Loss of Wt1 in the murine spinal cord alters interneuron composition and locomotion

Rhythmic and patterned locomotion is driven by spinal cord neurons that form neuronal circuits, referred to as central pattern generators (CPGs). Recently, dI6 neurons were suggested to participate in the control of locomotion. The dI6 neurons can be subdivided into three populations, one of which expresses the Wilms tumor suppressor gene Wt1. However, the role that Wt1 exerts on these cells is not understood. Here, we aimed to identify behavioral changes and cellular alterations in the spinal cord associated with Wt1 deletion. Locomotion analyses of mice with neuron-specific Wt1 deletion revealed that these mice ran slower than controls with a decreased stride frequency and an increased stride length. These mice showed changes in their fore-hindlimb coordination, which were accompanied by a loss of contralateral projections in the spinal cord. Neonates with Wt1 deletion displayed an increase in uncoordinated hindlimb movements and their motor neuron output was arrhythmic with a decreased frequency. The population size of dI6, V0 and V2a neurons in the developing spinal cord of conditional Wt1 mutants was significantly altered. These results show that the development of particular dI6 neurons depends on Wt1 expression and loss of Wt1 is associated with alterations in locomotion.


Introduction
In vertebrates, rhythmic activity is generated by a network of neurons, commonly referred to as central pattern generators (CPGs) (Grillner and Zangger 1979;Grillner 1985). CPGs do not require sensory input to produce rhythmical output; however, the latter is crucial for the refinement of CPG activity in response to external cues (Shik and Orlovsky 1976;Rossignol S 1988;Pearson 2003). The locomotor CPGs are located in the spinal cord and consist of distributed networks of interneurons and motor neurons (MN), which generate an organized motor rhythm during repetitive locomotor tasks like walking and swimming (Grillner 1985;McCrea and Rybak 2008).

4
The spinal cord develops from the caudal region of the neural tube. The interaction of secreted molecules including sonic hedgehog (Shh) and bone morphogenetic proteins (BMPs) provides instructive positional signals to the 12 progenitor cell domains that reside in the neuroepithelium (Alaynick et al., 2011).
Each domain is characterized by the expression of specific transcription factor encoding genes that are used to selectively identify these populations. The dI1-dI5 interneurons are derived from dorsal progenitors and primarily contribute to sensory spinal pathways. The dI6, V0-V3 interneurons and MN arise from intermediate or ventral progenitors and are involved in the locomotor circuitry (Goulding 2009).
Whereas the involvement of V0 -V3 neurons in locomotion has been well documented, the role for dI6 neurons in locomotion has only recently been investigated (Andersson et al. 2012;Dyck et al. 2012). In particular, a part of the dI6 population shows rhythmically active neurons (Dyck et al. 2012), and a more defined subpopulation of dI6 neurons expressing the transcription factor Dmrt3, is critical for normal development of coordinated locomotion (Andersson et al. 2012). Another group of dI6 neurons is suggested to express the Wilms tumor suppressor gene Wt1 but has not yet been characterized (Goulding 2009;Andersson et al. 2012).
Wt1 encodes a zinc finger transcription factor that is inactivated in a subset of Wilms tumors, a pediatric kidney cancer (Call et al. 1990;Gessler et al. 1990). Wt1 fulfills a critical role in kidney development; however, the function of Wt1 is not limited to this organ. Phenotypic anomalies of Wt1 knockout mice can be found, among others, in the gonads, heart, spleen, retina and the olfactory system (Kreidberg et al. 1993;Herzer et al. 1999;Moore et al. 1999;Wagner et al. 2002;Wagner et al. 2005).
In one of the first reports on Wt1 expression, the spinal cord was described as a prominent Wt1+ tissue (Armstrong et al. 1993;Rackley et al. 1993), however, until now there are no further reports regarding the function of Wt1 in the central nervous system (CNS).

