Respiratory motoneuron properties during the transition from gill to lung breathing in the American bullfrog

Abstract

Amphibian respiratory development involves a dramatic metamorphic transition from gill to lung breathing and coordination of distinct motor outputs. To determine whether the emergence of adult respiratory motor patterns was associated with similarly dramatic changes in motoneuron (MN) properties, we characterized the intrinsic electrical properties of American bullfrog trigeminal MNs innervating respiratory muscles comprising the buccal pump. In premetamorphic tadpoles (TK stages IX–XVIII) and adult frogs, morphometric analyses and whole cell recordings were performed in trigeminal MNs identified by fluorescent retrograde labeling. Based on the amplitude of the depolarizing sag induced by hyperpolarizing voltage steps, two MN subtypes (I and II) were identified in tadpoles and adults. Compared with type II MNs, type I MNs had larger sag amplitudes (suggesting a larger hyperpolarization-activated inward current), greater input resistance, lower rheobase, hyperpolarized action potential threshold, steeper frequency-current relationships, and fast firing rates and received fewer excitatory postsynaptic currents. Postmetamorphosis, type I MNs exhibited similar sag, enhanced postinhibitory rebound, and increased action potential amplitude with a smaller-magnitude fast afterhyperpolarization. Compared with tadpoles, type II MNs from frogs received higher-frequency, larger-amplitude excitatory postsynaptic currents. Input resistance decreased and rheobase increased postmetamorphosis in all MNs, concurrent with increased soma area and hyperpolarized action potential threshold. We suggest that type I MNs are likely recruited in response to smaller, buccal-related synaptic inputs as well as larger lung-related inputs, whereas type II MNs are likely recruited in response to stronger synaptic inputs associated with larger buccal breaths, lung breaths, or nonrespiratory behaviors involving powerful muscle contractions.

INTRODUCTION

An understanding of the evolutionary origins of neural networks that produce and regulate breathing offers the promise of revealing fundamental mechanisms at the core of this vital behavior. Among the most significant respiratory adaptations arising in the vertebrate lineage was the evolution of ventilated, air-breathing organs in bony fish and amphibians; these structural/functional changes were key to the invasion of terrestrial environments (35). While we can only speculate about the evolutionary changes in central respiratory control networks accompanying this transition (29), we can directly study relevant processes in extant amphibians as they undergo the remarkable developmental transition from aquatic to air-breathing environments (5).

Upon hatching, American bullfrog tadpoles (Lithobates catesbeianus) accomplish gas exchange by means of a continuous, low-amplitude motor rhythm driving the buccal musculature and rhythmic pumping of water across the gills (42). The dominant buccal motor rhythm is punctuated with occasional, large-amplitude bursts driving forceful buccal contractions that “push” air into the immature lungs via positive pressure (62). With development, environmental cues stimulate the release of thyroid hormones and trigger metamorphosis (17), which, at the level of the gas-exchange system, involves gill regression and lung maturation (11). During metamorphosis and into adulthood, the high-amplitude lung motor rhythm gains prominence, and there is a progressive reliance on lung ventilation to meet metabolic demands. Whereas ongoing buccal breathing in tadpoles is interrupted by single lung breaths, lung breathing in adults often occurs in episodes of multiple breaths, especially when respiratory drive (e.g., CO₂) is elevated (37). Importantly, the lung motor pattern does not replace the buccal breathing pattern; low-amplitude, high-frequency contractions in adult frogs produce buccal oscillations contributing to olfaction and refreshment of air in the oral cavity. Thus, tadpoles and adult frogs express two qualitatively distinct respiratory motor behaviors, with the relative prominence of each pattern shifting through development.

Breathing, like other rhythmic motor behaviors (e.g., locomotion and mastication), is produced by central pattern-generating networks comprising rhythm- and pattern-generating components. Gill and lung breathing in amphibians originates from the activities of spatially distinct rhythm generators located in rhombomeres 7/8 (buccal rhythm) and 4/5 (lung rhythm) (4, 20, 42, 54), which provide drive to downstream pattern-forming premotoneurons and motoneurons (MNs). Far from being passive input-output elements, respiratory MNs express multiple voltage- and ligand-gated conductances that are key determinants of motor output pattern. Importantly, the buccal and lung rhythm generators utilize the same MNs, pump muscles, and pumping mechanism to produce buccal and lung breaths (21, 56). In other words, both rhythm generators project to, and engage, a common MN pool to produce distinct respiratory behaviors. While developmental changes in rhythm generators likely contribute to patterns of respiratory MN output through ontogeny, potential changes at the MN level are yet to be addressed in nonmammalian vertebrates. With this in mind, the present study sought to determine whether intrinsic properties of respiratory MNs change in parallel with developmental changes in respiratory motor patterns that accompany the transition from aquatic to air breathing. We hypothesized that MN properties change developmentally in a manner facilitating generation of high-amplitude, rhythmic lung bursts of the adult frog over low-amplitude, higher-frequency buccal bursts of the tadpole.

Here, we focused on depolarizing “sag” responses, indicative of the hyperpolarization-activated cation current (I_h). The unique voltage dependence and slow activation/inactivation kinetics of depolarizing sag contribute to rhythmic behaviors through autorhythmic/pacemaker mechanisms and postinhibitory rebound (PIR) (41). Moreover, I_h could significantly impact MN responses to different patterns of respiratory drive originating from multiple rhythm generators. I_h is widely expressed in central neurons, and, in mammals, current magnitude increases severalfold during postnatal development (7, 15, 59). In amphibians, I_h occurs in locus coeruleus chemoreceptive neurons and paravertebral sympathetic ganglia, and I_h emerges with development in spinal MNs of Xenopus tadpoles (16, 51, 58). However, how I_h changes developmentally in amphibians, including through metamorphosis, and its role in respiratory MNs generating buccal versus lung motor patterns are not known. Thus our first objective was to characterize the intrinsic membrane properties, with a focus on I_h, of tadpole MNs that innervate key jaw muscles powering gill and, ultimately, lung breathing. We analyzed trigeminal MNs, because 1) they innervate the masseter major, m. temporalis, and m. submentalis muscles involved in buccal and lung breaths [as well as nonrespiratory functions that include movement of lymphatic fluid (27), vocalization, and prey capture], 2) respiratory motor output arising from the trigeminal nerve has been well characterized in most developmental stages of bullfrogs (21), and 3) owing to its size and location, this nerve can be easily identified for retrograde labeling, thus ensuring consistency between animals. The second objective was to assess developmental changes in membrane properties, including I_h, of trigeminal MNs that occur with the emergence of adult respiratory motor patterns. To this end, we performed morphometric analysis and whole cell recordings of trigeminal MNs in premetamorphic tadpoles (TK stages IX–XIII) and sexually mature adult frogs.

METHODS

Animals and ethical approval.

Experiments were performed on brain stem slice preparations from American bullfrog (Lithobates catesbeianus) premetamorphic tadpoles and sexually mature adult frogs of both sexes. Bullfrogs were obtained commercially (Island Bullfrog, Nanaimo, BC, Canada) and housed in aquaria containing flowing, filtered, and dechlorinated Québec City water at 19–22°C. A 12:12-h light-dark photoperiod was implemented. Tadpoles were fed goldfish food (Goldy Royal, Sera), frozen spinach, and naturally present algae, and adults were fed live crickets (CG Reptile, Saint-Apollinaire, QC, Canada), all on a daily basis. All protocols and methods used for anesthesia and euthanasia were approved by the Université Laval Animal Care Committee and performed in compliance with the guidelines of the Canadian Council on Animal Care.

Tadpole stages were determined according to the morphological criteria of Taylor and Kollros (57). Stage IX–XVIII tadpoles were assigned to the premetamorphic group (20 animals); fully developed, sexually mature frogs were assigned to the adult group (18 animals). Our choice of developmental stages was based on the timing of metamorphosis and concurrent emergence of the lungs as the primary gas-exchange surface. Specifically, metamorphic stages XX–XXV are characterized by a ~10- to 15-mmHg increase in systemic CO₂ tension (33), indicating a transition to aerial breathing with little to no reliance on gill ventilation. By considering stage IX–XVIII premetamorphic tadpoles and adult frogs, our data encompass two distinct physiological periods marked by reliance on gill and lung ventilation, respectively.

Retrograde labeling of MNs.

