Spinal motoneuron firing properties mature from rostral to caudal during postnatal development of the mouse

Key points Many mammals are born with immature motor systems that develop through a critical period of postnatal development. In rodents, postnatal maturation of movement occurs from rostral to caudal, correlating with maturation of descending supraspinal and local spinal circuits. We asked whether development of fundamental electrophysiological properties of spinal motoneurons follows the same rostro‐caudal sequence. We show that in both regions, repetitive firing parameters increase and excitability decreases with development; however, these characteristics mature earlier in cervical motoneurons. We suggest that in addition to autonomous mechanisms, motoneuron development depends on activity resulting from their circuit milieu. Abstract Altricial mammals are born with immature nervous systems comprised of circuits that do not yet have the neuronal properties and connectivity required to produce future behaviours. During the critical period of postnatal development, neuronal properties are tuned to participate in functional circuits. In rodents, cervical motoneurons are born prior to lumbar motoneurons, and spinal cord development follows a sequential rostro‐caudal pattern. Here we asked whether birth order is reflected in the postnatal development of electrophysiological properties. We show that motoneurons of both regions have similar properties at birth and follow the same developmental profile, with maximal firing increasing and excitability decreasing into the third postnatal week. However, these maturative processes occur in cervical motoneurons prior to lumbar motoneurons, correlating with the maturation of premotor descending and local spinal systems. These results suggest that motoneuron properties do not mature by cell autonomous mechanisms alone, but also depend on developing premotor circuits.


