Abstract
Locomotion is coordinated by neuronal circuits of the spinal cord. Recently, dI6 neurons were shown to participate in the control of locomotion. A subpopulation of dI6 neurons expresses the Wilms tumor suppressor gene Wt1. However, the function of Wt1 in 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 a slower walk 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 that 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) (Jessell, 2000; Grillner, 2003; Kiehn, 2006; Brownstone & Wilson, 2008; Goulding, 2009; Berkowitz et al, 2010). CPGs do not require sensory input to produce rhythmic output; however, the latter is crucial for the refinement of CPG activity in response to external cues (Rossignol & Drew, 1988; Jessell, 2000; Pearson, 2004). The locomotor CPGs are located in the spinal cord and consist of distributed networks of interneurons and motor neurons (MNs), which generate an organized motor pattern during repetitive locomotor tasks such as walking and swimming (Grillner, 1985; Kiehn 2006, 2016; Brownstone & Wilson, 2008; McCrea & Rybak, 2008; Goulding, 2009; Grillner & Jessell, 2009).
The spinal cord develops from the caudal region of the neural tube. The interaction of secreted molecules, including sonic hedgehog and bone morphogenetic proteins, 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).
The involvement of V0–V3 neurons in locomotion has been well documented: V0 (Lanuza et al, 2004; Talpalar et al, 2013; Bellardita & Kiehn, 2015), V1 (Zhang et al, 2014; Britz et al, 2015), V2a (Crone et al 2008, 2009; Dougherty & Kiehn, 2010; Zhong et al, 2010), and V3 (Zhang et al, 2008). The role for dI6 neurons in locomotion has only recently been addressed (Andersson et al, 2012; Dyck et al, 2012; Haque et al, 2018). A fraction of the dI6 population consists of 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). A group of dI6 neurons is suggested to express the Wilms tumor suppressor gene Wt1 (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 olfactory system (Kreidberg et al, 1993; Herzer et al, 1999; Moore et al, 1999; Wagner et al 2002, 2005). In one of the first reports on Wt1 expression, a particular region of the hindbrain below the fourth ventricle and the spinal cord were described as prominent Wt1+ tissues (Armstrong et al, 1993; Rackley et al, 1993). Very recent work focusing on Wt1-expressing cells in the spinal cord suggested those cells to be involved in locomotion (Haque et al, 2018). However, until now, there is no insight on the way that Wt1 determines the character of these cells.
Here, we have examined the importance of Wt1 for the developing spinal cord neurons. We performed locomotor analyses of conditional Wt1 knockout mice and used molecular biological and electrophysiological approaches to elucidate the role that Wt1 exerts on spinal cord 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.
Results
Wt1-expressing cells in the spinal cord are dI6 neurons
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 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 have so far only suggested its presence in a subpopulation of dI6 neurons expressing Dmrt3 (Andersson et al, 2012). To closely examine the nature of Wt1+ cells, we performed immunostainings of embryonic spinal cords at E12.5. All cells expressing Wt1 were positive for Pax2 and Lim1/2 labelling dI4, dI5, dI6, V0D, and V1 neurons (Tanabe & Jessell, 1996; Burrill et al, 1997), while being negative for the postmitotic V0V marker Evx1 (Moran-Rivard et al, 2001) (Fig 1D and E). Wt1 expression did not overlap with Lmx1b, a marker specific for dI5 neurons, but all Wt1+ cells exhibited Lbx1 (Gross et al, 2002) and Bhlhb5 labelling (Skaggs et al, 2011), which commonly occur in the ventral most dI4–dI6 Lbx1+ domain giving rise to dI6 neurons. Thus, these data support and extend on the previous observations that Wt1 is a marker for a subset of dI6 neurons.
Deletion of Wt1 affects locomotor behavior
Because a constitutive knockout of Wt1 is embryonically lethal, we made use of a conditional Nes-Cre;Wt1fl/fl mouse line to investigate the function of Wt1 in the spinal cord (Fig 2A). At E12.5, no Wt1 mRNA or protein was detected in neurons from this mouse line (Fig 2A and 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, and 0.35 m/s) were recorded to analyze different gait parameters (Fig S1A). Nes-Cre;Wt1fl/fl mice revealed a significant reduction in stride frequency for both the fore- and hindlimbs relative to control (Wt1fl/fl) animals at all speeds measured. Heterozygous Wt1 knockout mice (Nes-Cre;Wt1fl/+) did not differ significantly from controls. Stride length, accordingly, was significantly longer in Nes-Cre;Wt1fl/fl animals than in wild-type mice and Nes-Cre;Wt1fl/+. Thus, although Nes-Cre;Wt1fl/fl mice were slightly smaller than controls (body mass Wt1fl/fl versus Nes-Cre;Wt1fl/fl: males, 33 ± 3.9 versus 25 ± 3.7 g; females, 25 ± 3.2 g versus 22 ± 1.4 g; body length: males, 9.9 ± 0.4 g versus 9.4 ± 0.4 cm; females, 9.9 ± 0.4 cm versus 9.8 ± 0.3 cm), they made longer strides with lower frequency.