The fluorescent, lipophilic tracer DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindo-carbocyanine perchlorate) and methods modified slightly from those described by Vidal-Sanz et al. (61) were used to retrogradely label trigeminal MNs in vivo. Briefly, the injection solution was prepared as 3 mg of DiI salt (sonicated in darkness for 6–7 h; Life Technologies, Waltham, MA) dissolved in 1 ml of 0.9% NaCl and 20 µl of Triton X-100. The final DiI solution was sonicated for 4–6 h and stored in darkness at −20°C until use. Experimental animals were anesthetized via full immersion into tricaine methanesulfonate [a mixture of 2.8 g/l tricaine methanesulfonate (MS-222; Sigma-Aldrich, Oakville, ON, Canada) and 8.57 g/l NaHCO₃ diluted with 2 parts home tank water]. Tadpoles were injected bilaterally with 20 µl of prepared DiI (3.15 mM) through the skin into the ventrally situated intermandibularis muscle, ~0.5 cm from the mouth and adjacent to the mandible, where the mandibular branch of the trigeminal nerve passes (26). In adults, a small incision adjacent to the mandible permitted injection of DiI directly into the sheath surrounding the mandibular branch of the trigeminal nerve. Animals were allowed to recover for ≥10 days following the injections. For adults, the incision site was sutured closed (coated VICRYL 5-0, Ethicon, Somerville, NJ), and the animals were monitored daily for infection and to ensure feeding. [DiI is stable in vivo without loss of signal for >21 days (30).]

To confirm the effectiveness of our retrograde staining protocol, we initially imaged brain stem slices to locate and visualize trigeminal MNs using premetamorphic tadpoles (TK stages V–XIII, n = 5) separate from those utilized in soma area calculations and electrophysiology. Brain stems were dissected as described below (see In vitro slice preparations), fixed in 4% paraformaldehyde for 24 h at 4°C and then in 30% sucrose + paraformaldehyde for 48 h at 4°C, and frozen in embedding medium (HistoPrep, Fisher Scientific, Ottawa, ON, Canada) at −80°C until use. Serial cross sections (50 µm) of pontomedullary regions containing the trigeminal nerve were obtained using a cryostat (model CM1850, Leica) and plated onto slides with antifade mounting medium and micro (0.13–0.17 mm) coverslips (VWR, Mississauga, ON, Canada). To identify the anatomic location of DiI-labeled MNs, slides were first imaged at low power (×10) using a fluorescence microscope (Eclipse E600, Nikon, Mississauga, ON, Canada) with a Texas red filter (excitation 540–580) and a Lumenera camera (Infinity 3, Lumenera, Ottawa, ON, Canada). High-resolution confocal images were subsequently collected using a laser-scanning microscope (model LSM 800 with Airyscan, Zeiss, Toronto, ON, Canada) under the ×10 and ×20 objectives and Zen 2.1 software. In Fig. 1A1, fluorescence and phase-contrast images were merged, midtone contrast was enhanced, and background was lightened to delineate the brain stem slice. Midtone contrast was enhanced for better visualization of ×10 images in Fig. 1A2. In Fig. 1A3, z stacks were combined (65.04–73.18 µm, 1.12 µm thick) to show MN soma in the trigeminal motor nucleus. The background was darkened and midtone contrast was enhanced in Fig. 1, B1 and B2, to highlight the soma.

Fig. 1.

DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindo-carbocyanine perchlorate) fluorescence reveals morphology of amphibian trigeminal motoneurons (MNs). A1: in tadpoles, low-magnification (×4) fluorescence imaging revealed DiI-stained trigeminal MN cell bodies and axons in caudal pontine slices, ventrolateral to the 4th ventricle (IV Ven) and caudal to the cerebellum. A2: confocal differential interference contrast (DIC) image (×10) from a tadpole shows trigeminal MN soma (arrowheads). A3: DiI labeling of MNs in A2 at ×20 (arrowheads). Nuclei are not stained with DiI. B1: in live slice preparations, DiI-positive trigeminal MNs were identified for soma area analysis. B2: a DiI-positive MN was targeted with a patch electrode using DIC (×40). Scale bars = 50 nm. C: soma area, calculated using DIC images (as in B2), was significantly smaller in tadpoles (n = 19 MNs from 8 animals) than adults (n = 17 MNs from 8 animals). ○, Outliers. *P < 0.0001.

In vitro slice preparations.

Tadpoles were fully anesthetized by immersion into prepared MS-222. Adult frogs underwent the same procedure, but with MS-222 kept on ice to slow metabolism (64). The cranium was opened, and the intact brain stem (rostral to the optic lobes) and spinal cord (caudal to cranial nerve XII) were removed. During dissection, neural tissue was irrigated at regular intervals with cold (0–5°C) artificial cerebrospinal fluid (aCSF) to maintain tissue health and reduce neuronal activity. The aCSF used for dissection and experimentation for tadpoles contained (in mM) 90 NaCl, 4 KCl, 1.4 MgCl₂, 2.4 CaCl₂, 25 NaHCO₃, 1 NaH₂PO₄, and 7.5 d-glucose; the aCSF used for adult bullfrogs contained (in mM) 75 NaCl, 4.5 KCl, 1 MgCl₂, 2.5 CaCl₂, 40 NaHCO₃, 1 NaH₂PO₄, and 7.5 d-glucose. The aCSF for premetamorphic tadpoles was equilibrated with 1.8% CO₂-98.2% O₂ to pH 7.90 ± 0.15, while that for adult frogs was equilibrated with 2.5% CO₂-97.5% O₂ to pH 7.80 ± 0.15 (22, 36). The use of aCSF with a higher bicarbonate concentration in adult frogs is established (63) and reflects metabolic compensation of respiratory acidosis during the transition to air breathing (33). To ensure that differences in MN membrane properties between tadpole and adult frogs were not due to the different compositions of aCSF, we recorded membrane properties of tadpole and adult MNs during serial superfusion with tadpole vs. adult aCSF (order randomized). The aCSF bicarbonate composition had no significant effect on any of the electrophysiological properties measured in tadpole and adult MNs. Therefore, we were confident to perform data analysis on MNs perfused with their “native” developmental aCSF bicarbonate composition and compare results between tadpoles and adults.

Instant adhesive (Roti Coll 1, Techmate, Milton Keynes, UK) was used to attach isolated, intact brain stems caudal surface-down, to 4% agarose blocks, and a vibratome (model VT1000S, Leica, Concord, ON, Canada) was used to cut serial coronal sections, in ice-cold aCSF, in the rostral-to-caudal direction. Sections (250 µm thick) from the region encompassing the trigeminal motor nucleus (rostral medulla) were saved and allowed to recover for 1 h in oxygenated aCSF at room temperature (22°C) before imaging and whole cell-recording experiments.

Visualization of trigeminal MNs for electrophysiology and soma area calculations.

Brain-stem slices were placed in the head-stage well of a fluorescence microscope (model BX51WI, Olympus, Richmond Hill, ON, Canada) equipped for differential interference contrast (DIC) microscopy, secured with a nylon mesh, and perfused at 2 ml/min with native aCSF equilibrated with an O₂-CO₂ mixture as described above, and maintained at room temperature (22°C). Gas mixtures were controlled using a gas mixer (model GSM-3, CWE, Ardmore, PA). Trigeminal MNs were identified on the basis of their relatively large soma size, anatomic landmarks (4th ventricle, trigeminal nerve, medial to the main trigeminal sensory nucleus), and DiI labeling (Fig. 1A1) using a ×40 water-immersion objective and U-RFL-T mercury lamp with tetramethylrhodamine filter (excitation 549 nm, emission 565 nm). We selected healthy trigeminal MNs for imaging (using DiI fluorescence) and subsequent whole cell recordings; equal numbers of type I and II MNs were targeted to ensure that enough replicates were obtained. The soma of targeted MNs was imaged (using DIC) before electrophysiological recordings using an Olympus XM10 camera and corresponding CellSens software (ver. 1.12, equipment and software from Olympus). Soma area (µm²) was calculated from acquired DIC images using z stacks analyzed in ImageJ software (19 MNs from 8 tadpoles and 17 MNs from 8 adults).

Electrophysiology.

