor coordination of mildly ataxic Purkinje cell degeneration (pcd) mice, in which neural degeneration, particularly early in postnatal development, is largely restricted to cerebellar Purkinje cells, effectively disconnecting the output of the cerebellar cortex (Chen et al., 1996; Fernandez-Gonzalez et al., 2002; Le Marec and Lalonde, 1997). We found that locomotor deficits in pcd were restricted to specific aspects of multijoint, interlimb, and whole-body coordination, while the forward trajectories of individual paws were spared (Machado et al., 2015). We further found that the tail movements of pcd mice reflected the passive consequences of limb movement (Machado et al., 2015). However, it remained unclear to what extent these features represented fundamental features of cerebellar ataxia, or were specific to pcd mice.

Reeler mice are a classic ataxic mutant (Cendelin, 2014; Curran and D'Arcangelo, 1998; D'Arcangelo et al., 1999; D'Arcangelo et al., 1995; Falconer, 1951) with an autosomal recessive mutation in the reelin gene, which is important for neural cell migration (Beckers et al., 1994; Hack et al., 2002). Its loss causes several defects, in particular aberrant localization of neurons and failure of neuronal layer formation. Several brain regions are affected, including cerebellum (Hamburgh, 1963; Terashima et al., 1983), hippocampus (Stanfield et al., 1979), neocortex (Mikoshiba et al., 1980), inferior olive (Blatt and Eisenman, 1988) and substantia nigra (Kang et al., 2010). Neuropathology in these mice is particularly striking within the cerebellum, where severe irregularities in cellular localization are also associated with corresponding aberrant synaptic connectivity between cell types, abnormal foliation, and hypoplasia. Although their locomotor kinematics and whole-body coordination have not been reported, homozygous reeler mutants have been described as having a severely ataxic, ‘reeling’ gait, with difficulties in maintaining their hindquarters upright (Cendelin, 2014; Lalonde et al., 2004; Lalonde and Strazielle, 2019). Like most ataxic mutants, reelers also exhibit poor performance in rotarod, stationary beam and water maze tests (Lalonde et al., 2004).

Thus, pcd and reeler mice share grossly abnormal cerebellar circuitry, but exhibit marked differences in synaptic connectivity within the cerebellum and across the brain. We wondered whether these similarities and differences on the anatomical level might be associated with similarly shared and distinct features of motor behavior. Here we analyze the locomotor behavior of reeler mutants and compare it quantitatively to that of the more mildly ataxic pcd mice (Machado et al., 2015). Detailed comparison of locomotor kinematics and linear discriminant analysis reveals both shared and distinct features of gait ataxia in these two mutants. This approach provides a quantitative foundation for mapping specific locomotor impairments onto distinct neuropathologies.

When assessing motor coordination deficits in mice there is often a tradeoff between specificity and interpretability. For example, approaches such as rotarod testing provide measurable and intuitive low dimensional outputs, but lack specificity. Others, such as the CatWalk system (Gabriel et al., 2009) can provide many detailed measurements of locomotor behavior, but their meaning is not always readily apparent. With LocoMouse we have tried to provide both a comprehensive, quantitative description of locomotor behavior as well as a conceptual framework within which to interpret that high-dimensional data.

Here we analyzed the locomotor behavior of severely ataxic reeler mice and quantitatively compared it with that of Purkinje cell degeneration mutants (Machado et al., 2015). Detailed comparison of locomotor kinematics and linear discriminant analysis revealed both shared and distinct features of gait ataxia in the two mouse lines. Although the generality of the conclusions that can be reached from the analysis of two mutants is of course limited, the approach described here provides a comprehensive and quantitative way to map complex patterns of locomotor features onto partially overlapping neuropathologies. Extending it to additional models and manipulations could enable association of specific movement features with increasingly precise alterations in underlying neural circuits.

Our first finding lies in capturing specific quantitative differences in gait ataxia between pcd and reeler mice. Rather than simple differences in the level of severity in the same set of affected features, the visible gait differences in the two mutants (captured by the second linear discriminant) appear to stem specifically from an increase in movement variability and additional hindlimb involvement (Cendelin, 2014) in reelers. As a likely consequence of those deficits, analysis of support patterns and nose and tail movements suggests that reelers, unlike controls and pcd, use their front paws as their main supports while walking. This difference in support patterns fully accounts for differences in tail movements observed in the two mutants (Figure 4). It could also explain the inability of reelers to walk in a straight line (Video 1, Figure 4H), likely because the front limbs are unable to provide both support and steering control to keep the body moving forward. Thus, while all of these features in combination contribute to the ability of human observers to visually distinguish the overall differences in walking patterns in the two mice, the quantitative LocoMouse analysis is able to distill them down to reveal fundamental underlying alterations in variability and hindlimb control.

Capturing the essential differences in locomotor control between ataxic phenotypes provides a necessary starting point for understanding the contributions of individual circuit elements to this complex, whole-body behavior. In pcd, the main anatomical phenotype is cerebellar, with striking degeneration of Purkinje cells that effectively removes cerebellar cortical input to the cerebellar nuclei. In contrast, reeler mice exhibit aberrant cell localization throughout the brain (Terashima et al., 1983; Stanfield et al., 1979; Mikoshiba et al., 1980; Blatt and Eisenman, 1988; Kang et al., 2010), which could also contribute to their gait abnormalities. Moreover, in reeler, abnormal cell migration during development results in aberrant circuit wiring that alters, rather than removes, Purkinje cell activity (Dupont et al., 1983; Curran and D'Arcangelo, 1998; D'Arcangelo et al., 1995; Lalonde and Strazielle, 2019).

Locomotor differences between the two mutants (captured by LD2 in Figure 5) thus likely represent a combination of specific differences in circuitry within the cerebellum itself plus extracerebellar effects. The differences in variability may be of cerebellar origin (Walter et al., 2006; Medina and Lisberger, 2007). Movement variability is often considered to be a hallmark of cerebellar ataxia, and yet surprisingly, pcd mice were less variable than controls on several movement measures (Machado et al., 2015). We speculate that the opposite effects on variability in the two mutants may arise as a consequence of altered Purkinje cell output in reeler, compared to a lack of Purkinje cell output in pcd (Medina and Lisberger, 2007). Meanwhile, differences in hindlimb involvement, and the resulting reliance on front paw supports (Figure 3G,H), may be extracerebellar, perhaps resulting from cell positioning defects in reeler spinal cord (Yip et al., 2000; Phelps et al., 2002). Future experiments analyzing locomotor phenotypes in additional cerebellar and extracerebellar models will be crucial for establishing mechanistic links between specific features of locomotor behavior and precise alterations in underlying neural circuits.