To further explore gait alterations, we used X-ray fluoroscopy as a complementary method in a larger cohort of mice (Figs 2C and S1B and Videos 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;Wt1fl/fl (Fig 2D), but statistical significance is confirmed only for females. The changes were accompanied by a significant reduction of raw speed (animal velocity in m/s) and size-corrected speed (= Froude number) in Nes-Cre;Wt1fl/fl mice of both sexes (Fig S1C). Although both the duration of stance and swing phases and the distance covered by the trunk and the limbs, respectively, differ between controls and Nes-Cre;Wt1fl/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 (Fig S1D). Thus, the temporal coordination between stance and swing phases in adult Nes-Cre;Wt1fl/fl mice is normal.
X-ray fluoroscopy video shows skeleton of Wt1fl/fl control mouse running on a treadmill. Download video
X-ray fluoroscopy video shows skeleton of Nes-Cre;Wt1fl/fl mouse running on a treadmill. Download video
We tested whether changes in gait parameters are accompanied by changes in the phase relationships between the limbs (Fig 2E and F). The footfall pattern of control and Nes-Cre;Wt1fl/fl females did not show significant differences at the same speed of 0.21 m/s (Fig S1E). However, the different spread along the X-axis 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 (Fig S1F) was unaffected in the Nes-Cre;Wt1fl/fl mice (Fig 2E and F, 1 and 2). Also, the timing of forelimb footfalls relative to the ipsilateral and contralateral hindlimb cycles is very similar between Wt1fl/fl mice and Nes-Cre;Wt1fl/fl mice (Fig 2E and F, 3 and 4). Significant differences between Wt1fl/fl mice and Nes-Cre;Wt1fl/fl mice were observed in the timing of the hindlimb footfalls relative to the forelimb cycles (Fig 2E and F, 3 and 4). The touchdown of the ipsilateral and the contralateral hindlimb falls in a later fraction of the forelimb stride cycle in Nes-Cre;Wt1fl/fl mice compared with the Wt1fl/fl mice. The deviation cannot be explained by the differences in animal speed because the hind-to-forelimb coordination shows only small amount of speed-dependent variation: the time lag between the footfalls tend to increase with increasing speed (baseline ipsilateral: Wt1fl/fl males: F1,248 = 13.38, r2 = 0.051, Wt1fl/fl females: F1,248 = 18.63, r2 = 0.070; baseline contralateral: Wt1fl/fl males: F1,273 = 16.39, r2 = 0.057, Wt1fl/fl females: F1,274 = 8.14, r2 = 0.029). So far, the limb kinematics of adult Nes-Cre;Wt1fl/fl mice compared with the Wt1fl/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;Wt1fl/fl animals, we wondered whether gait also would be affected in younger mice. Indeed, Nes-Cre;Wt1fl/fl pups had more difficulty coordinating their fore- and hindlimbs than controls when performing air-stepping. Although there was no increase in hind-limb synchronous steps, left/right alternating steps were decreased and the number of uncoordinated steps was increased in Nes-Cre;Wt1fl/fl animals (Fig S2 and Videos 3 and 4). We next performed fictive locomotion experiments on isolated spinal cords from control and Nes-Cre;Wt1fl/fl mice (P0–P3). Fictive locomotor drugs induced a markedly slower, disturbed, more variable pattern of locomotor-like activity in Nes-Cre;Wt1fl/fl spinal cords than the stable, rhythmic pattern of locomotor-like activity in control mice. Control spinal cords had recorded activity bursts that showed clear left/right (L2 versus L2) and flexor/extensor (L2 versus L5) alternation that persisted throughout activity periods, whereas activity bursts in Nes-Cre;Wt1fl/fl spinal cords were uncoordinated and did not maintain strict left/right or flexor/extensor alternation (Fig 3A and B). The relationship between left/right and flexor/extensor alternation was examined, and control cords presented a reliable phase preference around 180° (Fig 3C; control average phase preference in l/r: 183.4°, R = 0.93; in f/e: 185.2°, R = 0.84). However, spinal cords from mice with Wt1 deletion showed an irregular locomotor pattern with inconsistent alternation as indicated by its short-phase vector (Fig 3C; Wt1fl/fl average phase preference in l/r: 165.3°, R = 0.60; in f/e: 155.2°, R = 0.44). Although there was no difference in the preferred phase across the two groups (l/r Watson's U2 = 0.10, P > 0.05; f/e Watson's U2 = 0.07, P > 0.05), the coupling strength, or R, as indicated by the vector length in the polar plots, was significantly decreased upon Wt1 deletion (l/r P = 0.031 and f/e P = 0.002, one-tailed Mann–Whitney U test). In addition, the frequency of the ventral root output was decreased (Fig 3D: control; 0.30 ± 0.024 Hz: Nes-Cre;Wt1fl/fl; 0.18 ± 0.08 Hz). This slower rhythm in Nes-Cre;Wt1fl/fl cords could be attributed to altered L2 and L5 activity burst parameters, as Nes-Cre;Wt1fl/fl mice had significantly longer burst, interburst, and cycle periods than control (Fig 3E and 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.