Introduction
Altricial mammals are born with immature nervous systems comprised of circuits that have neither the neuronal properties nor connectivity required to produce future behaviours. During the critical period of postnatal development, neuronal properties are tuned to participate in functional circuits. The degree to which these mature electrophysiological properties are determined by cell-intrinsic vs. extrinsic (e.g. circuit) factors is not clear. To understand how neural circuits are ultimately tuned requires an understanding of how properties of their component neurons develop.
A model system in which to understand this is the spinal cord, where two distinct regions -the cervical and the lumbar cord -include homologous circuits, with both regions containing circuitry to coordinate the intra-and inter-limb movements required for locomotion (Jessell, 2000a;Goulding, 2009). The output from each region is mediated by motoneurons, which have overlapping molecular profiles in the two regions. During embryonic development, cervical motoneurons are born a few days before lumbar motoneurons in mice, rats and chicks (Nornes & Das, 1974;Hollyday & Hamburger, 1977), and these regions develop sequentially from rostral to caudal (Sagner & Briscoe, 2019). Is this developmental trajectory maintained in the postnatal critical period such that the electrophysiological properties of cervical motoneurons are more mature at birth and then fully mature before lumbar motoneurons? Or do the development and maturation of these properties rely on the development of circuits, such as descending and sensory inputs; in which case the two populations would be born with similar properties that mature as movement develops during the postnatal critical period?
There is little question that both behaviour and motoneuron properties develop postnatally. Rodents are essentially immobile in the first 2 days following birth, until forelimb-propelled pivoting and crawling become the dominant forms of ambulation during the first week of life (Altman & Sudarshan, 1975;Sechzer et al. 1984). Quadrupedal locomotion subsequently emerges around P10-12 when hindlimbs consistently support the weight of the lower quadrant (Jiang et al. 1999), with locomotor maturity reached by the end of the third postnatal week (Altman & Sudarshan, 1975). Lumbar motoneuron properties change at least in the first postnatal week, the period (and region) in which electrophysiological experiments are usually done (Fulton & Walton, 1986;Nakanishi & Whelan, 2010;Quinlan et al. 2011). But do these changes in properties proceed from rostral to caudal?
Many aspects of spinal circuit development ensue from rostral to caudal. Descending supraspinal pathways arrive and mature in cervical segments earlier than in lumbar segments. In rodents, fibres originating in the brainstem are the first to arrive in the spinal cord, sprouting into the grey matter of the cervical cord prenatally, and reaching the lumbar spinal segments at or shortly after birth (Vinay et al. 2000). Descending modulatory systems follow a similar pattern: serotonergic innervation of the cervical spinal cord displays the adult profile by P14, whereas the fibre density in the lumbar cord does not mature until P21 (Bregman, 1987). Similarly, corticofugal axons innervate the cervical grey matter at P5-6 and arrive at the lumbar cord at approximately P9. But it is not until after P21 that all segments display the mature density and pattern of descending innervation (Donatelle, 1977;Gianino et al. 1999). Sensory innervation of the spinal cord follows a similar trend, as cutaneous reflexes can be evoked in the muscles of the neck and forelimbs (E16-17) before those in the hindlimbs (E17-18) (Narayanan et al. 1971). Furthermore, postnatal refinement of proprioceptive afferent input to motoneurons is dependent upon the maturation of descending input to the spinal cord and does not mature until the end of the third postnatal week in rats (Smith et al. 2017). Thus, key spinal circuit development occurs in the first three postnatal weeks and, where studied, proceeds from the cervical to the lumbar spinal cord.
Do motoneuron properties follow this same pattern? In addition to being born earlier, anatomical data indicate that cervical motoneurons reach adult motoneuron size before the lumbar motoneurons (Cameron et al. 1989). In rats, the appearance of spontaneous burst activity in embryonic cervical motoneurons precedes that in lumbar motoneurons by about one embryonic day (Nakayama et al. 1999). Importantly, it is clear that the transcriptional profile rather than circuit milieu (and limb movement) is sufficient for basic motoneuron properties to develop: motoneurons derived in culture from embryonic stem cells develop electrophysiological properties characteristic of spinal motoneurons (Miles et al. 2004).
We thus reasoned that, not only would motoneuron properties develop in the early postnatal period, but that if transcriptional profile alone were responsible for this maturation, then cervical motoneurons would be more mature at the time of birth and reach adult-like properties earlier than lumbar motoneurons. On the other hand, if circuit properties and limb movement are critical for maturation, then cervical and lumbar motoneurons would be born with similar properties that then mature postnatally. To study this question, we used whole-cell patch clamp recordings of mouse cervical and lumbar motoneurons to define and compare their electrophysiological properties at three time points during this critical period of development from birth to weaning (P2-3, P6-7 and P14-21). We demonstrate that the two populations are born with similar properties and then develop through the third postnatal week. Furthermore, cervical motoneuron properties mature earlier than those of lumbar motoneurons. We thus suggest that transcriptional profiles alone are insufficient for this maturation, and that the postnatal development of circuits contributes to the tuning of neuronal properties.

Confirmation of compliance
The authors confirm that they understand, and that this work complies with, the ethical principles under which the Journal of Physiology operates (Grundy, 2015).

Ethical approval
Experiments were approved by the University College London Animal Welfare and Ethical Review Body and performed under a licence granted under the Home Office Animals (Scientific Procedures) Act 1986.

