The acid-sensing ion channels (ASICs) form cation channels that are transiently activated by extracellular protons. They are expressed in dorsal root ganglia (DRG) neurons and in the periphery where they play a function in nociception and mechanosensation. Previous studies showed that FMRFamide and related peptides potentiate H+-gated currents. To better understand this potentiation, we examined the effect of FMRFamide-related peptides on DRG neurons from wild-type mice and animals missing individual ASIC subunits. We found that FMRFamide and FRRFamide potentiated H+-gated currents of wild-type DRG in a dose-dependent manner. They increased current amplitude and slowed desensitization following a proton stimulus. Deletion of ASIC3 attenuated the response to FMRFamide-related peptides, whereas the loss of ASIC1 increased the response. The loss of ASIC2 had no effect on FMRFamide-dependent enhancement of H+-gated currents. These data suggest that FMRFamide-related peptides modulate DRG H+-gated currents through an effect on both ASIC1 and ASIC3 and that ASIC3 plays the major role. The recent discovery of RFamide-related peptides (RFRP) in mammals suggested that they might also modulate H+-gated current. We found that RFRP-1 slowed desensitization of H+-gated DRG currents, whereas RFRP-2 increased the peak amplitude. COS-7 cells heterologously expressing ASIC1 or ASIC3 showed similar effects. These results suggest that FMRFamide-related peptides, including the newly identified RFRPs, modulate H+-gated DRG currents through ASIC1 and ASIC3. The presence of several ASIC subunits, the diversity of FMRFamide-related peptides, and the distinct effects on H+-gated currents suggest the possibility of substantial complexity in modulation of current in DRG sensory neurons.
Bioactive peptides with a C-terminus RFamide structure (such as FMRFamide) are abundant in invertebrates where they function as neurotransmitters and neuromodulators (Greenberg and Price 1992;Schneider and Taghert 1988). In mammals, genes for several FMRFamide-related peptides have been discovered, including neuropeptide FF (NPFF) and neuropeptide AF (Perry et al. 1997; Yang et al. 1985). Several observations suggested that mammalian FMRFamide-related peptides may play a role in sensory function and modulation of pain. NPFF was found in dorsal root ganglion (DRG) neurons, and inflammation induced its release (Allard et al. 1999; Ferrarese et al. 1986). Intracerebroventricular administration of FMRFamide or neuropeptide FF elicited hyperalgesia and reduced morphine-induced analgesia, suggesting an antimorphine effect of these peptides (Brussaard et al. 1989; Raffa 1988;Roumy and Zajac 1998; Tang et al. 1984). However, when injected intrathecally, these peptides induced a long-lasting analgesic effect (Gouardéres et al. 1993; Raffa 1988; Roumy and Zajac 1998). Novel mammalian RFamide-related peptides (RFRP) have recently been discovered; they include three human RFRPs and two rodent peptides, RFRP-1 (VPHSAANLPLRFamide) and RFRP-2 (SHFPSLPQRFamide) (Hinuma et al. 2000). These RFRPs are expressed in brain and spinal cord, but their function is unknown (Fukusumi et al. 2001; Ukena and Tsutsui 2001).
FMRFamide directly activated the FaNaCh (FMRFamide-activated Na+ channel) from Helix asperia andHelisoma trivolvis (Cottrell 1997;Jeziorski et al. 2000; Lingueglia et al. 1995). Application of FMRFamide to oocytes injected with FaNaCh RNA induced a large, partially desensitizing inward current. FMRFamide also activated FaNaCh in excised outside-out patches in the presence of GDP-β-S, which prevents G protein signaling, indicating direct gating by the peptide.
FaNaCh is a member of the degenerin/epithelial sodium channel (DEG/ENaC) family (Mano and Driscoll 1999;Waldmann and Lazdunski 1998; Welsh et al. 2002). Members of this family form voltage-insensitive cation channels in mammals where they have a variety of expression patterns and functional roles. In the mammalian neuronal system, there are three DEG/ENaC acid-sensing ion channels (ASICs), including acid-sensing ion channel 1 (ASIC1, also called ASIC and BaNaC2) (Garcı́a-Añoveros et al. 1997;Waldmann et al. 1997b), ASIC2 (also named BNC1, MDEG, and BaNaCl) (Garcı́a-Añoveros et al. 1997;Price et al. 1996; Waldmann et al. 1996), and ASIC3 (also called dorsal root acid-sensing ion channel, DRASIC) (Waldmann et al. 1997a). ASIC1 and ASIC2 also have alternatively spliced variants ASIC1a and 1b (Chen et al. 1998) and ASIC2a and 2b (Lingueglia et al. 1997). ASICs were found in both large and small DRG neurons and in the periphery they were localized in specialized cutaneous mechanosensory and nociceptor structures (Chen et al. 1998; Garcı́a-Añoveros et al. 2001;Olson et al. 1998; Price et al. 2000,2001). Studies of ASIC2 and ASIC3 knockout mice indicated that these channels were involved in both mechanosensation and nociception (Chen et al. 2002; Price et al. 2000,2001).