Video shows newborn Wt1fl/fl pup performing air-stepping after tail pinch. Download video
Video shows newborn Nes-Cre;Wt1fl/fl pup performing air-stepping after tail pinch. Download video
Wt1+ neurons receive various synaptic inputs and can project commissurally
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 labelled 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 (Fig 4C). Fluorescent dextran amine retrograde tracing of contralateral projections confirmed that at least part of the Wt1+ dI6 neurons project commissurally (Fig S3). We analyzed spinal cord commissural neurons in control (Fig 4D, left) and homozygous Wt1-mutant (Fig 4D, right) mice (P1–5) to determine whether the deletion of Wt1 alters the total number of commissural neurons and investigated ascending (aCIN), descending (dCIN), and bifurcating (adCIN) subpopulations (Fig 4D and E). All traced subpopulations were markedly reduced in Nes-Cre;Wt1fl/fl spinal cords compared with controls (Fig 4F–H), which suggests that Wt1 is crucial for proper axonal projection pattern.
Loss of Wt1 leads to altered interneuron composition
To assess the possible impact of Wt1 deletion for interneuron development, we analyzed dI6 and non-dI6 populations situated in the embryonic ventral spinal cord. The number of Dmrt3-expressing cells, which constitutes a distinct but partly overlapping dI6 population (Andersson et al, 2012), was significantly decreased in the embryos harboring a loss of Wt1 in the spinal cord already at E12.5 (Fig 5A) persisting throughout development (E16.5 and P1). At any investigated time point, neurons co-expressing both Wt1 and Dmrt3 were not detected in Nes-Cre;Wt1fl/fl embryos and neonates.
Loss of the transcription factor Dbx1 that is involved in differentiation of the V0 population results in a fate switch of some V0 neurons to become dI6 interneuron-like cells (Lanuza et al, 2004). Thus, we investigated whether populations flanking the dI6 population were affected in Nes-Cre;Wt1fl/fl mice. The Lmx1b+ dI5 population was similar in number when comparing Nes-Cre;Wt1fl/fl with wild-type embryos, whereas the number of Evx1+ V0V neurons was significantly increased already at E12.5 (Fig 5B). This increase was still detectable at E16.5. No differences could be seen in Foxp2+ V1 neurons, Chx10 (V2a) and Gata3 (V2b) neurons, and Islet 1/2+ MNs between conditional Wt1 knockout and control embryos at E12.5. However, at E16.5, Chx10+ V2a neurons showed a significant decrease in cell number.
To verify the changes in interneuron composition found in the developing Nes-Cre;Wt1fl/fl mice, we made use of a second mouse line, namely, Lbx1-Cre;Wt1fl/fl mice. At embryonic stage E16.5, we observed a decrease in the amount of dI6 neurons and an increase in the cell number of Evx1+ neurons similar to Nes-Cre;Wt1fl/fl mice (Fig 5C). This decline in the number of dI6 neurons and the concomitant increase in the amount of Evx1+ neurons might point to a change in the developmental fate from dI6 neurons into V0 neurons prompted by the deletion of Wt1. 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 suggest that the fate switch from dI6 neurons into Evx1+ V0 neurons occurs because of the deletion of Wt1. A postnatal phenotypic behavioral analysis of these mice was not possible because neonates died immediately after birth because of 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 V0V neurons that depends on the presence of the cells destined to express Wt1.
The transition of dI6 neurons into Evx1+ V0V neurons upon loss of Wt1 is not direct
To further investigate the cellular fate change upon deletion of Wt1, we combined Wt1-GFP and Nes-Cre;Wt1fl/fl animals to generate Nes-Cre;Wt1fl/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 Wt1fl/GFP control animals at E13.5, whereas GFP, but not Wt1, was detected in the spinal cords of Nes-Cre;Wt1fl/GFP embryos of the same age (Fig 6A). Thus, Nes-Cre;Wt1fl/GFP mice allowed us to inactivate Wt1, whereas the cells destined to express Wt1 are labelled by GFP.
To investigate whether Wt1 deletion leads to apoptosis in the respective cells, TUNEL was used. TUNEL+ cells were present in the ventrolateral spinal cords of Wt1fl/GFP control and Nes-Cre;Wt1fl/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.
