Reports of cholecystokinin (CCK) binding and expression of CCK receptors in neonatal rodent spinal cord suggest that CCK may influence neuronal excitability. In patch-clamp recordings from 19/21 ventral horn motoneurons in neonatal (PN 5–12 days) rat spinal cord slices, we noted a slowly rising and prolonged membrane depolarization induced by bath-applied sulfated CCK octapeptide (CCK-8s; 1 μM), blockable by the CCKB receptor antagonist L-365,260 (1 μM). Responses to nonsulfated CCK-8 or CCK-4 were significantly weaker. Under voltage clamp (VH −65 mV), 22/24 motoneurons displayed a CCK-8s-induced tetrodotoxin-resistant inward current [peak: −136 ± 28 pA] with a similar time course, mediated via reduction in a potassium conductance. In 29/31 unidentified neurons, CCK-8s induced a significantly smaller inward current (peak: −42.8 ± 5.6 pA), and I-V plots revealed either membrane conductance decrease with net inward current reversal at 101.3 ± 4.4 mV (n = 16), membrane conductance increase with net current reversing at 36.1 ± 3.8 mV (n = 4), or parallel shift (n = 9). Intracellular GTP-γ-S significantly prolonged the effect of CCK-8s (n = 6), whereas GDP-β-S significantly reduced the CCK-8s response (n = 6). Peak inward currents were significantly reduced after 5-min perfusion with N-ethylmaleimide. In isolated neonatal mouse spinal cord preparations, CCK-8s (30–300 nM) increased the amplitude and discharge of spontaneous depolarizations recorded from lumbosacral ventral roots. These observations imply functional postsynaptic G-protein-coupled CCKB receptors are prevalent in neonatal rodent spinal cord.
Neuropeptides are known to exert powerful effects on motor networks in both invertebrates (Harris-Warrick and Marder 1991) and vertebrates (LeBeau et al. 2005) and can modulate locomotor behavior generated by the rodent spinal cord (Barriere et al. 2005; Pearson et al. 2003). Several classes of spinal neuron including motoneurons express cholecystokinin (CCK) or the CCK receptor, suggesting that it has an important role in spinal cord function (Cortes et al. 1990; Schiffmann et al. 1991; Truitt et al. 2003).
CCK and the CCK family of peptides, originally isolated from the gastrointestinal tract, were among the first of the gastrointestinal peptides to be localized in the CNS (Vanderhaeghen et al. 1975). In brain, the predominant molecular form is the octapeptide (CCK-8) in its sulfated form, CCK-8s (Rehfeld 1978). As one of the most abundant peptides with a wide distribution in various regions of the CNS, CCK has been implicated in a diversity of brain functions (reviewed in Crawley and Corwin 1994). CCK's actions are mediated via two classes of receptor, designated as CCKA and CCKB subtypes, with a distribution that varies both among regions and between neurons (Honda et al. 1993; reviewed by Noble et al. 1999). In the spinal cord, CCK-like immunoreactive fibers originate in part from local CCK-containing neurons, but also from neurons in higher central regions (Abelson and Micevych 1991; Cortes et al. 1990, 1991).
Binding studies and receptor-expression analyses indicate plasticity in the expression profiles of CCK receptors. The high level of expression during the first and second postnatal week together with the ensuing changes during ontogeny (Cho et al. 1983) suggests a physiologically important role in CNS development. Although, functional receptors for CCK have been reported in neonatal spinal cord explants (e.g., Suzue et al. 1981) and culture preparations (e.g., Rogawski et al. 1985), there are few reports on the effects of CCK on neuronal excitability in the neonatal spinal cord. In an earlier study (Suzue et al. 1981), recordings from lumbar ventral roots or intracellular recordings from motoneurons in isolated hemisected spinal cord of 0- to 4-day-old rats revealed a depolarizing action of CCK-8. In another study CCK-8 depolarized neurons in Lamina X (Phelan and Newton 2000).
In the present study, we used patch-clamp and extracellular recordings to evaluate functional evidence for CCK receptors, define receptor subtype, and characterize possible mechanisms of CCK receptor-mediated conductances. We now report that motoneurons and unidentified ventral horn neurons in the neonatal rat spinal cord display CCK-8s induced membrane depolarization and inward currents that may differentially engage separate conductances: a potassium conductance in motoneurons and/or a presumed nonselective cationic conductance in unidentified neurons. Furthermore, we demonstrate that activation of CCK receptors increases the amplitude and discharge of spontaneous depolarizations recorded from the ventral roots of the isolated neonatal mouse spinal cord.