Despite the more severe behavioral and anatomical phenotype in reeler, we found the overall pattern of affected locomotor features in pcd and reeler to be surprisingly similar. The first linear discriminant pulled out a set of shared impairments in multi-joint, interlimb, and whole-body coordination that may reflect core quantitative features of mouse cerebellar ataxia. We propose that these features comprise the gestalt impression of clearly cerebellar gait phenotypes that trained observers can readily identify, despite substantial variation in specific manifestations across models.

In particular, like pcd (Machado et al., 2015), the tail and nose movements of reelers were also successfully modeled as passive consequences of limb movement, with a shift of supports to the front, rather than hind, paws. We had previously interpreted the pattern of coordination deficits in pcd mice, and particularly the passive tail oscillation, as consistent with the lack of an internal forward model that predicts and compensates for the consequences of movement (Bastian et al., 1996; Ebner and Pasalar, 2008; Ito, 2008; Kennedy et al., 2014; Wolpert et al., 1998). Specifically for locomotor control, a forward model can predict how the movement of one part of the body (e.g., a limb or a paw) will affect the movement of another part (e.g., another limb, or the tail), and inject a compensatory control signal to eliminate those consequences.

The idea that the cerebellum could provide a forward model for motor control is often invoked when considering the role of the cerebellum in motor learning (Ito, 2008; Shadmehr and Krakauer, 2008; Wolpert et al., 1998). Recently, we showed that both pcd and reeler mice were unable to learn to restore gait symmetry on a locomotor learning task that requires predictive control (Darmohray et al., 2019; Morton and Bastian, 2006; Reisman et al., 2005). That finding, together with the passive tail oscillations of both mutants, reinforces the idea that these animals lack the ability, provided by an intact cerebellum, to predict and compensate unintended movements during locomotion. The consistent lack of compensatory predictive mechanisms across mutants and behavioral paradigms suggests that they may represent core features of cerebellar contributions to coordinated locomotion. Extending the current findings and applying a similar approach across a broader range of circuit manipulations could become a key element in understanding how various neural components work together to control complex, whole-body behaviors.

Source data have been provided for all Figures.

1. 1. Anderson DJ
   2. Perona P

   (2014) Toward a science of computational ethology

   Neuron 84:18–31.

   https://doi.org/10.1016/j.neuron.2014.09.005

   • PubMed
   • Google Scholar
2. 1. Arshavsky YI
   2. Gelfand IM
   3. Orlovsky GN

   (1983) The cerebellum and control of rhythmical movements

   Trends in Neurosciences 6:417–422.

   https://doi.org/10.1016/0166-2236(83)90191-1

   • Google Scholar
3. 1. Bastian AJ
   2. Martin TA
   3. Keating JG
   4. Thach WT

   (1996) Cerebellar ataxia: abnormal control of interaction torques across multiple joints

   Journal of Neurophysiology 76:492–509.

   https://doi.org/10.1152/jn.1996.76.1.492

   • PubMed
   • Google Scholar
4. Software

   1. Bates D
   2. Maechler M
   3. Bolker B
   4. Walker S

   (2013)

   lme Linear mixed-effects models using Eigen and S4, version 1.0-5

   R Package Version.
5. 1. Batka RJ
   2. Brown TJ
   3. Mcmillan KP
   4. Meadows RM
   5. Jones KJ
   6. Haulcomb MM

   (2014) The need for speed in rodent locomotion analyses

   The Anatomical Record 297:1839–1864.

   https://doi.org/10.1002/ar.22955

   • Google Scholar
6. 1. Beckers MC
   2. Bar I
   3. Huynh-Thu T
   4. Dernoncourt C
   5. Brunialti AL
   6. Montagutelli X
   7. Guénet JL
   8. Goffinet AM

   (1994) A high-resolution genetic map of mouse chromosome 5 encompassing the reeler (rl) locus

   Genomics 23:685–690.

   https://doi.org/10.1006/geno.1994.1557

   • PubMed
   • Google Scholar
7. 1. Bellardita C
   2. Kiehn O

   (2015) Phenotypic characterization of speed-associated gait changes in mice reveals modular organization of locomotor networks

   Current Biology 25:1426–1436.

   https://doi.org/10.1016/j.cub.2015.04.005

   • PubMed
   • Google Scholar
8. 1. Berman GJ

   (2018) Measuring behavior across scales

   BMC Biology 16:23.

   https://doi.org/10.1186/s12915-018-0494-7

   • PubMed
   • Google Scholar
9. 1. Blatt GJ
   2. Eisenman LM

   (1988) Topographic and zonal organization of the olivocerebellar projection in the Reeler mutant mouse

   The Journal of Comparative Neurology 267:603–615.

   https://doi.org/10.1002/cne.902670412

   • PubMed
   • Google Scholar
10. 1. Brooks SP
    2. Dunnett SB

    (2009) Tests to assess motor phenotype in mice: a user's guide

    Nature Reviews Neuroscience 10:519–529.

    https://doi.org/10.1038/nrn2652

    • PubMed
    • Google Scholar
11. 1. Brown AEX
    2. de Bivort B

    (2018) Ethology as a physical science

    Nature Physics 14:653–657.

    https://doi.org/10.1038/s41567-018-0093-0

    • Google Scholar
12. 1. Cendelín J
    2. Voller J
    3. Vozeh F

    (2010) Ataxic gait analysis in a mouse model of the olivocerebellar degeneration

    Behavioural Brain Research 210:8–15.

    https://doi.org/10.1016/j.bbr.2010.01.035

    • PubMed
    • Google Scholar
13. 1. Cendelin J

    (2014) From mice to men: lessons from mutant ataxic mice

    Cerebellum & Ataxias 1:4.

    https://doi.org/10.1186/2053-8871-1-4

    • PubMed
    • Google Scholar
14. 1. Chen L
    2. Bao S
    3. Lockard JM
    4. Kim JK
    5. Thompson RF

    (1996) Impaired classical eyeblink conditioning in cerebellar-lesioned and purkinje cell degeneration (pcd) mutant mice

    The Journal of Neuroscience 16:2829–2838.

    https://doi.org/10.1523/JNEUROSCI.16-08-02829.1996

    • Google Scholar
15. 1. Curran T
    2. D'Arcangelo G

    (1998) Role of reelin in the control of brain development

    Brain Research. Brain Research Reviews 26:285–294.

    https://doi.org/10.1016/s0165-0173(97)00035-0

    • PubMed
    • Google Scholar
16. 1. D'Arcangelo G
    2. Miao GG
    3. Chen SC
    4. Soares HD
    5. Morgan JI
    6. Curran T

    (1995) A protein related to extracellular matrix proteins deleted in the mouse mutant reeler

    Nature 374:719–723.

    https://doi.org/10.1038/374719a0

    • PubMed
    • Google Scholar
17. 1. D'Arcangelo G
    2. Homayouni R
    3. Keshvara L
    4. Rice DS
    5. Sheldon M
    6. Curran T

