Electrical maturation of spinal neurons in the human fetus: comparison of ventral and dorsal horn

Abstract

The spinal cord is critical for modifying and relaying sensory information to, and motor commands from, higher centers in the central nervous system to initiate and maintain contextually relevant locomotor responses. Our understanding of how spinal sensorimotor circuits are established during in utero development is based largely on studies in rodents. In contrast, there is little functional data on the development of sensory and motor systems in humans. Here, we use patch-clamp electrophysiology to examine the development of neuronal excitability in human fetal spinal cords (10–18 wk gestation; WG). Transverse spinal cord slices (300 μm thick) were prepared, and recordings were made, from visualized neurons in either the ventral (VH) or dorsal horn (DH) at 32°C. Action potentials (APs) could be elicited in VH neurons throughout the period examined, but only after 16 WG in DH neurons. At this age, VH neurons discharged multiple APs, whereas most DH neurons discharged single APs. In addition, at 16–18 WG, VH neurons also displayed larger AP and after-hyperpolarization amplitudes than DH neurons. Between 10 and 18 WG, the intrinsic properties of VH neurons changed markedly, with input resistance decreasing and AP and after-hyperpolarization amplitudes increasing. These findings are consistent with the hypothesis that VH motor circuitry matures more rapidly than the DH circuits that are involved in processing tactile and nociceptive information.

the ability to interact appropriately with our environment arises, in part, from the capacity of the spinal cord to receive, modify, and relay sensory information and respond to this information via motor commands that produce coordinated movement. Establishing the underlying circuits requires a complex developmental plan that is first laid down in the embryo, consolidated in the fetus, and modified throughout life (Fitzgerald 2005; Petersson et al. 2003). Moreover, when this complex developmental plan is interrupted, it can result in adverse consequences for sensorimotor processing in the adult (Baccei 2010; Walker et al. 2009; Zouikr et al. 2014). Our understanding of the important developmental changes that occur in the early stages of spinal cord development come largely from studies in rodents. Collectively, these studies have demonstrated that primary afferents enter the spinal cord to make a myriad of connections with spinal neurons and then undergo substantial fine-tuning to generate a spatiotemporal map of the body within the spinal cord (Granmo et al. 2008; Petersson et al. 2004). Importantly, the intrinsic properties of these spinal neurons are also maturing during the late embryonic and early postnatal stages of development (Baccei 2014; Vinay et al. 2000b; Walsh et al. 2009). Few studies have investigated the functional development of neurons in spinal sensorimotor pathways in humans.

In rodents, specific neuronal subtypes can be identified in the spinal cord by their molecular signatures from approximately embryonic day (E) 10 (Guerout et al. 2014), and electrical activity can be observed soon after. Electrical activity is thought to exist in the embryonic spinal cords of mice at E12, based on recordings of spontaneous rhythmic activity in the sciatic nerve (Hanson and Landmesser 2003). Similarly, in the rat ventral horn (VH), presumptive motoneurons exhibit glycinergic postsynaptic currents from as early as E12.5 (Scain et al. 2010) and discharge antidromic action potentials (APs) by E18 (Di Pasquale et al. 1996). This occurs later in the dorsal horn (DH), where small and very broad APs have been recorded in neurons from E15 mice (Walsh et al. 2009). Taken together, these data suggest that functional development of spinal neurons in rodents occurs in the latter half of in utero development.

The capacity of human fetal spinal neurons to generate electrical activity has been examined using cultured spinal cords from 8- to 10-wk gestation (WG) fetuses. Electrophysiological studies have shown membrane depolarization in spinal neurons after glutamate application, as well as the presence of voltage-sensitive sodium, potassium, and calcium currents (Hosli et al. 1973, 1976). These data are supported by anatomical studies that have confirmed the presence of synaptic specializations between cultured neurons (Kim et al. 1988). Together, these findings suggest at least some of the fundamental elements for neurotransmission and AP generation are present in 8- to 10-WG fetal spinal neurons. However, the difficulties in establishing viable human neuronal cultures means data are only available over a very restricted time window during the first trimester of human development (Kato et al. 1985; Kennedy et al. 1980).

To address this deficit, we developed a new approach to studying electrophysiological properties of human fetal spinal neurons using acute spinal cord slices prepared from fetuses aged 10–18 WG. We used this preparation to investigate how the electrophysiological properties of neurons in the VH and DH of the gray matter mature over 8 wk of fetal development. Such investigations are important for translation of findings made in rodents to humans (see Davidson et al. 2014).

METHODS

Tissue Preparation

All procedures were approved by the University of Newcastle Human Research Ethics Committee. Informed, written consent was obtained from all tissue donors prior to termination, and all associated information remained at the clinic. Products of conception were only obtained from elective terminations, and there were no known abnormalities in our samples. Apart from gestational age (which ranged from 10 to 18 wk), no identifying information was supplied to researchers. Gestational age was determined using three criteria: 1) date of the last menstrual period; 2) ultrasound measurements of crown-rump length; and 3) foot length (Hern 1984). Products of conception were collected in cold glycerol-substituted artificial cerebrospinal fluid (gACSF) containing the following (in mM): 250 glycerol, 26 NaHCO₃, 11 glucose, 2.5 KCl, 1.2 NaH₂PO₄, 1.2 MgCl₂, and 2.5 CaCl₂, bubbled with 95% O₂ and 5% CO₂ (Ye et al. 2006). gACSF was used in place of the traditional sucrose-substituted ACSF (Walsh et al. 2009), as it improved slice viability and, importantly, allowed sharing of valuable tissue across multiple projects, including placental studies (Pringle et al. 2011) and investigation of vestibular hair cell development (Lim et al. 2014). All samples were delivered to the laboratory within 1 h of the termination procedure. Vertebral columns were isolated and transferred to a dissecting chamber containing ice-cold gACSF continuously bubbled with 95% O₂ and 5% CO₂. The spinal cord was removed, and transverse slices of the lumbosacral region (300 μm thick) were obtained using a vibrating-blade microtome (Leica VT1200S; Leica Microsystems, Wetzlar, Germany). Spinal cord slices were then transferred to a humidified storage chamber containing oxygenated ACSF (118 mM NaCl substituted for glycerol in the gACSF). Slices were allowed to incubate for 1 h at room temperature (22–24°C) before recording commenced.