To find out whether cells destined to express Wt1 would directly convert to V0V neurons upon Wt1 inactivation, we performed immunohistochemical analyses. The presence of Dmrt3 and Evx1 in GFP+ dI6 neurons was analyzed in Wt1fl/GFP control and Nes-Cre;Wt1fl/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 the 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 Wt1fl/GFP control and Nes-Cre;Wt1fl/GFP animals (below 1% for both). Thus, the increase in the amount of Evx1+ V0V neurons observed in mice lacking Wt1 does not seem to result from a direct transition of future Wt1+ dI6 neurons into Evx1+ V0V 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 during embryogenesis and that deletion of Wt1 results in locomotor aberrancies in neonates and adult mice.
Adult conditional Wt1 knockout animals (Nes-Cre;Wt1fl/fl) show an increased stride length and a decreased stride frequency, resulting in slower absolute walking speed. This supports a possible role of the Wt1+ dI6 neurons in both timing and limitation of the stride cycle. Although the CPG network is capable of producing accurate timing and phasing, proprioceptive and supraspinal input is needed to regulate the CPG activity (Pearson, 2004; Kiehn, 2016; Danner et al, 2017). The integration of this sensory information would require an integrative position in the locomotor CPGs, which is compatible with the observed multisynaptic input to Wt1+ dI6 neurons. However, it still has to be determined whether these various inputs come from proprioceptive sensors, supraspinal regions, or other spinal cord interneurons.
The timing of hindlimb footfalls relative to forelimb footfalls during walking differed between Wt1fl/fl and Nes-Cre;Wt1fl/fl mice, particularly at the diagonal fore- and hindlimbs, suggesting that Wt1 cells play a role in long-range coordination between various spinal cord segments. We could show that at least a fraction of Wt1+ dI6 neurons possesses commissural projections, which is in accordance to very recent published data (Haque et al, 2018). Our results further revealed that deletion of Wt1 leads to a decline in the number of commissural neurons. This suggests not only an involvement of Wt1 in establishing proper projections of the Wt1+ dI6 neurons, but might also explain the changes in the phase coupling between contralateral limbs that might not be related to the V0-based interlimb coordination between fore- and hindlimbs (Talpalar et al, 2013; Danner et al, 2017).
The locomotor alterations observed in adult mice were more subtle than the locomotion abnormalities seen in neonates, which could be due to various compensatory adaptations during postnatal maturation of neuronal circuits. Neonates lacking Wt1 in the spinal cord increased the number of uncoordinated steps, which was supported by a markedly slower and variable pattern of locomotor-like activity in isolated Nes-Cre;Wt1fl/fl spinal cords. They also exhibited a perturbed flexor–extensor and left–right alternation that might be a consequence of the loss of commissural and ipsilateral projections. The data from the fictive locomotion showing that a locomotor rhythm is established when Wt1 is deleted suggested that Wt1+ dI6 neurons are unlikely to participate directly in the kernel of neurons that generate the locomotor rhythm (Dougherty et al, 2013). But because an increase in the variability of bursts during fictive locomotion was observed, we hypothesize that the Wt1+ dI6 neurons are involved in the modulation of this rhythm.
Lack of Wt1 in the spinal cord caused alterations in the differentiation of dI6, V0, and V2a spinal cord neurons during embryogenesis (Fig 6C). The inverse alterations in the dI6 and V0 populations suggest a fate change from dI6 to V0-like neurons when Wt1 is inactivated. The putative transition from dI6 to V0-like neurons occurs at the time point when Wt1 expression would normally start. This instantaneous effect might be due to the derivation of both interneuron populations from neighboring progenitor domains sharing common transcription factors such as Dbx1 (Alaynick et al, 2011). Thus, loss of Wt1 might lead to a switch in developmental programs that are normally repressed; whether this repression occurs cell-autonomously or non–cell-autonomously still has to be determined. In any case, when future Wt1+ cells are ablated, an increase in V0-like neurons is no longer observed, suggesting that the fate switch requires the cells about to express Wt1.
The fate change of prospective dI6 to V0-like neurons is complex because dI6 neurons can be subdivided into at least three subsets based on the expression of the transcription factor–encoding genes Wt1 and Dmrt3 (Fig 6C). Loss of Wt1 affects not only the small number of dI6 neurons that possess Wt1 and Dmrt3 but also the number of neurons that only express Dmrt3. This population is significantly decreased. The presence of Wt1+ dI6 neurons, therefore, is essential to maintain the character of a subset of Dmrt3+ dI6 neurons. If Wt1 is inactivated, in addition to the cells that are programmed to express Wt1, possibly also Dmrt3+ dI6 neurons may differentiate into V0-like neurons.