    (1999) Reelin is a ligand for lipoprotein receptors

    Neuron 24:471–479.

    https://doi.org/10.1016/S0896-6273(00)80860-0

    • PubMed
    • Google Scholar
18. 1. Darmohray DM
    2. Jacobs JR
    3. Marques HG
    4. Carey MR

    (2019) Spatial and temporal locomotor learning in mouse cerebellum

    Neuron 102:217–231.

    https://doi.org/10.1016/j.neuron.2019.01.038

    • PubMed
    • Google Scholar
19. 1. Datta SR
    2. Anderson DJ
    3. Branson K
    4. Perona P
    5. Leifer A

    (2019) Computational neuroethology: a call to action

    Neuron 104:11–24.

    https://doi.org/10.1016/j.neuron.2019.09.038

    • PubMed
    • Google Scholar
20. 1. Dupont JL
    2. Gardette R
    3. Crepel F

    (1983) Bioelectrical properties of cerebellar purkinje cells in Reeler mutant mice

    Brain Research 274:350–353.

    https://doi.org/10.1016/0006-8993(83)90717-5

    • PubMed
    • Google Scholar
21. 1. Ebner TJ
    2. Pasalar S

    (2008) Cerebellum predicts the future motor state

    The Cerebellum 7:583–588.

    https://doi.org/10.1007/s12311-008-0059-3

    • PubMed
    • Google Scholar
22. 1. Falconer DS

    (1951) Two new mutants, 'trembler' and 'reeler', with neurological actions in the house mouse (Mus musculus L.)

    Journal of Genetics 50:192–205.

    https://doi.org/10.1007/BF02996215

    • PubMed
    • Google Scholar
23. 1. Fernandez-Gonzalez A
    2. La Spada AR
    3. Treadaway J
    4. Higdon JC
    5. Harris BS
    6. Sidman RL
    7. Morgan JI
    8. Zuo J

    (2002) Purkinje cell degeneration (pcd) phenotypes caused by mutations in the axotomy-induced gene, Nna1

    Science 295:1904–1906.

    https://doi.org/10.1126/science.1068912

    • PubMed
    • Google Scholar
24. 1. Fisher RA

    (1936) The use of multiple measurements in taxonomic problems

    Annals of Eugenics 7:179–188.

    https://doi.org/10.1111/j.1469-1809.1936.tb02137.x

    • Google Scholar
25. 1. Fortier PA
    2. Smith AM
    3. Rossignol S

    (1987) Locomotor deficits in the mutant mouse, lurcher

    Experimental Brain Research 66:271–286.

    https://doi.org/10.1007/BF00243304

    • PubMed
    • Google Scholar
26. 1. Gabriel AF
    2. Marcus MA
    3. Walenkamp GH
    4. Joosten EA

    (2009) The CatWalk method: assessment of mechanical allodynia in experimental chronic pain

    Behavioural Brain Research 198:477–480.

    https://doi.org/10.1016/j.bbr.2008.12.018

    • PubMed
    • Google Scholar
27. 1. Goldowitz D
    2. Cushing RC
    3. Laywell E
    4. D'Arcangelo G
    5. Sheldon M
    6. Sweet HO
    7. Davisson M
    8. Steindler D
    9. Curran T

    (1997) Cerebellar disorganization characteristic of Reeler in scrambler mutant mice despite presence of reelin

    The Journal of Neuroscience 17:8767–8777.

    https://doi.org/10.1523/JNEUROSCI.17-22-08767.1997

    • PubMed
    • Google Scholar
28. 1. Gorska T
    2. Zmysłowski W
    3. Majczyński H

    (1998)

    Overground locomotion in intact rats: interlimb coordination, support patterns and support phases duration

    Acta Neurobiologiae Experimentalis 59:131–144.

    • Google Scholar
29. 1. Hack I
    2. Bancila M
    3. Loulier K
    4. Carroll P
    5. Cremer H

    (2002) Reelin is a detachment signal in tangential chain-migration during postnatal neurogenesis

    Nature Neuroscience 5:939–945.

    https://doi.org/10.1038/nn923

    • PubMed
    • Google Scholar
30. 1. Hamburgh M

    (1963) Analysis of the postnatal developmental effects of “reeler,” a neurological mutation in mice A study in developmental genetics

    Developmental Biology 8:165–185.

    https://doi.org/10.1016/0012-1606(63)90040-X

    • Google Scholar
31. 1. Hoogland TM
    2. De Gruijl JR
    3. Witter L
    4. Canto CB
    5. De Zeeuw CI

    (2015) Role of synchronous activation of cerebellar purkinje cell ensembles in Multi-joint movement control

    Current Biology 25:1157–1165.

    https://doi.org/10.1016/j.cub.2015.03.009

    • PubMed
    • Google Scholar
32. 1. Ito M

    (2008) Control of mental activities by internal models in the cerebellum

    Nature Reviews Neuroscience 9:304–313.

    https://doi.org/10.1038/nrn2332

    • PubMed
    • Google Scholar
33. Book

    1. James G
    2. Witten D
    3. Hastie T
    4. Tibshirani R

    (2013)

    An Introduction to Statistical Learning: With Applications in R

    Springer.

    • Google Scholar
34. 1. Kang WY
    2. Kim SS
    3. Cho SK
    4. Kim S
    5. Suh-Kim H
    6. Lee YD

    (2010) Migratory defect of mesencephalic dopaminergic neurons in developing reeler mice

    Anatomy & Cell Biology 43:241.

    https://doi.org/10.5115/acb.2010.43.3.241

    • PubMed
    • Google Scholar
35. 1. Kennedy A
    2. Wayne G
    3. Kaifosh P
    4. Alviña K
    5. Abbott LF
    6. Sawtell NB

    (2014) A temporal basis for predicting the sensory consequences of motor commands in an electric fish

    Nature Neuroscience 17:416–422.

    https://doi.org/10.1038/nn.3650

    • PubMed
    • Google Scholar
36. 1. Kiehn O

    (2016) Decoding the organization of spinal circuits that control locomotion

    Nature Reviews Neuroscience 17:224–238.

    https://doi.org/10.1038/nrn.2016.9

    • PubMed
    • Google Scholar
37. 1. Lalonde R
    2. Hayzoun K
    3. Derer M
    4. Mariani J
    5. Strazielle C

    (2004) Neurobehavioral evaluation of Relnrl-orl mutant mice and correlations with cytochrome oxidase activity

    Neuroscience Research 49:297–305.

    https://doi.org/10.1016/j.neures.2004.03.012

    • Google Scholar
38. 1. Lalonde R
    2. Strazielle C

    (2007) Spontaneous and induced mouse mutations with cerebellar dysfunctions: behavior and neurochemistry

    Brain Research 1140:51–74.