Electrophysiology

Spinal cord slices were transferred to a recording chamber and held in place using nylon netting fixed to a U-shaped platinum frame (Fig. 1A). The recording chamber was continually superfused with oxygenated ACSF (4–6 bath volumes/min) and maintained at near-physiological temperature (32°C) using an in-line temperature control device (TC-324B; Warner Instruments, Hamden, CT). Whole cell patch-clamp recordings were obtained from spinal neurons in either VH or DH cells, visualized using infrared differential contrast optics (Fig. 1B) and an infrared-sensitive camera (Rolera-XR; QImaging, Surrey, BC, Canada). For VH cells, we deliberately targeted the largest neurons, as these are considered presumptive motoneurons (Carlin et al. 2000). At the end of each recording session, the location of the recorded neuron was documented using still images of the spinal cord, while the patch pipette remained in place (Fig. 1C).

Fig. 1.

Location of recorded neurons. A: image of a transverse spinal cord slice (300 μm thick) at 10 wk gestation (WG), held in place by nylon netting (vertical lines). B: images of representative dorsal horn (DH; left) and ventral horn (VH; right) neurons at 16 WG. Note the clear difference in neuron soma size for neurons at microelectrode tip. It was difficult to obtain recordings in DH neurons younger than 16 WG (see text for details). C: schematic representation of the location of each recorded neuron. Recordings were made in DH (open circles) and VH (solid circles) neurons across all age groups examined.

Patch pipettes (3–5 MΩ resistance) were pulled from thin-walled borosilicate glass (PG150T-15; Harvard Apparatus, Kent, UK) and filled with a potassium-based internal solution containing the following (in mM): 135 KCH₃SO₄, 6 NaCl, 2 MgCl₂, 10 HEPES, 0.1 EGTA, 2 MgATP and 0.3 NaGTP (pH 7.3, using KOH). Whole cell patch-clamp recordings were made using a Multiclamp 700B Amplifier (Molecular Devices, Sunnyvale, CA). The whole cell recording configuration was first established on visualized neurons in voltage-clamp (holding potential −60 mV). Series resistance and input resistance (R_IN) were measured from the averaged response (5 trials) to a 5-mV hyperpolarizing pulse. These parameters were measured at the beginning and end of each recording session, and data were rejected if values changed by >20%. After determining series resistance and R_IN values, the recording mode was switched to current clamp, and the membrane potential observed after ∼15 s was designated as resting membrane potential (RMP). All subsequent current clamp recordings were made from this potential. The capacity of neurons to discharge APs was assessed by injecting a series of depolarizing current steps (Tadros et al. 2012, 2014). Due to generally lower R_IN in VH neurons, we used 50-pA increments, 1-s duration, while for higher resistance DH neurons we used 20-pA increments, 800-ms duration. Similarly, the potential for rebound discharge was assessed by hyperpolarizing current steps, −50-pA increments, 1-s duration for VH neurons, and −20-pA increments, 800 ms for DH neurons.

Data Capture and Analysis

Data were digitized on-line (sampled at 20 kHz, filtered at 6 kHz) via an ITC-16 computer interface (Instrutech, Long Island, NY) and captured on a Macintosh computer using Axograph X software (Axograph X, Sydney, NSW, Australia). All data were analyzed offline using the Axograph software. RMP and AP threshold were corrected for a calculated liquid junction potential of ∼10 mV (Barry and Lynch 1991). Individual APs were detected using the derivative threshold method (range of 15–20 mV/ms; Graham et al. 2007b). AP threshold was measured at the inflection point during the rising phase of the AP [i.e., when the first derivative of voltage over time (dV/dt) exceeded 15–20 mV/ms]. Rheobase current was defined as the minimum step-current that could evoke at least one AP. The amplitude of each AP was measured between threshold and maximum positive peak, and after-hyperpolarization (AHP) amplitude was measured as the difference between AP threshold and the maximum negative peak. AP half-width was calculated at 50% of the maximum positive peak, and AHP half-width was measured at 50% of the maximum negative peak. AHP half-width was chosen as a measure of AHP duration as it was often hard to determine precisely when membrane potential returned to rest, especially in tonic firing neurons.

SPSS version 21 software package (SPSS, Chicago, IL) was used for statistical analysis. Student's T-tests were used to compare VH and DH data when such comparisons could be made, and one-way ANOVA was used to compare means across age groups. Scheffé's post hoc tests were used to determine where means differed. Data sets that failed Levene's test of homogeneity of variance were compared using nonparametric Mann-Whitney or Kruskal-Wallace tests, followed by Tamhame's T2 post hoc test. G-tests, with Williams' correction, were used to determine whether the prevalence of AP discharge categories, responses to hyperpolarizing steps and spontaneous activity differed between ages and regions of the spinal cord. Statistical significance was set at P < 0.05, and all data are presented as means ± SE.