Two main subpopulations exist within the V0 population (Alaynick et al, 2011): the Evx1+, more ventrally derived V0V, and the Evx1 negative, more dorsally derived V0D population, for which no distinct marker has yet been described. The knockout of Dbx1 results in trans-differentiation of the whole V0 population, whereby Evx1+ V0V neurons acquire a more ventral fate and become V1 neurons, whereas Evx1-negative V0D neurons acquire characteristics of dI6 neurons (Lanuza et al, 2004). This suggests that the V0D, rather than the V0V, neurons closely resemble the dI6 neurons and poses the question whether the fate change from Wt1+ dI6 neurons to Evx1+ V0V-like neurons represents a direct or an indirect transition. The investigations using the Nes-Cre;Wt1fl/GFP mice suggest that the Wt1-deficient dI6 cells do not change their fate directly into Evx1+ V0V-like neurons, suggesting an indirect transition. This points to the possibility that the fate change might be achieved by a transition of Wt1+ dI6 neurons into the more closely related Evx1-negative V0D-like neurons, which leads to a putative increase in the V0D population (Fig 6C). The Evx1+ V0V population might, in turn, increase its number to compensate for a higher proportion of V0D-like neurons.
In addition to the changes in the dI6 and V0 populations 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 populations, which occur already at E12.5. It was reported that V2a neurons directly innervate V0V neurons (Crone et al, 2008). This secondary decrease in the number of these V2a neurons might thus be due to a potential adaptation to the increased number of V0V neurons.
The trans-differentiation of spinal cord neurons observed upon Wt1 knockout might also have an effect on the locomotor phenotype detected in adults and neonates. Excitatory V0V and inhibitory V0D commissural neurons and the ipsilateral excitatory V2a interneurons built up a dual-inhibitory commissural system that is involved in left–right alternation of locomotion (Crone et al, 2008; Talpalar et al, 2013). This dual-inhibitory system works in a speed-dependent manner and allows switching between different gaits. At low speed, V0D neurons are active and cause walking and at higher speed, V2a and V0v neurons become active and cause trotting (Talpalar et al, 2013; Bellardita & Kiehn, 2015; Kiehn, 2016). In Wt1 knockout animals, the trans-differentiation of neurons probably interferes with this system. This can be seen by the left–right perturbation observed in neonates, suggesting that the increase in V0 neurons is not sufficient to compensate for the loss of V2a neurons and commissural projections upon Wt1 deletion. On the other hand, adult Nes-Cre;Wt1fl/fl animals refused to run even when they were forced to move faster by increasing the speed of the treadmill. This suggests that they might have difficulties to run at high speeds and to switch from walk to other running gaits (Bellardita & Kiehn, 2015). However, their ability to change between different gaits still has to be investigated to elucidate the mechanism by which the increase in V0 neurons and the decrease in V2a neurons act in particular on the CPG output.
The approach to ablate a gene to investigate the function of a particular cell population in the spinal network comes with limitations. As already observed for the V0 neurons, inactivating a gene that is crucial for the differentiation of spinal cord neurons might lead to trans-differentiation. This is often accompanied by gain and loss of function of several neuron populations, which can make it challenging to assign distinct functions to a particular neuronal cell type (Lanuza et al, 2004; Talpalar et al, 2013). However, we have chosen the knockout of Wt1 to investigate its role in the differentiation of spinal cord neurons and its influence on locomotion. If the scope was to determine the position of Wt1+ neurons in the locomotor CPG, silencing of the respective neurons would have been a more direct way as very recently performed by Haque et al (2018).. The authors report that acute silencing of these cells revealed their role for appropriate left–right alternation during locomotion (Haque et al, 2018). They also showed that Wt1+ dI6 neurons are inhibitory neurons with contralateral projecting axons terminating in close proximity to other commissural interneuron subtypes. Thus, our data and the results by Haque et al (2018) are complementary. We could not only confirm that Wt1+ dI6 neurons are commissural projecting but were also able to further show that Wt1 is crucial for the formation of these projections. Although the locomotor phenotype of the conditional Wt1 knockout is more diverse than the altered left–right alternation seen in neonates with functionally silenced Wt1+-positive cells, the analyses of the locomotor behavior of adult Wt1 knockout mice revealed a further involvement of Wt1+ dI6 neurons in modulating the gait rhythm. This modulation may be achieved because of an integrative position of the Wt1+ dI6 neurons with multisynaptic input.
In sum, the results obtained in this study shed light not only on the so far undescribed necessity for Wt1 in the development of spinal cord neurons but also on their functional implementation in circuits responsible for locomotion.