5
Here, we have examined the role of Wt1 in the developing spinal cord. We performed locomotor analyses of conditional Wt1 knockout mice and used molecular biological and electrophysiological approaches to elucidate the function of Wt1 expressing neurons for locomotion. Our data suggest that Wt1 expressing dI6 neurons contribute to the coordination of locomotion and that Wt1 is needed for proper dI6 neuron specification during development.

Wt1 expressing cells in the spinal cord are dI6 neurons
In order to determine the spatial and temporal pattern of Wt1 expressing cells in the spinal cord, we performed immunohistochemical analyses. Wt1+ cells were detected in the medioventral mantle zone of the developing spinal cord at embryonic day (E) 12.5 (Fig. 1A). Until E15.5, embryonic spinal cords showed a constant amount of Wt1+ cells; thereafter, their number gradually decreased until they could no longer be detected in adult mice (Fig. 1B).
We next wanted to determine the birthdate of Wt1+ cells, defined as the time point when progenitor cells cease to proliferate, leave the ventricular zone and start to differentiate. Using Bromodeoxyuridine (BrdU), the proliferative cells in the ventricular zone were labelled at different embryonic stages (E9.5, E10.5 and E11.5).
Immunostaining of these cells for Wt1 at E12.5 revealed that prospective Wt1 expressing cells still proliferate at E9.5 and even at E10.5 (Fig. 1C). At E11.5 Wt1+ cells no longer showed incorporation of BrdU suggesting that they had left the ventricular zone and started their migration and differentiation in the mantel zone at this time-point. Wt1 has been proposed to label dI6 neurons (Goulding 2009), however, the only available primary data has so far only suggested its presence in a subpopulation of dI6 neurons expressing Dmrt3 (Andersson et al., 2012). In order to closer examine the nature of Wt1+ cells, we performed immunostainings of embryonic spinal cords at 6 E12.5. Cells expressing Wt1 were positive for Pax2 and Lim1/2 labelling dI4, dI5, dI6, V0 D and V1 neurons (Tanabe and Jessell 1996;Burrill et al. 1997) while being negative for the post-mitotic V0 V marker Evx1 (Moran-Rivard et al. 2001) (Fig. 1D, E).
Wt1 expression did not overlap with Lmx1b, a marker specific for dI5 neurons, but did coincide with Lbx1 (Gross et al. 2002) and Bhlhb5 (Skaggs et al. 2011), which commonly occur in the ventral most dI4-dI6 Lbx1+ domain giving rise to dI6 neurons.
Thus, these data supports and extends on the previous observations that Wt1 is a marker for a subset of dI6 neurons.

Deletion of Wt1 affects locomotor behavior
To investigate the function of the Wt1+ neurons in the spinal cord, we made use of a Nes-Cre;Wt1 fl/fl mouse line ( Fig. 2A). At E12.5, no Wt1 mRNA or protein was detected in neurons from this mouse line ( Fig. 2A, B). Given the location of the Wt1+ neurons within the ventral dI6 population that has been shown to be involved in regulating locomotion, we performed behavioral tests associated with locomotion to investigate potential phenotypic consequences of deleting Wt1 in spinal cord neurons. Footprints of adult mice walking on a transparent treadmill at fixed speeds (0.15, 0.25, 0.35 m/s) were recorded to analyze different gait parameters (Supplemental Fig. S1A). Nes-Cre;Wt1 fl/fl mice revealed a significant reduction in stride frequency for both the fore-and hindlimbs relative to control (Wt1 fl/fl ) animals at all speeds measured. Heterozygous Wt1 knockout mice (Nes-Cre;Wt1 fl/+ ) did not differ significantly from controls. Stride length, accordingly, was significantly longer in Nes-Cre;Wt1 fl/fl animals compared to wild type mice and Nes-Cre;Wt1 fl/+ . Thus, although Nes-Cre;Wt1 fl/fl mice were slightly smaller compared to controls (body mass Wt1 fl/fl vs Nes-Cre;Wt1 fl/fl : males 33 +/-3.9 vs 25 +/-3.7 g; females 25 +/-3.2 g vs 22 +/-1.4 g; body length males 9.9 +/-0.4 g vs 9.4 +/-0.4 cm; females 9.9 +/-0.4 cm vs 9.8 +/-0.3 cm), they made longer strides with lower frequency.