Spinal cord isolation
Lumbar spinal cords: Animals were deeply anaesthetized by intraperitoneal injection of a ketamine (100 mg kg -1 ) and xylazine (20 mg kg -1 ) mixture. Upon loss of paw withdrawal, mice were then decapitated and the vertebral column was quickly isolated and pinned down (ventral side up) in a dissecting dish containing ice-cold (0-4°C) normal artificial cerebrospinal fluid (nACSF) saturated with 95% carbogen. The composition of the nACSF was as follows (in mM): 113 NaCl, 3 KCL, 25 NaHCO 3 , 1NaH 2 PO 4 , 2 CaCl, 2 MgCl 2 and 11 D-glucose, pH 7.4 (Mitra & Brownstone, 2012). The vertebral bodies were removed to reveal the spinal cord, the roots were cut and dura matter removed. The spinal cord was isolated from rostral thoracic to caudal sacral levels, placed upon tissue paper to soak up excess liquid and then the dorsal side glued (3M Vetbond, No.1469SB) to a pre-cut block of agarose mounted on a cutting chuck.
Cervical spinal cords: For cervical slices (also under deep anaesthesia), a craniotomy was done to expose and remove the cerebellum, and the brainstem was transected at the level of the obex. All nervous tissue rostral to the transection was immediately removed. This process was critical for preserving cervical spinal tissue. The vertebral column was then transferred to a dissecting dish (as with the lumbar preparation), pinned dorsal side up and a laminectomy performed to a mid-thoracic level. Roots were cut, the dura removed and cervical cord glued to the chuck in the same fashion as for the lumbar cord.
Large eGFP-positive neurons in the ventral lumbar and cervical spinal cord were identified as motoneurons (Wilson et al. 2005). Motoneurons were patched at room temperature (approx. 21°C) using infrared-differential interference contrast (IR-DIC) optics on a DMLFSA microscope (Leica DMLFSA; Leica Microsystems, Wetzlar, Germany). All lumbar motoneurons were selected from the lateral motor nuclei of segments L1-6 and all cervical cells were selected from the lateral motor nuclei of slices made from C4-8. It is therefore likely that the majority (if not all) of cells included in our study were limb-innervating motoneurons (Watson et al. 2009;Mohan et al. 2014;Mohan et al. 2015).

Data analysis
All data were captured and analysed with CED Signal software (RRID:SCR 01 7081). Motoneurons were only analysed if they had a resting membrane potential (RMP) more hyperpolarized than -60 mV that did not deviate by more than 5 mV during recording.
All experiments were performed in current clamp mode. The bridge was balanced and capacitance neutralized prior to commencing recording. Motoneurons were injected with a small negative rectangular pulse (500 ms duration) and the voltage response of 15-30 traces was averaged to measure input resistance and whole-cell capacitance (WCC). Resistance was measured as the peak voltage change to the injected current and τ calculated from an exponential curve fitted to the response (automated in Signal). WCC was calculated using resistance and τ values and cross-checked against the values automatically recorded by the software during the experiment. I min was defined as the minimum amount of current needed to evoke ࣙ 2 action potentials (APs). This was tested with 500 ms duration, rectangular current pulses of increasing magnitude (0.03 nA steps) starting at -0.3 nA. Sag potentials were recorded by injecting 0.03 nA hyperpolarizing current steps (500 ms) from 0 to -1 nA. The sag amplitude was measured as the difference between the peak of the negative voltage deflection at the start of the 500 ms pulse and the steady state at the end. Motoneuron f-I graphs were generated by injecting depolarizing current steps increasing from 0 nA until maximum firing was observed. The excitability of the cell (gain) was determined by measuring the slope of the linear portion of f-I plots for spike number, initial frequency (first two spikes instantaneous frequency), and final frequency (steady state, final two spikes instantaneous frequency).
Action potential half widths (AP HWs), spike amplitude, maximum rate of depolarization/repolarization and fast afterhyperpolarization (fAHP) were measured from 15-30 averaged single APs evoked with a 20 ms rectangular current pulse. The AP HW was calculated as the time between the 50% rise and 50% fall in amplitude of the AP. Spike height was measured as the voltage difference between the threshold (voltage at maximum positive value of the second derivative of membrane potential of AP) and the peak of the AP. The fAHP was measured as the difference between the voltage baseline and the most negative point on the first trough of the AP. Phase-plane plots of the single APs were generated to calculate maximum rates of depolarization and repolarization. Afterpotential measurements (mAHP, mAHP half decay time and afterdepolarization (ADP) amplitude) were taken from single APs evoked with a 1 ms duration current pulse. The mAHP amplitude was calculated from baseline (held at -65 mV) to the most negative point on the trough. The mAHP half decay time is calculated as half the time taken (ms) from the most negative point of the mAHP to baseline. ADP amplitude was measured from the fAHP peak to the most positive value before the start of the mAHP repolarization. All cells were held at -65 mV for single AP experiments.