Materials and Methods
Mouse husbandry
All mice were bred and maintained in the Animal Facility of the Leibniz Institute on Aging—Fritz Lipmann Institute, Jena, Germany, according to the rules of the German Animal Welfare Law. Sex- and age-matched mice were used. Animals were housed under specific pathogen-free conditions, maintained on a 12-h light/dark cycle, and fed with mouse chow and tap water ad libitum. Mice used for analysis of fictive locomotion and projection tracing were kept according to the local guidelines of Swedish law. Wt1fl/fl mice were maintained on a mixed C57B6/J × 129/Sv strain. Wt1-GFP mice (Hosen et al, 2007) were maintained on a C57B6/J strain. Conditional Wt1 knockout mice were generated by breeding Wt1fl/fl females (Gebeshuber et al, 2013) to Nes-Cre;Wt1fl/fl (Tronche et al, 1999) or Lbx1-Cre;Wt1fl/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 Wt1fl/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 euthanized by CO2 inhalation at specific time points during embryo development and embryos were dissected. Typically, female mice between 2 and 6 mo 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 DTA was incorporated. Before Cre induction, the internal ribosomal entry site (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.
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. The injected blastocysts were reimplanted 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 Wt1 knockin allele.
Immunohistochemistry
Embryonic and postnatal spinal cords were dissected. They were either frozen unfixed after 15-min dehydration with 20% sucrose (in 50% TissueTec/PBS) (postfix) or fixed for 75 min in 4% paraformaldehyde in PBS (prefix). Prefixed tissue was cryoprotected in 10%, 20%, and 30% sucrose (in PBS) before freezing in cryoembedding medium (Neg-50; Thermo Fisher Scientific). Post- and prefixed samples were sectioned (12 μm). Postfixed samples were fixed for 10 min after sectioning and washed with 2% Tween in PBS (PBS-T). For prefixed samples, antigen retrieval was performed by incubation in sub-boiling 10 mM sodium citrate buffer (pH 6.0) for 30 min. After blocking with 10% goat serum and 2% BSA in PBS-T (postfix or prefix), the sections were incubated with primary antibodies (in blocking solution) using the following dilutions: gBhlhb5 1:50 (Santa Cruz Biotechnology, Inc.), BrdU 1:100 (abcam), shChx10 1:100 (abcam), gpDmrt3 1:5,000 (custom made [Andersson et al, 2012]), mEvx1 1:100 (1:3,000 prefix) (Developmental Studies Hybridoma Bank, University of Iowa), chGFP 1:1,000 prefix (abcam), mGFP 1:100 (Santa Cruz Biotechnology, Inc.), rFoxP2 1:800 (abcam), mIslet1/2 1:50 (Developmental Studies Hybridoma Bank, University of Iowa), gpLbx1 1:20,000 (gift from C. Birchmeier, MDC), Lim1/2 1:50 (Developmental Studies Hybridoma Bank, University of Iowa), NeuN 1:500 (Merck), rbPax2 1:50 (Thermo Fisher Scientific), rbLmx1b 1:100 (gift from R. Witzgall, University of Regensburg), and rbWt1 1:100 (Santa Cruz Biotechnology, Inc.). 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 t test.
BrdU injection
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 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, the sections were washed with PBS. Antigen retrieval was performed by incubation in 98°C sub-boiling 10 mM sodium citrate buffer (pH 6.0) for 30 min. After treatment with 2N HCl at 37°C for 30 min, the sections were incubated with primary antibodies using the dilutions mentioned above (see the Immunohistochemistry section). Secondary antibodies were applied according to the species specificity of primary antibodies.
RNA isolation and qRT–PCR analysis
Total RNA was isolated from E12.5 embryonic spinal cords using Trizol (Invitrogen) according to the manufacturer's protocol. Subsequently, 0.5 μg of RNA was reverse transcribed with iScript cDNA synthesis kit (Bio-Rad) and used for quantitative real-time PCR (qRT–PCR). The primer sequences used for RT–PCR analyses are as follows: TGT TAC CAA CTG GGA CGA CA (Act_for); GGG GTG TGG AAG GTC TCA AA (Act_rev); AGT TCC CCA ACC ATT CCT TC (Wt1_qRT_for); TTC AAG CTG GGA GGT CAT TT (Wt1_qRT_rev). Real-time PCR was carried out in triplicates for each sample using SyberGreenER (Thermo Fisher Scientific) and Bio-Rad iCycler (Bio-Rad). PCR efficiencies of primer pairs were calculated by the linear regression method. Ct values were normalized to the mean of the reference gene Actin. Relative expression was determined by comparing normalized Ct values of Wt1 conditional knockout and control samples (Pfaffl et al, 2002). Significance was determined by using pairwise reallocation randomisation test.