7
To further explore gait alterations, we used X-ray fluoroscopy as a complementary method in a larger cohort of mice (Fig. 2C;Supplemental Fig. S1B; Supplemental Movie 1 and 2). When animals walked voluntarily at their preferred speed, deviations in stride frequency and stride length from the expected value (control baseline) for the given speed were again observed in Nes-Cre;Wt1 fl/fl (Fig.   2D), but statistical significance is confirmed only for females. The changes were accompanied by a significant reduction of raw speed and size-corrected speed (= Froude number) in Nes-Cre;Wt1 fl/fl mice of both sexes (Supplemental Fig. S1C).
While both the duration of stance and swing phase and the distance covered by the trunk and the limbs, respectively, differ between controls and Nes-Cre;Wt1 fl/fl by more than 10 percent in males and more than 15 percent in females, the ratio between the two phases, expressed by the Duty factor, remains unaffected (Supplemental Fig.   S1D). Thus, the temporal coordination between stance and swing phase in adult Nes-Cre;Wt1 fl/fl mice is normal.
We tested whether changes in gait parameters are accompanied by changes in the phase relationships between the limbs (Fig. 2E, F). The footfall pattern of control and Nes-Cre;Wt1 fl/fl females did not show significant differences at the same speed of 0.21 m/s (Supplemental Fig 1E). However, the different spread along the Xaxis indicates the evenly elongated stance and swing phases.
The symmetry of left and right limb movements expressed as the time-lag between footfalls in percent stride duration of a reference limb was unaffected in the So far, the limb kinematics of adult Nes-Cre;Wt1 fl/fl mice compared to the Wt1 fl/fl mice shows subtle differences in gait parameters and interlimb coordination with a high degree of variation. In sum, these differences result in a performance reduction indicated by the overall lower walking velocities.

Deletion of Wt1 results in a disturbed and irregular postnatal locomotor pattern
After having observed altered gait parameters in adult Nes-Cre;Wt1 fl/fl animals, we wondered whether gait also would be affected in younger mice. Indeed, Additionally, the frequency of the ventral root output was decreased (Fig 3D: control; 0.30 ± 0.024 Hz: Nes-Cre;Wt1 fl/fl ; 0.18 ± 0.08 Hz ). This slower rhythm in Nes-Cre;Wt1 fl/fl cords could be attributed to altered L2 and L5 activity burst parameters, as Nes-Cre;Wt1 fl/fl mice had significantly longer burst, interburst and cycle periods compared to control (Fig. 3E, F). Thus, the deletion of Wt1 results in a disturbed and irregular locomotor pattern, which suggests that there are changes to the neuronal locomotor circuitry that occur following Wt1 deletion.

Wt1+ neurons receive various synaptic inputs and can project commissurally
In order to assess how Wt1+ dI6 neurons are connected within the CPG network, we focused on the innervation pattern of these cells. We used the Wt1-GFP reporter mouse line (Hosen et al. 2007) where Wt1+ neurons are labeled by GFP. In contrast to the restricted localization of Wt1 in the nucleus, GFP is distributed throughout the cytoplasm and labels the soma and major processes ( Fig 4A). In combination with antibodies against particular vesicular synaptic transporters, we observed that excitatory (VGLUT2), inhibitory (VGAT) and modulatory (VMaT2) synapses contact the soma of Wt1+ dI6 neurons (Fig. 4B). This shows that Wt1+ dI6 neurons receive excitatory, inhibitory and modulatory inputs suggesting that Wt1+ neurons are positioned to receive a multitude of signals and could act during locomotion to integrate different CPG signals.
Using the Wt1-GFP reporter mouse, we found GFP+ fibers crossing the spinal cord midline beneath the central canal suggesting that Wt1+ neurons project commissural fibers (