RESULTS

Recordings were obtained from 91 neurons in 17 fetuses aged 10–18 WG. To compare the properties of neurons with the best available behavioral (reflex/movement) data, we assigned recorded neurons into one of three age groups: 10–12 WG, 13–15 WG, and 16–18 WG. These groups were selected as they represent times when mouth opening reflexes elicited by cutaneous stimulation (touching or stroking) with “…a hair tipped with smooth inert material…” fall into three phases (Humphrey 1969). Reflexes first appear at 10–12 WG, become more complex at 13–14 WG, before appearing less stereotyped and “more graceful and flowing” after 16 WG (Humphrey 1969). Recordings were made from visualized neurons in either the VH or DH, and the location of recorded neurons in representative transverse spinal cord sections for each of the three age groups is shown in Fig. 1.

We were able to obtain recordings from VH neurons in all three age groups as they were easily identified and more securely embedded in the spinal cord (Fig. 1B). In contrast, obtaining patch-clamp recordings from DH neurons in spinal cords younger than 16 WG proved to be technically challenging: few recordings were obtained in the 10–12 WG and 13–15 WG groups (Fig. 1C, left and middle, n = 2 and 1, respectively). Before 16 WG, DH neurons were extremely small, appeared to lack dendrites, and were weakly tethered in the spinal cord neuropil (Fig. 1B). This meant the positive pressure required for successful patch-clamp recordings pushed neurons away from the recording electrode and often ejected them out of the spinal cord slice. Importantly, optimal recordings of VH neurons were obtained in the same slices where these features were observed in DH neurons. Furthermore, none of the small neurons, which dominated in the DH, were observed in the VH at any age. This suggests slice quality was not an issue in achieving adequate patch-clamp recordings of DH neurons prior to 16 WG.

Spinal Neurons' Capacity to Discharge APs

The capacity of neurons to discharge APs was assessed by injecting a series of depolarizing current steps of increasing amplitude. One of four types/patterns of AP discharge was observed in each recorded neuron (Fig. 2A). Tonic firing neurons discharged APs for the entire duration of the current injection and increased their discharge frequency with increasing current amplitude. Initial bursting neurons were characterized by a brief, high-frequency burst of APs at the onset of the current step. Single spiking neurons discharged a single AP immediately upon depolarization, regardless of current intensity. Reluctant firing neurons did not discharge APs in response to current injection, despite depolarization well above mean AP threshold (see Fig. 2A, bottom right). These neurons were included in our analysis as their intrinsic properties fell within the range observed in those that discharged APs in response to current injection in their respective age groups. We have also observed reluctant firing neurons in previous studies of spinal neurons in postnatal rodent models, both in vitro and in vivo (Graham et al. 2004; Walsh et al. 2009).

Fig. 2.

Action potential (AP) discharge patterns in fetal spinal neurons. A: neurons responded to depolarizing square current step injection (20- or 50-pA increments, 800- or 1,000-ms duration; inset bottom right) in one of four ways: tonic firing (TF) neurons discharged APs for the entire duration of the injected current; initial bursting (IB) neurons discharged a burst of APs at the onset of the current step; single spiking (SS) neurons discharged a single AP at the onset of each current step; and reluctant firing (RF) neurons did not discharge APs in response to square current step injection. Representative traces are from VH neurons aged 16–18 WG in response to 100- and 200-pA current steps; dashed line indicates −60 mV. B: the incidence of each of the responses displayed in A for DH neurons aged 16–18 WG. No AP discharge was observed in DH neurons prior to 16 WG (see text). C: the incidence of each response profile displayed in A for VH neurons at each age group. TF was the most commonly observed discharge pattern, and its incidence increased with age. The prevalence of RF neurons decreased with age. RMP, resting membrane potential.

There were notable differences in the capacity of VH and DH neurons to generate APs following depolarizing current injection (Fig. 2, B and C). For VH neurons, the four discharge categories were observed at all ages (Fig. 2C). Tonic firing was the most commonly observed discharge pattern, and its incidence increased significantly with age (12/22, 55% at 10–12 WG and 15/23, 65% at 16–18 WG, P = 0.025). This was accompanied by a significant decrease in the prevalence of reluctant firing neurons (7/22, 32% at 10–12 WG vs. 1/23, 4% at 16–18 WG, P = 0.014). The proportion of initial bursting and single spiking neurons was similar across the three age groups (initial bursting: 1/22, 5% at 10–12 WG vs. 2/23, 9% at 16–18 WG, P = 0.57; single spiking: 2/22, 9% at 10–12 WG vs. 5/23, 22% at 16–18 WG, P = 0.21; see Fig. 2C).

In slices younger than 16 WG, we did not record APs in DH neurons mainly due to the difficulties encountered in establishing reliable seals on neurons in these two younger age groups (described above). From 16 WG, however, over one-half of the DH neurons discharged APs and exhibited either tonic firing (3/15, 20%) or single spiking (6/15, 40%, Fig. 2B) profiles. A comparison of the various AP discharge patterns in VH and DH at 16–18 WG revealed a significant difference in the proportions of these patterns (P = 0.003, Fig. 2, B and C), with VH neurons displaying less single spiking or reluctant firing neurons compared with the DH sample. We never observed delayed firing neurons in the DH, which have been considered to be excitatory interneurons in the mature rodent DH (Lu et al. 2006). Together, these data suggest the adult proportions of excitatory and inhibitory DH neurons is yet to be established because the mature rodent DH exhibits all major discharge patterns (Lu et al. 2006; Walsh et al. 2009).

