Role of Intrinsic Properties in Drosophila Motoneuron Recruitment During Fictive Crawling

Abstract

Motoneurons in most organisms conserve a division into low-threshold and high-threshold types that are responsible for generating powerful and precise movements. Drosophila 1b and 1s motoneurons may be analogous to low-threshold and high-threshold neurons, respectively, based on data obtained at the neuromuscular junction, although there is little information available on intrinsic properties or recruitment during behavior. Therefore in situ whole cell patch-clamp recordings were used to compare parameters of 1b and 1s motoneurons in Drosophila larvae. We find that resting membrane potential, voltage threshold, and delay-to-spike distinguish 1b from 1s motoneurons. The longer delay-to-spike in 1s motoneurons is a result of the shal-encoded A-type K⁺ current. Functional differences between 1b and 1s motoneurons are behaviorally relevant because a higher threshold and longer delay-to-spike are observed in MNISN-1s in pairwise whole cell recordings of synaptically evoked activity during bouts of fictive locomotion.

INTRODUCTION

Motoneurons must integrate diverse inputs from sensory networks, central pattern generators (CPGs), and higher-order brain centers into coherent output to generate coordinated movements. In most organisms the motoneurons conserve a division into low-threshold and high-threshold types that are responsible for generating powerful and precise movements, respectively. Mammalian low-threshold motor units are comprised of small, quickly recruited motoneurons that innervate a small number of muscle fibers and generate low forces (Burke et al. 1973; Enoka 1995; Henneman and Mendell 1981). High-threshold motor units consist of slowly recruited, large motoneurons that innervate a large number of muscle fibers and generate high levels of force. High-threshold, phasic motoneurons in the crayfish abdomen project to large, powerful muscle fibers and are silent except when recruited for escape activity. Low-threshold, tonic motoneurons project to thin muscle fibers and exhibit spontaneous activity that is useful for maintaining normal locomotor and postural functions, similar to mammalian low-threshold motor units (Atwood 2008). Insect motoneurons innervating the hind leg of the locust and the ventrolateral abdominal muscles of the blowfly larva also display a division into low-threshold and high-threshold types (Hardie 1976; Hoyle and Burrows 1973).

Drosophila melanogaster larval glutamatergic motoneurons may also be divided into low-threshold and high-threshold types. The divergence in Drosophila motoneurons has been well studied at the neuromuscular junction. Type 1b (“big”) neurons have “bigger” synaptic terminals and project to a single muscle. Type 1s (“small”) neurons have “smaller” synaptic terminals and project to groups of muscles (Kurdyak et al. 1994; Lnenicka and Keshishian 2000; Lnenicka et al. 1986). Therefore 1b neurons may be specialized for low-threshold, precise movements and recruited prior to 1s neurons that are specialized for high-threshold, powerful movements.

This positions the Drosophila larval locomotor system as a potentially powerful model in which to investigate the recruitment pattern of motoneurons, the importance of intrinsic properties to recruitment, and the behavioral relevance of recruitment order. Genetic tools allow consistent identification and manipulation of the intrinsic properties of neuronal populations and individual motoneurons. Further, the larval Drosophila CNS generates a peristaltic crawling movement in segmentally repeated muscles (Fox et al. 2006), a relatively simple movement compared with the left/right and extensor/flexor alternation generated by the mammalian locomotor CPG.

It is now possible to investigate the role of intrinsic properties in Drosophila motoneuron recruitment using in situ patch-clamp recordings (Choi et al. 2004; Rohrbough and Broadie 2002; Worrell and Levine 2008). Previous studies have examined motoneurons referred to as MN1-1b (also known as aCC), MNISN-1s (RP2), and RP3 (MN6/7-1b). We use the terminology set forth by Hoang and Chiba (Hoang and Chiba 2001) for all neurons except RP3 (see methods for explanation of RP3 terminology). In previous patch-clamp studies, MN1-1b and RP3 exhibited a lower current threshold for spiking and a higher spike frequency than those of MNISN-1s (Choi et al. 2004; Worrell and Levine 2008). The intrinsic properties of 1b and 1s motoneurons studied to date appear to correlate with neuromuscular junction physiology, yet it is unknown whether these properties are characteristic of neuron type (1b/1s) or simply the individual neuron (MN1-1b; MNISN-1s; RP3) because descriptions of 1s intrinsic properties are based on a single motoneuron. Importantly, it is also unknown whether differences in intrinsic properties of Drosophila motoneurons are behaviorally relevant. Therefore we set out to compare 1b and 1s motoneurons projecting to dorsal muscles (MN1-1b and MNISN-1s) and ventral muscles (RP3 and MNSNb/d-1s) using whole cell current-clamp protocol and whole cell recordings of synaptically evoked firing behavior during fictive locomotion.