Loss of Wt1 leads to altered interneuron composition
To assess the possible impact of Wt1 deletion for interneuron development, To test this hypothesis, we ablated the cells destined to express Wt1. We used Lbx1-Cre;Wt1-GFP-DTA mice in which the Diphtheria toxin subunit A (DTA) is expressed from the endogenous Wt1 locus after Cre-mediated excision of a GFP cassette harboring a translational STOP-codon. Cre expression driven by the Lbx1 promoter targets the dI4 to dI6 interneuron populations (Müller et al. 2002). In Lbx1-Cre;Wt1-GFP-DTA embryos, nearly all Wt1+ neurons were ablated at E16.5 (Fig. 5D). The ablation of Wt1+ neurons coincided with a significantly decreased number of Dmrt3+ neurons in Lbx1-Cre;Wt1-GFP-DTA embryos, but did not affect the number of Evx1+ neurons ( Fig 5D). Taken together, the results from the Wt1 deletion and the ablation of the Wt1 neurons suggests that the fate switch from dI6 neurons into Evx1+ V0 neurons occurs due to the deletion of Wt1. A postnatal phenotypic behavioral analysis of these mice was not possible because neonates died immediately after birth due to serious respiratory deficits (data not shown).
The analyses of the interneuron composition in developing conditional Wt1 knockout mice and embryos with an ablation of Wt1+ neurons suggest a fate switch within a specific subset of dI6 and V0 V neurons that depends on the presence of the cells destined to express Wt1.

The transition of dI6 neurons into Evx1+ V0 V neurons upon loss of Wt1 is not direct
In order to further investigate the cellular fate change upon deletion of Wt1 we combined Wt1-GFP and Nes-Cre;Wt1 fl/fl animals to generate Nes-Cre;Wt1 fl/GFP mice.
These mice harbor a constitutive knockout allele of Wt1 due to the insertion of a GFP coding sequence and another conditional Wt1 knockout allele. GFP and Wt1 were co-localized in the ventral spinal cord of Wt1 fl/GFP control animals at E13.5, whereas GFP, but not Wt1, was detected in spinal cords of Nes-Cre;Wt1 fl/GFP embryos of the same age ( Fig 6A). Thus, Nes-Cre;Wt1 fl/GFP mice allowed us to inactivate Wt1 while the cells destined to express Wt1 are labelled by GFP.
To investigate whether Wt1 deletion leads to apoptosis in the respective cells, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) was used. TUNEL+ cells were present in the ventrolateral spinal cords of Wt1 fl/GFP control and Nes-Cre;Wt1 fl/GFP embryos ( Fig 6A). However, TUNEL signals never overlapped with GFP+ dI6 neurons destined to express Wt1, suggesting that Wt1 inactivation in dI6 neurons did not result in cell death.
In order to find out whether cells destined to express Wt1 would directly convert to V0 V neurons upon Wt1 inactivation, we performed immunohistochemical analyses. The presence of Dmrt3 and Evx1 in GFP+ dI6 neurons was analyzed in Wt1 fl/GFP control and Nes-Cre;Wt1 fl/GFP embryos at E12.5 ( Fig 6B). The number of GFP+ cells per hemicord was determined and set to 100%. The proportion of Dmrt3+ cells was approximately 13% of all GFP+ cells in spinal cord of E12.5 control embryos. When Wt1 was absent, the amount of Dmrt3+ GFP cells significantly decreased to 4%. In contrast, the proportion of GFP+ dI6 neurons that also showed Evx1 staining was not changed between Wt1 fl/GFP control and Nes-Cre;Wt1 fl/GFP animals (below 1% for both). Thus, the increase in the amount of Evx1+ V0 V neurons observed in mice lacking Wt1, does not seem to result from a direct transition of future Wt1+ dI6 neurons into Evx1+ V0 V neurons.