We next compared the intrinsic and AP properties across the discharge categories for each age group. Although statistics were limited in some instances, we noted several significant differences between certain populations (Table 1). At all three age groups, tonic firing neurons in the VH displayed more hyperpolarized RMPs compared with the other discharge categories, with greater AP and AHP amplitudes than single spiking neurons at both 13–15 WG and 16–18 WG, as well as initial bursters at 16–18 WG. These differences between tonic firing and single spiking neurons have been observed in rodent DH neurons during development (Tadros et al. 2012; Walsh et al. 2009), suggesting the characteristics of each discharge category are consistent across species.

Table 1.

Intrinsic and AP properties for each of the observed discharge categories

Discharge Category/Region	Age, WG	n	Input Resistance, MΩ	RMP, mV	Rheobase Current, pA	AP Threshold, mV	AP Amplitude, mV	AP Half-width, ms	AHP Amplitude, mV
Tonic firing
Dorsal	10–12	0
	13–15	0
	16–18	3	533 ± 234	−67.3 ± 8.6	20 ± 0	−52.1 ± 3.2†	46.0 ± 16.7	2.7 ± 0.7	−9.2 ± 2.0
Ventral	10–12	12	795 ± 145	−62.1 ± 2.9†‡	50 ± 0	−45.0 ± 1.3	44.8 ± 3.6	3.0 ± 0.6	−9.2 ± 2.4
	13–15	12	279 ± 140	−70.5 ± 2.6†‡	99 ± 19*	−48.1 ± 1.6†	48.2 ± 3.7†	1.44 ± 0.2†	−23.8 ± 2.9†
	16–18	15	238 ± 83	−69.1 ± 2.4*†	117 ± 30	−45.5 ± 1.3	56.5 ± 1.8*†	1.3 ± 0.2*	−23.3 ± 1.7*†
Initial bursting
Dorsal	10–12	0
	13–15	0
	16–18	0
Ventral	10–12	1	849	−85.6	100	−41.7	52.9	1.4	−16.7
	13–15	3	429 ± 88	−60.8 ± 7.4	48 ± 2.5	−52.6 ± 3.0†	53.5 ± 10.3†	2.1 ± 0.7	−12.6 ± 3.1
	16–18	2	332 ± 195	−51.0 ± 2.7	50 ± 0	−36.2 ± 4.4	35.8 ± 11.0	3.9 ± 0.2	−6.6 ± 6.4
Single spiking
Dorsal	10–12	0
	13–15	0
	16–18	6	730 ± 148	−65.9 ± 4.1	87 ± 31	−39.3 ± 1.5	21.5 ± 3.4	2.7 ± 0.4	−5.1 ± 1.8
Ventral	10–12	2	1488 ± 377	−44.5 ± 3.8	50 ± 0	−39.7 ± 3.9	37.0 ± 12.2	4.0 ± 2.4	−4.0 ± 7.7
	13–15	6	450 ± 240	−51.0 ± 3.1	90 ± 21	−41.9 ± 2.1	27.8 ± 4.2	2.9 ± 0.4	−10.8 ± 3.6
	16–18	5	653 ± 323	−52.4 ± 6.1	100 ± 27	−43.8 ± 2.6	24.2 ± 3.4	3.7 ± 1.2	−10.1 ± 3.0
Reluctant firing
Dorsal	10–12	2	2,928 ± 1,080	−33.1 ± 4.1
	13–15	1	873 ± 115	−56.1 ± 2.9
	16–18	6	1,470 ± 363	−64.6 ± 7.8
Ventral	10–12	7	836 ± 228	−45.0 ± 2.4
	13–15	4	861 ± 285	−40.9 ± 3.6
	16–18	1	322	−91.4

Values are means ± SE; n, no. of fetuses. WG, weeks gestation; RMP, resting membrane potential; AP, action potential; AHP, after-hyperpolarization. Significance between discharge categories within age groups are denoted as follows:

^*significantly different to initial bursting;

^†significantly different to single spiking;

^‡significantly different to reluctant firing.

Responses to Hyperpolarizing Current

Our laboratory has shown previously that developing spinal and brain stem neurons in rodents respond to hyperpolarizing current in one of four ways (Tadros et al. 2012, 2014). Similar types of responses were observed in human fetal VH and DH neurons (Fig. 3A). Passive responses were characterized by the absence of active conductances during or after the hyperpolarization step. Rebound spiking responses featured AP discharge upon release from hyperpolarization. Such responses have been associated with the low-threshold T-type calcium current in adult rodent spinal neurons (Graham et al. 2007b; Yoshimura and Jessell 1989). In some neurons, we observed a “sag” in the voltage trace during the hyperpolarization step, which is typically associated with the hyperpolarization-activated mixed cationic inward current (I_h) (Graham et al. 2007b; Yoshimura and Jessell 1989). Finally, some neurons exhibit a “sag” indicative of I_h, as well as an extended hyperpolarization following release from current injection. This extended hyperpolarization is known to be dependent on extracellular potassium concentration (Yoshimura and Jessell 1989) and could be driven by either a calcium-activated potassium conductance, or A-type potassium currents (Callister et al. 1997; Graham et al. 2007b).

Fig. 3.