Recordings were made from thoracic motoneurons because a recent study indicated differences in thoracic and abdominal locomotor patterns (Dixit et al. 2008). Therefore it is possible that the intrinsic properties of thoracic motoneurons are specialized to facilitate a specialized role of thoracic muscles. This possibility remains untested because the majority of whole cell recordings (Choi et al. 2004; Rohrbough and Broadie 2002) have been performed on abdominal motoneurons.

We find that resting membrane potential, voltage threshold, and delay-to-spike differ between 1b and 1s motoneurons. The longer delay-to-spike in 1s motoneurons is not due to passive properties of the cells, but is likely a result of active currents. The Shal A-type K⁺ current (I_A) is the primary determinant of delay-to-spike and recruitment patterns. Functional differences between 1b and 1s motoneurons are behaviorally relevant, given that a longer delay-to-spike is seen in MNISN-1s compared with that in MN1-1b when dual recordings of synaptically evoked activity during bouts of fictive locomotion are compared. Additionally, based on comparisons of drive potentials recorded simultaneously from the cells, it appears that MN1-1b and MNISN-1s share a large portion of their synaptic input.

METHODS

Drosophila stocks

Whole cell current-clamp recordings were obtained in situ from late third instar Drosophila larval motoneurons in ventral ganglion thoracic segments 2 and 3. The GAL4-UAS system was used to drive expression of green fluorescent protein (GFP) in identified motoneurons. Lines used included: heterozygous dHb9-GAL4 (Broihier and Skeath 2002) to label RP3 (thoracic MN6/7-1b homologue; see following text) [w;UAS-cd8-GFP;dHb9-GAL4], homozygous RRA-GAL4 (Fujioka et al. 2003) to label MN1-1b and MNISN-1s [w;UAS-cd8-GFP;RRA-GAL4], heterozygous vGLUT-GAL4 to label MNSNb/d-1s [w;UAS-cd8-GFP;vGLUT-GAL4], homozygous OK371-GAL4 (Mahr and Aberle 2006) to label MN1-1b, MNISN-1s, and RP3 [w;OK371-GAL4;UAS-cd8-GFP], and homozygous w;RN2-GAL4,UAS-mcd8-GFP;act<cd2<GAL4,UAS-flp for shal knockdowns. Vienna Stock Center #103363 (ShalRNAi) was used for shal knockdown experiments. vGLUT-GAL4 was a generous gift from the laboratory of S. Birman. The UAS-cd8-GFP source was Bloomington Stock Center #5137. For some recordings, Alexa 568 and/or rhodamine-dextran were included in the intracellular solution to confirm motoneuron identity based on neurite morphology and axon trajectories.

Names of motoneurons are designated according to peripheral targets (1b) and host nerve (1s) (Hoang and Chiba 2001), with the exception of the motoneuron referred to as RP3. “RP3” is the embryonic name of this motoneuron (Landgraf et al. 1997). In the larval abdomen, RP3 forms the 1b neuromuscular junction shared by muscles 6 and 7. Therefore the abdominal motoneuron is called MN6/7-1b. The muscle target of the thoracic motoneuron, however, is a previously undescribed ventral muscle (L. Feng and M. Landgraf, personal communication). Therefore we use the embryonic terminology RP3 for lack of a described and named muscle target.

To examine the possibility that differences in genetic background contributed to the differences that were observed in the functional properties among motoneurons, recordings were obtained from MN1-1b and MNISN-1s using OK371-GAL4 to drive GFP expression. The firing behavior and passive properties of these cells were not different from recordings made from cells labeled with GFP using the RRA-GAL4 driver line (data not shown). The parental line w;RN2-GAL4,UAS-mcd8-GFP;act<cd2<GAL4,UAS-flp served as the control line for the RNAi knockdown experiments.

Preparation

Dissection and desheathing of the preparation were performed in Ca²⁺-A solution (in mM: 118 NaCl, 2 NaOH, 2 KCl, 4 MgCl₂, 1.8 CaCl₂, 25 sucrose, 5 trehalose, 5 HEPES; ∼ 295 mOsm; pH 7.1) (Jan and Jan 1976). Osmolarity of all solutions was verified with a 5520 VAPRO osmometer (Wescor). Recordings obtained during fictive locomotion used a modified Ca²⁺-A solution containing 3.3 mM CaCl₂ and 2.5 mM MgCl₂. Protease 14 added to the Ca²⁺-A solution (2 mg/ml) was applied to the ventral ganglion to degrade the glial sheath through a micropipette, with the tip broken to a diameter of about 10 μm. Alternating sets of positive and negative pressures were administered to apply protease and to remove debris.