All experiments conformed to the Canadian Council for Animal Care guidelines and to the guidelines of Ottawa Health Research Institute and National Institutes of Health on the ethical use of animals in research and were approved by institutional ethics committees.
Neonatal spinal cord slice preparations
The spinal cords of methoxyflurane-anesthetized Sprague-Dawley rats of either sex (5–12 days old) were removed after a dorsal thoracolumbar laminectomy and placed in ice-cold (4°C) artificial cerebrospinal fluid (ACSF) composed of (in mM) 127 NaCl, 26 NaHCO3, 3.1 KCl, 1.2 MgCl2, 2.4 CaCl2, and 10 d-glucose (pH 7.35; osmolarity, 290–305 mosmol) and gassed with 95% O2-5% CO2. Transverse 350- to 450-μm sections from the T7 to L5 segments were cut on a vibratome, equilibrated in ACSF at room temperature for 1 h and continuously superfused at 4–6 ml/min in a recording chamber (volume, 500 μl).
Using the blind whole cell patch-clamp technique, neurons were recorded using Axopatch 1A or Axopatch 1D amplifiers (Molecular Devices, Sunnydale, CA). Micropipettes were filled with (in mM) 130 K-gluconate, 10 KCl, 10 NaCl, 1 MgCl2, 10 HEPES, 1 EGTA, 1 GTP, and 2 Mg-ATP, adjusted to pH 7.3 with Tris buffer. Lucifer yellow (1 mg/ml) was included for later visualization and morphological identification using methods described earlier (Oz et al. 2001). Corrections for liquid-junction potentials (approximately −10 mV) were performed off-line. Data were filtered on-line at 2 kHz. The Digidata 1200 interface (Molecular Devices) and version 7 of pCLAMP software were used on-line to generate clamp commands. Motoneurons were identified by their all-or-nonantidromic responses to stimulation of ventral rootlets using a bipolar electrode (1–10 V, duration: 0.02 s) and/or by their morphology and evidence of an axon projecting toward the ventral root. In the present study, we applied the term “unidentified neuron” to any other neuron.
Sulfated (CCK-8s) and nonsulfated CCK-8, GTP, GTP-γ-S, GDP-β-S, pertussis toxin, and tetrodotoxin were from Sigma-RBI (St. Louis, MO). CCK-4 was obtained from American Peptide (Sunnyvale, CA). L-365,260 and L-364,718 were kindly supplied by the Research Division of Merck (Rahway, NJ). Agents were dissolved in ACSF at their final concentrations and delivered by bath application at a perfusion rate of 4–6 ml/min. For statistical evaluation, we used paired, unpaired Student's t-test, or ANOVA as it is indicated in the text. Results are presented as means ± SE.
Mouse spinal cord preparation
Neonatal (1–4 days old) Swiss Webster mice (Taconic Laboratory, Germantown, NY) were used to investigate the effects of CCKs on the motor output of the isolated spinal cord preparation (Whelan et al. 2000). Briefly, pups were anesthetized by hypothermia, decapitated rapidly, and eviscerated. The remaining tissue was placed in a dissection chamber filled with oxygenated ACSF, the spinal cord was transected at T1–T3, removed, and transferred to the recording chamber. The ACSF contained (in mM) 128 NaCl, 4 KCl, 1.5 CaCl2, 2 MgSO4, 0.5 NaH2PO4, 21 NaHCO3, and 30 d-glucose. The bath solution was gradually heated from room temperature to ∼27°C, and motoneuron electrical activity was recorded using plastic suction electrodes into which segmental L1–L5 ventral roots were drawn. Electrophysiological recordings were amplified, filtered (DC to 1 kHz), digitized, and recorded for further analysis.