    https://doi.org/10.1016/j.brainres.2006.01.031

    • PubMed
    • Google Scholar
39. 1. Lalonde R
    2. Strazielle C

    (2019) Motor performances of spontaneous and genetically modified mutants with cerebellar atrophy

    The Cerebellum 18:615–634.

    https://doi.org/10.1007/s12311-019-01017-5

    • Google Scholar
40. 1. Le Marec N
    2. Lalonde R

    (1997) Sensorimotor learning and retention during equilibrium tests in purkinje cell degeneration mutant mice

    Brain Research 768:310–316.

    https://doi.org/10.1016/S0006-8993(97)00666-5

    • PubMed
    • Google Scholar
41. 1. Machado AS
    2. Darmohray DM
    3. Fayad J
    4. Marques HG
    5. Carey MR

    (2015) A quantitative framework for whole-body coordination reveals specific deficits in freely walking ataxic mice

    eLife 4:e07892.

    https://doi.org/10.7554/eLife.07892

    • PubMed
    • Google Scholar
42. 1. Medina JF
    2. Lisberger SG

    (2007) Variation, signal, and noise in cerebellar sensory-motor processing for smooth-pursuit eye movements

    Journal of Neuroscience 27:6832–6842.

    https://doi.org/10.1523/JNEUROSCI.1323-07.2007

    • PubMed
    • Google Scholar
43. 1. Mikoshiba K
    2. Kohsaka S
    3. Takamatsu K
    4. Aoki E
    5. Tsukada Y

    (1980) Morphological and biochemical studies on the cerebral cortex from Reeler mutant mice: development of cortical layers and metabolic mapping by the deoxyglucose method

    Journal of Neurochemistry 34:835–844.

    https://doi.org/10.1111/j.1471-4159.1980.tb09655.x

    • PubMed
    • Google Scholar
44. 1. Morton SM
    2. Bastian AJ

    (2006) Cerebellar contributions to locomotor adaptations during splitbelt treadmill walking

    Journal of Neuroscience 26:9107–9116.

    https://doi.org/10.1523/JNEUROSCI.2622-06.2006

    • PubMed
    • Google Scholar
45. 1. Morton SM
    2. Bastian AJ

    (2007) Mechanisms of cerebellar gait ataxia

    The Cerebellum 6:79–86.

    https://doi.org/10.1080/14734220601187741

    • PubMed
    • Google Scholar
46. 1. Mullen RJ
    2. Eicher EM
    3. Sidman RL

    (1976) Purkinje cell degeneration, a new neurological mutation in the mouse

    PNAS 73:208–212.

    https://doi.org/10.1073/pnas.73.1.208

    • PubMed
    • Google Scholar
47. Book

    1. Orlovsky GN
    2. Deliagina TG
    3. Grillner S
    4. Orlovskii GN

    (1999)

    Neuronal Control of Locomotion: From Mollusc to Man

    Oxford University Press.

    • Google Scholar
48. 1. Phelps PE
    2. Rich R
    3. Dupuy-Davies S
    4. Ríos Y
    5. Wong T

    (2002) Evidence for a cell-specific action of reelin in the spinal cord

    Developmental Biology 244:180–198.

    https://doi.org/10.1006/dbio.2002.0580

    • PubMed
    • Google Scholar
49. 1. Powell K
    2. Mathy A
    3. Duguid I
    4. Häusser M

    (2015) Synaptic representation of locomotion in single cerebellar granule cells

    eLife 4:e07290.

    https://doi.org/10.7554/eLife.07290

    • Google Scholar
50. 1. Reisman DS
    2. Block HJ
    3. Bastian AJ

    (2005) Interlimb coordination during locomotion: what can be adapted and stored?

    Journal of Neurophysiology 94:2403–2415.

    https://doi.org/10.1152/jn.00089.2005

    • PubMed
    • Google Scholar
51. 1. Sarnaik R
    2. Raman IM

    (2018) Control of voluntary and optogenetically perturbed locomotion by spike rate and timing of neurons of the mouse cerebellar nuclei

    eLife 7:e29546.

    https://doi.org/10.7554/eLife.29546

    • PubMed
    • Google Scholar
52. 1. Schiffmann SN
    2. Cheron G
    3. Lohof A
    4. d'Alcantara P
    5. Meyer M
    6. Parmentier M
    7. Schurmans S

    (1999) Impaired motor coordination and purkinje cell excitability in mice lacking calretinin

    PNAS 96:5257–5262.

    https://doi.org/10.1073/pnas.96.9.5257

    • PubMed
    • Google Scholar
53. 1. Shadmehr R
    2. Krakauer JW

    (2008) A computational neuroanatomy for motor control

    Experimental Brain Research 185:359–381.

    https://doi.org/10.1007/s00221-008-1280-5

    • PubMed
    • Google Scholar
54. 1. Stanfield BB
    2. Caviness VS
    3. Cowan WM

    (1979) The organization of certain afferents to the Hippocampus and dentate gyrus in normal and reeler mice

    The Journal of Comparative Neurology 185:461–483.

    https://doi.org/10.1002/cne.901850304

    • PubMed
    • Google Scholar
55. 1. Stroobants S
    2. Gantois I
    3. Pooters T
    4. D'Hooge R

    (2013) Increased gait variability in mice with small cerebellar cortex lesions and normal rotarod performance

    Behavioural Brain Research 241:32–37.

    https://doi.org/10.1016/j.bbr.2012.11.034

    • PubMed
    • Google Scholar
56. 1. Terashima T
    2. Inoue K
    3. Inoue Y
    4. Mikoshiba K
    5. Tsukada Y

    (1983) Distribution and morphology of corticospinal tract neurons in reeler mouse cortex by the retrograde HRP method

    The Journal of Comparative Neurology 218:314–326.

    https://doi.org/10.1002/cne.902180307

    • PubMed
    • Google Scholar
57. 1. Udo M
    2. Matsukawa K
    3. Kamei H
    4. Oda Y

    (1980) Cerebellar control of locomotion: effects of cooling cerebellar intermediate cortex in high decerebrate and awake walking cats

    Journal of Neurophysiology 44:119–134.

    https://doi.org/10.1152/jn.1980.44.1.119

    • PubMed
    • Google Scholar
58. 1. Vinueza Veloz MF
    2. Zhou K
    3. Bosman LW
    4. Potters JW
    5. Negrello M
    6. Seepers RM
    7. Strydis C
    8. Koekkoek SK
    9. De Zeeuw CI

    (2015) Cerebellar control of gait and interlimb coordination

    Brain Structure and Function 220:3513–3536.

    https://doi.org/10.1007/s00429-014-0870-1

    • PubMed
    • Google Scholar
59. 1. Walter JT
    2. Alviña K
    3. Womack MD
    4. Chevez C
    5. Khodakhah K