Discussion
In this study, we have examined Wt1, which marks a subset of dI6 neurons.
We found that Wt1 is required for proper differentiation of spinal cord neurons and that deletion of Wt1 results in locomotor aberrancies in neonate and adult mice. negative V0 D -like neurons, which leads to a putative increase of the V0 D population ( Fig 6C). The Evx1+ V0 V population might, in turn, increase its number to compensate for a higher proportion of V0 D -like neurons.
In addition to the changes in the dI6 and V0 population that occur upon Wt1 deletion in the spinal cord, Chx10+ V2a neurons show a slight but significant decrease in their cell number at E16.5 (Fig 6C). This might represent a secondary effect of the alterations in the dI6 and V0 population, which occur already at E12.5. It was reported that V2a neurons directly innervate V0 V neurons (Crone et al., 2008).
This secondary effect might thus be due to a potential adaptation to the altered interneuron composition in the spinal cord and the necessity to form proper contacts with target cells to build up the neuronal circuits responsible for locomotion.
In sum, the results obtained in this study not only shed light on the so far undescribed necessity for Wt1 in the development of spinal cord neurons and their 16 functional implementation in circuits responsible for locomotion. The data also broadens our view on the complex interplay of the various neuron subpopulations within the spinal cord.  (Tronche et al. 1999) or Lbx1-Cre;Wt1 fl/fl mice (Sieber et al. 2007). To generate mice with Wt1 ablated cells, Wt1-GFP-DTA mice were bred with Lbx1-Cre mice. Control mice were sex-and age-matched littermates (wild type or Wt1 fl/fl ). For plug mating analysis, females of specific genotypes were housed with males of specific genotypes and were checked every morning for the presence of a plug. For embryo analysis, pregnant mice were sacrificed by CO 2 inhalation at specific time points during embryo development and embryos were dissected. Typically, female mice between 2 and 6 months were used.

Generation of Wt1-GFP-DTA mice
The Wt1-GFP-DTA mouse line bares an IRES-lox-GFP-lox-DTA cassette that was inserted into intron 3 of the Wt1 locus. This cassette consists of a GFP encoding sequence that ends in a translational STOP-codon and is flanked by loxP sites.
Downstream of GFP, the coding sequence for the Diphtheria toxin subunit A (DTA) was incorporated. Before Cre-induction, the IRES cassette ensures the generation of a functional GFP protein. After Cre-mediated excision of the floxed GFP sequence, the DTA is expressed from the endogenous Wt1 promotor.

18
The Wt1-GFP-DTA model was generated by homologous recombination in embryonic stem (ES) cells. After ES cell screening using PCR and Southern Blot analyses, recombined ES cell clones were injected into C57BL/6J blastocysts.
Injected blastocysts were re-implanted into OF1 pseudo-pregnant females and allowed to develop to term. The generation of F1 animals was performed by breeding of chimeras with wild type C57BL/6 mice to generate heterozygous mice carrying the Biotechnology, Inc., Santa Cruz, California, USA). Secondary antibodies were applied according to species specificity of primary antibodies. Hoechst was used to stain nuclei. Quantitative analysis of the antibody staining was statistically analyzed using student t-test and two-way ANOVA followed by Tukey's post hoc test.

Bromodeoxyuridine labeling
To label proliferating cells in the embryonic spinal cord, pregnant mice at E9.5, E10.5 and E11.5 were injected intraperitoneally with 100 µg/g of Bromodeoxyuridine (BrdU) dissolved in 0.9% sodium chloride solution. Embryos were harvested at E12.5 to isolate spinal cords and stain for BrdU and Wt1. Spinal cords were frozen unfixed after 15 min dehydration with 20% sucrose (in 50 % TissueTec/PBS) and sectioned (12 µm). After any of the following treatments, sections were washed with PBS. Antigen retrieval was performed by incubation in 98°C sub boiling 10 mM sodium citrate buffer pH6.0 for 30 min. After treatment with 2N HCl at 37°C for 30 min, sections were incubated with primary antibodies using dilutions mentioned above (Immunohistochemistry). Secondary antibodies were applied according to species specificity of primary antibodies.