Data obtained from 45 motoneurons located within Rexed laminae VIII and IX displayed a mean resting membrane potential of −74.6 ± 2.1 mV, mean input resistance of 64.3 ± 4.2 MΩ, and mean resting conductance of 14.2 ± 2.3 nS (mean age = 7.8 ± 0.6 days). In 19/21 neurons tested in current-clamp mode, bath application of CCK-8s (1 μM; 30 s) initiated a slowly rising (60–90 s to peak) membrane depolarization that reached a plateau of 14.1 ± 1.8 mV (n = 17) sufficient to trigger a burst of action potentials in two cells (Fig. 1 A). Washout intervals of 20–30 min were required to obtain full recovery. Whereas application of CCKB receptor antagonist L-365,260 (1 μM; n = 5) was without effect on resting membrane properties, this agent significantly suppressed subsequent responses to CCK-8s (1.4 ± 1.2 vs. 14.2 ± 3.1 mV controls, paired t-test, P < 0.05; Fig. 1A). The CCKA receptor antagonist L-364,718 (1 μM; n = 4) did not affect CCK-8s-induced depolarizations (12.4 ± 1.6 vs. 15.3 ± 3.4 mV controls, paired t-test, P > 0.05; Fig. 1B). Compared with CCK-8s controls (13.4 ± 2.1 mV, n = 9 cells), responses to applications of nonsulfated CCK-8 (CCK-8ns, 1 μM, n = 4 cells, 2.2 ± 1.4 mV; paired t-test, P < 0.05) or to the tetrapeptide CCK-4 (1 μM, n = 4 cells, 1.7 ± 0.9 mV; paired t-test, P < 0.05) were significantly weaker (Fig. 1C).
Under voltage clamp (VH −65 mV) and in the presence of 1 μM TTX, 22/24 motoneurons tested (mean age = 7.4 ± 0.7 days) responded to bath-applied CCK-8s (1 μM) with a slowly rising inward current with a mean peak of −136 ± 28 pA and slow recovery over 7–10 min (Fig. 2 A). Responses to CCK-8s were concentration dependent, displaying an EC50 of 347 nM (Fig. 2B; n = 22 cells). Comparison of instantaneous I-V plots before and at the peak of the CCK-8s responses revealed net CCK-8s currents (difference between control and peak CCK-8s effect) the slope of which indicated a 16.2% reduction in membrane conductance (from 12.1 ± 1.8 to 10.3 ± 1.6 nS; n = 7 cells, paired t-test; P < 0.05). The mean net CCK-8s current reversed at −94.9 ± 4.3 mV, approximating the estimated equilibrium potential for potassium ions under these conditions (Fig. 2C, ░). In four cells, increasing the extracellular concentration of K+ to 10 mM resulted in a shift of the reversal potential to −67 ± 2 mV (Fig. 2C, ▪), close to the Nernstian predicted equilibrium potential, further implying mediation via potassium channels.
A population of 97 cells (mean age = 7.2 ± 0.4 days) that we collectively labeled as unidentified based on the absence of antidromic activation displayed a significantly less negative resting membrane potential of −61.7 ± 1.7 mV (–74.6 ± 2.1 mV for motoneurons, ANOVA, P < 0.05), a higher input resistance of 173.6 ± 12.8 MΩ (64.3 ± 4.2 MΩ for motoneurons, ANOVA, P < 0.05), and lower resting conductance of 5.6 ± 0.8 nS (14.2 ± 2.3 nS for motoneurons, ANOVA, P < 0.05) when compared with the motoneuron population. When recorded in voltage-clamp mode in the presence of TTX, 29/31 tested neurons responded to CCK-8s (1 μM) with a mean inward peak current of 42.8 ± 5.6 pA. Although different in magnitude from the response in motoneurons, the time course of these CCK-8s-induced currents was virtually identical (compare Figs. 2A and 3A). Whereas the collective data on voltage-current relationships obtained at the peak of the responses indicated a decrease in membrane conductance (from a control value of 5.3 ± 0.7 to 4.9 ± 0.6 nS, paired t-test; P < 0.05), a comparison of the net CCK-8s-induced currents for 29 individual cells revealed three significantly different patterns (Fig. 3 B). In 16 neurons, net CCK-8s-induced conductance decreased from a control of 5.4 ± 0. 8 to 4.6 ± 0.6 nS (14.8% decrease, paired t-test; P < 0.05; mean age = 6.9 ± 0.4 days) and reversed at 101.3 ± 4.4 mV, similar to data from the motoneuron population that implied mediation via reduction in conductance for potassium ions. By contrast in four neurons, the slope of the net CCK-8s current reflected an increase in conductance from 4.7 ± 0.6 to 5.4 ± 0.9 nS (15.1% increase, paired t-test; P < 0.05; mean age = 6.8 ± 0.8 days) with current reversal at 36.1 ± 3.8 mV, a value that could reflect mediation of the CCK-8s response via increase in a nonselective cationic conductance. In the remaining nine cells, the mean net CCK-8s-induced current displayed a slight (5.7%) decrease in membrane conductance (from 5.5 ± 1.1 to 5.2 ± 1.2 nS, paired t-test; P > 0.05; mean age = 7.1 ± 0.7 days) with no reversal within the voltage range tested.