Statistics
Data from all cells were initially exported to a Microsoft Excel file. All subsequent data processing and analysis was done using Python (RRID:SCR 0 08394) running through the Jupyter notebooks (RRID:SCR 01 8413) environment. All data processing, graphing and statistical procedures can be found, edited and re-run in the 'myBinder' link below (see Data deposition section). Given the diversity of motoneurons in the spinal cord and in line with the three Rs for animal research, measures from individual cells were treated as the experimental unit (N).
There were three questions we wanted to ask of our data, which advised our statistical procedures. They were: 1. Is there sustained change in an electrophysiological parameter over development? 2. If there is a change, is that change localized to a particular time point? 3. If it is localized, when does that change occur?
To answer these questions we took the following approach: 1. Most of our parameters were expected to change monophasically throughout development. We therefore decided to use Spearman's correlation coefficient to determine the effect of age within each nominal class (lumbar and cervical). For sag slope, which showed a V shape development profile, we used a two-way ANOVA with the two independent variables being age and region and the dependent variable sag slope. 2-3. Subsequently, in order to determine if the change was localized to a particular time point (2) and when that is (3), we performed corrected multiple pairwise comparisons using either Student's t tests (Gaussian distribution) or Mann-Whitney U tests (non-Gaussian distribution). Distribution was assessed using the Shapiro-Wilk test. The Holm-Sidak method was used to correct P values for multiple comparisons.
Following two-way ANOVAs we used the Tukey method for pairwise comparisons.
All data are reported as means ± standard deviations. Graphs were produced using Seaborn and Python open-source software. Split violin plots with individual observations (grey lines) and means (red) were used to show and compare the distribution of the data. Joint-plots are based on the means of the data and were included to give a clear comparison of the developmental profile between cervical and lumbar regions. Final figures were produced using CorelDraw Home & Student X8 software (RRID:SCR 01 4235).

Principal component analysis
Principal component analysis was performed in python using the Scikit-learn package (Pedregosa et al. 2011). All data were scaled and fit-transformed before analysis (see myBinder link below for full code).

Results
To effect movement, motoneurons must have the machinery to integrate synaptic inputs and convert them to trains of APs of appropriate frequencies for the intended behaviour. The ability to do this processing arises from a combination of passive, transition and repetitive firing properties. Here, we report our findings on postnatal development in each of these categories, comparing limb-innervating motoneurons within the cervical and lumbar spinal cord. For each measure (except sag slope) we first report whether there is a sustained change with age for cervical and lumbar motoneurons based on the results of a Spearman's rank test. If a significant correlation is found, we proceed to test for normality (Shapiro-Wilk) and then perform corrected pairwise comparisons using t tests or Mann-Whitney U tests between six groups (lumbar and cervical, P2-3, P6-7 and P14-21).
At P2-3 no differences between cervical and lumbar motoneurons were observed in AP HW (p = 0.34), or maximum rate of depolarization (p = 0.32) or repolarization (p = 0.55). At P6-7, cervical motoneurons had shorter duration APs than lumbar motoneurons mainly due to a higher rate of repolarization (AP HW: p = 0.007, maximum rate of depolarization: p = 0.071 and maximum rate of repolarization: p = 0.023). There was no difference between regions in the P14-21 group for AP HW (p = 0.151), despite a higher maximum rate of repolarization in cervical motoneurons: p = 0.035). Overall, these results suggest significant changes in the AP HW, with cervical motoneurons developing earlier than lumbar motoneurons.
Because spike amplitude is measured as the difference between threshold and peak, threshold changes could contribute to changes in amplitude. However, we found no correlation between age and AP threshold in either region (lumbar: Spearman's ρ = 0.062, p = 0.65; cervical: Spearman's ρ = 0.148, p = 0.29; Fig. 2E, e, G). The apparent lack of developmental change for AP threshold suggests that development of Na + channels may not contribute as much as later activated channels to the development of spike morphology.