Responses to hyperpolarizing current in DH and VH neurons. A: neurons responded to hyperpolarizing square current step injection (−20- or −50-pA increments, 800- or 1,000-ms duration, inset bottom right) in one of four ways: passive (P) responses were characterized by a lack of any active conductances during or after hyperpolarization; rebound spiking (R) was evident with AP discharge upon release from hyperpolarization; sag (S) was observed as a slow depolarization occurring during the hyperpolarizing step; and a few neurons exhibited a sag during the hyperpolarizing step plus an extended hyperpolarization (H) at the end of the current step. Representative traces are from VH neurons aged 16–18 WG, with the defining characteristic for each category indicated by a solid arrowhead. B: the incidence of each response in A, for DH neurons aged 16–18 WG. Prior to 16 WG, DH neurons displayed the passive response (see text). C: the incidence of each response in A, for VH neurons of each age group. Rebound spiking decreased, and the other response types emerged with increasing age.

Responses to hyperpolarizing current injection displayed greater variability in VH neurons than in DH (Fig. 3C) and could be compared across the three age groups. Rebound spiking was observed in 17/22 (77%; 10–12 WG) and 10/23 (44%; 13–15 WG) VH neurons. In the 16–18 WG age group, all four responses were observed with 6/23 (26%) neurons showing a passive response, 12/23 (52%) displaying rebound spiking, 4/23 (18%) exhibiting the voltage-sag characteristic of I_h current and only 1/23 (4%) neurons showing both the “sag” and extended hyperpolarization. Prior to 16 WG, DH neurons exhibited passive responses during hyperpolarizing current injection (n = 3). Such responses also dominated in the 16–18 WG group (9/15; 60%); however, rebound spiking (3/15; 20%) and the extended hyperpolarizing response (3/15; 20%) were also observed (Fig. 3B). Interestingly, the proportions of each response to hyperpolarizing current differed significantly between the VH and DH neurons at 16–18 WG (P = 0.01, Fig. 3, B and C). These data further emphasize the differences between VH and DH at 16–18 WG and suggest the responses to hyperpolarizing current injection become more variable with age in VH neurons.

Differences between VH and DH Neurons

We next compared the intrinsic and AP properties of VH and DH neurons that were capable of AP discharge at 16–18 WG (Fig. 4A). VH neurons had lower R_IN values (mean = 341 vs. 664 MΩ, P = 0.05), and higher rheobase currents (141 vs. 64 pA, P = 0.29) than DH neurons. RMP was similar (approximately −75 mV) in VH and DH neurons (Fig. 4A). It is notable that including all neurons (those that did and did not discharge APs) in the above analysis increased variability and minimized statistical differences, because both intrinsic and AP properties differed across the various discharge categories observed in spinal neurons (Graham et al. 2007a; Tadros et al. 2012).

Fig. 4.

Intrinsic and AP properties of DH and VH neurons. A: bar plots for intrinsic properties of DH (n = 9) and VH (n = 22) neurons capable of AP discharge. Input resistance (left) was lower in VH neurons (P = 0.05). Rheobase current (center) was trending toward larger values in VH neurons; however, this was not significant (P = 0.29). RMP (right) was similar in DH and VH neurons. B: representative trains of AP discharge (left) from spontaneously active dorsal (top) and ventral (bottom) neurons. Dashed line indicates −70 mV. Single APs from each spinal cord location are expanded on the right; arrowheads denote AP threshold. C: bar plots for AP properties measured on DH (n = 9) and VH (n = 22) neurons capable of AP discharge. AP threshold (top left) did not differ between the two spinal cord locations. AP amplitude (top right) was greater in VH neurons (P = 0.015). AP half-with (bottom left) was unchanged between DH and VH neurons. After-hyperpolarization (AHP) amplitude (bottom right) was greater in VH neurons (P = 0.001). The number of cells in each age group is denoted on each bar. *Significant difference, P < 0.05.

AP discharge was not observed in the limited recordings made in 10–12 WG and 13–15 WG DH neurons. There was also a significant difference in the proportion of various AP discharge patterns (P = 0.003, Fig. 2, B and C) and responses to hyperpolarizing current (P = 0.01, Fig. 3, B and C) in VH and DH neurons at 16–18 WG. Spontaneous AP discharge was observed in some neurons (Fig. 4B). In VH, spontaneous activity was observed in all three age groups. Predictably, spontaneous activity was not observed in DH until 16–18 WG, when 3/15 (20%) neurons displayed spontaneous AP discharge. The prevalence of spontaneous AP discharge at 16–18 WG in DH was similar to that observed in VH (7/23, 30%, P = 0.6). These data provide further evidence that VH neurons develop more rapidly than DH neurons in the human fetal spinal cord.

AP properties were measured on rheobase APs, that is, the first AP elicited in response to depolarizing current injection (Fig. 4C). AP threshold was similar in VH and DH neurons (approximately −53 mV), although AP amplitude (VH 47.3 ± 3.4 mV vs. DH 29.7 ± 6.7 mV, P = 0.015) and AHP amplitude (VH −18.8 ± 2.0 mV vs. DH −6.5 ± 1.5 mV, P = 0.001) were greater in VH neurons. AP half-width was not different between VH and DH neurons (2.1 ± 0.4 vs. 2.7 ± 0.3 ms, P = 0.32). Even though AP half-width did not differ, we found a significantly greater dV/dt on the rising phase of the AP in VH neurons (126.0 ± 16.2 vs. 38.9 ± 16.0 mV/ms, P = 0.005), suggesting the underlying depolarizing currents contributing to the AP differ between VH and DH neurons. We also found a greater AHP half-width (i.e., duration at 50% of the maximum negative peak) in VH neurons (VH 83.1 ± 15.7 ms vs. DH 25.9 ± 11.0 ms, P = 0.02). This is consistent with the larger AHP amplitude in VH neurons. Taken together, these differences in intrinsic and AP properties imply larger, faster APs in VH neurons, which is clearly illustrated by the example APs shown in Fig. 4C, right.