Because CCK receptors belong to the family of G-protein-coupled receptors, we first compared the control data where pipettes contained GTP (1 mM) with data obtained 10 min after establishment of seals using pipettes that contained GTP-γ-S (0.5 mM), a nonhydrolyzable derivative of GTP that activates G protein in an irreversible manner (Gilman 1987). For 6/6 cells responding to CCK-8s (1 μM), the mean amplitude of response (–42.1 ± 4.8 pA) was not significantly different from controls (–40.7 ± 4.1 pA, unpaired t-test; P > 0.05), but the recovery phase was greatly prolonged by the presence of GTP-γ-S in the pipette (Fig. 4 A). Of the six cells treated with GTP-γ-S, membrane conductance decreased in three cells (reversed close to EK), increased in one cell (reversed close to −40 mV), and was unchanged in the remaining two cells (with no reversal potential detected in the range of −10 to −120 mV). We also dialyzed eight cells with GDP-β-S (1 mM for 10 min), a stable analog of GDP that competitively inhibits G protein binding by GTP (Gilman 1987). In 6 responding cells, mean CCK-8s-induced inward current was significantly reduced (–12.7 ± 3.1 pA) by contrast with the −42.4 ± 5.6 pA recorded in the 29 control cells (P < 0.05, Student's unpaired t-test, Fig. 4B).
Treatment with pertussis toxin (PTX) or N-ethylmaleimide (NEM) has been reported to uncouple the receptors from their associated Gi/o-type G proteins (Gilman 1987; Shapiro et al. 1994). To evaluate the role of PTX-sensitive G proteins on CCK-8s-induced currents, slices were incubated for 12–18 h either in ACSF or in ACSF containing PTX (5 μg/ml) and the magnitudes of the CCK-8s-induced currents compared. In similarly incubated control slices, CCK-8s induced a mean inward current of −39.3 ± 3.7 pA (n = 5). By contrast, the mean CCK-8s-induced inward current was significantly reduced to −22.4 ± 2.9 pA in neurons from PTX treated slices (Fig. 4B, n = 7, P < 0.05, Student's unpaired t-test). The mean amplitude of CCK-8s-induced inward currents in motoneurons (–127.1 ± 24.3 pA in controls; n = 5) was also significantly reduced to −53.7 ± 12.6 pA in motoneurons (n = 6; P < 0.05, Student's unpaired t-test) from PTX-treated slices.
We also tested bath-applied NEM (50 μM), a sulfhydryl-alkylating agent shown to block G-protein-effector interactions by alkylating the α-subunits of PTX-sensitive GTP-binding protein (Shapiro et al. 1994). The advantage with NEM is that it allows examination of CCK-8s-induced currents before and after inhibition of PTX sensitive G proteins within the same cell. In six control cells, the current induced by a second application of CCK within 20–30 min of the first application showed no significant reduction (–43.7 ± 3.3 vs. −40.3 ± 3.9 pA, paired t-test; P > 0.05, Fig. 5 A). In another 22 cells thus evaluated, the mean CCK-8s-induced inward current of −36.4 ± 2.8 pA was significantly reduced to −19.8 ± 2.2 pA (paired t-test; P < 0.05) when re-tested with CCK 20–30 min later, after a 5-min pretreatment with NEM (Fig. 5A). When I-V relationships were further examined, these results applied to 12 cells where the net CCK-induced current reversed at −101.4 ± 5.1 mV with the CCK-induced membrane conductance decreasing from 5.2 ± 0.4 to 4.4 ± 0.3 nS (paired t-test; P < 0.05) after NEM (Fig. 5B). In four other cells where the net CCK-induced current reversed at −38.7 ± 4.3 mV, mean membrane conductance increased from 4.9 ± 0.5 to 5.3 ± 0.7 nS (paired t-test; P < 0.05) after NEM (Fig. 5C). In the remaining six cells with no obvious net current reversals, mean membrane conductance decreased marginally from 5.5 ± 1.1 to 5.2 ± 0.9 nS (paired t-test; P > 0.05) after NEM (Fig. 5D).