Figure 3. Postnatal development of afterpotentials
A-B, split violin plots showing the changes in mAHP amplitude and mAHP half decay time at each age for lumbar (white) and cervical (grey) motoneurons. The mAHP amplitude was measured from baseline (all cells held at -65 mV) to the most negative point of the mAHP. Half decay time was measured from the mAHP peak to baseline. C, strip-plots showing development of ADP amplitude with age. D, averaged trace (15-30 sweeps) of the action potential evoked with a 1 ms square current pulse for a representative motoneuron (cervical) at P3 (black) and P19 (grey). Action potential rising phases are truncated. ADP amplitude was measured from the most negative point of the fAHP to the peak of the ADP; see inset for expanded ADPs. (a-c) Joint-plots illustrating the developmental profile of each measure. Overlaid white dashed lines represent statistically significant changes between age groups. Dashed circles represent statistically significant difference between P2-3 and P14-21. * represents a statistically significant difference (p < 0.05) between regions for a particular age group. Mann-Whitney U tests with Holm-Sidak corrected P values used for pairwise comparisons. were no differences in ADP amplitude between lumbar and cervical motoneurons at P2-3 (p = 0.58), P6-7 (p = 0.16), or P14-21 (p = 0.58).
Given the importance of ADPs in generating high initial firing frequencies and doublets, these data suggest that development of ADPs enables motoneurons to fire at higher initial frequencies as they mature. Furthermore, this increase occurred earlier in cervical motoneurons, suggesting that development of high initial firing frequencies would develop earlier than in lumbar motoneurons (see below).

Postnatal development of motoneuron firing properties: minimum current for repetitive firing (I min )
For function, motoneurons must fire repetitive trains of APs. The frequencies of firing are graded, with increased synaptic (or injected) current leading to higher rates of firing. By plotting the frequency of firing vs. the current injected, various key measures can be quantified, including the minimum current needed for repetitive firing (I min ), the maximum firing rate obtainable (F max ) and the slopes for initial, steady state, and overall relationships.
Comparisons between cervical and lumbar motoneurons at P2-3 showed no differences for any of the maximum firing measurements (number of spikes per 500 ms current pulse, p = 0.73, maximum initial frequency, p = 0.52; maximum final frequency, p = 0.71). At P14-21, cervical motoneurons had greater maximum initial frequencies (p = 0.025), but maximum spike output (p = 0.11) and final firing frequency (p = 0.24) were not different between regions. However, at P6-7 cervical motoneuron values were greater than lumbar motoneurons for all measures (number of spikes per 500 ms current pulse: p = 0.003; maximum initial frequency: p = 0.001; maximum final frequency: p = 0.008). Thus, repetitive firing parameters seem to mature earlier in cervical than in lumbar motoneurons.
Overall, motoneuron repetitive firing function matured over postnatal development, with cervical motoneurons increasing their firing capacity earlier than lumbar motoneurons. These results are consistent with the developmental profile of AP characteristics described above.
In summary, for almost all measures of repetitive firing, it can be seen that at P2-3 lumbar and cervical motoneurons are the same, but there is a reduction in excitability for cervical motoneurons at P6-7 (Fig. 6a-c). In lumbar motoneurons, this reduction is seen later, between P6-7 and P14-21.