Analysis of locomotor behavior
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 the male Wt1fl/fl and Nes-Cre;Wt1fl/fl mice (Wt1fl/fl: 28 g ± 3 g versus Nes-Cre;Wt1fl/fl: 23 g ± 3 g; Fs = 31.98; ts = 3.28, P > 0.001) and moderate differences between the female Wt1fl/fl and Nes-Cre;Wt1fl/fl mice (Wt1fl/fl: 25 g ± 5 g versus Nes-Cre;Wt1fl/fl: 22 g ± 4 g; Fs = 3.80; ts = 1.62, not significant). We recorded the voluntary walking performance of this larger cohort using high-resolution X-ray fluoroscopy (biplanar C-arm fluoroscope Neurostar; Siemens AG). Strides defined as running gait according to the hindlimb duty factor (Hildebrand, 1985; Herbin et al, 2004) occasionally occur in some male Wt1fl/fl mice and were excluded from the analysis. 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; and 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 1,536 dpi × 1,024 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, the durations of stance and swing phases, and the distances that trunk or limb covered during these phases were computed from the landmark coordinates collected at touchdown and liftoff 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 (Fig S1F). 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 and all temporal and spatial gait parameters were then scaled to body size following the formulas published by Hof (1996): nondimensional speed = v/gl0, where v is raw speed, g is gravitational acceleration, and l0 is the cube root of body mass as characteristic linear dimension, which scales isometrically to body mass; nondimensional frequency = f/gl0, where f is raw frequency; and nondimensional stride length = l/l0, where l is raw stride length. The scaled spatiotemporal gait parameters change as a function of nondimensional speed. Therefore, linear regression analyses were computed for each parameter in the male and the female Wt1fl/fl group. The power formulas obtained from regression computation (Y = a + bX) were then used to calculate the expected value for a given nondimensional speed for each gait parameter (baseline) in each animal of all four groups. The coefficient of determination r2 was computed. The deviations of the measured values of Y from the expected values, the residuals, were determined and are given in percent of deviation. Using these residuals, one-way ANOVA was computed to establish the significance of the differences between the means of Wt1fl/fl and Nes-Cre;Wt1fl/fl in males and females. Group means were calculated from the means of 10 animals. Sample size per mouse and limb ranged between 5 and 41 stride cycles, with an average sample size of 22 ± 9.
Fictive locomotion
Animals (P0–P3) were euthanized and the spinal cords eviscerated in ice-cold cutting solution containing (in mM) 130 K-gluconate, 15 KCl, 0.05 EGTA, 20 Hepes, and 25 glucose (pH adjusted to 7.4 by 1M KOH) and then equilibrated in artificial cerebrospinal fluid (Perry et al, 2015) for at least 30 min before the beginning of experimental procedures. Suction electrodes were attached to left and right lumbar (L) ventral roots 2 and 5 (L2 and L5). A combination of NMDA (5 μM) + 5-HT (10 μM) + dopamine (50 μM) were added to the perfusing artificial cerebrospinal fluid to induce stable locomotor-like output. All chemicals were obtained from Sigma-Aldrich. 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 Inc.) and recorded using Axoscope 10.2 (Axon Instruments Inc.) for later off-line analysis. The data were rectified and low-pass filtered using a third-order Butterworth filter with a 5-Hz cutoff frequency before further analysis. Coherence plots between L2 and L2/L5 traces were analyzed using a mortlet wavelet transform in SpinalCore (Version 1.1). Preferential phase alignment across channels are shown in the circular plots and burst parameters were analyzed for at least 20 sequential bursts, as previously described (Kiehn & Kjaerulff, 1996) using an in-house designed program in Matlab (R2014b; MathWorks). Ventral root recording preferential phase alignment was assessed by means of circular statistics from five control cords and seven Nes-Cre;Wt1fl/fl cords (Rayleigh test and Watson's U2 test) for 20 consecutive cycles as described (Kiehn & Kjaerulff, 1996). Burst parameters, including frequency, are presented as the mean ± SD. Burst parameters from five control cords and seven Nes-Cre;Wt1fl/fl cords were compared using the Mann–Whitney U test.
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;Wt1fl/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 L5. Fluorescent dextran amine (FDA, 3,000 MW; Invitrogen) was applied at L1 and rhodamine dextran amine (RDA, 3,000 MW; Invitrogen) was applied between the L4/L5 ventral roots. Spinal cords were incubated overnight at room temperature, subsequently fixed in 4% formaldehyde, and stored in the dark at 4°C until transverse sectioning (60 μm) on a vibratome (Leica).
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). Captured images were auto-leveled using Adobe Photoshop software. Only cords with an intact midline, as assessed during imaging, were used for analysis.
Traced neurons in Wt1fl/fl control and Nes-Cre;Wt1fl/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 ± SEM using 3,975 total cells, 215 sections, and nine spinal cords (Wt1fl/fl control); 3,421 total cells, 228 sections, and seven spinal cords (Nes-Cre;Wt1fl/fl).