Ventral root recordings
To assess whether the depolarizing effects of CCK-8s on ventral horn neurons are related to functionally relevant spinal cord outputs, the effects of CCK-8s on motor output were examined in the isolated spinal cord of the neonatal mouse. Under control conditions, we recorded spontaneous depolarizations and discharge from the lumbar ventral roots as described in an earlier report (Whelan et al. 2000). Figure 6 shows that bath application of CCK-8s (300 nM) increased the amplitude and discharge of spontaneous depolarizations recorded from the lumbosacral ventral roots.
The observation that exogenous CCK induces membrane depolarization and inward currents in neonatal spinal ventral horn neurons and increases ventral root activity implies the presence of functional postsynaptic CCK receptors. These receptors are of the CCKB subtype because responses were sensitive to pretreatment with a selective CCKB receptor antagonist, L-365,260. In the present study, the observation that CCK-8ns and CCK-4 had weak activity compared with CCK-8s also suggests that the actions of CCK-8 on ventral horn neurons are mediated by a receptor similar to that characterized in brain (Innis and Snyder 1980; Saito et al. 1980) and spinal cord (Rogawski 1982; Rogawski et al. 1985) by radioligand binding where CCK-8ns displayed a 10- to 50-fold lower affinity than CCK-8s. Collectively, these observations provide support for a functional role of postsynaptic CCK-receptors and indicate that their activation can enhance the excitability of both motoneurons and other unidentified neurons in the neonatal rat and mouse ventral spinal cord.
Our analyses of I-V relationships suggest that more than one membrane conductance underlies the depolarizing action of CCK-8s. In both motoneurons and a subpopulation of unidentified neurons, the CCK-8s–induced inward currents were associated with reduction in a membrane conductance that reversed close to the potassium equilibrium potential and shifted appropriately with the transmembrane potassium gradient. In these neurons, the linearity and reversal potential for the net CCK-8s-induced currents indicate involvement of a voltage-independent potassium conductance that contributes to resting membrane potential, often referred to as a leak conductance. Similar potassium-mediated responses to CCK-8s can be seen in rat solitary complex neurons (Branchereau et al. 1993), thalamic neurons (Cox et al. 1995), hippocampal interneurons (Miller et al. 1997), nucleus accumbens neurons (Kombian et al. 2004), brain stem motoneurons (Zheng et al. 2005), cultured nodose ganglion neurons (Peters et al. 2006), and cultured guinea pig sympathetic neurons (Xian and Kreulen 1994). A strong candidate for leak conductances is the two-pore domain family of potassium channels, in particular TASK-1 and TASK-3 members, which are highly expressed in the ventral horn (Talley et al. 2000, 2001).
By contrast with the responses observed in motoneurons, the CCK-8s-induced inward currents in a population of unidentified neurons were associated with an increase in membrane conductance that reversed around −30 mV, suggestive of a nonselective cationic conductance. This is similar to the CCK-8s-induced responses in rat supraoptic neurons where net currents also reverse close to −40 mV (Jarvis et al. 1992). In rat neostriatal neurons, net CCK-8s currents reverse around −10 mV (Wu and Wang 1996). Whether these differences reflect variations in underlying cationic conductances remains to be clarified, but they do suggest cell specificity in coupling of CCK receptors to ion channels. Moreover, we speculate that in some neurons, CCK receptors can couple to both potassium and cationic conductances in which situation the effects of the conductance increase and decrease are balanced causing no net change (a nearly parallel shift) as observed here in a subpopulation of ventral horn unidentified neurons. We presently lack data over a broader age range that might indicate a role for development influences on how these CCK receptors couple to the different conductances throughout and beyond the neonatal period. Although in the present work the current in motoneurons was completely abolished by the CCKB receptor antagonist, L-365,260, CCK-8s may bind to either CCKA or CCKB receptors in different regions examined. This binding can activate a number of intracellular signaling pathways, mediated through heterotrimeric G proteins (see Noble et al. 1999 for review). These may include Gqα to stimulate phosphoinositide hydrolysis, subsequent activation of protein kinase C and a rise in intracellular calcium, and/or Gi to inhibit adenylate cyclase. Although the identity of the G proteins linking CCK receptors to the conductances described in the preceding text remains undefined, the linkage may involve PTX-sensitive G proteins given the marked reduction in both types of conductances after treatment with PTX or NEM (Figs. 4 and 5).