Figure 5. Postnatal development of repetitive firing
A-B, violin plots showing the minimum current required for repetitive firing and maximum number of spikes in lumbar (white) and cervical (grey) motoneurons at each age group. C, representative traces of maximum firing of neurons from each age group, illustrating the increased ability of motoneurons to produce high frequency trains of action potentials as they mature. Depolarizing current pulses (500 ms) of increasing magnitude were injected until the cell reached its maximum firing frequency or depolarizing block. D, max initial firing frequency was calculated from the first two action potentials at maximum firing rate. E, the final firing frequency was calculated from the last two action potentials at maximum firing rates. (a-e) Joint-plots of means illustrating the developmental profile for each measure. Overlaid white dashed lines represent statistically significant changes between age groups. Dashed circles represent statistically significant difference between P2-3 and P14-21. * represents a statistically significant difference (p = <0.05) between regions for a particular age group. Mann-Whitney U (a, b, e) or t tests (
In summary, these data further support suggestions from analyses of individual firing characteristics that lumbar and cervical motoneurons are similar at birth and undergo significant postnatal development. Importantly, they also provide further evidence for earlier postnatal development of cervical compared with lumbar motoneurons.

Discussion
Altricial mammals are born in an immature state, and their nervous systems and musculoskeletal systems must develop such that there is sufficient motor independence prior to the time of weaning. These two systems, which develop hand in hand, are connected by motoneurons and proprioceptive afferents. Here, we ask how the properties of motoneurons develop during this period, and ask whether the evidence supports cell autonomous or circuit factors as being the leading instigators for this development. To do this, we compare lumbar and cervical spinal motoneuron properties through this postnatal period, examining passive, transition and repetitive firing properties from a stage where the animal is virtually immobile (P2-3) to the point when motor function matures, just prior to weaning (P14-21). We show that there is ongoing development of both cervical and lumbar motoneuron properties throughout this period and that despite the fact that cervical motoneurons are embryonically born prior to lumbar motoneurons (Nornes & Das, 1974), their properties at birth are similar. In the first 3 weeks of postnatal life, the properties of both sets mature, but those of cervical motoneurons develop earlier than those of lumbar motoneurons. This development correlates with the arrival of descending axons, which is known to be crucial to the maturation of spinal sensorimotor circuits. We therefore suggest that maturation of these circuits contributes to development of motoneuron firing properties.

The critical period: postnatal development of motoneuron properties continues into the third postnatal week
Weight bearing and fundamental aspects of motor control are established to a degree by P10-12, but it is clear that motor functional output in rodents does not mature until the third postnatal week (Altman & Sudarshan, 1975). Indeed, anatomical studies show that developmental organization of membrane proteins and synaptic input to spinal motoneurons begin to reach maturity in the third week postnatally in rodents (Wilson et al. 2004;Jean-Xavier et al. 2017;Smith et al. 2017). A profile of changes in passive properties and firing characteristics of motoneurons has been described for rodent hypoglossal and genioglossal motoneurons from birth to adulthood, but equivalent studies in the spinal cord have been limited to ages younger than P12 (Fulton & Walton, 1986;Nunez-Abades et al. 1993;Berger et al. 1996;Nakanishi & Whelan, 2010;Quinlan et al. 2011). Due to the ready decline in motoneuronal viability in slices after P10, early neonatal spinal cord preparations have become the dominant tool to study motoneuron physiology, and therefore much of our knowledge is limited to the first postnatal week in rodents and to the adult in cats (Kernell, 1999). Our recordings from spinal motoneurons in the cervical and lumbar spinal cord up to P21 confirm that maturation of electrophysiological properties continues into the third postnatal week, and demonstrates the importance of using older preparations to study neuronal properties.
Changes in motoneuron properties are dependent upon the expression and function of many membrane proteins. Although we did not study specific motoneuron ion channels, our results indicate that many channels undergo developmental changes in this critical period. During this time, WCC increases and input resistance is reduced, with a corresponding increase in I min . These changes are expected with growth of the motoneuron soma and dendritic tree, but also indicate that there is maturation of conductances that are active near resting potential (Fleshman et al. 1981;Vinay et al. 2000) . Analysis of AP morphology shows reduced half width due to increased rates of both depolarization and repolarization phases, suggesting that expression of Na + (Barrett & Crill, 1980), Ca 2+ (Hounsgaard & Mintz, 1988;Mynlieff & Beam, 1992;Viana et al. 1993a), and K + channels undergoes maturation (Viana et al. 1993b). Thus, widespread changes to the motoneuron membrane occur during this period.