Whole cell recordings were performed on DiI-labeled trigeminal MNs from tadpole and adult slice preparations (15 type I and 10 type II MNs from 10 tadpoles and 28 type I and 28 type II MNs from 18 adults). Gigaseal formation was performed using DIC visualization under voltage-clamp configuration. Briefly, 4 to 6.5 MΩ resistance patch pipettes were pulled (model PC-10 vertical puller, Narishige, Amityville, NY) from 1.12-mm inside-diameter filamented borosilicate glass (World Precision Instruments, Sarasota, FL) and filled with solution containing (in mM) 110 K-gluconate, 10 HEPES, 2 MgCl₂, 1 ATP-Mg₂, 0.1 GTP-Na, and 2.5 EGTA, with pH adjusted to 7.2. Data were recorded using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA), digitized using Axon Digidata 1550 (Molecular Devices), sampled at 10 kHz, filtered at 2 kHz, and acquired using pCLAMP 10 software (ver.10.7, Molecular Devices). Series resistance was monitored under voltage-clamp throughout the protocol, and recordings showing a >20% variation in series resistance between baseline and subsequent measurements were not analyzed. The average series resistances were as follows: mean 20.5 MΩ [95% confidence interval 19.2 MΩ (lower), 21.8 MΩ (upper)] for tadpoles and mean 16.1 MΩ [95% confidence interval 15.0 MΩ (lower), 17.2 MΩ (upper)] for adults. Resting membrane potential was measured <30 s after a stable whole cell configuration was established and before any manipulations were performed. To determine the fidelity of the recording and input resistance, a series of identical square-wave current pulses (+5 mV, 40 ms) were injected into the cell. Input resistance was calculated by averaging the resulting traces and calculating the change in steady-state current as follows: resistance (MΩ) = 5 × 10⁻³ V/ΔI, where ΔI is change in current (in pA).

A series of hyperpolarizing and depolarizing current steps (20-pA steps over a range of −200 to +920 pA, 1,000 ms) were applied to determine membrane properties. These properties, and their significance for MN activity and motor behavior, are summarized in Table 2. Electrophysiological data were analyzed offline using Clampfit software (ver. 10.7, Molecular Devices). Sag amplitude was measured at −100 mV as the difference between peak and steady-state membrane responses; PIR amplitude was measured as the peak membrane response within 300 ms following termination of the −100-mV hyperpolarizing current pulse. Rheobase was measured as the current step magnitude (1,000 ms) from resting membrane potential required to elicit a single action potential. Action potential properties were measured from the first elicited spike as follows: before the action potential upstroke (threshold), from threshold to peak (amplitude), at half the calculated amplitude (half-width), and from threshold to maximum hyperpolarization [afterhyperpolarization (AHP) fast/medium]. Excitatory postsynaptic currents (EPSCs) were recorded for 5 min at −60 mV under voltage clamp, and frequency [Hz (number of events/s)] was analyzed using custom event detection templates in Clampfit. Additional criteria for including MNs in the analysis included a stable resting membrane potential (initially −40 mV or more hyperpolarized) and action potentials that were >40 mV in amplitude and peaked above 0 mV.

Table 2.

Functional significance of membrane properties recorded from trigeminal MNs of L. catesbeianus

Property	Definition	Predicted Outcome for In Vivo Motor Activity
Neuronal excitability	Manner in which inputs are transformed by neurons into patterns of AP output; it is determined by interactions between membrane (passive and active) and synaptic properties	Compared with neurons with low excitability, highly excitable neurons respond to the same input, but with greater depolarization and greater AP discharge. Common features of highly excitable neurons include higher input resistance, lower rheobase, and steeper f/I relationships. Excitability of a specific neuron is not constant; it is highly modulated by neurochemicals that can increase or decrease excitability
Input resistance, mV	A function of neuron area and specific membrane conductance; it is measured by steady-state voltage response to a current pulse or current response to a voltage pulse	Excitability of MNs within a pool varies widely. The more excitable MNs (↑ input resistance, ↓ rheobase) tend to be recruited first in response to depolarizing synaptic input; they are typically smaller motor units driving smaller gradations in muscle force. Stronger synaptic inputs tend to recruit MNs with progressively lower excitability (↓ input resistance, ↑ rheobase); these are often larger motor units that generate larger increments in force
Rheobase, pA	Magnitude of depolarizing current needed to elicit an action potential 50% of the time
Hyperpolarization-activated inward current (Ih) amplitude, mV	A noninactivating inward (depolarizing) current that is progressively activated with increasing membrane hyperpolarization	Ih influences MN excitability by 1) contributing a small depolarization to the resting membrane potential; 2) activation of Ih by a hyperpolarizing input results in a slow-onset depolarizing sag (that increases with hyperpolarization), reflecting slow activation kinetics (this counteracts inhibitory inputs but is also important in controlling the time between APs or bursts of APs); 3) withdrawal of inhibition can result in PIR. Properties 2 and 3 contribute to the timing and pattern of APs during rhythmic behaviors
Postinhibitory rebound (PIR) amplitude, mV	The short-lasting depolarizing “overshoot” in membrane potential following termination of a hyperpolarizing input; it occurs in neurons with Ih (an inward depolarizing current), because Ih remains active (deactivates slowly) for a short time after removal of the hyperpolarizing input that first activated Ih
Action potential (AP) half-width, ms	Duration of the AP measured at half-maximal amplitude	AP duration, measured as half-width, is most important in determining the time at which the presynaptic terminal is depolarized and the amount of Ca2+ that can enter (proportional to the vesicular release of transmitter); increased AP duration results in a stronger postsynaptic potential
Fast and medium afterhyperpolarization (fAHP, mAHP), mV	Afterhyperpolarization is a measure of maximum membrane hyperpolarization from resting membrane potential that is observed following the repolarizing downstroke phase of the AP. fAHP occurs within milliseconds and is attributed to current through voltage-gated K+ channels; mAHP is longer-lasting, peaking in 15–25 ms, and is attributed to current through Ca2+-activated K+ channels	Time course and amplitude of the AHP, especially the mAHP, are key determinants of repetitive firing frequency. Small-amplitude, short-duration AHPs will allow faster firing and increase neuronal excitability. Faster MN AP discharge results in faster generation of greater forces in the innervated muscle fibers
Frequency/current (f/I) slope	Slope of the f/I plot describes the relationship between injected depolarizing current and AP firing frequency	MNs with a ↑ f/I slope show a greater increase in firing rate for a given increase in depolarizing current (see “neuronal excitability”)
Excitatory postsynaptic current (EPSC) amplitude, pA	Peak amplitude of the inward (depolarizing) current evoked in a postsynaptic neuron by the neurotransmitter released at a synapse in response to a spontaneous AP	↑ EPSC amplitude indicates a greater postsynaptic response to AP-evoked transmitter release; ↑ EPSC frequency indicates that the frequency of APs in excitatory presynaptic neurons has increased. Both will increase MN excitation and AP output, resulting in more forceful muscle contractions
EPSC frequency, Hz	Frequency of EPSCs evoked by endogenous APs arriving at excitatory presynaptic terminals

Predicted outcomes for each measurement assume the absence of confounding changes in other properties. MNs, motoneurons.

Statistical analyses.

For all statistical analyses, sample size was based on the number of neurons. In cases where multiple neurons were recorded per animal, we confirmed that the MN properties from that animal did not differ from the group average. Soma area across developmental stages (tadpole vs. adult) was tested using an independent samples t-test, and data are presented as box plots (Fig. 1C). Electrophysiological data across developmental stage were analyzed using a two-way ANOVA (with cell type and developmental stage considered as independent factors), and data are presented as box plots. Values >2 SDs from the mean were considered outliers and excluded from analysis; total replicates analyzed and number of outliers removed before analysis are listed for each measurement in Table 1. Data sets with skewed distributions violating the normality assumption or showing unequal variances were log₁₀-transformed before analysis [sag amplitude (see Fig. 4A), input resistance (see Fig. 4C), rheobase (see Fig. 4D), and PIR amplitude (see Fig. 7B)]. The two-way ANOVA results are further presented with the mean (SD) of each intrinsic property in Table 1. Developmental changes in type I sag activation potential violated the normality assumption and were assessed using the Mann-Whitney U-test. The percentage of type I cells showing PIR action potentials was determined using the χ²-test and plotted as a bar graph (see Fig. 7C). Correlations between intrinsic property and tadpole stage (IX–XVIII) were determined for each cell type using Pearson’s (parametric) or Spearman’s (nonparametric) correlation tests and are presented as scatterplots. Outliers, identified as values >2 SDs from the mean, were removed before correlation analysis. For intrinsic properties that correlated significantly with tadpole stage, we identified the cell type (I and/or II) showing the correlation and removed its early-stage (IX-XIII) data from the data set (thus leaving only data from late stages XIV–XVIII for that cell type). We then performed additional two-way ANOVAs (with stage and cell type as factors) on this new data set to confirm that comparison of late-stage tadpole data with adult data did not alter the statistical significance of any developmental changes. All statistical tests were performed using SPSS 13.0 for Windows (IBM, Markham, ON, Canada).