Electrophysiology

Thin-walled borosilicate electrodes were pulled on a PP-83 (Narishige Scientific Instruments) to a resistance of 4–7 MΩ and fire-polished using an MF-35 microforge (Narishige). Electrodes were filled with intracellular solution containing (in mM): 120 potassium gluconate, 10 HEPES, 1.1 EGTA, 2 MgCl₂, 0.1 CaCl₂, and 20 KCl (pH: 7.2; ∼290 mOsm). An Axoclamp 1B amplifier (Axon Instruments) with Clampex 10.1 software (Axon Instruments) was used for single-cell recordings. A Multiclamp 700B amplifier (Axon Instruments) was used for dual-cell recordings. Motoneurons with resting membrane potentials that were more depolarized than −40 mV or with seals <1 GΩ were not used. After verification of a GΩ seal, whole cell configuration was achieved. In voltage-clamp mode, a 40 mV hyperpolarizing voltage command was administered and the resulting current measurement was used to calculate input resistance. For current-clamp experiments, resting membrane potentials were brought to −60 mV through bias current injection into the cell body unless otherwise noted. Current injections were provided in intervals of 10 pA, from −10 to 80 pA, lasting 400 ms. Mean series resistances were: 13.95 ± 0.30 mΩ (MN1-1b), 15.38 ± 0.22 mΩ (RP3), 14.20 ± 0.39 mΩ (MNISN-1s), and 14.18 ± 0.54 mΩ (MNSNb/d-1s).

Tetrodotoxin (TTX, 1 μM; Sigma–Aldrich) was added to Ca²⁺-free A solution (modified Ca²⁺-A solution containing 0 mM CaCl₂) to isolate voltage-dependent K⁺ currents for voltage-clamp recording. A holding potential of −80 mV was used for voltage-clamp studies. The linear leakage current was subtracted from all records. Current amplitude was normalized to whole cell capacitance and reported as current density in current–voltage (I–V) plots. Whole cell capacitance was calculated from the charging transient following a 40 mV hyperpolarizing voltage command. I_A was isolated by subtracting the currents evoked by a series of voltage-clamp commands made from a holding potential of −40 mV from those obtained from a holding potential of −80 mV (see results).

The large degree of similarity in the excitatory postsynaptic potentials (EPSPs) and firing responses of cells in dual recordings could have reflected electrical coupling of the cells and/or cross talk between electrodes. However, current injected into one cell produced no response in the other cell (data not shown) and vice versa, ruling out the possibility of electrical coupling or reciprocal synaptic connections. Current injection into the bath also produced no response in the recording from the other electrode, ruling out the possibility of cross talk between electrodes (data not shown).

Statistical analysis

Statistical analyses were performed using a standard t-test with Excel software (Microsoft) except for measurement of delay-to-spike with and without a depolarizing prepulse (Fig. 3B), where a paired t-test was used. SE values are reported. Significance was assumed when P ≤ 0.05.

Fig. 3.

Delay-to-spike distinguishes 1b vs. 1s motoneurons and is altered by active properties. A: mean delay-to-spike is longer in 1s motoneurons per current injection level. MNISN-1s (n = 16) is significantly different from MN1-1b (n = 16) and RP3 (n = 21) between 50 and 110 pA. MNSNb/d-1s (n = 12) is significantly different from MN1-1b at 60, 80, and 90 pA, and from RP3 at 60, 70, 90, and 110 pA. Error bars are large at low current injection levels in 1s motoneurons because many individual neurons do not spike at low current injection levels and those that do often generate a spike only near the end of the 400 ms current injection. B: delay-to-spike for 80 pA current step after 1 s depolarizing prepulse. Decrease in delay-to-spike after +20 pA prepulse is significant for MN1-1b (n = 11) and MNSNb/d-1s (n = 11); decrease in delay after +40 pA prepulse is significant for MNISN-1s (n = 12) and MNSNb/d-1s. RP3 (n = 13) decreases were not significant. C: MNISN-1s firing response with no prepulse (top) and with a +40 pA prepulse (bottom); 20 pA current injection steps beginning at −10 pA are shown.