Maturation of repetitive firing
A motoneuron's raison d'être is to ensure that muscle fibres contract appropriately to produce the behaviour, and to do so the motoneuron must be able to fire graded trains of APs. The maturation of spike properties, including increasing amplitude and reduced half width resulting from increasing rates of depolarization and repolarization, can facilitate faster firing (McCormick et al. 1985). There is also an increase in the expression and amplitude of the ADP, a potential likely dependent upon high voltage activated Ca 2+ channels (Granit et al. 1963b;Kobayashi et al. 1997;Vinay et al. 2000). ADPs promote doublet firing in motoneurons, significantly increasing the rate and magnitude of muscle force generation (Parmiggiani & Stein, 1981;Sandercock & Heckman, 1997). The increase in ADP that we report here may underlie the increase of maximum initial firing frequency across this developmental period. These changes in APs and ADPs thus likely underpin the changes in motoneuron repetitive firing capabilities.
The mAHP is also an important factor in regulating the frequency of firing, or interspike interval (Granit et al. 1963a;Kernell & Monster, 1982;Bean, 2007;Deardorff et al. 2013). This afterpotential is mediated by small conductance Ca 2+ -activated potassium channels (SK2 & SK3; Deardorff et al. 2013). We report a small developmental change in the mAHP amplitude in cervical, but not lumbar, motoneurons. Although greater changes in the mAHP might have been expected based upon the reported development in hypoglossal nuclei (Viana et al. 1994;Berger et al. 1996), our results are largely consistent with previous work in young spinal cord slices and brainstem slices throughout development (Carrascal et al. 2005;Nakanishi & Whelan, 2010;Quinlan et al. 2011).
The ability of motoneurons to fire trains of APs at high frequencies increased with development, while excitability, as measured by the gain of f-I plots, decreased. This reduction in f-I slope is likely related to the change in passive properties such as size of the soma and dendritic tree , and active properties such as potassium channels associated with spike repolarization (Gao & Ziskind-Conhaim, 1998;Martin-Caraballo & Greer, 2000). The combination of higher F max and reduced f-I slope leads to a broader range of input currents to which the motoneuron responds, an increase in signal-to-noise ratio of the response, and an enhancement in the tunability of motoneuron firing rates. It is perhaps intuitive that these properties continue developing throughout the third postnatal week and beyond, as the speed, force, and control of motor output continues to mature across this period (Altman & Sudarshan, 1975).

Possibility of selection bias?
Could some of the differences in properties that we report result from selection bias? For example, SK2/3 expression and thus mAHP characteristics are different in different motoneuron types (i.e. those innervating fast vs. slow twitch muscle fibres, which correspond to large vs. small motoneurons (Deardorff et al. 2013). While in older preparations there is an inherent bias to record smaller motoneurons (because they survive; see Mitra & Brownstone (2012) where average input resistance was 123 M ), we were clearly not biased to smaller motoneurons: the mean input resistances we report in the oldest age group were approximately 30 M . In fact, in this study, most motoneurons at P6-7 were under 50 M and at P2-3 most were under 100 M , both in the range of larger motoneurons. Furthermore, it seems that GFP expression decreases in Hb9::eGFP mice during this time period, persisting primarily in large motoneurons (and possibly even a specific subset of these). It is thus likely that we sampled the largest motoneurons in each age group. But there is also no doubt that motoneurons are growing during this period, and there is no way, at this point in time, to target neurons in younger mice that may be "destined" to grow up to be large motoneurons. That is, in order to reliably track the development of properties such as the mAHP, motoneuron types must be identified throughout development, which is a difficult proposition considering that motoneurons undergo significant diversification during postnatal development (Navarrete & Vrbová, 1993;Kanning et al. 2010).