Interestingly, FMRFamide, NPFF, and the synthetic FRRFamide affected currents generated by heterologously expressed ASIC1 and ASIC3 channels (Askwith et al. 2000; Catarsi et al. 2001). None of these peptides activated ASIC channels on their own. However, the peptides altered current generated by extracellular acidosis. The peptides slowed the desensitization rate following application of a low pH stimulus and in some cases they increased the peak current amplitude (Askwith et al. 2000;Catarsi et al. 2001). Consistent with these heterologous expression studies, FMRFamide also modulated endogenous H+-gated currents in DRG neurons (Askwith et al. 2000). However, the contribution of individual ASIC subunits was not certain. To understand better how FMRFamide-related peptides modulate endogenous DRG currents, we examined the response in neurons from animals bearing disrupted genes for ASIC1, ASIC2, and ASIC3.
Mice with disruptions in the genes encoding ASIC1, ASIC2, and ASIC3 were generated as previously described (Price et al. 2000,2001; Wemmie et al. 2002). Wild-type or knockout mice were anesthetized with halothane. Following decapitation, the spine was opened, the spinal cord was rapidly removed, and thoracic and lumbar DRGs were dissected. All procedures were approved by the local Institutional Animal Care and Use Committee.
DRG neurons were cultured as described with some modifications (Mannsfeldt et al. 1999; Price et al. 2000). In short, dissected ganglia were digested with dispase (Roche Diagonostics, Indianapolis, IN) and collagenase IV (Worthington, Lakewood, NJ) followed by trypsin (Gibco BRL, Rockville, MD) in 37°C Hank's balanced buffer. Following repeated washing and trituration, cells were placed in petri dishes coated with poly-l-lysine and laminin (Sigma Chemical, St. Louis, MO).
We studied large DRG neurons with a diameter of 30–35 μm and focused specifically on cells with transient acid-evoked currents. Cells were obtained from 18 wild-type, 8 ASIC1−/−, 10 ASIC2−/−, and 8 ASIC3−/− animals.
COS-7 cells were transfected with cDNAs encoding mouse ASIC1 or mouse ASIC3 cloned into pMT3 (2 μg/ml) (Askwith et al. 2000;Swick et al. 1992) using Transfast reagent (Promega). They were cotransfected with a cDNA for enhanced green fluorescent protein as a marker to identify transfected cells.
Whole cell patch-clamp recording
Glass pipettes (Fisher Scientific, Palatine, IL) were prepared (3–6 MΩ) with a Flaming/Brown micropipette puller (model P-97/VF, Sutter Instrument, Novato, CA). Whole cell recordings were made using an Axopatch 200 (Axon Instruments, Union City, CA) amplifier. pH stimuli were applied by a rapid solution change with a series of capillaries closely positioned to the cells. Digital recordings were captured with a TL-1 DMA Interface and pClamp6.0 software (Axon Instruments). Membrane voltage was maintained at −70 mV. Series resistance was compensated (approximately 60%) in some but not all studies; because the acid-induced current changes are relatively slow, series resistance compensation did not change the results. Experiments were performed at room temperature (20–23°C). Most cells were studied within 1 day after seeding, although some were studied on day 2.
During the course of experiments, we found that, with repeated stimuli, the amplitude of H+-gated currents sometimes progressively declined (i.e., rundown). As we previously reported, this was especially the case in ASIC3−/− neurons (Xie et al. 2002). To minimize the effect of rundown on the analysis, in all studies we bracketed the intervention (for example, pH 5 application in the presence of peptide) with two controls (for example, two pH 5 applications without peptide). We then compared currents obtained in response to an intervention to the average of currents obtained before and after the intervention.