RESULTS

We were able to target and perform in situ whole cell patch-clamp recordings from GFP-labeled MN1-1b, MNISN-1s, RP3, and MNSNb/d-1s in the second and third thoracic segments of the larval ventral ganglion. Whereas MN1-1b and MNISN-1s innervate muscles in the dorsal body wall, the others innervate ventral targets. Examination of the cellular responses to current injection indicated that all motoneurons exhibited spike shapes comparable to those described in previous studies (Choi et al. 2004; Rohrbough and Broadie 2002; Worrell and Levine 2008) (Fig. 1). Large spikes were separated by noticeable afterhyperpolarization, with no evidence of accommodation or adaptation during 400 ms current injections. Measurement parameters included resting membrane potential, voltage threshold for spiking, current threshold for spiking (rheobase current), delay-to-spike, firing frequency, and passive membrane properties.

Fig. 1.

Current-clamp recordings of motoneuron spiking behavior; 1b motoneurons are blue, 1s motoneurons are red. Current injections in 20 pA intervals beginning at −20 pA are shown. Images show motoneuron cell bodies in thoracic segments 2 and 3 of the ventral ganglion except for MNSNb/d-1s image showing thoracic segment 3 and abdominal segment 1. Scale bars: 50 μm. A: arrow indicates MN1-1b. Asterisk indicates MNISN-1s.

The resting membrane potential of 1s motoneurons was significantly more hyperpolarized than 1b motoneurons (Table 1) and the voltage threshold was more depolarized. Voltage thresholds were: −23.81 ± 0.80 for MN1-1b; −24.04 ± 1.03 for RP3; −15.68 ± 0.92 for MNISN-1s; and −21.03 ± 1.12 for MNSNb/d-1s (Fig. 2A). MNISN-1s was significantly different from all other neurons, whereas MNSNb/d-1s displayed a trend toward difference from RP3 (P = 0.059) and MN1-1b (P = 0.057).

Table 1.

Properties of identified motoneurons

Motoneuron	Resting Membrane Potential, mV	Input Resistance, mΩ	Capacitance, pF
MN1-1b (n = 16)	−46.6 ± 1.3	666.0 ± 29.9	32.2 ± 2.0
RP3 (n = 21)	−47.1 ± 1.0	496.0 ± 17.8*	14.4 ± 0.8*
MNISN-1s (n = 16)	−50.6 ± 1.0*,**	1,051.7 ± 57.9*,**	24.9 ± 1.1*,**
MNSNb/d-1s (n = 12)	−51.4 ± 0.9*,**	1,090.8 ± 83.9*,**	22.9 ± 0.9*,**

Values are means ± SE, n values in parentheses. Type 1s and 1b motoneurons are significantly different in resting membrane potential, input resistance, and capacitance.

^*Indicates significant difference with MN1-1b;

^**indicates significant difference with RP3.

Fig. 2.

Voltage threshold distinguishes 1b vs. 1s motoneuron firing behavior. A: 1b motoneurons have a more hyperpolarized spike threshold than that of 1s motoneurons. MNISN-1s (n = 16) is significantly different from all other motoneurons, whereas MNSNb/d-1s (n = 12) shows a trend toward difference with MN1-1b (n = 16, P = 0.059) and RP3 (n = 21, P = 0.066). B: MN1-1b has a higher firing frequency than that of all other neurons for current injections ≥40 pA. C: current threshold does not distinguish 1b and 1s neurons. MN1-1b has a significantly lower current threshold than that of all other motoneurons. D: current injections produce significantly less depolarization in RP3 than that in other motoneurons at some current levels, but there are no systematic differences among the type 1b and 1s motor neurons.

Neither current threshold nor firing frequency distinguished 1b from 1s motoneurons. MN1-1b fired at a significantly higher frequency than that of all other neurons at current injections ≥40 pA (Fig. 2B). These data, combined with the observation that MN1-1b had a lower current threshold than that of all other cells (Fig. 2C), indicate that MN1-1b is the most excitable of the four cells.

MNISN-1s exhibited a significantly longer delay-to-spike for a given current injection than did 1b motoneurons at current injection levels between 60 and 110 pA. MNSNb/d-1s exhibited a significantly longer delay-to-spike than that of MN1-1b at 60, 80, and 90 pA injections and a longer delay than RP3 at 60, 70, 90, and 110 pA injections (Figs. 1 and 3A). Additionally, the delay-to-spike relationship for MN1-1b had a significantly more shallow slope than that of the 1s motoneurons (MN1-1b: −2.54 ms/pA vs. MNISN-1s: −4.65 ms/pA and MNSNb/d-1s: −5.38 ms/pA), likely a result of having an already short delay-to-spike at low current injections that excludes substantial shortening at higher injections. The longer delay-to-spike in 1s motoneurons suggests that 1s motoneurons may be recruited after 1b motoneurons during normal behavior (see following text).