RNA isolation and qRT-PCR analysis
Total RNA was isolated from E12.5 embryonic spinal cords using Trizol

Analysis of locomotor behavior
In order to characterize gait parameters, 10 animals per sex and genotype were used. Body masses of the mice varied considerably within the groups and among the groups with significant differences between male Wt1 fl/fl and Nes-Cre;Wt1 fl/fl mice (Wt1 fl/fl : 28 g  3 g vs. Nes-Cre;Wt1 fl/fl : 23 g  3 g; F s = 31.98; t s = 3.28, P> 0.001) and moderate differences between the female Wt1 fl/fl and Nes-Cre;Wt1 fl/fl mice (Wt1 fl/fl : 25 g  5 g vs. Nes-Cre;Wt1 fl/fl : 22 g  4 g; F s = 3.80; t s = 1.62, n.s.). We recorded the voluntary walking performance of this larger cohort using high-resolution X-ray fluoroscopy (biplanar C-arm fluoroscope Neurostar, Siemens AG, Erlangen, Germany). Because of body size variation within and among groups, we adjusted treadmill speed dynamically to the individual preferences and abilities of the mice. This method of motion analysis has been described in detail in several recent publications (e.g. Böttger et al., 2011;Andrada et al., 2015;Niederschuh et al., 2015) and will be only briefly summarized here: The X-ray system operates with high-speed cameras and a maximum spatial resolution of 1536 dpi x 1024 dpi. A frame frequency of 500 Hz was used. A normal-light camera operating at the same frequency and synchronized to the X-ray fluoroscope was used to document the entire trial from the lateral perspective. Footfall sequences and spatiotemporal gait parameters were quantified by manual tracking of the paw toe tips and two landmarks on the trunk (occipital condyles, iliosacral joint) using SimiMotion 3D.
Speed, stride length, stride frequency, as well as the durations of stance and swing phases, as well as the distances that trunk or limb covered during these phases were computed from the landmark coordinates collected at touchdown and lift-off of each limb. The phase relationships between the strides of left and right limbs as well as fore-and hindlimbs were determined from footfall sequences as expression of temporal interlimb coordination. As the animals frequently accelerated or decelerated relative to the treadmill speed, the actual animal speed was obtained by offsetting trunk movement against foot movement during the stance phase of the limb. The resulting distance was divided by the duration of the stance phase. Animal speed as well as all temporal and spatial gait parameters were then scaled to body size following the formulas published by Hof (1996) Recorded signals containing compound action potentials were amplified 10,000 times, and band-pass filtered (100-10 kHz) before being digitized (Digidata 1322A, Axon instruments) and recorded using Axoscope 10.2 (Axon Instruments Inc.) for later off-line analysis. The data was rectified and low-pass filtered using a third-order Butterworth filter with a 5 Hz cut-off frequency before further analysis. Coherence plots between L2 and L2/L5 traces were analysed using a mortlet wavelet transform in SpinalCore (Version 1.1, (Mor and Lev-Tov, 2007)). Preferential phase alignment across channels are shown in the circular plots and burst parameters were analysed for at least 20 sequential bursts, as previously described (Kiehn and Kjaerulff 1996) using an in-house designed program in Matlab (Mathworks R2014b). Ventral root recording preferential phase alignment was assessed by means of circular statistics (Rayleight test) for 20 consecutive cycles as described (Kiehn and Kjaerulff 1996).
Burst parameters, including frequency, are presented as the mean ± standard deviation (SD). Burst parameters were compared using the two-tailed Mann-Whitney test or the Kruskal-Wallis analysis of variance test followed by a Dunns post-test comparing all groups.