Table 1.

Developmental changes in intrinsic membrane properties in type I vs. type II trigeminal MNs of L. catesbeianus

	Tadpole		Adult		Significance
	Type I	Type II	Type I	Type II	Cell type	Stage	Interaction
Ih amplitude, mV	5.8 (2.1) [14/1]	1 ( 1) [10]	4.5 (1.9) [28]	0.4 (0.6) [28]	P < 0.0001, F(1,76) = 86.38	NS	NS
Input resistance, mV	289 (119) [12]	90 (20) [10]	119 (45) [22]	45 (20) [28]	P < 0.0001, F(1,69) = 97.29	P < 0.0001, F(1,69) = 59.96	NS
Rheobase, pA	33 (28) [15]	196 (97) [10]	67 (38) [26/2]	296 (285) [28]	P < 0.0001, F(1,75) = 76.34	P = 0.011, F(1,75) = 6.86	NS
Action potential
Amplitude, mV	65 (14) [14/1]	62 (10) [10]	84 (10) [28]	66 (14) [28]	P = 0.001, F(1,77) = 11.70	P < 0.0001, F(1,77) = 14.71	P = 0.008, F(1,77) = 7.30
Half-width, ms	0.6 (0.1) [15]	0.51 (0.09) [10]	0.59 (0.09) [28]	0.5 (0.2) [28]	P = 0.01, F(1,77) = 6.94	NS	NS
AHP, mV
Fast	−32 (6) [15]	−33 (4) [10]	−26 (7) [(28]	−33 (5) [28]	P = 0.004, F(1,77) = 8.69	NS	P = 0.016, F(1,77) = 6.08
Medium	−20 (6) [13]	−19 (3) [10]	−18 (6) [21]	−17 (4) [27]	NS	NS	NS
Threshold, mV	−41 (10) [14/1]	−34 (5) [10]	−44 (9) [28]	−40 (10) [28]	P = 0.024, F(1,77) = 5.33	P = 0.047, F(1,77) = 4.079	NS
Resting membrane potential, mV	−57 (10) [15]	−57 (9) [10]	−52 (8) [27/])	−58 (11) [28]	NS	NS	NS
f/I slope, number of events/pA	0.18 (0.06) [11/1]	0.08 (0.01) [5/1]	0.09 (0.03) [26/2]	0.06 (0.02) [21]	P < 0.0001, F(1,59) = 43.00	P < 0.0001, F(1,59) = 25.88	P = 0.003, F(1,59) = 9.58
PIR amplitude, mV	6 (3) [14]	1 (1) [10]	6 (3) [25/1]	0.7 (0.6 [18/2]	NS	P < 0.0001, F(1,63) = 132.6	NS
EPSC
Amplitude, pA	−30 (4) [13]	−23 (5) [7/1]	−26 (7) [15/1]	−27 (7) [13/1]	NS	NS	P = 0.04, F(1,45) = 2.338
Frequency, Hz	5.4 (2.7) [13]	8.0 (5.0) [8]	5.5 (2.8) [16]	8.2 (3.5) [14]	P = 0.01, F(1,47) = 7.27	NS	NS

Values are means (SD); number of motoneurons (MNs) analyzed/outliers removed are shown in square brackets. I_h, hyperpolarization-activated inward current; AHP, afterhyperpolarization; f/I slope, frequency/current slope; PIR, postinhibitory rebound; EPSC, excitatory postsynaptic current. Statistics represent results of 2-way ANOVAs; NS, not significant.

Fig. 4.

Input resistance and rheobase show developmental changes in type I and II motoneurons (MNs). A: sag amplitude was greater in type I than type II MNs, with no developmental changes following metamorphosis. B: activation voltage for sag was depolarized in adults. C: input resistance was higher for tadpole and adult type I than type II MNs but decreased in both MN subtypes during development. Inset: developmental changes in type I and II input resistance during tadpole stages IX–XVIII were assessed by Pearson correlations. Type II input resistance correlated significantly with tadpole stage (trend line), but this did not alter results in C. D: rheobase maintained a higher value in type II MNs, but development increased rheobase magnitude in both MN subtypes. Inset: type II rheobase correlated with tadpole stage (trend line) but did not alter results in D. Outliers are shown as ○ in boxplots and ! in correlation plots. Replicates are listed in Table 1. *P < 0.0001 vs. type I. †P < 0.01 vs. tadpoles.

Fig. 7.

Percentage of type I motoneurons (MNs) expressing postinhibitory rebound (PIR) action potentials increases developmentally. A1 and A2: in tadpole and adult type I MNs, hyperpolarizing current pulses to −100 mV revealed a PIR that, in some cases, elicited action potentials (A2). In A1, PIR magnitude is indicated as the segment above the dashed line (arrow). B: PIR amplitude was significantly greater for type I MNs, with no developmental changes, whereas type II MN PIR amplitude was small or absent. C: percentage of type I MNs expressing a PIR sufficient to generate action potentials (APs) increased significantly in adults. For tadpoles, n = 14 MNs from 9 animals (type I) and n = 10 MNs from 7 animals (type II); for adults, n = 26 MNs from 10 animals (type I) and n = 18 MNs from 14 animals (type II). Outliers are shown as ○ in boxplots. *P < 0.0001 vs. type I; †P < 0.05 vs. tadpole type I.

RESULTS

Targeting of trigeminal MNs using DiI retrograde labeling.

In amphibians, the trigeminal motor nuclei are situated bilaterally within the rostral medulla, ventral to the fourth ventricle, and medial to the trigeminal nerve and main sensory nucleus (24). To identify and visualize trigeminal MNs in slice preparations, these cells were labeled retrogradely with fluorescent DiI. In pilot experiments, confocal imaging confirmed the presence of DiI-labeled neurons in the trigeminal motor nucleus, identified using neuroanatomic landmarks (i.e., 4th ventricle, cerebellum, and trigeminal nerve) (Fig. 1, A1–A3).

In live tissue slices obtained from a distinct cohort, DiI-labeled trigeminal MNs were targeted for analysis of soma area (ImageJ analysis; Fig. 1, B1 and B2) and subsequent whole cell recording experiments. A range of soma areas were observed for each stage group, and average soma cross-sectional area increased significantly (t-test: P < 0.0001, F_1,30 = 35.55) between tadpoles (range 170–1,400 µm², n = 19) and adults (range 670–1,780 µm², n = 15) (Fig. 1C).

Depolarizing sag amplitude discriminates two MN subtypes in the tadpole trigeminal motor nucleus.

Sag depolarization evoked by hyperpolarizing pulses that increase with hyperpolarization is a subthreshold signature commonly associated with activation of I_h. This current contributes a small amount to resting membrane potential, counteracts hyperpolarizing inputs, underlies PIR, and plays important roles in multiple rhythmic behaviors, including heart rate control and thalamic oscillations (19, 41, 55) (Table 2). To assess the presence of sag depolarization in tadpole trigeminal MNs, we measured the membrane potential responses of MNs to a series of hyperpolarizing and depolarizing square-wave current steps (holding potential = −70 mV, 20-pA steps; Fig. 2, A1 and A2).

Fig. 2.

Whole cell recordings in tadpoles characterize motoneuron (MN) subtypes based on sag depolarization, input resistance, and rheobase. A1 and A2: representative voltage responses of tadpole trigeminal MNs to incrementing hyperpolarizing current pulses (20-pA steps, 1,000 ms). A1: in a subset of MNs, hyperpolarization to −100 mV revealed a sag depolarization, measured as the difference between peak (●) and steady-state (○) voltages (Δ12.88 mV, only current steps from 0 to −60 pA are shown for clarity). MNs exhibiting sag responses were classified as “type I.” A2: MNs without sag responses (i.e., Δ >2 mV; Δ0.73 mV) were classified as “type II.” B1 and B2: current-voltage (I-V) curves of peak vs. steady-state responses (means ± SE) show voltage dependence of sag amplitude in type I MNs (B1). Overlap between peak and steady-state curves in B2 suggests little to no sag depolarization for type II MNs. C1 and C2: sag amplitude was voltage-dependent in type I MNs (C1) and significantly greater for type I than type II MNs (C2). D: rheobase and input resistance correlated significantly with sag amplitude, making these properties good predictors of cell type. E and F: type I MNs exhibited higher input resistance (E) and lower rheobase (F) than type II MNs. Outliers are shown as ○ in boxplots. G: input resistance and rheobase showed a typical inverse relationship in tadpole trigeminal MNs. For tadpole type 1, n = 15 MNs from 10 animals; for tadpole type II, n = 10 MNs from 8 animals. *P < 0.0001 vs. type I.