    (2006) Decreases in the precision of purkinje cell pacemaking cause cerebellar dysfunction and ataxia

    Nature Neuroscience 9:389–397.

    https://doi.org/10.1038/nn1648

    • PubMed
    • Google Scholar
60. 1. Wolpert DM
    2. Miall RC
    3. Kawato M

    (1998) Internal models in the cerebellum

    Trends in Cognitive Sciences 2:338–347.

    https://doi.org/10.1016/S1364-6613(98)01221-2

    • PubMed
    • Google Scholar
61. 1. Yip JW
    2. Yip YP
    3. Nakajima K
    4. Capriotti C

    (2000) Reelin controls position of autonomic neurons in the spinal cord

    PNAS 97:8612–8616.

    https://doi.org/10.1073/pnas.150040497

    • PubMed
    • Google Scholar
62. 1. Zörner B
    2. Filli L
    3. Starkey ML
    4. Gonzenbach R
    5. Kasper H
    6. Röthlisberger M
    7. Bolliger M
    8. Schwab ME

    (2010) Profiling locomotor recovery: comprehensive quantification of impairments after CNS damage in rodents

    Nature Methods 7:701–708.

    https://doi.org/10.1038/nmeth.1484

    • PubMed
    • Google Scholar

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Shared and specific signatures of locomotor ataxia in mutant mice" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

The work seems like a useful extension of Machado et al., 2015, from a benchmarking perspective – this paper helps to show that LocoMouse "works" at least in the sense of being able to identify commonalities and differences between mutants and to make semantic sense of those similarities and differences. However, after extensive consultation the consensus was reached that the paper does not contain enough new and impactful science for an advance on Machado et al., 2015, especially given reeler data in Darmohray et al., 2019, and the methodology used is that previously used by the authors, albeit in a different context and with a different mutant. The attempt to tie the behavioral defects quantified to circuit level defects were not convincing.

Reviewer #1:

This manuscript is a very well written manuscript describing the differences and similarities in the ataxic leg movements during walking in two mutant mouse models that have cerebellar neuropathy. Namely these mice are the Purkinje Cell Generation mice and the Reeler mice. They did a nice set of recordings and data analysis. However, this manuscript is purely descriptive, without any conceptual conclusion. I don't think this work, although well done, is not appropriate for eLife.

Reviewer #2:

We are still at the proof-of-concept phase for computational neuroethology. One goal of these methods is to identify relationships between complex patterns of behavior by – in a task or context specific manner – balancing (as the discussion says) specificity (e.g., high dimensional measures) with interpretability (e.g., a conceptual framework). This paper puts this hypothesis to the test using the LocoMouse framework developed by this lab to compare ataxic phenotypes among pcd and reeler mice. Here the authors do a very nice job of demonstrating that – like the pcd mice – many of the apparent phenotypes in the reeler mouse are a simple consequence of the slowed speed and smaller size of the mutants. In addition, the observations that phenotypes are now largely found in the front paws – and that tail and nose dynamics can be predicted from this effect – effectively distinguishes the two mutants. The data are clear, and the findings of significant import and utility to those interested in ataxias and in motor control.

Given that the technology and analytics associated with LocoMouse are at this point well vetted, I have no specific comments or concerns about the data in the paper. I would like only a few small adjustments. It would be nice to move some of the data from Figure 1—figure supplement 1 into main Figure 1 (or maybe to make Figure 1—figure supplement 1 a new Figure 2) so that the baseline comparisons between the mutants is more front stage in the figures. In addition, it would be clarifying to generate some simplified graphs (maybe focused on specific speeds) that, in at least some cases, enable comparisons of reeler, pcd and controls all at once. The compromises used in Figure 3 could work here (where you have adjacent panels with two of the three comparisons), but on the whole it would be generally nice to see more direct comparisons across the data in the main figures. Finally, I am left with a sense of the power of the approach to both relate and distinguish, but the one thing that isn't made completely explicit is why reeler mice are so much more apparently ataxic when you just look at them. Is this largely the collective consequence of the "high-stepping" phenotype, the increased variability and the steering phenotype? There is obviously not a simple answer to this, but I would appreciate some discussion of what might account for the fact that the reeler mice look worse by eye, and how quantitative measures might highlight or obscure the gestalt we have as researchers inspecting our mice.

Reviewer #3:

The authors use established methods from their lab to compare locomotor kinematics in wildtype mice and two mutant mouse lines: pcd and reeler. This work builds on some of their impressive previous work where they already showed the results for the pcd mice (Machado et al., 2015), as well as evaluations of both pcd and reeler mice in a different paradigm testing locomotor kinematics, the split-belt treadmill (Darmohray et al., 2019). With the current manuscript, the authors aim to provide a quantitative foundation for mapping specific locomotor impairments onto distinct neuropathologies. The experiments seem to be competently performed, and the manuscript is well written. However, the limitation of the current manuscript is that it does not provide much new insight to the analysis of body part kinematics which hasn't already largely been covered in their previous work.

The main concerns are:

1) Lack of novelty compared to their previous papers (Machado et al., 2015; Darmohray et al., 2019). Basic locomotor kinematics of both pcd and reeler mice (as shown here in Figures 1-3) are already described in these papers. Figure 4 does contain new information, although it largely depends on the results from Figure 2. Similar observations to those in Figure 2 of the current manuscript, e.g. different paw supports for the two mutant lines compared to controls, have already been shown in Figure 3D and H of Darmohray et al., 2019, albeit in a different behavioral paradigm.

2) The authors claim to provide a quantitative foundation for mapping specific locomotor impairments onto distinct neuropathologies. However, there is not enough biological validation for the claim that the data presented in the current manuscript can link behavioral phenotypes with underlying neural circuits, as there is no detailed description of the anatomical differences between the two mutant lines within the cerebellum. Certainly the found behavioral differences between the two are not linked to their neural circuits. The authors would need to revise their conclusions based on the data they present here.

3) An attempt to possibly rescue the phenotype would help to improve the biological understanding of which behavioral phenotypes are linked to specifically which anatomical underlying structures. I.e., is it possible to intervene in one of the animal models, so that their phenotype switches on the LD1 axis presented in Figure 4 to the side of the control mice? Such conclusions would then truly link behavioral phenotypes to neural circuits, and would help to significantly improve the biological understanding of mouse locomotor kinematics.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Shared and specific signatures of locomotor ataxia in mutant mice" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Ronald Calabrese as the Senior Editor and Reviewing Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

In this manuscript, Machado and colleagues use LocoMouse, their behavioural tracking software package published previously as part of a study in eLife (Machado et al., 2015), to reveal similarities and differences in the locomotor kinematic deficits of two mouse models of ataxia, specifically pcd and reeler mice. A behavioural description of locomotion in pcd mice was already included in Machado et al., 2015, so the additional data provided here are for reeler mice. The analyses provided reveal that some locomotor parameters are similarly impaired in these two mutant models while other parameters are unique to each mutant.