Possible factors leading to advanced maturation of cervical vs. lumbar motoneurons
Sequential rostro-caudal development of many components of the nervous system has been observed both pre-and postnatally. During embryonic development, spinal motoneurons are born first in the cervical spinal cord, and then sequentially in the more caudal segments (Nornes & Das, 1974). This could mean that delayed postnatal maturation of motoneuron firing characteristics in the lumbar cord may be set from the point of neurogenesis. We find this unlikely to be the case, however, as there are few differences between diverse sets of motoneuron properties in cervical vs. lumbar regions at P2-3 (see, e.g. Fig. 7). However, by P6-7, cervical motoneurons had more mature properties than lumbar motoneurons. We therefore propose an alternative hypothesis suggesting that regional differences in states of motor circuit maturity and activity patterns during postnatal development underlie the differences seen between cervical and lumbar motoneuron firing characteristics.
The delayed arrival and maturation (including myelination and synaptic refinement) of supraspinal descending systems is perhaps the most obvious difference between the two regions during postnatal development (Donatelle, 1977;Gianino et al. 1999;Vinay et al. 2005). And following arrival of these systems, their termination patterns as well as their synaptic effects on spinal motoneurons also mature -with cervical effects thus preceding lumbar changes (Commissiong, 1983;Tanaka et al. 1992;Floeter & Lev-Tov, 1993;Brocard et al. 1999b). These developmental changes are parallel to behavioural changes, as can be seen in the development of postural control (Skoglund, 1960;Brocard et al. 1999a), which occurs in a proximo-distal gradient corresponding to the more caudal location of motor nuclei innervating distal muscles (Nicolopoulos-Stournaras & Iles, 1983). Furthermore, normal development of sensory afferents and spinal premotor circuits are dependent upon descending innervation (Chakrabarty & Martin, 2011b;Smith et al. 2017), and these afferents themselves may affect motoneuron maturation (Woolley et al. 1999). The importance of these systems on motoneuron properties has been seen when descending tracts are prevented from growing: there is disruption of the development of inhibitory systems such as Renshaw cells and GABA-pre circuits, and motoneuron hyperexcitability emerges (Martin, 2005;Chakrabarty et al. 2009;Chakrabarty & Martin, 2011a, b;Basaldella et al. 2015;Smith et al. 2017). We thus propose that while cell autonomous factors may play an important role, the differences in the states of maturation of cervical vs. lumbar motoneurons largely result from differences in the timing of local sensorimotor circuit development in the two regions, which is affected by innervation by descending systems.

Conclusion
We show that between birth and weaning, cervical motoneuron properties mature earlier than those of lumbar motoneurons. We suggest that this maturation results from the development of circuits that engage these motoneurons. It has been shown that the properties of many different neuronal types mature in this postnatal period, including thalamocortical neurons (Warren & Jones, 1997), somatosensory cortical inhibitory neurons (Kinnischtzke et al. 2012), medial prefrontal cortex pyramidal neurons (Favuzzi et al. 2019), and auditory cortex pyramidal neurons (Oswald & Reyes, 2008), amongst others. While of course cell autonomous factors such as transcription factor expression respond to environmental cues such as morphogens to govern cell fate, including the fate to become a motoneuron (Jessell, 2000b;Dasen et al. 2008), the degree to which these molecules set a maturating course is not clear (Harb et al. 2016). In fact, many fundamental motoneuron properties can develop in motoneurons derived in a dish from embryonic stem cells (Miles et al. 2004), demonstrating the strength of intrinsic programmes in determining electrophysiological properties. From our study of geographically separated motoneurons, however, we argue that circuit factors play an important role in physiological maturation. That is, it could be argued that cell autonomous factors are sufficient for the acquisition of repetitive firing, and that motor circuit development and activity contribute to maturation. It seems likely that the combination of cell autonomous factors and circuit development are required to ultimately produce a mature, functional neuron.