In a subset of MNs, hyperpolarizing current steps (to −100 mV) revealed a depolarizing sag in membrane potential that increased in amplitude with increased hyperpolarization (Fig. 2A1). The maximum hyperpolarization occurred within ~250 ms of pulse onset (Fig. 2A1), after which the membrane depolarized throughout the remainder of the pulse. During the 1,000-ms pulse, the maximum sag depolarization amplitude was measured as the difference between membrane potential at peak and steady state (Δ12.88 mV; Fig. 2A1). Interestingly, depolarizing sag was not present in all tadpole trigeminal MNs [i.e., sag depolarization <2 mV was defined as no sag (Δ0.73 mV; Fig. 2A2)]. Plots of the frequency distribution of sag amplitudes for all tadpole and adult MNs revealed two peaks (at 1 and 6 mV) within each stage (data not shown). We interpreted these bimodal histograms as evidence that trigeminal MNs of tadpoles and adults could be classified into two distinct “subtypes” based on presence or relative absence of sag responses (n = 10 animals, n = 24 MNs). Trigeminal MNs showing depolarizing sag >2 mV were defined as type I MNs (n = 14 of 24) and those with sag depolarization ≤2 mV as type II MNs (n = 10 of 24).

Figure 2, B1 and B2, shows mean current-voltage (I-V) plots corresponding to type I and II MNs, with sag depolarization represented by the gap between the peak and steady-state I-V curves that increases with hyperpolarization. Overlap of peak and steady-state I-V curves in type II MNs at hyperpolarized potentials indicates minimal or no sag depolarization and, presumably, no I_h (Fig. 2, B2 and C1). In tadpoles, sag amplitude evoked by current pulses that hyperpolarized MNs to −100 mV was significantly greater in type I than type II MNs [P < 0.0001 (value taken from 2-way ANOVAs reported in Table 1); Fig. 2C2].

Input resistance and rheobase co-vary with tadpole MN sag amplitude.

Sag amplitude co-varies with other intrinsic membrane properties (3, 47), including input resistance (a measure of excitability determined by measuring the steady-state voltage response to a current pulse) and rheobase (current magnitude required to reach firing threshold from resting membrane potential), which are key indicators of neuronal excitability: neurons with high input resistance and low rheobase have high excitability. Thus we next analyzed the relationship between sag amplitude, input resistance, and rheobase. Input resistance and rheobase co-varied with sag amplitude: type I MNs with larger sag potentials exhibited higher input resistance and lower rheobase, whereas type II MNs with small sag potentials had lower input resistance and higher rheobase (Fig. 2D; Pearson: P = 0.001, sag vs. rheobase; P < 0.01, sag vs. input resistance). Additional statistical analysis confirmed that input resistance was significantly higher in type I than type II MNs [ANOVA (cell type): P < 0.0001; Table 1; Fig. 2E] and rheobase was significantly lower in type I than type II MNs [ANOVA (cell type): P < 0.0001; Table 1; Fig. 2F]; i.e., excitability was greater for type I than type II MNs. The inverse relationship between input resistance and rheobase for type I and II MNs is illustrated in Fig. 2G. Therefore, tadpole type I and II MNs could be discriminated not only by the presence of sag potential but also on the basis of input resistance and rheobase.

Type I and II MNs are preserved throughout development.

As observed in tadpole MNs, adult MNs responded to hyperpolarizing current steps (to −100 mV) with (n = 28 of 56 MNs) or without (n = 28 of 56 MNs) a depolarizing sag in membrane potential, as shown for a type I MN in Fig. 3A1 (sag = Δ16.9 mV) and a type II MN in Fig. 3A2 (sag = Δ1.1 mV, Fig. 3B2). Mean I-V responses for each cell type are shown in Fig. 3, B1 and B2. Sag amplitude was greater in adult type I than type II MNs [ANOVA (cell type): P < 0.0001; Table 1; Fig. 3, C1 and C2]. Input resistance and rheobase also co-varied with sag amplitude in adult MNs (Pearson: P < 0.01, sag vs. rheobase; P < 0.0001, sag vs. input resistance; Fig. 3D). While the range of input resistance and rheobase values for type I and II MNs overlapped to some degree, the average input resistance was significantly higher for type I than type II MNs [ANOVA (cell type): P < 0.0001; Table 1; Fig. 3E] and rheobase was significantly lower [ANOVA (cell type): P < 0.0001; Table 1; Fig. 3F], indicating that, in adults, like tadpoles, type I MNs are more excitable than type II MNs. Importantly, high values of input resistance or rheobase accurately distinguished type I from type II MNs (i.e., type II MNs did not exhibit high input resistance, while type I MNs did not exhibit high rheobase; Fig. 3, D and G). Based on these data, trigeminal MNs of adult frogs, like those of tadpoles, can be differentiated into two subtypes, primarily on the basis of sag potential amplitude, input resistance, and rheobase.

Fig. 3.

Motoneuron (MN) subtypes characterized by sag depolarization persist throughout development. A: in adult frogs, hyperpolarizing current pulses to −100 mV revealed MNs with sag responses [type I, Δ16.88 mV (A1)] and MNs with marginal or no sag [type II, Δ1.10 mV (A2)]. B and C: sag amplitude, measured as peak minus steady-state, was voltage-dependent in type I MNs (B1 and C1), whereas sag responses were not apparent from type II current-voltage curves (B2; also see C1). Sag amplitudes were significantly greater in type I than type II MNs (C2). D–G: rheobase and input resistance correlated significantly with sag amplitude. Type I MNs exhibited a higher input resistance (E) and lower rheobase (F) than type II MNs. In adult MNs, input resistance and rheobase showed a typical inverse relationship (G). For adult type I, n = 28 MNs from 13 animals; for adult type II, n = 28 MNs from 15 animals. *P < 0.0001 vs. type I.

Cell type-specific changes in sag, input resistance, and rheobase during development.

As our primary aim was to identify trigeminal MN properties showing significant developmental changes between premetamorphic tadpoles and adult frogs, we next performed two-way ANOVAs on sag amplitude, input resistance, and rheobase, with “cell type” and “stage” as factors (results for all 2-way ANOVAs are presented in Table 1).

Group data presented in Fig. 4A illustrate that sag amplitude did not change with development/metamorphosis in type I or II MNs [ANOVA (stage): P > 0.05]. Since a range of tadpole stages were included in the “premetamorphic” group, we also confirmed that sag amplitude did not change during premetamorphic tadpole development between stages IX and XVIII. Pearson correlation analyses of the relationship between sag amplitude and tadpole stage for type I and II MNs did not show any significant correlations (data not shown). Interestingly, the membrane potential for sag activation was significantly depolarized in type I MNs postmetamorphosis [−93.3 ± 1.7 mV (tadpole) and −86.0 ± 1.6 mV (adult), U = −2.995, P < 0.01; Fig. 4B].

In type II MNs, input resistance increased transiently in late tadpole stages (+Δ40 MΩ, n = 10, Pearson: P < 0.01; Fig. 4C, inset). However, input resistance of type I and II MNs decreased overall with development [ANOVA (stage): P = 0.011, ANOVA (cell type): P < 0.0001; Table 1; Fig. 4C]. While the transient increase in type II input resistance during tadpole stages was unexpected, studies of MN development have identified transient electrophysiological changes apparent on shorter time scales (25). Given this transient developmental change in tadpole type II input resistance, we questioned whether the results of our previous ANOVA comparing tadpoles (with all stages pooled together) with adults (Fig. 4C, left) would change if type II data from only late-stage (XIV–XVIII) tadpoles were included in the analysis (effectively eliminating the confounding effects of the correlation). A subsequent two-way ANOVA (type I: stages IX–XVIII, adult; type II: stages XIV-XVIII, adult) produced the same results as the original analysis [ANOVA (cell type): P < 0.0001, ANOVA (stage): P < 0.0001], indicating that, despite a gradual increase in input resistance of type II MNs during tadpole development, the most important change occurred as a decrease in input resistance following metamorphosis.