Cells were superfused in buffer solutions containing (in mM) 120 NaCl, 1 MgCl2, 2 CaCl2, 10 HEPES, 10 MES, 5 KCl, and 5.55 glucose, adjusted to pH 7.4 with TMA·OH. Osmolarity of the solution was adjusted with TMA·Cl. The pipette solution contained (in mM) 100 KCl, 5 MgCl2, 10 EGTA, 40 HEPES, and 1 Na2ATP, adjusted to pH 7.4 with KOH. Solutions containing peptides were perfused onto cells for 15 to 20 s prior to a low pH stimulus. FMRFamide was purchased from Sigma. FRRFamide, mouse RFRP-1 (VPHSAANLPLRFamide), and mouse RFRP-2 (SHFPSLPQRFamide) were synthesized by Research Genetics.
We measured the rate of desensitization to study the kinetics of DRG H+-gated currents. For wild-type DRG neurons, current desensitization could be fit with a double exponential function. However, after FMRFamide or FRRFamide treatment the current was not well fit with exponential functions. Therefore we measured the time for the current to decay to 50% of the maximum (T 1/2).
Data are mean ± SE. For analysis of the dose–response curves, we used a four-parameter logistic function y =d + (a − d)/[1 + (X/c)b], wherey is the response variable (amplitude orT 1/2); X is the concentration of peptide; a and d estimate the minimum and maximum response, respectively; c is the concentration that would give a response that is halfway between the two limits (i.e., the EC50), and b is the Hill slope. A paired t-test was used to compare data from the same cell. When the parameters from different genotypes of neurons were compared, a one-way ANOVA and least significant difference tests were used (Armitage 1971). A value of P < 0.05 was considered statistically significant.
FMRFamide and FRRFamide modulated H+-evoked currents in cultured DRG neurons
Application of a pH 5 solution to wild-type DRG neurons elicited a current that rapidly activated and then desensitized in the continued presence of the stimulus (Fig.1 A). Pretreatment with FMRFamide or FRRFamide increased the peak amplitude of the H+-gated current. The increased current amplitude is consistent with our previous qualitative results studying mASIC3 expressed in Xenopus oocytes (Askwith et al. 2000) and with studies of heteromultimeric hASIC2a and hASIC3 expressed in Xenopus oocytes (Catarsi et al. 2001). In a previous qualitative study of current in rat DRG neurons (Askwith et al. 2000), we did not report an increase in H+-gated current amplitude; that may represent a difference between rat and mouse DRG neurons or the rapid solution changes that were used in the present study. To assess the response of DRG H+-gated currents to FMRFamide-related peptides, we examined the effects of increasing concentrations of FMRFamide and FRRFamide (Fig. 1 B). Both peptides increased current amplitude. At equivalent concentrations, FRRFamide produced larger effects than FMRFamide.
In addition to effects on amplitude, FMRFamide and FRRFamide also affected the rate of desensitization. To quantitatively examine this effect, we measured the time for the current to fall to 50% of the maximum (T 1/2). We calculated the ratio of T 1/2 in the presence to that in the absence of FMRF-amide and related peptides. Both FMRFamide and FRRFamide slowed desensitization of H+-gated currents, increasing the T 1/2 ratio (Fig. 1 C). The threshold effect was observed at approximately 1 μM. FRRFamide produced larger effects than FMRFamide.
Effects of FMRFamide-related peptides on DRG H+-gated currents are mediated by ASIC channels
To identify the ASIC subunits responsible for the FMRFamide response, we studied H+-gated currents from wild-type, ASIC1−/−, ASIC2−/−, and ASIC3−/− DRG neurons. We used a nearly maximal concentration of peptide (100 μM). Examples are shown in Fig. 2.