Development of Intrinsic and AP Properties in VH Neurons

Intrinsic and AP properties of VH neurons were also compared across the three age groups. Spontaneous AP discharge was observed at all three ages, and its prevalence did not change significantly with age [8/22 (36%) at 10–12 WG; 6/26 (23%) at 13–15 WG; and 7/23 (30%) at 16–18 WG; P = 0.6, Fig. 4B]. Comparisons of intrinsic and AP properties in VH neurons were restricted to neurons that exhibited the tonic discharge pattern for two reasons. First, intrinsic and AP properties differ in spinal neurons according to AP discharge type (Tadros et al. 2012; Walsh et al. 2009). Second, tonic firing neurons were the dominant AP discharge category at all three ages examined [n = 12 (55%) for 10–12 WG; n = 12 (50%) for 13–15 WG; n = 15 (65%) for 16–18 WG].

Several intrinsic properties of tonically firing VH neurons changed during the developmental period examined (Fig. 5). R_IN decreased from 795 ± 145 MΩ at 10–12 WG to 238 ± 83 MΩ at 16–18 WG (P = 0.004), and rheobase current increased from 50 pA at 10–12 WG to 117 ± 30 pA at 16–18 WG (P = 0.04). The difference in rheobase current is likely to be an underestimate, since the smallest current step we used (i.e., 50 pA) elicited APs in all VH neurons at 10–12 WG. RMP and AP threshold were unchanged at approximately −70 mV and approximately −45 mV, respectively (Fig. 5A, RMP; Fig. 5B, AP threshold) between the three age groups. Several AP properties changed dramatically between 10–12 WG and 16–18 WG. Specifically, AP amplitude increased (10–12 WG, 44.8 ± 3.6 mV vs. 16–18 WG, 56.5 ± 1.8 mV, P = 0.01) and AP half-width decreased (10–12 WG, 3.03 ± 0.61 ms vs. 16–18 WG, 1.30 ± 0.21 ms, P = 0.001). AHP amplitude also increased over this developmental period (10–12 WG, −9.2 ± 2.4 mV vs. 16–18 WG, −23.3 ± 1.7 mV, P = 0.001; Fig. 5B) and was accompanied by an increase in the AHP half-width (10–12 WG, 39.7 ± 13.5 ms vs. 16–18 WG, 107.0 ± 19.5 ms, P = 0.016). Together, these data suggest the intrinsic and AP properties of tonic firing VH neurons undergo significant maturation between 10 and 18 WG in the human fetus.

Fig. 5.

Changes in the properties of TF VH neurons during development. A: bar plots displaying developmental changes in intrinsic properties of TF VH neurons. Input resistance (left) decreased, rheobase (center) increased, and RMP (right) was similar across the age groups examined. B: some AP properties differed during the developmental period examined. AP threshold (far left) was unchanged, whereas AP amplitude (center left) increased from 10–12 WG to 16–18 WG. AP half-width decreased (center right) and AHP amplitude increased (far right) from 10 to 18 WG. *P < 0.05, 10–12 WG vs. 13–15 WG and 16–18 WG. #P < 0.05, 10–12 WG vs. 16–18 WG. The number of cells in each age group is denoted on each bar.

DISCUSSION

Here, we provide functional data on the intrinsic properties of VH and DH neurons in acute slices prepared from first and early second-trimester human fetal spinal cords. We document dramatic changes in the intrinsic and AP properties of VH spinal neurons during this developmental period. In contrast, we were unable to record APs from DH neurons prior to 16 WG, which may indicate differences in the development of electrical activity in the DH and VH of the human fetal spinal cord. These findings are consistent with classical anatomical (Okado et al. 1979, 1981; Ray and Wadhwa 1999; Sarnat et al. 1998) and clinical studies (de Vries et al. 1985; Slater et al. 2010b) that indicate functional connectivity develops earlier within motor circuits compared with those in the DH involved in processing tactile and nociceptive information.

Previous Studies Investigating Electrical Properties of Human Fetal Spinal Neurons

To our knowledge, the functional properties of human spinal neurons have only been examined in cultured fetal neurons. Early functional studies, using sharp microelectrode recording in long-term cultures (22 days) prepared from a 17-WG fetus, reported R_IN values of ∼5 MΩ (3 neurons) and RMPs of approximately −44 mV (25 neurons; Hosli et al. 1973). These values are much lower than those we recorded at any gestational age (Figs. 4A and 5A). Notably, the authors stated “the membrane potential of most cells decayed within 20–40 sec after impalement” (Hosli et al. 1973). In that study, the response of neurons to depolarizing or hyperpolarizing current injection could not be examined. Later work by the same group recorded dose-dependent depolarizations to bath-applied excitatory amino acids (glutamate and aspartate) in neurons cultured (8–56 days) from 7- to 17-WG fetuses (Hosli et al. 1976). These important data show human spinal neurons in culture express ligand gated ion channels for major excitatory neurotransmitters. However, the neurons sampled in the above study came from fetuses of varying ages and different times in culture. This makes it difficult to establish precisely the spatiotemporal expression of receptor channels and their subunits in situ. In future studies, it will be possible to use the recording techniques used in our study, combined with sophisticated electrophysiological methodologies, to answer these important questions using in situ spinal neurons in VH and DH populations.