Rheobase current showed an inverse change with respect to input resistance, increasing significantly in type I and II MNs of adults [ANOVA (stage): P < 0.01; Fig. 4D]. During tadpole development, rheobase transiently decreased between stages IX and XVIII, as indicated by the negative correlation coefficient (n = 9, Pearson: P < 0.05; Fig. 4D, inset), but additional analysis established that this correlation did not alter interpretation of results presented in Fig. 4D, left [type I: stages IX-XVIII, adult; type II: stages XIV-XVIII, adult; ANOVA (cell type): P < 0.0001, ANOVA (stage): P < 0.01]. Together, these data suggest that, despite an overall decrease in excitability in all MNs during development, type I MNs are more excitable than their type II counterparts.

Action potential properties differ between type I and II MNs throughout development.

The relatively high input resistance and low rheobase of type I MNs are indicative of greater intrinsic excitability. To further explore differences in excitability, we compared action potential and repetitive firing properties of the two MN subtypes in tadpoles and frogs.

Current pulses (1,000 ms) were injected to elicit action potentials, as shown for an adult type I MN in Fig. 5, A1 and A2. When trains of action potentials were elicited, only the first was analyzed. Action potential amplitude, important for propagation at terminal axon branches, increased significantly between tadpoles and adults in type I MNs only [ANOVA (stage × cell type): P = 0.008; Table 1; Fig. 5, B1, B2, and C]. Similarly, the action potential amplitude of type I MNs increased significantly during tadpole development (stages IX–XVIII, +Δ9 mV, n = 12, Pearson: P < 0.05; Fig. 5C, inset). While type I action potential amplitude increases began during tadpole stages, a subsequent two-way ANOVA confirmed an additional significant increase between tadpoles and adults [73 ± 7 mV (tadpole stages XIV–XVIII) vs. 84 ± 2 mV (adults), ANOVA (stage × cell type): P < 0.01].

Fig. 5.

Developmental changes in action potential properties are specific to type I motoneurons (MNs). A1: a typical action potential of an adult type I MN. fAHP and mAHP, fast and medium afterhyperpolarization, respectively. A2: trace in A1 on a longer time scale to illustrate fAHP and mAHP in the context of repetitive firing. B1 and B2: type I action potential amplitude increased significantly postmetamorphosis. C: whereas correlation analysis (inset) revealed a significant increase in type I action potential (AP) amplitude with tadpole stage (inset, trend line), this did not alter the results in C. D: action potential half-width was prolonged in type I MNs, with no developmental changes (see also B1 and B2). E and F: type I peak fAHP amplitude decreased significantly during development, whereas mAHP amplitude showed no differences between cell types or developmental stages (values are on a negative scale). For tadpoles, n = 15 MNs from 10 animals (type I) and n = 10 MNs from 8 animals (type II); for adults, n = 28 MNs from 13 animals (type I) and n = 28 MNs from 15 animals (type II). Outliers are shown as ○ in boxplots and ! in correlation plots. #P < 0.05 vs. type I tadpole and all type II MNs. *P < 0.05 vs. type I.

Action potential half-width, a measure of action potential duration, was greater in type I MNs in both tadpoles and adults [ANOVA (cell type): P = 0.01; Table 1; Fig. 5, B1, B2, and D]. Action potential duration has a minor influence on action potential frequency. At neuron-neuron synapses, increased action potential duration is typically associated with greater presynaptic Ca²⁺ entry, increased transmitter release, and EPSC potentiation. The consequences of greater transmitter release at the neuromuscular junction, where the EPSC is already suprathreshold, are not clear. No significant correlations were found between action potential half-width and tadpole stage (data not shown).

All single action potentials featured a fast and a medium AHP (fAHP and mAHP; Fig. 5, A1 and A2), the amplitudes of which influence the rate of membrane repolarization after an action potential. The AHP, especially the mAHP, is a major determinant of repetitive firing frequency. The fAHP amplitude decreased following metamorphosis in type I MNs only [ANOVA (cell type × stage): P = 0.016, Table 1; Fig. 5E]. The mAHP was not significantly different between cell type or developmental stage (Fig. 5F), nor did fAHP or mAHP amplitude show any correlations with tadpole stages (data not shown). Thus the mAHP alone is not likely to significantly impact differences in the repetitive firing behavior of type I and II MNs or changes in behavior with development. It is important to emphasize, however, that the sensitivity of the mAHP to various modulators could change developmentally and in a MN subtype-specific manner (49).

Repetitive firing behavior is faster in type I MNs throughout development.

Injection of incrementing square-wave depolarizing currents (20-pA steps) elicited tonic action potential firing in all MNs and revealed differences in the action potential threshold and repetitive firing properties of type I and II MNs (Fig. 6, A1–A4). Repetitive firing was generated at much lower levels of current injection in type I than type II MNs (Fig. 6, A1 and A3 vs. A2 and A4), as expected based on their lower rheobase values.

Fig. 6.

Type I motoneurons (MNs) demonstrate a higher repetitive firing frequency throughout development. A1–A4: voltage traces representing typical responses of tadpole and adult trigeminal MNs to incrementing depolarizing current steps (20 pA steps, 1,000 ms). Note larger amplitude of adult type I MNs (A3), as described in Fig. 5, B and C. Action potential threshold was hyperpolarized in type I MNs but further hyperpolarized in both MN types following metamorphosis. B and inset: type II threshold correlated significantly with tadpole stage, suggesting that threshold hyperpolarization may be initiated during tadpole stages. C: despite a correlation between type I MNs and tadpole stage (inset), resting membrane potential (RMP) was similar for both MN subtypes and stages. For tadpoles, n = 15 MNs from 10 animals (type I) and n = 10 MNs from 8 animals (type II); for adults, n = 28 MNs from 13 animals (type I) and n = 28 MNs from 15 animals (type II). D1: slope of action potential (AP) frequency vs. current (f/I, means ± SE) was calculated to assess changes in firing frequency across MN subtypes and stages. D2: firing rates were higher for tadpole and adult type I than type II MNs (evidenced by a higher f/I slope). The f/I slope decreased developmentally in both MN subtypes. For tadpoles, n = 12 MNs from 9 animals (type I) and n = 6 MNs from 5 animals (type II); for adults, n = 27 MNs from 13 animals (type I) and n = 21 MNs from 14 animals (type II). Outliers are shown as ○ in boxplots and ! in correlation plots. *P < 0.05 vs. type I; †P < 0.05 vs. tadpoles.

The threshold potential for action potential firing was hyperpolarized in type I compared with type II MNs [ANOVA (cell type): P = 0.024], indicating that type I MNs require less depolarization to reach threshold and fire an action potential. This, in combination with their higher input resistance, will contribute to the lower rheobase and greater excitability of type I MNs. The threshold potential was more hyperpolarized in type I and II MNs of adult frogs than tadpoles [ANOVA (stage): P = 0.047; Table 1; Fig. 6B]. Analysis of correlations between threshold and tadpole stage revealed a significant negative correlation for type II MNs (−Δ8 mV, n = 10, Pearson: P = 0.05; Fig. 6B, inset) and a negative trend for type I MNs (−Δ6 mV, n = 15, Pearson: P = 0.18; Fig. 6B, inset). When these correlations were taken into account, a subsequent two-way ANOVA (type I: stages IX–XVIII, adult; type II: stages XIV–XVIII, adult) suggested that threshold hyperpolarization is likely initiated before metamorphosis in trigeminal MNs, as “stage” was no longer a significant factor [subsequent ANOVA (stage): P = 0.59].

No significant differences were detected for resting membrane potential between cell types or developmental stages (Fig. 6C, left). Despite a significant positive correlation between type I resting membrane potential and tadpole stages IX–XVIII (+Δ8 mV, n = 14; Pearson: P < 0.05; Fig. 6C, inset), a two-way ANOVA (type I: stages XIV–XVIII, adult; type II: stages IX–XVIII, adult) found no significance for resting membrane potential [ANOVA (cell type): P > 0.05, ANOVA (stage): P > 0.05].

The slope of the action potential frequency-current (f/I) relationship reports the increment in action potential discharge frequency associated with a specific increment in excitatory current: the greater the slope, the greater the increase in frequency for a given current input and the greater the excitability. The slope of the f/I relationships averaged for type I and II MNs from tadpoles and adult frogs was greatest for tadpole type I MNs (Fig. 6D1) and decreased postmetamorphosis. The slope of the f/I relationships was lowest in adult type II MNs [ANOVA (stage): P < 0.001, ANOVA (cell type): P < 0.001; Table 1; Fig. 6D2], indicating the lowest excitability based on this measure. No correlations were observed between the slope of the f/I relationship and tadpole stage (data not shown).