Although they were more easily recruited by current injection, 1b motoneurons had a significantly lower input resistance than that of 1s motoneurons (Table 1). Additionally, 1b motoneurons did not achieve more depolarized membrane potentials per current injection than 1s motoneurons (Fig. 2D). There was no consistent pattern of difference between 1b and 1s whole cell capacitance, an indicator of cell size (Table 1). Therefore passive properties cannot explain the longer delay-to-spike in 1s neurons.

To determine whether thoracic motoneurons display firing properties similar to those of previously studied abdominal motoneurons, current-clamp protocols were administered to MN1-1b (n = 6) and MNISN-1s (n = 3) in abdominal segments 1–3. These cells exhibited the same relationships in firing behavior as those of thoracic MN1-1b and MNISN-1s (data not shown). Interestingly, the input resistance of abdominal MN1-1b (1,028.74 ± 127.95) was significantly higher than that of thoracic MN1-1b (665.99 ± 29.88), whereas its capacitance was significantly lower (abdominal MN1-1b: 24.363 ± 2.72; thoracic MN1-1b: 32.210 ± 2.01). Therefore abdominal MN1-1b uses a different strategy to achieve the same firing behavior as that of thoracic MN1-1b (see discussion).

Role of A-type current in firing properties

Active conductances may play an important role in the recruitment pattern of Drosophila motoneurons. Specifically, an inactivating K⁺ current I_A has been proposed to be important for regulating the delay-to-spike (Choi et al. 2004). To examine this hypothesis, depolarizing prepulses of +20 or +40 pA lasting 1 s, which are known to inactivate I_A (Salkoff and Wyman 1983), were administered prior to a current injection protocol (Fig. 3C). The delay-to-spike was decreased in all motoneurons, with the most dramatic reduction seen in 1s motoneurons (Fig. 3B). Firing frequency was not altered in any of the motoneurons (data not shown). These results are consistent with the hypothesis that I_A is responsible for at least a portion of the delay-to-spike and that it plays a relatively larger role in the firing behavior of 1s neurons.

Drosophila A-type currents are encoded by two genes, shaker and shal (Tsunoda and Salkoff 1995). Therefore RNA interference (RNAi) or mutations resulting in reduced function of one or both genes should result in a decreased delay-to-spike. shal was targeted for this study because shaker mutants were previously found to have no effect on delay-to-spike in abdominal motoneurons (Choi et al. 2004). A shal RNAi construct was therefore expressed in MN1-1b and MNISN-1s using a recombination-induced “flipout” strategy in which GAL4 activation in selected motoneurons is controlled by the actin promoter throughout embryonic and larval development (Hartwig et al. 2008). The parental line w;RN2-GAL4,UAS-mcd8-GFP;act<cd2<GAL4,UAS-flp served as the control line in these experiments. In motoneurons expressing the RNAi construct there was a decreased delay-to-spike in MN1-1b and MNISN-1s without a significant change in firing frequency compared with that of controls (Fig. 4, A and B).

Fig. 4.

ShalRNAi expression decreases delay-to-spike and increases firing frequency. ShalRNAi expression in MN1-1b and MNISN-1s significantly modified firing behavior. A: current-clamp recordings from control and ShalRNAi-expressing cells. Obvious decreases in delay-to-spike are observed in MN1-1b and MNISN-1s. B: ShalRNAi decreased delay-to-spike in MN1-1b ShalRNAi (n = 3) and MNISN-1s ShalRNAi (n = 5) compared with control MN1-1b (n = 6) and MNISN-1s (n = 5), respectively. Significant decreases for MN1-1b were observed at current injection levels ≥80 pA. Significant differences for MNISN-1s were observed at 70, 100, and 110 pA current injection levels. Firing frequency was not significantly changed in MN1-1b or MNISN-1s ShalRNAi-expressing cells.

Voltage-clamp measurements of I_A in MN1-1b and MNISN-1s were performed to verify the effect of ShalRNAi knockdown on the A-type current (Fig. 5, A and B). I_A was isolated by subtracting the record obtained at −40 mV holding potential from the record obtained at −80 mV holding potential and measuring the resulting peak transient outward current. Control MN1-1b and MNISN-1s did not differ significantly in I_A or sustained outward current densities (see discussion). I_A density was decreased in ShalRNAi-expressing MN1-1b and MNISN-1s compared with that of controls (Fig. 5, A and B). Sustained current density was not different in ShalRNAi compared with that in control cells (Fig. 5B).

Fig. 5.