TUNEL assay
To detect apoptosis in situ, the TUNEL assay was performed before antibody binding. Slides were incubated with TUNEL reaction solution (1× reaction buffer TdT and 15 U TdT in ddH2O from Thermo Fisher 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). 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) and Volocity software (Improvision). 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 mean ± SEM. Groups were compared using one-way ANOVA or two-tailed two-sample t test depending on the number of groups and sample size. If normal distribution of a sample was not confirmed, sample means are compared by using nonparametric Mann–Whitney U test. All statistical analyses were performed using GraphPad Prism Software (GraphPad Software Inc.), IBM SPSS Statistics 24 (IBM Corporation), Microsoft Excel (Microsoft Corporation), or Matlab (R2014b; MathWorks). Normal distribution was assessed using the D'Agostino-Pearson normality test or Kolmogorov–Smirnov test. Significance was determined as *P < 0.05, **P < 0.01, and ***P < 0.001.
Treadmill gait analyses
This approach involved two groups of mixed sexes of Wt1fl/fl and Nes-Cre;Wt1fl/fl mice (Wt1fl/fl: 10 males and 10 females; Nes-Cre;Wt1fl/fl: 4 males and 8 females; age 24 wk). Locomotor performance was investigated at the German Mouse Clinic—Helmholtz Center, Munich, Germany (www.mouseclinic.de). Treadmill gait analysis was performed with the DigiGait Imaging System (Mouse Specifics, Inc.), which performs ventral plane videography to obtain digital footprints of a mouse walking on a transparent treadmill at different fixed speeds and subsequent analyses of gait patterns using DigiGait software. The DigiGait software determines treadmill contacts of individual paws that were used to quantify spatial (stride length) and temporal indices of gait parameters (stride, stance, and swing time) in walking or running animals. Paw placement of each limb is monitored throughout the gait cycle at up to 150 frames per second with a spatial resolution of more than 5,000 pixels per cm2. For statistical analysis, at least 10 strides for each limb are included in the data set. The mean values for pairs of fore- and hindlimbs were used. Each speed was analyzed separately. Linear regression models using R (version 3.2.3) were used to determine the statistical significance between the groups (RCoreTeam, 2015). Because of strong influence of body weight and body length on the gait performance, those factors are also included into the model to dissect their combined effects on the data.
Air-stepping analysis
P1 mice were used for air-stepping behavioral test according to Andersson et al (2012) (n = 9 Wt1fl/fl control, n = 7 Nes-Cre;Wt1fl/fl) to investigate limb movements in neonates. The hindlimb steps were manually recorded for each animal over 20 s. The number of alternating, synchronous, and uncoordinated hindlimb steps was determined. Each parameter was then statistically tested using the t test. The experimenter was blind to genotype while performing and analyzing experiments.
Supplementary Information
Supplementary Information is available at https://doi.org/10.26508/lsa.201800106.
Acknowledgements
We thank D Kruspe and R Peterson for technical assistance; C. Birchmeier (Max Delbrück Center for Molecular Medicine, Berlin, Germany) for providing the Lbx1-Cre mouse line; and C Hübner, H Heuer, and G Zimmer for critical discussion. This project was supported by grants from the German Federal Ministry of Education and Research (Infrafrontier grant 01KX1012) to L Becker and the Swedish Medical Research Council, Hållsten, Ländells, Swedish Brain Foundations to K Kullander. D Schnerwitzki received a scholarship from the Leibniz Graduate School on Ageing and Age-Related Diseases. FV Caixeta was funded by a scholarship from CNPqi–Brazil. The Fritz Lipmann Institute is a member of the Leibniz Association and is financially supported by the Federal Government of Germany and the State of Thuringia.
Author Contributions
D Schnerwitzki: conceptualization, formal analysis, investigation, visualization, and writing—original draft, review, and editing.
S Perry: investigation, visualization, and writing—original draft, review, and editing.
A Ivanova: conceptualization and investigation.
F Caixeta: formal analysis, methodology, and writing—review and editing.
P Cramer: investigation and visualization.
S Guenther: investigation.
K Weber: investigation.
A Tafreshiha: investigation.
L Becker: formal analysis, supervision, and visualization.
I Vargas Panesso: investigation.
T Klopstock: supervision and project administration.
M Hrabě de Angelis: supervision and project administration.
M Schmidt: conceptualization, resources, formal analysis, supervision, investigation, visualization, methodology, and writing—original draft, review, and editing.
K Kullander: conceptualization, supervision, project administration, and writing—review and editing.
C Englert: conceptualization, supervision, funding acquisition, project administration, and writing—original draft, review, and editing.
Conflict of Interest Statement
The authors declare that they have no conflict of interest.
- Received June 12, 2018.
- Revision received August 9, 2018.
- Accepted August 10, 2018.
- © 2018 Schnerwitzki et al.
This article is available under a Creative Commons License (Attribution 4.0 International, as described at https://creativecommons.org/licenses/by/4.0/).