The presence of I_h in a neuronal membrane can have several effects on repetitive firing behavior that depend on the size of the I_h, the input resistance of the neuron, and the presence of additional voltage-dependent conductances. One of the most dramatic effects of I_h on discharge behavior is PIR. An inhibitory input (or hyperpolarizing current) activates I_h, causing a slow-onset sag depolarization of membrane potential. However, because I_h deactivates slowly as well, when the inhibitory input is removed, the inward depolarizing I_h remains on, causing a depolarization beyond resting membrane potential, which is referred to as PIR. This rebound depolarization then slowly decays back to resting membrane potential as I_h deactivates. In neurons with large I_h and high input resistance, PIR can generate single action potentials or bursts of actions potentials and assist with rapid onset of firing and phase transitions during rhythmic behaviors. To assess the presence of PIR, type I MNs were presented with a 1,000-ms negative current injection of sufficient magnitude to hyperpolarize the MN to −100 mV, and the membrane potential responses to withdrawal of the current were observed. Only type I MNs of tadpoles and adults exhibited PIR. Figure 7 shows traces from tadpole (Fig. 7A1) and adult (Fig. 7A2) type I MNs with PIR sufficient to elicit an action potential (Fig. 7A2) or overshooting baseline potential without generating an action potential (Fig. 7A1, arrow). PIR amplitude (measured following hyperpolarization to −100 mV) was significantly greater in type I than type II MNs, with type II MNs showing little to no PIR [ANOVA (cell type): P < 0.0001; Table 1; Fig. 7B]. No developmental changes in PIR amplitude were observed for type I MNs; however, the percentage of type I MNs in which the rebound was sufficient to generate an action potential increased significantly during development (2 of 16 = 13% of tadpoles and 6 of 29 = 21% of adults; χ²(2) = 46.312, P < 0.0001; Fig. 7C).

EPSC frequency is higher in type II MNs throughout development.

Networks and their output neurons (in this case, MNs) undergo synaptogenesis and become active over different developmental time frames. To quantify developmental changes in the pattern of synaptic activity in type I and II MNs, we measured the amplitude and frequency of endogenous, spontaneous postsynaptic currents across developmental stages. Traces in Figs. 8, A1 and A2 show ~2-min voltage-clamp recordings from tadpole trigeminal type I and II MNs to illustrate that the recording is stable and that MNs receive significant, primarily EPSCs (evident as downward deflections in the current trace). EPSCs were also recorded from type I and II MNs of adults (data not shown). Developmental changes in EPSC amplitude were cell type-specific, with type II EPSC amplitude increasing postmetamorphosis [ANOVA (cell type × stage): P = 0.04; Table 1; Fig. 8B]. No correlations were detected between EPSC amplitude and tadpole stage (data not shown).

Fig. 8.

Postsynaptic current frequency is higher in type II motoneurons (MNs). A1 and A2: voltage-clamp recordings over 2 min (−60 mV) revealed primarily excitatory postsynaptic currents (EPSCs), shown as downward deflections in the activity trace, in type I and II MNs of tadpoles and adults (data for adults not shown). B: EPSC amplitude increased significantly in type II MNs following metamorphosis (data shown on a negative scale). C: EPSC frequency [Hz (number of events/s)] was higher for type II MNs in tadpoles and adults but, in type II MNs, showed a significant negative correlation with tadpole stage (inset; trend line), suggesting a transient developmental change for this MN subtype. For tadpoles, n = 13 MNs from 9 animals (type I) and n = 8 MNs from 6 animals (type II); for adults, n = 17 MNs from 9 animals (type I) and n = 14 MNs from 11 animals (type II). Outliers are shown as ○ in boxplots and ! in correlation plots. *P < 0.05 vs. type I; #P < 0.05 vs. adult type II and all type I MNs.

EPSC frequency [Hz (number of events/s)] was consistently higher for type II cells of tadpoles and frogs [ANOVA (cell type): P = 0.01; Table 1; Fig. 8C]. Type II MN EPSC frequency also correlated significantly with tadpole stages IX–XVIII (−Δ10 Hz, n = 8, Pearson: P < 0.05; Fig. 8C, inset); the number of excitatory synaptic inputs per unit time decreased as the tadpoles matured. A subsequent two-way ANOVA (type I: stages IX–XVIII, adult; type II: stages XIV–XVIII, adult) revealed in type II MNs a potential developmental increase in spontaneous EPSC frequency from a mean of 2.3 ± 0.4 in late-stage tadpoles to 8.1 ± 3.4 in adults [ANOVA (cell × stage): P < 0.05]. Some caution must be exercised in this interpretation, however, as the correlation was based on data from a small number of type II MNs (tadpole stages XIV–XVIII, n = 2). The relatively higher amplitude and frequency of EPSCs recorded from type II MNs suggest that, under these in vitro conditions, type II MNs receive greater synaptic drive and that this drive increases through metamorphosis.

DISCUSSION

The developmental transition from gill to lung breathing has been studied in amphibians in vivo (53) and in vitro (21, 23). This transition involves a dramatic change in the main pattern of respiratory MN activity from small-amplitude, high-frequency gill/buccal bursts to large-amplitude, low-frequency lung bursts that are coordinated with ongoing low-amplitude buccal bursts. Developmental processes underlying the transition in the pattern of MN output will include changes in the properties of rhythm generating networks and synaptic properties, as well the intrinsic properties of the MNs themselves. Contributions of MNs to the maturation of respiratory behavior have received little attention in nonmammalian vertebrates, including the American bullfrog. In Lithobates, metamorphosis does not involve trigeminal MN neurogenesis (1, 2). Thus the properties of trigeminal MNs in the tadpole must mature to support the emergence of adult respiratory motor behaviors. Here we report on how the intrinsic properties of trigeminal MNs, which innervate major respiratory pump muscles important for both gill and lung breathing, change with the metamorphic transition to air breathing.

Two types of trigeminal MNs in tadpoles and adult frogs.

MNs within a specific pool are not uniform (49). Here, in tadpole and adult frogs, we distinguished two subtypes of trigeminal MNs based on sag amplitude: type I and II MNs had large and minimal sag depolarizations, respectively. Sag amplitude did not change with development (7, 18). Based on sag as the initial discriminating parameter, additional analysis revealed that, at a population level, type I and II MNs could also be differentiated based on input resistance (higher in type I), rheobase (lower in type I), and pattern of spontaneous synaptic activity (lower in type I). We did not pharmacologically identify the current underlying the sag potential in trigeminal MNs, but the voltage dependence, slow activation/inactivation kinetics, and PIR we report are suggestive of I_h (7, 15, 59).

Significance of I_h.

I_h is widespread throughout the nervous systems of invertebrates and vertebrates (52). The genetics, channel kinetics, and physiological properties of I_h are well characterized (7, 9), and the influence of I_h on repetitive firing and the role of I_h in various rhythmic behaviors have been studied extensively (6, 41, 48, 65). Its functional role in type I trigeminal MNs of tadpoles and frogs remains to be fully determined.

In hypoglossal MNs, I_h reduces the duration of the mAHP and enhances repetitive firing (60). The mAHPs were similar in type I and II MNs, but, given the apparent I_h in type I MNs, our measurement of the mAHP in this subtype likely underestimates the actual mAHP amplitude. To assess directly whether the mAHP is greater in type I MNs will require direct comparison of the mAHP in the two groups of MNs after blockade of I_h. The presence of I_h is expected to be a factor in the faster repetitive firing behavior of type I MNs.

I_h also has a strong influence on phase timing and phase transition in many rhythmic networks where alternation between phases involves strong, phasic GABAergic or glycinergic inhibition, both of which contribute to buccal and lung motor patterns in tadpoles and adult frogs (10). We predict that, in type I trigeminal MNs from tadpoles and frogs, I_h activation will counteract the actions of inhibitory inputs through the sag depolarization and facilitate rapid phase transitions through PIR and generation of action potentials. Conversely, type II MNs lacking sag responses would remain susceptible to prolonged inhibition at all developmental stages.

The proportion of type I MNs that produced action potentials during PIR responses increased in adults. However, this is most likely due to developmental increases in channels other than I_h, such as low-voltage-activated (T-type) Ca²⁺ channels, since sag amplitude did not change with metamorphosis and input resistance actually decreased (meaning that a larger I_h would be required to produce the same PIR depolarization). The physiological significance of the enhanced PIR in adults is not known, but this will likely contribute to the stronger activation of MNs that occurs during air breaths; the PIR could also contribute to other adult-specific behaviors, including the generation of clusters or bursts involved in consecutive lung breaths, vocalization, and feeding.

Potential significance of type I vs. type II MNs.