ShalRNAi expression decreases A-type K⁺ current (I_A). A: voltage-clamp recordings of K⁺ currents from control and ShalRNAi-expressing MN1-1b and MNISN-1s. Voltage steps of 20 mV starting at −120 mV from a holding potential of −80 mV are shown. The transient peak outward current is absent from ShalRNAi cells. B: voltage-clamp record obtained from a holding potential of −40 mV was subtracted from voltage-clamp record obtained from a −80 mV holding potential to isolate I_A. Representative subtracted records are shown for a MNISN-1s control cell (left) and a MNISN-1s ShalRNAi cell (right). C: current–voltage (I–V) plot of I_A density and sustained outward current density in control MN1-1b (n = 5), ShalRNAi MN1-1b (n = 3), MNISN-1s (n = 5), and ShalRNAi MNISN-1s (n = 4). Significant decreases in MNISN-1s I_A density were observed at −80, 20, 40, and 60 mV steps. No differences in sustained current density were observed.

Responses of motoneurons to synaptic drive during fictive locomotion

Whole cell patch-clamp recordings were obtained from semiintact preparations undergoing fictive locomotion to determine whether the differences observed between 1b and 1s motoneurons in current-clamp experiments are behaviorally relevant. Extended periods of rhythmic activity were frequently observed in whole cell records from freshly dissected preparations. This activity correlated with visible rhythmic peristaltic contractions of the intact body wall muscles. The cycle period of these rhythmic bouts segregated into three groups: short cycle periods of <2 s (consistent with twitches, rejected from analysis), medium cycle periods between 5 and 12 s (consistent with backward fictive locomotion), and long cycle periods >15 s (consistent with forward fictive locomotion) (Fox et al. 2006). Simultaneous recordings from MN1-1b on the right and left sides of abdominal segment 1 during fictive crawling indicate that the cells fire bursts of action potentials in response to rhythmic synaptic input (Fig. 6, top). When the cells were held at similar membrane potentials, they displayed similarly shaped drive potentials and fired nearly coincident action potentials (Fig. 6, bottom). Right and left pairs of MNISN-1s also displayed simultaneous depolarizations, although not every cycle of synaptic input elicited bursts of action potentials. If one MNISN-1s did not fire action potentials, neither did its contralateral homologue. Depolarizing one of the cells via bias current injection caused it to fire more action potentials per burst than its paired homologue.

Fig. 6.

Dual recordings from MN1-1b right and left homologues. Dual whole cell patch-clamp recordings from contralateral MN1-1b homologues in the first abdominal segment of the ventral ganglion during fictive locomotion. Cells receive coincident excitation and fire nearly simultaneously.

Simultaneous recordings were also obtained from MN1-1b and MNISN-1s projecting to the same body wall hemisegment (Fig. 7A). The underlying drive potentials displayed similar time courses and amplitudes, although MN1-1b fired earlier than MNISN-1s and fired more action potentials per burst than did MNISN-1s (Fig. 7B). In many recordings, the drive potentials of MNISN-1s and MN1-1b followed the same time course and elicited firing in MN-1b but not in MNISN-1s (data not shown). When a current step protocol was administered to both cells during the interburst interval, MN1-1b required less current injection to initiate spiking and MNISN-1s displayed its characteristic delay to first spike. Therefore the cells embedded in a functioning network display similar firing properties as when driven by current injection.

Fig. 7.

Dual recordings from MN1-1b and MNISN-1s projecting to same body wall hemisegment. A: dual whole cell patch-clamp recordings from MNISN-1s (red) and MN1-1b (blue) display coincident rhythmic drive during fictive locomotion. B: MNISN-1s fires later than MN1-1b in response to synaptic input during fictive locomotion. C: current-clamp recordings from pair of cells shown in A. Current injections of −20, 10, 50, 110, 130, and 150 pA are shown.

The synaptic drive to MN1-1b and MNISN-1s during fictive locomotion was examined in simultaneous recordings from cells that were hyperpolarized to prevent activation of voltage-dependent channels. Under these conditions, the drive potentials of MN1-1b and MNISN-1s were nearly identical (Fig. 8B). In some recordings the rhythmic synaptic drive did not evoke action potentials, but took the form of individual EPSPs that occurred nearly one-for-one in MN1-1b and MNISN-1s (Fig. 8C), indicating a common source of synaptic drive.

Fig. 8.

MN1-1b and MNISN-1s receive common synaptic input. A: dual patch-clamp recording from MNISN-1s (red, top) and MN1-1b (blue, bottom) during fictive locomotion display nearly identical drive potentials. B: overlay of simultaneous bursts generated in MNISN-1s and MN1-1b, highlighting similar drive potential amplitude and shape when resting at −60 mV (left) and when brought to −80 mV to block spiking (right). C: in some recordings, it was possible to discriminate one-for-one excitatory postsynaptic potentials (EPSPs) that did not elicit action potentials in MNISN-1s and MN1-1b.