Tracing of commissural neurons
To examine whether the loss of Wt1 affects spinal cord populations, tracing experiments were conducted as previously described (Rabe et al. 2009;Andersson et al. 2012). Nes-Cre;Wt1 fl/fl and Nes-Cre;Wt1 +/+ littermate control mice P0-P5, were prepared as described above (Fictive locomotion). Two horizontal cuts (intersegmental tracing targeting commissural ascending/descending/bifurcating neurons) were made in the ventral spinal cord at lumbar (L) level 1 and between L4 and 5. Fluorescent dextran-amine (FDA, 3000 MW, Invitrogen) was applied at L1 and rhodamine-dextran amine (RDA, 3000 MW, Invitrogen) was applied between the L4/5 ventral roots. Spinal cords were incubated overnight at room temperature, subsequently fixed in 4% formaldehyde (FA) and stored in the dark at 4°C until transverse sectioning (60µm) on a vibratome (Leica, Germany).
Fluorescent images were acquired on a fluorescence microscope (Olympus BX61W1). For quantitative analyses of traced cords, consecutive images were taken between the two tracer application sites using Volocity software (Improvision, Lexington, USA). Captured images were auto-levelled using Adobe Photoshop software. Only cords with an intact midline, as assessed during imaging, were used for analysis.
Traced neurons in Wt1 fl/fl control, Nes-Cre;Wt1 fl/+ and Nes-Cre;Wt1 fl/fl cords were examined for significance using the Kruskal-Wallis analysis of variance test followed by a Dunns post-test comparing all groups. Tracing data are presented as the mean ± standard error of the mean (SEM).

TUNEL-Assay
To detect apoptosis in situ, the TUNEL assay was performed prior to antibody binding. Slides were incubated with TUNEL reaction solution (1x Reaction Buffer TdT and 15 U TdT in ddH 2 O from Thermo Scientific; 1 mM dUTP-biotin from Roche) at 37 °C for 1 h and washed with PBS afterwards.

Imaging and picture processing
Fluorescent images were viewed in a Zeiss Axio Imager and a Zeiss Observer Z1 equipped with an ApoTome slider for optical sectioning (Zeiss, Germany). Images were analyzed using the ZEISS ZEN2 image analysis software.
For quantitative analyses of traced spinal cords, the application sites were identified and consecutive photographs were taken between the two application sites using the OptiGrid Grid Scan Confocal Unit (Qioptiq, Rochester, USA) and Volocity software (Improvision, Lexington, USA). Confocal images were captured on a ZEISS LSM 710 ConfoCor 3 confocal microscope and analyzed using the ZEISS ZEN2 image analysis software. Captured images were adjusted for brightness and contrast using ZEN2 image analysis software and Adobe Photoshop software.

Statistical Analyses
Data are expressed as mean ± SD or as indicated. Groups were compared using two-way ANOVA or two-tailed two-sample equal variance student t-test as    (1) and between hindlimbs (2) illustrate overall symmetry of the walk. Timing of forelimb touchdown relative to hindlimb touchdown for ipsilateral (3) and contralateral (4) limbs show only minor differences between Wt1 fl/fl and Nes-Cre;Wt1 fl/fl mice. The timing of hindlimb footfalls relative to forelimb footfalls (5,6) differ between Wt1 fl/fl and Nes-Cre;Wt1 fl/fl mice, particularly at the contralateral limbs. Box plots indicate the median (bold white or black line), the 25 th and the 75 th percentile (box), and the data range (whiskers). Significance level: *** P < 0.001; ** P < 0.01; * P < 0.05.   Significance level:*P < 0.05, **P < 0.01, ***P < 0.001. Dmrt3 (dI6 D ), Wt1 (dI6 W ) or both (dI6 DW ). Due to the knockout of Wt1, no dI6 W and dI6 DW are detectable and the number of dI6 D neurons is decreased. In contrast, the number of Evx1+ V0 V neurons increases, which is an indirect effect as potential dI6 W cells that lack Wt1 did not show a Evx1 signal. This effect might be explained by a hypothetical fate change of dI6 neurons into V0 D like neurons (dashed light grey circle). The increased number of V0 D neurons would thus prompt the pV0 progenitor cells to differentiate preferentially into V0 V neurons, which would compensate the excess amount of V0 D neurons and lead to an increase in the population size of Evx1+ V0 V neurons. As a secondary effect, the number of V2a neurons, which innervate the V0 V neurons, declines at later developmental stages when neurons start to connect to each other potentially compensating the increased number of V0 V neurons. Only the subsets of interneuron populations are shown that are affected by the tissue specific Wt1 knockout. Red indicates decrease in population size, green indicates increase in population size.