Loss of ASIC1 and ASIC3 produced opposite effects on the way that FMRFamide altered the response to a pH 5 solution (Figs. 2 and 3,A and B). Without ASIC1, FMRFamide generated a larger H+-gated current amplitude and a longer T 1/2 of desensitization (Figs. 2 and 3). In contrast, without ASIC3, the FMRFamide-induced increase in amplitude and prolongation of the desensitized rate were markedly attenuated. However, loss of ASIC3 did not abolish the peptide effect; in ASIC3-null neurons FMRFamide and FRRFamide slowed the desensitization rate by 1.208 ± 0.096- and 1.819 ± 0.348-fold, respectively (Fig. 3 B). However, the effect on H+-gated current amplitude was eliminated. Amplitude was 1.00 ± 0.04- and 1.05 ± 0.05-fold with FMRFamide and FRRFamide, respectively (Fig.3 A). These data suggest that ASIC3 plays a major role in mediating the response to FMRFamide and FRRFamide; eliminating ASIC3 attenuated the response, whereas eliminating ASIC1 and retaining ASIC3 augmented the response. However, as we discuss further below, ASIC1 also appeared to contribute to the FMRFamide and FRRFamide effect. In other words, when ASIC3 was disrupted we saw the smaller ASIC1 response, and when ASIC1 was disrupted we saw the larger ASIC3 response.
The loss of ASIC2 did not significantly change the effect of FMRFamide or FRRFamide on peak current amplitude or desensitizationT 1/2 (Fig. 3, A andB). These results suggest that ASIC2 did not play a major role in mediating the DRG current response to these peptides. This result is consistent with the observation that these peptides did not alter current generated by ASIC2 expression in heterologous systems (Askwith et al. 2000).
Mouse RFRPs modulated H+-gated currents from cultured DRG neurons
The recently identified rodent RPRF-1 and RPRF-2 are candidates for mammalian modifiers of DEG/ENaC currents. We tested their effects on mouse DRG neurons and found that applying the peptide immediately before lowering the pH altered the current response. RFRP-1 (100 μM) slowed the desensitization of DRG H+-gated currents (Fig. 4), but the peak amplitude of H+-gated currents did not significantly change. In contrast, RFRP-2 (100 μM) increased the peak amplitude of pH 5–evoked currents but did not alter the desensitization rate (Fig.4). Consistent with earlier studies, neither peptide elicited currents on its own (not shown) (Askwith et al. 2000;Catarsi et al. 2001). Moreover, as previously reported for FMRFamide, it was necessary to apply the peptide prior to the pH stimulus; when we applied simultaneously the pH 5 stimuli and RPRF-1 or RPRF-2, there was no alteration of the currents (Askwith et al. 2000). These data indicate that RFRPs alter H+-gated DRG currents and suggest that each may have a specific effect.
Recent data have shown that FMRFamide-related peptides also activate heptahelical G protein–coupled receptors (Dong et al. 2001; Elshourbagy et al. 2000). This observation raised the question of whether FMRFamide and related peptides might act on DRG H+-gated currents through a second messenger effect. To test this possibility, we included GDP-β-S, a competitive G protein inhibitor, in the internal pipette solution at a 1–2 mM concentration (Oz and Renaud 2002); then, after obtaining the whole cell configuration and allowing 5 min for the GDP-β-S to diffuse into the cell, we tested the effects of FRRFamide and RFRP-2 on H+-gated currents. GDP-β-S did not alter the current response to either peptide (Fig.5). These results suggest that, as previously reported for FMRFamide (Askwith et al. 2000;Catarsi et al. 2001), RFRP-2 may directly interact with the ion channel.
RFRP-1 and RFRP-2 altered H+-gated currents of heterologously expressed ASIC1 and ASIC3
Both FMRFamide and RFRPs altered DRG H+-gated currents. Because the effects of FMRFamide are mediated primarily through ASIC1 and ASIC3, but not ASIC2, we hypothesized that RFRP-1 and RFRP-2 also modulate DRG H+-gated currents through ASIC1 and ASIC3 subunits. To test this, we expressed ASIC1 and ASIC3 in COS-7 cells and examined the peptide response. RFRP-1 slightly slowed the desensitization of ASIC3, but not ASIC1 (Table1). Although statistically significant, this effect was more subtle than that in DRG neurons. RFRP-2 increased the peak amplitude of both ASIC1 and ASIC3, with no significant effects on desensitization rates, consistent with RFRP-2 activity in DRG neurons (Table 1). These results suggest that, as with FMRFamide, RFRPs modified DRG H+-gated currents through both ASIC1 and ASIC3. ASIC3 was the only channel in which RFRP-1 slowed desensitization. Thus, as with FMRFamide, ASIC3 may play a more significant role than ASIC1 (Table 1).
