Previous Article |
Table of Contents
| Next Article 
BIOLOGICAL SCIENCES / NEUROSCIENCE
Contribution of the receptor guanylyl cyclase GC-D to chemosensory function in the olfactory epithelium
,**
*Department of Physiology, University of Saarland School of Medicine, 66421 Homburg/Saar, Germany; Department of Anatomy and Neurobiology and Program in Neuroscience, University of Maryland School of Medicine, Baltimore, MD 21201; Munich Center for Integrated Protein Science and Department of Pharmacy, Center for Drug Research, Ludwig-Maximilians-Universität München, 81377 Munich, Germany; Department of Pharmacology and the Cecil H. and Ida Green Center for Reproductive Biology Sciences, University of Texas Southwestern Medical Center, Dallas, TX 75390; and ¶Department of Molecular Biology and Genetics and Center for Sensory Biology, Johns Hopkins School of Medicine, Baltimore, MD 21205
Edited by Linda B. Buck, Fred Hutchinson Cancer Research Center, Seattle, WA, and approved July 26, 2007 (received for review May 25, 2007)
 |
Abstract |
|---|
 Top Abstract ResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
cGMP | natriuretic peptide | transduction | Gucy2d | CNGA3
olf, which stimulates type III adenylyl cyclase to increase intracellular levels of cAMP (4–9). This second messenger then directly opens cAMP-sensitive, cyclic nucleotide-gated (CNG) channels in the plasma membrane, resulting in the initial depolarization of the cell (3, 4). OSNs that use this cAMP-mediated transduction cascade respond to a wide variety of odors, including food odors, volatile pheromones, and peptides that bind major histocompatibility complex proteins (4, 10–12). A recently discovered second family of G protein-coupled receptors expressed in the MOE, the trace amine-associated receptors, may also function in the recognition of some odors or pheromones by coupling to the canonical cAMP pathway (13). However, there has been intense debate as to whether the cAMP cascade is the only excitatory sensory transduction mechanism in the MOE (11, 14, 15).
A subpopulation of ciliated MOE neurons that express the orphan receptor guanylyl cyclase GC-D (16) are prime candidates to mediate cAMP-independent odor recognition. GC-D cells lack key components of the canonical OSN odor transduction cascade, including Golf, type III adenylyl cyclase, the Ca2+/calmodulin-dependent phosphodiesterase PDE1C2, the cAMP-specific phosphodiesterase PDE4A, and the cAMP-sensitive CNG channel subunits CNGA2 and CNGB1b (17, 18). Instead, these neurons express a cGMP-specific CNG channel subunit, CNGA3, and a cGMP-stimulated phosphodiesterase, PDE2 (17, 18). Although these cells were identified more than a decade ago (16), there is no functional evidence that they play a role in odor recognition. On the basis of their distinct molecular profiles and axonal targets in the olfactory bulb (OB) (16–18), GC-D cells have been suggested to respond to hormones or pheromones (18). However, it has not been possible to obtain physiological recordings from identified GC-D neurons. Consequently, no sensory stimuli detected by these cells are known, and the proposed chemosensory function of GC-D and the cells that express it remains elusive.
|
Results |
|---|
|
TopAbstractResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
GC-D expression is specifically abolished in Gucy2d –/– mice. The GC-D message was amplified by RT-PCR from MOE cDNA of Gucy2d +/– but not –/– mice (SI Fig. 6C). The expression of two other markers of GC-D neurons, PDE2 (18) and CNGA3 (17), was unaffected. Only a small subset of MOE neurons in Gucy2d +/– and –/– mice express the 
-galactosidase (-gal) reporter (SI Fig. 6D), consistent with the small number of GC-D neurons reported previously (16). The axons of GC-D neurons innervate a small number of glomeruli in the caudal main OB, the necklace glomeruli (18), although the projection path of GC-D neuron axons to the necklace glomeruli was unknown. Whole-mount X-Gal histochemistry of OBs from Gucy2d +/– and –/– mice revealed a diffuse group of -gal-positive axons that project caudally from the MOE and along the ventrolateral aspects of the OB (Fig. 1A and SI Fig. 8). Upon reaching the caudal OB, these axons course dorsally on the medial and lateral surfaces to terminate in 9–11 superficial glomeruli that are consistent in number and position with the necklace glomeruli (20, 21). These -gal-positive glomeruli exhibit a rough symmetry across the two OBs within animals (Fig. 1A Left). The level of -gal expression is greater in Gucy2d –/– mice than in +/– mice (Fig. 1A Right), as would be expected by the presence of a second lacZ allele. However, the number and positions of -gal-positive glomeruli in Gucy2d +/– and –/– mice are nearly identical (Fig. 1A Right), indicating that functional GC-D is not required for normal axonal targeting.