The contribution of type I and II trigeminal MNs to specific trigeminal MN-supported behaviors (e.g., gill vs. lung breathing) in tadpoles and adult bullfrogs is not known. An answer to this question will require analysis of MNs connected to spontaneously breathing networks. However, we can comment on how the intrinsic properties of the two populations will help shape MN output, which ultimately is transformed into muscle contraction and force generation. Table 2 lists some of these properties and their general impact on MN output and muscle activity. The discussion that follows focuses on how the changes reported here are likely to affect the breathing behavior of tadpoles/frogs (Table 2).

The generation of quiet vs. forceful breathing patterns in mammals is largely based on principles of ordered motor unit recruitment, with smaller, high-input-resistance MNs (corresponding to smaller motor units) recruited before the larger, low-input-resistance MNs (8, 12, 13, 40, 49). Similar principles may apply in amphibians. As in mammalian hypoglossal (8) and phrenic (12, 13, 40) MN pools, trigeminal MNs from both tadpoles and frogs had a broad range of the input resistances important in graded motor unit recruitment. Developmentally, MN input resistance decreased for type I and II MNs, likely reflecting, as in mammals, increased MN size (28) as well as increased membrane conductance that comes with developmental increases in the expression of leak, voltage, and ligand-gated ion channels (49). Based on these principles and their lower action potential threshold, high-input-resistance type I MNs would likely be recruited first, with relatively small excitatory synaptic input driving small, low-force-generating motor units and small-amplitude buccal breaths. Lower-input-resistance type I and II MNs would be recruited as the strength of synaptic drive increased, engaging higher-force-generating motor units to produce larger buccal breaths and, ultimately, lung breaths. In addition to serving a direct respiratory function, recruitment of high-force motor units for lung ventilation in adult frogs assists lymphatic fluid movement and, thus, contributes to regulation of blood volume (27). We determined that input resistance/rheobase and sag amplitude are correlated (i.e., greater sag in high-input-resistance type I MNs), as reported for mammalian spinal MNs (47), suggesting that sag depolarization and PIR are likely to be more important in shaping activity of buccal breaths.

Connectivity is an additional factor that may be important in determining the order and pattern of MN recruitment in buccal vs. air breathing. In mammals, small and large inspirations are generated by the same network; i.e., the single network is connected to small and large motor units, such that synaptic drive and input resistance/rheobase are key factors in establishing recruitment order. In amphibians the small buccal and large lung breaths are generated by different oscillators, each with its own pattern of connectivity. Thus, while there is clearly convergence of these two networks onto the trigeminal MN pool (and, most likely, significant overlap between the MNs used for buccal and lung breaths), type I MNs may be preferentially connected to the buccal oscillator and type II MNs to the lung oscillator. We cannot determine from our data the connectivity of type I and II MNs with buccal and lung oscillators, because these oscillators are located caudal to the trigeminal MN pool and are absent from our slices (4). The synaptic inputs (EPSCs) we recorded are more likely to have arisen from cholinergic and glutamatergic premotoneurons, interneurons, or central MN collaterals (43, 49). Nevertheless, the higher frequency of EPSCs recorded in type II than type I MNs is consistent with different connectivity. Finally, while it is not clear whether developmental increases in EPSC amplitude in type II MNs result from pre- or postsynaptic modifications, it is tempting to speculate that this increase is associated with the emergence of the lung motor pattern at metamorphosis.

Metamorphosis.

Intrinsic membrane properties of trigeminal MNs changed significantly during metamorphosis, in a manner consistent with general developmental changes in mammalian MNs indicative of an overall reduction in excitability, i.e., increased soma size, decreased input resistance, increased rheobase, and reduced slope in the f/I relationship. A progressive hyperpolarization of resting membrane potential contributes to developmental reductions in the excitability of some mammalian respiratory MN pools, such as phrenic MNs (40), but this is not universal (14, 44, 60). In the present study, resting membrane potential did not differ between MN types and did not change developmentally. The main change that would contribute to an increase in excitability with development was the hyperpolarization of the action potential threshold in type I MNs.

Developmental changes in the action potential profile were relatively modest. The main factor determining repetitive firing frequency is the mAHP, which did not change significantly with metamorphosis. Changes in action potential half-width and amplitude would have only minor effects on firing frequency. Increases in action potential amplitude, especially in type I MNs, could be important in reducing conduction failure at axonal branch points.

Perhaps one of our most striking observations was that, despite the dramatic changes in breathing pattern and, therefore, MN output that occur with the transition from gill to air breathing, developmental changes in the intrinsic membrane properties were relatively modest. The modest nature of the changes may, in part, reflect the reduction in neuromodulatory tone that occurs in slices in vitro. Modulatory inputs change developmentally and can profoundly alter membrane excitability through actions on leak and voltage-gated (including I_h) and ligand-gated ion channels (31). Serotonin, for example, increases MN excitability through multiple mechanisms (32, 45, 46, 49), including increased input resistance via inhibition of a K⁺ leak conductance, potentiating I_h, increasing action potential frequency (31, 38), and increasing the frequency and amplitude of excitatory postsynaptic potentials in frog lumbar MNs (34) and rat spinal MNs (39). Serotonin application in developing bullfrog brain stems also produces distinct and stage-dependent modulation of buccal and lung motor patterns (36).

An alternative hypothesis is that the dramatic transition from the buccal to the lung pattern of trigeminal MN activity is due primarily to developmental changes in the pattern of synaptic drive originating from buccal and lung oscillators. Future research in behaving preparations is needed to determine the specific role of type I versus type II MNs for buccal versus lung (and other behavioral) motor output, as well as the role of neuromodulators in altering membrane properties and responses to synaptic drive.

An important caveat with any whole cell recording study that uses DIC microscopy to visually target neurons in slices for electrophysiological analysis is that selected neurons are close to the slice surface (within 80 µm). Thus, while some pruning of the dendritic tree is inevitable, it will be mitigated by the fact that the dendritic field of trigeminal MNs is largely confined to the medial and lateral planes in bullfrogs (tadpoles and adults) (50). The risk and extent of pruning will increase for neurons with larger dendritic fields. While the cross-sectional area of trigeminal MNs increased between tadpoles and frogs, the dendritic field also likely expanded in the medial-lateral plane. We consider it highly unlikely that a difference in dendritic pruning is a factor in the developmental changes in MN membrane properties reported here. First, loss of distal dendrites will have only modest effects on postsynaptic MN properties measured at the soma due to space-clamp issues associated with morphologically complex neurons. Second, synaptic properties (EPSC frequency for type I and II MNs and amplitude for type I MNs) were similar in tadpoles and frogs, strongly suggesting that their dendritic trees were not differentially affected by the slicing procedure.

Perspectives and Significance

In tadpoles and adult frogs, we found two types of trigeminal MNs distinguished largely on the basis of amplitude of depolarizing sag potentials that are suggestive of I_h. Those with large sag, termed type I MNs, had higher input resistance, lower rheobase, and overall greater excitability than type II MNs. With metamorphosis, type I and II MNs increased in size and showed a reduction in input resistance, an increase in rheobase, and an overall reduction in the peak repetitive firing frequency, consistent with observations in many mammalian MN pools. We propose that type I trigeminal MNs are recruited during buccal breathing in tadpoles and adults. Type I MNs are also likely to be recruited during lung breaths in adults. In contrast, type II MNs, with lower input resistance, higher rheobase, and greater susceptibility to prolonged inhibition due to lack of sag depolarization, may only be recruited by strong excitatory drive to engage forceful breathing, as occurs during lung breaths or other behaviors requiring powerful contractions; e.g., trigeminal MNs innervate buccal musculature involved in respiration, but these muscles are also involved in lymphatic fluid movement (27), vocalization, and prey-capture behaviors.

GRANTS

This work was supported by a Discovery Grant and a Research Tools and Instruments Grant from the Natural Sciences and Engineering Research Council of Canada awarded to R. Kinkead. The work of T. A. Janes is supported by a Fonds de Recherche du Québec-Santé Collaborative Scholarship.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.F., S.C., and R.K. conceived and designed research; T.A.J., S.F., and S.C. performed experiments; T.A.J., S.F., and S.C. analyzed data; T.A.J., S.F., G.D.F., and R.K. interpreted results of experiments; T.A.J. and S.C. prepared figures; T.A.J., G.D.F., and R.K. drafted manuscript; T.A.J., S.F., S.C., G.D.F., and R.K. edited and revised manuscript; T.A.J., S.F., S.C., G.D.F., and R.K. approved final version of manuscript.