Contribution of ASIC subunits to the FMRFamide response in DRG neurons
Earlier studies showed transcription of ASIC1, ASIC2, and ASIC3 and several alternatively spliced isoforms in large DRG neurons (Chen et al. 1998; Garcı́a-Añoveros et al. 2001; Lingueglia et al. 1997;Price et al. 2000, 2001; Waldmann et al. 1997a). ASIC1, ASIC2, and ASIC3 proteins have also been identified in these neurons (Garcı́a-Añoveros et al. 2001; Olson et al. 1998; Price et al. 2000, 2001). Functional data indicated that all three subunits contribute to H+-gated currents in these large neurons (Benson et al. 2002; Xie et al. 2002), with heteromultimeric channels composed of at least two, and probably all three, subunit types (Benson et al. 2002; de La Rosa et al. 2002; Price et al. 2001; Xie et al. 2002). In this study, we examined the quantitative aspects of FMRFamide modulation of these H+-gated currents. By examining the effect of FMRFamide and FRRFamide on H+-gated DRG currents from mice bearing disrupted ASIC genes, we were able to assess the relative contribution of the different channel subunit.
The contribution of ASIC3 was the most readily apparent. In ASIC3-null neurons, FMRFamide treatment elicited very little increase in H+-gated current amplitude or desensitization rate. The response to FRRFamide was also markedly attenuated. These data indicate that ASIC3 is required for the maximal FMRFamide response. Because, as we describe above, H+-gated currents represent heteromultimers of ASIC subunits, ASIC3 is likely a component that is critical for the normal response to FMRFamide and related peptides. They are also consistent with the observation that FMRFamide has a large effect on H+-gated currents in heterologous cells expressing homomultimeric ASIC3 channels (Askwith et al. 2000; Catarsi et al. 2001). Thus ASIC3 appears to play a major role in conferring the response to FMRFamide.
In ASIC1-null DRG neurons, FMRFamide and FRRFamide potentiated H+-gated current amplitude and prolonged desensitization to a greater extent than in wild-type, ASIC2−/− or ASIC3−/− neurons. These data could be interpreted to indicate that ASIC1 plays little or no role in the response of heteromultimeric channels to the RFamide peptides. In either case, the absence of ASIC1 subunits might allow an even greater contribution from ASIC3 subunits and hence a greater response to the RFamide peptides. The data suggest that ASIC1 makes a significant contribution because FMRFamide and related peptides enhanced H+-gated current from homomultimeric ASIC1 channels expressed in heterologous cells (Askwith et al. 2000). We also found that, when ASIC3 was missing, FMRFamide and FRRFamide still prolonged the desensitization rate (the effect was small but statistically significant). Because ASIC2 does not appear to respond to FMRFamide (Askwith et al. 2000 and see following text), these results suggest that ASIC1 was responsible for the response in ASIC3−/− neurons. Our data do not allow us to distinguish the relative contributions of ASIC1a or ASIC1b to the FMRFamide response, although previous data indicate that FMRFamide modulates H+-gated current produced by both isoforms (Askwith et al. 2000).
In an earlier study, we showed that FMRFamide did not alter ASIC2 homomultimeric currents (Askwith et al. 2000). In ASIC2-null neurons, the FMRFamide and FRRFamide augmentation of H+-gated currents could not be distinguished from wild-type. There are at least two potential explanations for the absence of an effect with ASIC2 disruption. It is possible that heteromultimeric channels composed of the remaining ASIC3 and ASIC1 subunits balance the response to FMRFamide in ASIC2-null neurons so that there is no net change. Alternatively, it is possible that ASIC2 makes a very minor contribution to DRG H+-gated current, hence eliminating it would not alter the response to FMRFamide. We think this alternative is unlikely because we previously showed the absence of ASIC2 altered the biophysical properties of DRG H+-gated current in a manner most consistent with its significant contribution to heteromultimeric channels (Benson et al. 2002). Thus ASIC2 probably does not contribute to the FMRFamide binding site even though it is an important component of heteromultimeric H+-gated DRG channels. Although additional studies of FMRFamide and related peptides applied to combinations of ASIC subunits expressed in heterologous cells could be of interest, our study has the advantage that we studied the peptide effect on DRG neurons. Thus we could evaluate the consequences of eliminating individual subunits in the context of any posttranslational modifications or associated proteins that may influence ASIC functions in neurons.