Revisions for this paper:

The reviewers found considerable merit in this advance but have further revisions they recommend that can be accomplished quickly and evaluated by the RE. Two of the reviewers thought that further experiments were needed to sustain the authors claims but realizing that this advance was considerable recommend revision in consultation.

Reviewer 3:

"The analyses provided reveal that some locomotor parameters are similarly impaired in these two mutant models while other parameters are unique to each mutant. While the additional findings presented in this study are sound, it is unclear how much we can generalize from analysis of one additional ataxia model, particularly because the neuropathology in these two models is quite different. Without a deeper evaluation of the generality and specificity of locomotor deficits in mouse models of ataxia, this study only provides a relatively slim additional advance over their previous study, and the more general conclusions they draw are not sufficiently justified. I have the following suggestions for strengthening the present manuscript. The authors show nicely in Figure 5 that combinations of behavioral features can be used to distinguish not only wild type from mutant mice but also pcd and reeler mice from each other. The authors' interpretation of these results is that the locomotor features that commonly distinguish wild type animals from the two mutants have a common underlying neuropathological origin (i.e. a cerebellar deficit) while the features that distinguish pcd mice from reeler mice reflect differences between these two mutants (i.e. absence of Purkinje cell output vs. miswiring in the cerebellum plus the additional extracerebellar defects in reeler mice). This is an important idea that could be bolstered with some additional data/analysis: 1. To show that aberrant cerebellar function is the root cause of the variability along LD1, the authors could use perturbation experiments to specifically affect locomotion by manipulating activity in interpositus-projecting Purkinje cells. This could be done acutely (as in Sarnaik and Raman, 2018) and/or chronically (as in Darmohray et al., 2018). 2. To test if LD1 does indeed distinguish wild type from ataxic mice and LD2 (or higher dimensions) distinguish different types of ataxic mice from each other, the authors could analyze locomotion in other models of ataxia that have different neuropathologies."

Reviewer 2:

"On a related note, a major claim is that the LD analysis defines a core set of behavioral features associated with cerebellar dysfunction. These are: impairments in multijoint, interlimb, and whole-body coordination, as well as in predictive mechanisms (i.e. forward models). It is true that these behavioral impairments are observed in both pcd and reeler mice, which share cerebellar neuropathology. However, it is possible that pathology in other brain circuits (basal ganglia, for example) may also be associated with some of the same behavioral impairments observed in the pcd and reeler mice, which would indicate that these behavioral impairments are not unique to the cerebellum after all. The impact of the paper, and the degree to which it is possible to claim that particular abnormalities are hallmarks of cerebellar dysfunction, would be much higher if the LD analysis could be repeated in mice with gait abnormalities caused by non-cerebellar dysfunction: Amende et al., 2005. Gait dynamics in mouse models of Parkinson's disease and Huntington's disease. If it is not feasible at this time to use the Locomouse system to analyze gait abnormalities in the mouse models in the Amende paper, the paper should at least discuss some of the limitations in terms of interpreting results as capturing behavioral hallmarks of cerebellar dysfunction. The distinction is between the impairments being "specific" markers of cerebellar dysfunction (which is supported by the data) vs "unique" markers of cerebellar dysfunction (which is not supported by the data). In other words, if a mouse has cerebellar dysfunction then we would be able to predict, based on the current paper, some of the abnormalities we may expect. On the other hand, if a mouse displays some of the abnormalities, it would not be possible to assume, based on the current results, that the underlying neuropathology is cerebellar."

In light of these reservations, the authors should scale back some of their claims and discuss the limitations of the analysis more.

[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]

Reviewer #1:

This manuscript is a very well written manuscript describing the differences and similarities in the ataxic leg movements during walking in two mutant mouse models that have cerebellar neuropathy. Namely these mice are the Purkinje Cell Generation mice and the Reeler mice. They did a nice set of recordings and data analysis. However, this manuscript is purely descriptive, without any conceptual conclusion. I don't think this work, although well done, is not appropriate for eLife.

The current manuscript was submitted to eLife as a Research Advance – a format unique to eLife in which a paper is closely associated with a previous eLife paper. In this case, the current manuscript describing reeler mice is a follow up of our 2015 paper that developed the LocoMouse analysis framework to quantify the gait impairments of Purkinje cell degeneration mice (Machado et al., 2015). Yes, the data in both papers are descriptive. But this does not mean that they have no conceptual conclusion. Many researchers have told us that our 2015 paper fundamentally transformed the way they think about mouse locomotion and the role of the cerebellum. The current manuscript provides solid evidence for some of the ideas we could only speculate about then, as well as capturing shared and specific features of two forms of gait ataxia in a broadly applicable way, for the first time. We have revised the paper, with substantive changes to the Abstract, Introduction, and Discussion, to make our motivation and conceptual points more clear. In particular, the Discussion now highlights conceptual conclusions that are only made possible by taking both of Machado et al., 2015 and the current study into account.

Reviewer #2:

We are still at the proof-of-concept phase for computational neuroethology. One goal of these methods is to identify relationships between complex patterns of behavior by – in a task or context specific manner – balancing (as the discussion says) specificity (e.g., high dimensional measures) with interpretability (e.g., a conceptual framework). This paper puts this hypothesis to the test using the LocoMouse framework developed by this lab to compare ataxic phenotypes among pcd and reeler mice. Here the authors do a very nice job of demonstrating that – like the pcd mice – many of the apparent phenotypes in the reeler mouse are a simple consequence of the slowed speed and smaller size of the mutants. In addition, the observations that phenotypes are now largely found in the front paws – and that tail and nose dynamics can be predicted from this effect – effectively distinguishes the two mutants. The data are clear, and the findings of significant import and utility to those interested in ataxias and in motor control.

Given that the technology and analytics associated with LocoMouse are at this point well vetted, I have no specific comments or concerns about the data in the paper. I would like only a few small adjustments. It would be nice to move some of the data from Figure 1—figure supplement 1 into main Figure 1 (or maybe to make Figure 1—figure supplement 1 a new Figure 2) so that the baseline comparisons between the mutants is more front stage in the figures.

We thank the reviewer for this suggestion. We have converted Figure 1—figure supplement 1 into a main figure (now Figure 2).

In addition, it would be clarifying to generate some simplified graphs (maybe focused on specific speeds) that, in at least some cases, enable comparisons of reeler, pcd and controls all at once. The compromises used in Figure 3 could work here (where you have adjacent panels with two of the three comparisons), but on the whole it would be generally nice to see more direct comparisons across the data in the main figures.