Subsequent studies using patch-clamp recordings from spinal neurons, grown in long-term cultures (19–20 days) from 8- to 9-WG fetuses report RMPs of approximately −40 mV (Kato et al. 1985). Here at later stages of development, we report substantially more hyperpolarized RMP (approximately −75 mV). In contrast, we have shown similar AP half-width (∼2 ms; Fig. 4C) to previous findings where short duration APs were recorded in cultured neurons in response to current injection. In those cultured neurons, tetrodotoxin-sensitive sodium, tetraethylammonium and 4-aminopyridine-sensitive potassium, and cobalt-sensitive calcium currents are also present (Kato et al. 1985).

It is difficult to determine in the above study whether the cultured neurons were DH or VH in origin, although attempts were made to identify motoneurons using acetylcholinesterase immunohistochemistry (Kato et al. 1985). APs have been examined in VH neurons grown in long-term cultures (9 wk) from 8-WG fetuses; however, these VH neurons only discharged single and slow APs in response to depolarizing current injection (Kim et al. 1988). These results are in contrast to ours that show VH neurons have robust, repetitive AP discharge from 10 WG (Fig. 4). It is therefore likely that neurons recorded in spinal cord slices from fetuses aged 10–18 WG are more representative of neurons in situ than those in culture, as they display more diverse electrical properties.

Comparison of the Electrical Properties of Rodent and Fetal Spinal Neurons

VH neurons.

The most comprehensive data investigating the development of intrinsic properties in VH neurons prior to birth come from rodent studies. In rats, sodium, calcium, and potassium conductances can be recorded in VH neurons at E14, and large overshooting sodium-dependent APs are generated a few days later (Ziskind-Conhaim 1988). In respiratory motoneurons, electrical properties, including RMP, R_IN, and rheobase current have been shown to change during the course of development (Di Pasquale et al. 1996). Here, we show similar developmental changes in human fetal VH neurons, with dramatic changes in both intrinsic and AP properties occurring between 10 and 18 WG (Fig. 4). We cannot conclusively determine whether our VH neurons were motoneurons. In early experiments, we attempted to fill recorded neurons and then stain for choline acetyltransferase to determine their identity. This could not be achieved, as our spinal cord slices did not survive fixation and subsequent processing. We assume this is due to the lack of connective tissue in the developing nervous system (Weidenheim et al. 1993). However, we deliberately targeted large neurons in the VH, and so it would be reasonable to assume these were nascent motoneurons. Tonic (or repetitive) firing has long been considered a distinguishing feature of adult spinal motoneurons (Brownstone 2006); however, early postnatal spinal motoneurons can generate bursting and even single spiking AP discharge patterns in rats (Vinay et al. 2000a). In our sample of fetal human VH neurons, we observed each of these three discharge patterns (Fig. 2B), suggesting that the electrophysiological profile observed in the human fetus is similar to rodent VH neurons during postnatal development. It appears that the greatest difference between human and rodent VH neurons is when these discharge patterns are expressed.

DH neurons.

Our recent study examined the intrinsic properties of neurons in DH outer laminae in mice during late embryonic (E15-17) development (Walsh et al. 2009). After birth, AP properties continue to mature, with increases in AP and AHP amplitudes and decreases in AP half-width, until they exhibit adult-like characteristics at approximately postnatal day 10. AP discharge patterns also changed from predominately single spiking at E15-17 to bursting and tonic firing discharge patterns at postnatal day 10 (Walsh et al. 2009). Although we could only measure AP properties and discharge at one time point (16–18 WG) for human DH neurons (Fig. 4), it appears they mature much earlier than their rodent counterparts (i.e., in the first vs. latter half of gestation). The predominant discharge patterns observed in human fetal DH neurons were single spiking and reluctant firing (Fig. 2C), similar to that observed in embryonic mice (Walsh et al. 2009).

Functional Connectivity within the Developing Sensorimotor System

During development, axons of VH neurons exit the spinal cord and ultimately innervate skeletal muscle (Vermeren et al. 2003). The timing of this process has been studied extensively in rodents. For example, the phrenic nerve enters the diaphragm as early as E15, and neuromuscular junctions are formed by E18-19 (Bennett and Pettigrew 1974). Based on the presence of end plate potentials recorded in the diaphragm, these junctions are functional by E17-22 (Diamond and Miledi 1962). As the rat gestation period is 21–23 days, functional connections between nerve and muscle occur very late during intrauterine development in this species. In contrast, motor end plates are present very early in human development: in quadriceps at 9 WG (Fidzianska 1980), intercostal muscles at 8.5 WG, tibialis anterior at 10 WG (Juntunen and Teravainen 1972) and masseter at 12 WG (Ezure 1996). To our knowledge, end-plate potentials have not been recorded from developing human neuromuscular junctions. However, studies conducted over half a century ago considered the emergence of movement and spinal reflexes in the human fetus as evidence of functional connectivity between nerve and muscle. Specifically, reflex responses to cutaneous stimulation of the perioral region can be observed as early as 7.5 WG, and movement complexity to the same stimuli increases over the next 10 wk (summarized in Humphrey 1969). Furthermore, histological studies on fiber projections and spinal cord cytology in early stage human fetuses, coupled to clinical observations, have proposed that spinal reflex mechanisms are established by 8 WG (Windle and Fitzgerald 1937; Hooker 1936). Ultrasound imaging has also shown that the human fetus is capable of reflex movement very early in development (de Vries et al. 1985). Our data showing VH neurons discharge trains of APs from 10 WG (Fig. 2C), coupled with the presence of neuromuscular junctions, is consistent with the existence of well-established connections between VH neurons and skeletal muscle in the first trimester of pregnancy.