Modulation of mouse DRG H+-gated current by mouse FMRFamide-related peptides
The relationship of mouse RFRP-1 and -2 to FMRFamide suggested that they might activate or modify DRG H+-gated currents. As with previous reports on FMRFamide (Askwith et al. 2000; Catarsi et al. 2001), these peptides did not activate the channel on its own. However, they modified the response of ASIC channels to acidification. Interestingly, the two peptides had slightly different effects on H+-gated current in DRG neurons: RFRP-1 slowed desensitization, whereas RFRP-2 increased the peak amplitude. These effects were specific, with different channels showing distinct responses to the peptides. Results with heterologously expressed channels were consistent with this conclusion, RFRP-1 did not alter H+-gated current from heterologously expressed ASIC1 but did slow desensitization of homomultimeric ASIC3 currents. Conversely, RFRP-2 increased amplitude of both ASIC1 and ASIC3 currents without significantly prolonging desensitization.
A number of FMRFamide-related peptides modulate ASIC H+-gated current. Our studies combined with earlier work begin to provide insight into structure–function relationships for these peptides. For example, RFRP-1 (VPHSAANLPLRFamide) and RFRP-2 (SHFPSLPQRFamide) both ended with PXRFamide, but they showed different effects on DRG H+-gated currents. RFRP-2 and NPFF (FLFQPQRFamide) share the same C-terminal sequence (PQRF), but RFRP-2 only increased the peak amplitude, whereas NPFF slowed desensitization (Askwith et al. 2000). These results indicate that peptide regions other than the four C-terminal residues influence the response. Moreover, RFRP-1 had effects similar to those of NPFF, suggesting that several sites of interaction between ASIC subunits and the peptides are important in determining the functional consequences. Future studies examining the effect of multiple peptides will be required to develop detailed structure–function relationships for the effects of FMRFamide-related peptides on ASIC channels.
ASIC subunits have been shown to play a role in sensory function. For example, in a skin-nerve preparation, the loss of ASIC3 blunted the response to acid application (Price et al. 2001). Behavioral assays also indicated that ASIC3 was required for a normal response to noxious stimuli (Chen et al. 2002;Price et al. 2001). Moreover, inflammation induced the release of NPFF in spinal cord (Kontinen et al. 1997) and increased ASIC3 expression in DRG (Voilley et al. 2001; Yiangou et al. 2001).
Mammalian FMRFamide-related peptides also play an important role in sensory function (Raffa 1988; Roumy and Zajac 1998). For example, NPFF has been shown to modulate nociception (Brussaard et al. 1989; Gouardéres et al. 1993; Raffa 1988; Roumy and Zajac 1998; Tang et al. 1984). These considerations suggest that interactions between ASIC subunits and FMRFamide-related peptides might modulate nociceptive responses. These peptides might also modulate ASIC channel function in the CNS. Recent work has shown that ASIC1 is important for normal hippocampal synaptic plasticity and for normal learning and memory (Wemmie et al. 2002). Immunolocalization data have also shown RFRP-like immunoreactivity in multiple brain and spinal cord regions, suggesting the possibility of interactions.
In different cells and tissues, H+-gated currents are composed of different ASIC subunits and/or different proportions of subunits. For example, some CNS neurons express ASIC1 and ASIC2, but not ASIC3 (Chen et al. 1998; Lingueglia et al. 1997; Waldmann et al. 1997a; Wemmie et al. 2002). It seems likely that, in distinct DRG neurons, the relative proportion of the subunits will vary. Our data also show that RFRP-1 and -2 produce specific effects on H+-gated current generated by different ASIC subunits. Thus the combination of subunits and the diversity of FMRFamide-related peptides offer the possibility of substantial complexity in modulation of currents generated by ASIC channels.
We thank A. Nekoomand, P. Weber, T. Nesselhauf, P. Karp, L. Field, C. Pruess, J. Hensen, E. Lamani, L. Momchilov, T. Rokhlina, H. Carmichael, R. Smith, and T. Mayhew for excellent technical support and assistance. We thank C. Benson, A. Berger, P. Snyder, and L. Qin for advice and comments, and we thank B. Zimmerman for help with the statistical analysis.
This study was supported by the Howard Hughes Medical Institute (HHMI). M. J. Welsh is an Investigator of the HHMI. J. A. Wemmie was supported by the Veterans Administration Research Career Development Award.
Address for reprint requests: M. J. Welsh, Howard Hughes Medical Institute, Roy J. and Lucille A. Carver College of Medicine, 500 EMRB, Iowa City, IA 52242 (E-mail:).
- Copyright © 2003 The American Physiological Society