We agree. What makes this tricky is the need to both speed and size-match the data from the two mutants with the controls. In Figure 1—figure supplement 1 (now Figure 2), we had accomplished that by plotting the differences between each mutant and their respective size-matched controls, but this does not work for every feature. To do our best to satisfy this reviewer’s request, which is similar to some of the points raised by reviewer 3, we have a) included Figure 1—figure supplement 1 as a main figure (now Figure 2), and b) included Summary plots like those from Figure 2 for the 43 features that provided inputs to the LDA, in the new Figure 5—figure supplement 1.

Finally, I am left with a sense of the power of the approach to both relate and distinguish, but the one thing that isn't made completely explicit is why reeler mice are so much more apparently ataxic when you just look at them. Is this largely the collective consequence of the "high-stepping" phenotype, the increased variability and the steering phenotype? There is obviously not a simple answer to this, but I would appreciate some discussion of what might account for the fact that the reeler mice look worse by eye, and how quantitative measures might highlight or obscure the gestalt we have as researchers inspecting our mice.

We thank the reviewer for this suggestion and have expanded our discussion of this issue, and of the relationship between hindlimb involvement, variability, and front paw supports, along with their possible neural basis, in the Discussion.

Reviewer #3:

The authors use established methods from their lab to compare locomotor kinematics in wildtype mice and two mutant mouse lines: pcd and reeler. This work builds on some of their impressive previous work where they already showed the results for the pcd mice (Machado et al., 2015), as well as evaluations of both pcd and reeler mice in a different paradigm testing locomotor kinematics, the split-belt treadmill (Darmohray et al., 2019). With the current manuscript, the authors aim to provide a quantitative foundation for mapping specific locomotor impairments onto distinct neuropathologies. The experiments seem to be competently performed, and the manuscript is well written. However, the limitation of the current manuscript is that it does not provide much new insight to the analysis of body part kinematics which hasn't already largely been covered in their previous work.

We thank the reviewer for the nice comments about our previous work. Below, we clarify some misunderstandings about what we have and have not shown before. We also note that this paper’s tight relationship to Machado et al., 2015 is precisely why we submitted this paper as a Research Advance – a format, unique to eLife, that allows a manuscript to be formally associated with a previous eLife paper.

The main concerns are:

1) Lack of novelty compared to their previous papers (Machado et al., 2015; Darmohray et al., 2019). Basic locomotor kinematics of both pcd and reeler mice (as shown here in Figures 1-3) are already described in these papers. Figure 4 does contain new information, although it largely depends on the results from Figure 2. Similar observations to those in Figure 2 of the current manuscript, e.g. different paw supports for the two mutant lines compared to controls, have already been shown in Figure 3D and H of Darmohray et al., 2019, albeit in a different behavioral paradigm.

First, yes, the pcd data were published in Machado et al., 2015 (which is part of why we are submitting this as a Research Advance). But all of the reeler data in this paper is entirely new. The statement “Basic locomotor kinematics of both pcd and reeler mice (as shown here in Figures 1-3) are already described in these papers” is just not true.

That claim appears to be supported by the statement “Similar observations to those in Figure 2 of the current manuscript, e.g. different paw supports for the two mutant lines compared to controls, have already been shown in Figure 3D and H of Darmohray et al., 2019, albeit in a different behavioral paradigm.” We know that the locomotor terminology can be confusing, and so here we aim to clarify an unfortunate misunderstanding. Darmohray et al., 2019 dealt exclusively with locomotor learning. Figure 3D and H of that paper quantified the extent of spatial vs. temporal components of learning in those mutants. Changes in double support symmetry after learning were one way that we assessed the temporal component of learning, but that measure has no correspondence to any parameter reported in the current manuscript. We invite reviewers to consider of Figure 3D and H of Darmohray et al. alongside Figure 2H from our current manuscript. Immediate inspection of these two plots reveals that while they both contain the words “% double support”, they contain entirely different data. While the reeler datapoints hover around 0 in the learning experiments (because there was no temporal learning in these mice), the opposite is true in the current manuscript – the reelers never approach 0 double support.

The reason for this discrepancy is that double support aftereffects, which measure learning, are a completely different measure from double support itself. Given that reeler mice were not included at all in Machado et al., 2015, and that the only reeler data in Darmohray et al., 2019 were a few datapoints concerned only with learning, there is simply no basis for the claim that we have previously reported the locomotor kinematics of reeler mice, or that the reeler data in the first 3 figures of this paper are not new.

To avoid additional confusion, we have added a statement in the Introduction to clarify that “the locomotor kinematics and whole-body coordination of reeler mice have not been reported”.

Finally, the reviewer acknowledges that “Figure 4 does contain new information, although it largely depends on the results from Figure 2.” In fact, Figure 4 incorporates results from all figures of this paper and those of Machado et al., 2015, in order to synthesize this vast amount of locomotor analysis into a digestible format. This approach is entirely new and, as we discuss in the Discussion, not only does it allow us to capture shared and specific features of gait ataxia here, it will form an essential component of our ability to extract meaning from the kinds of behavioral quantification that we first described in Machado et al., 2015 and that have become much more common since.

2) The authors claim to provide a quantitative foundation for mapping specific locomotor impairments onto distinct neuropathologies. However, there is not enough biological validation for the claim that the data presented in the current manuscript can link behavioral phenotypes with underlying neural circuits, as there is no detailed description of the anatomical differences between the two mutant lines within the cerebellum. Certainly the found behavioral differences between the two are not linked to their neural circuits. The authors would need to revise their conclusions based on the data they present here.

We sincerely apologize if we mistakenly gave the impression that we were trying to link specific aspects of the altered cerebellar circuitry to specific aspects of behavior. This was completely unintentional! We have made a number of revisions to the text to more clearly convey our actual meaning, including:

– Major changes to Abstract and Introduction to provide additional context and motivation for the current study .

– Changes in language throughout the paper to favor use of “structures”,

“neuropathologies”, “anatomical features”, and similar in place of “circuitry”, “circuits”, etc.

– A clarified and extended discussion of how we think the gait alterations we observe may related to gross pathology (related to the final comment of reviewer 2).

We hope that these changes have helped clarify that rather than claiming to have mapped specific locomotor features onto specific circuit abnormalities within the cerebellum, we are trying to demonstrate the ability to map shared and specific gait abnormalities onto shared and specific anatomical features. The shared locomotor phenotypes between the two mutants are almost by definition cerebellar, for example, as the neuropathology in pcd is largely specific to the cerebellum (particularly at the youngest ages we tested in Machado et al., 2015) and that is the only overlapping anatomical feature between the two lines. We speculate in the Discussion (and have now expanded this section) about possible anatomical foundations for the differences between the two mutants.

Finally, although we hope it is now clear that we are not attempting to make specific claims about their altered cerebellar circuitry giving rise to specific behavioral differences (other than speculating in the Discussion about the difference between a lack of Purkinje cell output (pcd) vs. aberrant Purkinje cell output (reeler)), we note that reeler and pcd are two of the bestdescribed mutants in the literature – their neuropathologies within the cerebellum and beyond have been described and replicated in countless papers, many of which we cite and describe in some detail here.