|
-gal-positive neurons express a key molecular marker of GC-D neurons, PDE2. This phosphodiesterase, present in the somata, cilia, dendrites, and axons of GC-D neurons, is a specific marker of both GC-D neurons and the necklace glomeruli (18). We performed double-label immunohistochemistry (IHC) for -gal and PDE2 in OBs of Gucy2d +/– and –/– mice. All glomeruli that were immunopositive for -gal were also PDE2-positive (and vice versa) in both Gucy2d +/– and –/– mice (Fig. 1B and data not shown). Thus, expression of the -gal reporter is restricted to GC-D neurons and the necklace glomeruli and serves as a specific marker for GC-D neurons in these gene-targeted mice.
The presence of normal afferent activity from the MOE to most OB glomeruli, although not the necklace glomeruli, depends on an intact cAMP signaling cascade (22). To provide an initial assessment of the necessity of GC-D expression for the maintenance of afferent activity in the necklace glomeruli, we took advantage of the observation that tyrosine hydroxylase (TH) expression in OB periglomerular cells is a correlate of afferent activity (22, 23). In Gucy2d +/– mice, both -gal-positive and -gal-negative glomeruli express TH (Fig. 1C Left). However, TH immunoreactivity is nearly absent in -gal-positive glomeruli of Gucy2d –/– mice (Fig. 1C Right), suggesting that GC-D neurons provide functional innervation to the necklace glomeruli and that GC-D itself is necessary for normal afferent activity to these glomeruli.
MOE Responses to Uroguanylin Family Peptides Require GC-D. We next sought to determine whether GC-D neurons recognize distinct chemosensory cues. We focused on peptides found in urine (Fig. 2A) because other receptor guanylyl cyclases function as peptide receptors (24) and nonvolatile peptide stimuli have been shown to reach and activate the MOE after direct contact with a stimulus (10, 12). Because GC-D neurons lack components of the canonical olfactory cAMP signaling cascade, we reasoned that any candidate ligand would continue to activate the MOE even when the cAMP cascade has been disrupted. By recording local field potentials [the electroolfactogram (EOG)] (12, 25) in Cnga2-null mice (26), we found that uroguanylin, a peptide hormone present in urine and associated with the regulation of sodium balance in the gut and kidney through the receptor guanylyl cyclase GC-C (27), elicits a robust electrical response (Fig. 2B). The related peptide guanylin (27) produced a similar response (Fig. 2B). However, another member of this family, the heat-stable bacterial enterotoxin STp (27), failed to activate the MOE (Fig. 2B). The renal natriuretic peptide urodilatin (28), a GC-A agonist, also failed to produce an EOG response (Fig. 2B). Uroguanylin- and guanylin-evoked potentials were unaffected by the addition of the adenylyl cyclase inhibitor SQ22536 (300 µM, Fig. 2C), confirming their independence from a cAMP signaling mechanism. However, the responses were inhibited by the addition of the CNG channel blocker L-cis-diltiazem (29) (100 µM, Fig. 2C), implicating another CNG channel, perhaps CNGA3 (17) (see below), in the transduction of uroguanylin and guanylin.
|
MOE Responses to Uroguanylin Family Peptides Require CNGA3. In the MOE, the cGMP-specific CNG channel subunit CNGA3 is expressed specifically in GC-D neurons (17). To examine the role of CNGA3 in GC-D neuron function, we used Cnga3 –/– mice (30), which express GC-D and PDE2 but not CNGA3 in the MOE (SI Fig. 6C). Consistent with our observations in Gucy2d –/– mice, TH immunoreactivity is nearly absent in PDE2-positive glomeruli in Cnga3 –/– mice (Fig. 3A). Thus, it appears that CNGA3, like GC-D (Fig. 1C), is necessary for normal afferent activity to the necklace glomeruli. We next tested the effects of Cnga3 deletion on uroguanylin and guanylin responses in the MOE. Again consistent with our observations in Gucy2d –/– mice, EOG responses to these peptides were abolished in the MOE of Cnga3 –/– mice (Fig. 3 B and C). Together, these data indicate that CNGA3 is required for the transduction of uroguanylin and guanylin stimulation of the MOE.