The present observations bear a striking resemblance to the response of neonatal spinal neurons to other peptides including thyrotropin releasing hormone (Kolaj et al. 1997), angiotensin (Oz and Renaud 2002), vasopressin (Kolaj and Renaud 1998; Oz et al. 2001), and substance P (Ptak et al. 2000). Issues that arise from these observations include the source of the endogenous ligands for these receptors, and the possible physiological roles for these peptide receptors on neuronal function in the neonatal period. One source may derive from the cerebrospinal fluid, which is known to contain a variety of neuropeptides, some of which (e.g., vasopressin) exhibit a circadian rhythmicity in their concentrations (see Reppert et al. 1987). Although information is lacking about their development in the neonatal period, fibers displaying CCK-like immunoreactivity that originate from brain stem and midbrain sources have been observed in adult spinal cord (e.g., Maciewicz et al. 1984; Mantyh and Hunt 1984; Skirboll et al. 1983). Regardless of origin, CCK receptors may have specific functional links to neuronal development and survival (reviewed in Tirassa et al. 2002). This is suggested by their ability to undergo developmental changes (Cho et al. 1983), to modulate neurite outgrowth and by their neurotrophic action in cultured ventral spinal cord neurons (Iwasaki et al. 1989a,b), and to participate in recovery after axotomy in adults (Palkovits 1995; Saika et al. 1991; Schiffmann et al. 1991). It is also possible that their influence on excitability of neonatal spinal cord neurons is important for their subsequent development (see following text).
CCK-8s administration increased the frequency of spontaneous depolarizations and discharge recorded from the ventral roots. Because the ventral roots contain preganglionic autonomic and somatic motoneuronal axons, ventral root recordings may reflect activity from both sources. However, we believe that somatic motoneuronal activity provides the dominant contribution to the slow ventral root potentials. This is because the slow potentials are electrotonically propagated population synaptic potentials generated in neuronal somata and proximal dendrites. The preganglionic autonomic somata are ∼100–200 μm further than motoneuronal cell bodies from the recording site at the ventral roots, which will result in a correspondingly greater attenuation of their slow potentials. Our results, therefore demonstrate that CCK-8s-induced depolarization of ventral horn neurons can modulate the functional output of spinal networks during the neonatal period when locomotor networks are becoming functional. It remains to be determined if their actions during this period can influence network assembly, either indirectly by altering activity or directly through their trophic effects.
Several lines of evidence indicate that synaptic activity is an important component of motoneuron development. Differentiation of motoneurons is delayed in rat embryonic spinal cords cultured in the presence of TTX, indicating that electrical activity influences the time course of their development (Xie and Ziskind-Conhaim 1995). Similar to TTX action, motoneuron development is retarded when synaptic release is chronically blocked with tetanus toxin or omega-conotoxin, a Ca2+ channel blocker. Moreover, incubating spinal cords in medium containing high K+, which increases the frequency of spontaneous potentials, reverses the inhibitory effect of TTX, suggesting that electrical activity modulates motoneuron differentiation via Ca2+-dependent synaptic release of neurotransmitters or neurotrophic factors (Xie and Ziskind-Conhaim 1995). Indeed, activity-dependent developmental processes have been demonstrated to have profound effects not only on synaptic connectivity but also on the morphological and electrophysiological properties of motoneurons (for review, see Kalb and Hockfield 1992). Further experiments on the endogenous source of CCK and functional role of CCK receptors in development of neonatal spinal cord may provide some insight on these issues.
This study was supported by the funds from the Intramural Research Programs of the National Institutes of Health, DHHS, and by Canadian Institutes of Health Research Grant FRN 13244 and Heart and Stroke Foundation of Canada Grant T5643. L. P. Renaud holds the J. David Grimes Research Chair at the University of Ottawa.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 by the American Physiological Society