3) An attempt to possibly rescue the phenotype would help to improve the biological understanding of which behavioral phenotypes are linked to specifically which anatomical underlying structures. I.e., is it possible to intervene in one of the animal models, so that their phenotype switches on the LD1 axis presented in Figure 4 to the side of the control mice? Such conclusions would then truly link behavioral phenotypes to neural circuits, and would help to significantly improve the biological understanding of mouse locomotor kinematics.

While we agree that it would be quite powerful if such a rescue were possible, it would essentially amount to curing ataxia, which is clearly outside the scope of the current manuscript. Instead, what we propose is that applying this same approach to a broad range of mutants and circuit manipulations could achieve a similar goal. As we outline in the Discussion, we would argue that LD1 here represents “cerebellar dysfunction”, which is the shared anatomical feature in the two mutants, whereas LD2 would represent a combination of a) specific differences in the circuitry within the cerebellum itself, and b) extracerebellar effects. With the future addition of more mouse models and manipulations, the LDA would acquire more axes that could capture the contributions of other brain areas or cell types. We absolutely were not trying to imply that we had already achieved that broader goal, and we have tried to be more circumspect in our conclusions as suggested in major point #2. However, we emphasize that our approach has successfully captured and distilled the shared and specific locomotor phenotypes of two mutants with very clear, and previously well described, shared and specific neuropathologies – and to our knowledge this is the first time that this has been done. We hope to share the success of this approach with a wide audience in the hopes that others will build on the foundation we have established here – as they have already started to do with the LocoMouse system.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Revisions for this paper:

The reviewers found considerable merit in this advance but have further revisions they recommend that can be accomplished quickly and evaluated by the Reviewing Editor. Two of the reviewers thought that further experiments were needed to sustain the authors claims but realizing that this advance was considerable recommend revision in consultation.

Reviewer 3:

"The analyses provided reveal that some locomotor parameters are similarly impaired in these two mutant models while other parameters are unique to each mutant. While the additional findings presented in this study are sound, it is unclear how much we can generalize from analysis of one additional ataxia model, particularly because the neuropathology in these two models is quite different. Without a deeper evaluation of the generality and specificity of locomotor deficits in mouse models of ataxia, this study only provides a relatively slim additional advance over their previous study, and the more general conclusions they draw are not sufficiently justified. I have the following suggestions for strengthening the present manuscript. The authors show nicely in Figure 5 that combinations of behavioral features can be used to distinguish not only wild type from mutant mice but also pcd and reeler mice from each other. The authors' interpretation of these results is that the locomotor features that commonly distinguish wild type animals from the two mutants have a common underlying neuropathological origin (i.e. a cerebellar deficit) while the features that distinguish pcd mice from reeler mice reflect differences between these two mutants (i.e. absence of Purkinje cell output vs. miswiring in the cerebellum plus the additional extracerebellar defects in reeler mice). This is an important idea that could be bolstered with some additional data/analysis: 1. To show that aberrant cerebellar function is the root cause of the variability along LD1, the authors could use perturbation experiments to specifically affect locomotion by manipulating activity in interpositus-projecting Purkinje cells. This could be done acutely (as in Sarnaik and Raman, 2018) and/or chronically (as in Darmohray et al., 2018). 2. To test if LD1 does indeed distinguish wild type from ataxic mice and LD2 (or higher dimensions) distinguish different types of ataxic mice from each other, the authors could analyze locomotion in other models of ataxia that have different neuropathologies."

Reviewer 2:

"On a related note, a major claim is that the LD analysis defines a core set of behavioral features associated with cerebellar dysfunction. These are: impairments in multijoint, interlimb, and whole-body coordination, as well as in predictive mechanisms (i.e. forward models). It is true that these behavioral impairments are observed in both pcd and reeler mice, which share cerebellar neuropathology. However, it is possible that pathology in other brain circuits (basal ganglia, for example) may also be associated with some of the same behavioral impairments observed in the pcd and reeler mice, which would indicate that these behavioral impairments are not unique to the cerebellum after all. The impact of the paper, and the degree to which it is possible to claim that particular abnormalities are hallmarks of cerebellar dysfunction, would be much higher if the LD analysis could be repeated in mice with gait abnormalities caused by non-cerebellar dysfunction: Amende et al., 2005. Gait dynamics in mouse models of Parkinson's disease and Huntington's disease. If it is not feasible at this time to use the Locomouse system to analyze gait abnormalities in the mouse models in the Amende paper, the paper should at least discuss some of the limitations in terms of interpreting results as capturing behavioral hallmarks of cerebellar dysfunction. The distinction is between the impairments being "specific" markers of cerebellar dysfunction (which is supported by the data) vs "unique" markers of cerebellar dysfunction (which is not supported by the data). In other words, if a mouse has cerebellar dysfunction then we would be able to predict, based on the current paper, some of the abnormalities we may expect. On the other hand, if a mouse displays some of the abnormalities, it would not be possible to assume, based on the current results, that the underlying neuropathology is cerebellar." In light of these reservations, the authors should scale back some of their claims and discuss the limitations of the analysis more.

Thank you for these insightful comments and suggestions. We absolutely acknowledge that the conclusions that can be drawn from two mutants are limited and fully agree that the inclusion of additional cerebellar and non-cerebellar models within the framework we provide here will be hugely informative as to the specific contributions of both cerebellar and other circuits to locomotor phenotypes. Indeed, this has always been our plan, and we hope to follow up on this study with exactly those analyses in the near future. We have revised our language throughout the manuscript to soften our claims, address limitations, and make more explicit distinctions between what we have actually shown here, vs. predictions for future studies.

Author details

1. Ana S Machado

   Champalimaud Neuroscience Program, Champalimaud Center for the Unknown, Lisbon, Portugal

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

2. Hugo G Marques

   Champalimaud Neuroscience Program, Champalimaud Center for the Unknown, Lisbon, Portugal

   Contribution

   Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8709-4841

3. Diogo F Duarte

   Champalimaud Neuroscience Program, Champalimaud Center for the Unknown, Lisbon, Portugal

   Contribution

   Conceptualization, Software, Formal analysis, Visualization, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

4. Dana M Darmohray

   Champalimaud Neuroscience Program, Champalimaud Center for the Unknown, Lisbon, Portugal

   Contribution

   Conceptualization, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

5. Megan R Carey

   Champalimaud Neuroscience Program, Champalimaud Center for the Unknown, Lisbon, Portugal

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Validation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   megan.carey@neuro.fchampalimaud.org

   Competing interests

   Reviewing Editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0002-4499-1657

• 1,919

  Page views

• 190

  Downloads

• 9

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.