DISCUSSION

We find that resting membrane potential and voltage threshold distinguish 1b and 1s motoneurons and provide evidence that shal is the gene encoding the A-type K⁺ channel that is largely responsible for the characteristic delay-to-spike. When MN1-1b and MNISN-1s were recorded simultaneously during fictive locomotion, MNISN-1s displayed its characteristically longer delay-to-spike, although the time course and amplitude of the drive potentials were similar. MNISN-1s often failed to fire action potentials in response to the shared synaptic drive. Therefore it appears that the increased delay-to-spike and higher voltage threshold in 1s motoneurons play a functional role in the recruitment of motoneurons during locomotion.

Type 1b motoneurons exhibit a more hyperpolarized voltage threshold than that of 1s motoneurons. Type 1b neurons also have a shorter delay-to-spike following current injection. These characteristics, together with the more depolarized resting potential of 1b motoneurons, lead to the prediction that 1b motoneurons are more easily recruited than 1s motoneurons. Other firing behavior parameters, such as firing frequency, failed to differentiate between 1b and 1s motoneurons, indicating that recruitment order is the primary distinction between the two types. The short delay-to-spike in1b motoneurons indicates that they are functionally analogous to easily recruited, low-threshold motoneurons described in other organisms and the long delay-to-spike in1s motoneurons indicates that they are functionally analogous to slowly recruited, high-threshold motoneurons. Our results provide no support for a division of 1b and 1s motoneurons into tonic and phasic types because neither type responded to current injection or synaptic input in a phasic manner.

It is of interest that although 1b and 1s motoneurons displayed significant differences in voltage threshold, there was also variation between the dorsally and ventrally projecting 1s motoneurons. Interestingly, there appears to be an offset in ventral muscle contraction compared with their dorsal counterparts during crawling (see Supplemental videos from Fox et al. 2006; Hughes and Thomas 2007). It is thus either the case that the crawling CPG provides offset input to dorsally and ventrally projecting motoneurons or that the intrinsic properties of dorsally and ventrally projecting motoneurons impart distinct delays to recruitment. Our studies revealed few differences in the firing properties of dorsally and ventrally projecting motoneurons, suggesting the former. Nevertheless, the voltage threshold for firing of MNISN-1s was significantly higher than that of MNSNb/d-1s. Such a difference in threshold may cause MNISN-1s to fire after MNSNb/d-1s during crawling, resulting in an offset in dorsal and ventral muscle group recruitment. Focal patch recordings of synaptic currents at the neuromuscular junction (Fox et al. 2006) and extracellular nerve recordings from dorsally and ventrally projecting nerves (E. McKiernan and C. Duch, personal communication) indicate an offset in dorsal and ventral muscle/nerve activation during fictive locomotion, supporting this idea. Simultaneous intracellular recordings from MNISN-1s and MNSNb/d-1s during fictive locomotion could resolve this question.

The resting membrane potential of 1b motoneurons was significantly depolarized compared with 1s motoneurons. It is likely that the more depolarized resting state of 1b motoneurons facilitates 1b motoneuron recruitment by reducing the drive necessary to initiate spiking. This may contribute to the more reliable recruitment of MN1-1b than MNISN-1s during bouts of fictive locomotion. Indeed, depolarizing bias current during fictive locomotion increased the reliability of MNISN-1s recruitment.

In mammals, ordered motoneuron recruitment is explained by the size principle (Henneman and Mendell 1981). According to the size principle, the longer delay-to-spike in 1s motoneurons should be a consequence of a lower input resistance. However, 1s motoneurons have a significantly higher input resistance than that of 1b motoneurons. Additionally, rheobase current (current injection to spike) was not a parameter that distinguished 1b from 1s motoneurons in this study. If the size principle were guiding recruitment, a higher rheobase would be expected in 1s motoneurons. Therefore larval Drosophila motoneurons must use a strategy other than the size principle to generate ordered recruitment. Interestingly, Drosophila motoneurons of the same identity may also use different strategies to generate appropriate recruitment and firing behavior because abdominal MN1-1b exhibits recruitment timing and firing behavior like that of thoracic MN1-1b, but has a significantly higher input resistance and lower whole cell capacitance. Future studies to compare active currents in thoracic and abdominal MN1-1b may be useful to understand how the higher input resistance in abdominal MN1-1b fails to alter recruitment timing, measured as delay-to-spike, compared with thoracic MN1-1b.