In contrast to VH neurons, our study suggests DH neurons are very immature and incapable of discharging APs prior to 16 WG (Fig. 2B). As a consequence, it is unlikely DH neurons are capable of processing sensory inputs before 16 WG. The lack of APs in our study does not necessarily confirm that DH neurons are unexcitable prior to 16 WG. However, our findings are consistent with previous anatomical studies on human fetal spinal cords that demonstrate a lack of neuronal nuclear antigen staining in the DH, despite strong staining in the VH at 10 WG (Sarnat et al. 1998). Similarly, very low Nissl staining has been observed in the DH at 12–13 WG (Ray and Wadhwa 1999). In addition, a “sparse” number of sensory axon terminals have been observed in DH superficial lamina as late as 19 WG (Konstantinidou et al. 1995). However, the authors were unable to confirm whether the lack of terminals in the DH arose from technical considerations associated with DiI transport in small axons or true developmental mechanisms. Nevertheless, the lack of responsiveness of DH neurons in our study (Fig. 2) indicates that the components of DH sensory pathways are not fully mature prior to 16 WG in the human. Further evidence suggesting that DH circuits develop later than those in the VH come from electron microscopic studies of the human fetal spinal cord (Okado 1981; Okado et al. 1979). These studies show that synaptic junctions in the human spinal cord are visible at 5 WG and are present in the motor neuropil of the VH well before other spinal regions.

Spinal Reflexes and Pain Responses in the Human Fetus

Meaningful movement depends, in part, on how sensory feedback from receptors located in muscles, joints and skin is processed in spinal circuits. Anatomical studies in rodents demonstrate sensory afferent fiber entry to the spinal cord is dependent upon afferent axon diameter. The largest peripheral afferents enter the spinal cord at ∼E15 in rats and reach the VH by E16 (Snider et al. 1992). Functional studies have shown that monosynaptic connections are established with presumed motoneurons in VH by E18 (Kudo and Yamada 1987). By E17, other large-diameter fibers occupy the entire DH (Snider et al. 1992), and deep DH neurons respond to stimulation of skin afferents from E18 (Fitzgerald 1991). In contrast, small-diameter Aδ- and C-fiber nociceptive afferents enter the DH later, at ∼E17, and neurons in the superficial DH only appear to connect with these peripheral afferents close to birth (Fitzgerald 2005). Together these data suggest that classical spinal pain circuits involving small-diameter fibers that terminate in the superficial DH in rodents are poorly developed, if at all, before birth (Fitzgerald and Jennings 1999; Woodbury and Koerber 2003).

Our data show that DH neurons are capable of discharging APs after 16 WG, suggesting that ascending pain pathways are incomplete in the human fetus prior to 16 WG. Importantly, the “pain experience” requires a functional ascending pain pathway, whereby peripherally generated noxious signals reach the spinal cord and are then relayed to the somatosensory cortex. Electrophysiological studies on newborn humans have demonstrated cortical potentials in response to noxious procedures that break the skin (Slater et al. 2010b). These data suggest most components of the pain neuroaxis are fully formed at birth. Importantly, repeated noxious stimulation of the skin can lead to persistent changes in this pathway and result in exaggerated cortical responses to acute noxious stimulation (Slater et al. 2010a). One important question arising from these observations is when does the ascending pain pathway assemble in utero? Recent work using cerebral oxygen changes, as a measure of cortical activity, has shown that acute noxious stimuli (heel lances for routine blood sampling) is transmitted to the somatosensory cortex in the fetus as young as 25 WG (Slater et al. 2006). Taken together, our study and the available data on cortical processing in the developing human suggest the ascending pain pathway is assembled between 16 WG and 25 WG.

Major Conclusions and Future Directions

This study presents the first data acquired from VH and DH neurons in acute spinal cord slices from the human fetus. We have shown that patch-clamp recordings can be made from in situ VH neurons from 10 WG, and that at this age they can discharge APs in response to depolarizing current. In contrast, recordings from DH neurons before 16 WG proved to be difficult: we could not record APs in DH neurons until 16 WG. Comparisons made between the intrinsic properties of VH and DH neurons at 16–18 WG suggest VH neurons, and VH circuitry, develop weeks before their counterparts in DH. This conclusion is consistent with available anatomical data for the human fetus, which shows peripheral afferents reach the VH and arborize well before they do so in the DH. Our data also have important implications for estimating when the human fetus can experience pain. Neurons in the DH are the first central target for peripheral afferents that carry nociceptive and tactile information. As these neurons are electrically/functionally immature prior to 16 WG it is unlikely that nociceptive information could be processed in the DH and subsequently passed onto supraspinal sites within the pain neuroaxis at this time during development.

GRANTS

This research was supported by National Health and Medical Research Council of Australia Grant GNT1022717.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: M.A.T., R.L., D.I.H., A.M.B., and R.J.C. conception and design of research; M.A.T. performed experiments; M.A.T. analyzed data; M.A.T., R.L., D.I.H., A.M.B., and R.J.C. interpreted results of experiments; M.A.T. prepared figures; M.A.T. and R.J.C. drafted manuscript; M.A.T., R.L., D.I.H., A.M.B., and R.J.C. edited and revised manuscript; M.A.T., R.L., D.I.H., A.M.B., and R.J.C. approved final version of manuscript.