Role of A-type current in firing properties

Active properties of motoneurons, specifically A-type K⁺ currents, appear to determine motoneuron recruitment order in Drosophila larvae. Two A-type channels are present in Drosophila: Shaker and Shal. Genetically targeted expression of ShalRNAi in MN1-1b and MNISN-1s significantly reduced delay-to-spike. Voltage-clamp recordings indicate that I_A is decreased in MN1-1b and MNISN-1s cells expressing ShalRNAi. The decreased A-type current correlates with the shorter delay-to-spike observed in these cells. Therefore the shal-encoded A-type current is the primary determinant of the timing of larval motoneuron recruitment.

However, differences between 1b and 1s motoneuron delay-to-spike cannot be explained solely by differences in A-current density because measurements of A-current density were equal in MN1-1b and MNISN-1s. Importantly, our results agree with those described by Choi et al. (2004), indicating that when an inactivating prepulse protocol was administered prior to a −40 mV test pulse, I_A in MN1-1b exhibited greater inactivation than that in MNISN-1s at voltages at least −60 mV (data not shown). Therefore the voltage dependence of I_A inactivation is different in MN1-1b and MNISN-1s, such that a greater portion of I_A undergoes closed-state inactivation near rest in MN1-1b. The more depolarized resting potential of MN1-1b would also tend to encourage I_A inactivation. The greater inactivation of I_A in 1b motoneurons appears to cause the shorter delay-to-spike. It cannot be concluded, however, that this result applies to all 1b and 1s motoneurons because recordings were not performed in MNSNb/d-1s and RP3.

Responses of motoneurons to synaptic drive during fictive locomotion

We obtained recordings of MN1-1b and MNISN-1s firing behavior during fictive crawling to interpret the behavioral relevance of the firing responses of these cells to current injection. Dual patch-clamp recordings were performed from pairs of MN1-1b, MNISN-1s, and MN1-1b/MNISN-1s motoneurons during fictive locomotion to determine the relative timing and amplitude of the synaptic drive to MN1-1b and MNISN-1s as well as their firing patterns in response to this drive. Recordings from contralateral pairs of MN1-1b or MNISN-1s motoneurons displayed nearly one-to-one firing patterns as well as synaptic drive potentials with nearly identical time courses and amplitudes. Although MN1-1b fired bursts of action potentials during each bout of rhythmic drive, MNISN-1s often did not. When the cells were hyperpolarized to block activation of voltage-dependent channels, drive potentials were similar in amplitude, duration, and timing between cells. This suggests a higher voltage threshold for recruitment of MNISN-1s than MN1-1b in response to behaviorally relevant synaptic input. Further, dual recordings from MN1-1b and MNISN-1s indicated that, when MNISN-1s did fire a burst of action potentials, it displayed its characteristic delay-to-spike relative to MN1-1b, supporting the role of an A-type K⁺ current carried by Shal in patterning recruitment.

The observation of coincident EPSPs in MN1-1b and MNISN-1s suggests that at least some synaptic input is shared between the two cells. Anatomical studies of motoneuron dendritic morphology within the ventral ganglion have shown that dendritic arborizations of motoneurons projecting to a shared body wall region are segregated to a distinct neuropil region (Landgraf et al. 2003) and show extensive overlap (Kim et al. 2009). MN1-1b and MNISN-1s project to shared dorsal internal muscles and therefore are likely to exhibit overlapping dendritic arborizations in a shared neuropil region displayed in Fig. 1.

Overall, it appears that the A-type current carried by Shal channels is important in determining the delay-to-spike in motoneurons. The delay-to-spike, then, imparts a recruitment order to 1b and 1s motoneurons that would support an initial development of low force in single muscles driven by 1b motoneurons followed by the recruitment of high-threshold 1s motoneurons to drive the contraction of groups of muscles to generate higher levels of force. The importance of the A-type current for delay-to-spike does not exclude the contribution of other aspects of motoneuron identity and intrinsic properties to motoneuron function. Rather, it is likely that the distinction between 1b versus 1s motoneuron type is one facet of individual motoneuron identity. Transcription factor expression pattern determines individual motoneuron identity, including electrical properties (Pym et al. 2006). The unique electrical properties of a neuron can function to produce properties that are shared by 1b and 1s motoneurons, such as delay-to-spike, whereas other parameters may be unique to the individual motoneuron based on its genetic identity.

GRANTS

This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-28495 to R. B. Levine, National Institute of General Medical Sciences Fellowship T32 GM-008400 to J. Schaefer, and National Science Foundation Fellowship 0638744 to J. Schaefer.

DISCLOSURES

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