|
-gal-positive neurons were visualized with the fluorogenic -gal substrate fluorescein digalactoside and were clearly distinguished from those surrounding knobs that did not express the transgene (Fig. 4 A and B). Action potentials were measured by obtaining loose patch clamp recordings (33, 34) of labeled knobs. GC-D neurons in Gucy2d +/– mice were spontaneously active, with an average spike frequency at rest of 5.7 ± 2.2 Hz (n = 14). Spontaneous firing was significantly reduced or eliminated (1.2 ± 1.1 Hz; n = 6; P < 0.001) in the absence of GC-D, suggesting that basal enzymatic activity of GC-D contributes to setting the resting potential of these cells. These results also support a role for GC-D in the maintenance of normal afferent activity in the necklace glomeruli (Fig. 1C).
|
-gal-positive neurons from Gucy2d –/– mice (Fig. 4 D and F; n = 6), consistent with the field potential recordings (Fig. 2D). Thus, GC-D neurons are chemosensory neurons activated by the novel stimuli uroguanylin and guanylin and by a biologically important odor source, urine. Furthermore, these sensory cells, in contrast to mammalian photoreceptors, use an excitatory, cGMP-mediated sensory transduction mechanism.
Optical Imaging of GC-D Cell Function Reveals Unique Receptive Properties.
The results described thus far indicate that stimulus encoding by GC-D cells differs sharply from that of canonical OSNs and vomeronasal sensory neurons, where a given molecular cue stimulates only a small fraction of the entire population (4, 12, 33, 34) because of the expression of large numbers of receptors with different receptive properties (2). To provide a quantitative analysis of the diversity of responses across the GC-D neuron population, we next used confocal microscopy of the epithelial sheet (33, 34) to image changes in intracellular Ca2+ levels in single dendritic knobs of -gal-positive neurons. The neuroepithelium was loaded with the Ca2+ indicator fluo-4/AM (33, 34), and -gal-positive knobs were visualized with the fluorogenic substrate resorufin galactoside (Fig. 5 A–D). Consistent with our patch clamp results (Fig. 4) and the role of CNGA3 in GC-D-mediated sensory transduction (Figs. 2C and 3), dendritic knobs of -gal-positive neurons in Gucy2d +/– mice displayed a rise in intracellular Ca2+ upon stimulation with KCl, 8-bromo-cGMP or two highly membrane-permeant PDE inhibitors, the PDE2-selective erythro-9-[2-hydroxy-3-nonyl]-adenine (EHNA) and 3-isobutyl-1-methylxanthine (IBMX) (Fig. 5 E and F). However, in contrast to its effects on canonical OSNs (12), the adenylyl cyclase activator forskolin did not produce a Ca2+ transient in GC-D neurons (Fig. 5 F and G). This observation is consistent with the absence of adenylyl cyclase and cAMP-sensitive CNG channels in GC-D cells. EHNA- and IBMX-induced Ca2+ transients were absent in -gal-positive neurons of Gucy2d –/– mice, although these neurons were still stimulated by KCl and 8-bromo-cGMP (Fig. 5 E and G). Therefore, we conclude that GC-D is responsible for the cGMP signal in these cells and that the transduction cascade downstream of GC-D remains intact in Gucy2d –/– mice.
|
-gal-positive dendritic knobs to chemosensory stimuli in Gucy2d +/– and –/– mice (58 and 74 knobs, respectively; Fig. 5 H–J). In Gucy2d +/– mice, nearly 100% of the GC-D neurons tested produced a transient Ca2+ elevation to a mixture of uroguanylin and guanylin (Fig. 5 H–J). Dilute urine evoked a response with the same high probability (Fig. 5I). Even when we applied only a single peptide, the probability of evoking a response was still remarkably high (70–76%; Fig. 5I). However, in no case did we observe a Ca2+ response to urine or peptide ligands in Gucy2d –/– mice (Fig. 5I), although -gal-positive cells were KCl- and 8-bromo-cGMP-responsive in these mice (Fig. 5G). We also never observed a Ca2+ response to uroguanylin or guanylin in any -gal-negative neuron, attesting to the exceptional selectivity of these two peptides for GC-D cells. Although GC-D cells have been suggested to detect some small volatile urinary pheromones such as 2-heptanone and 2,5-dimethylpyrazine (11), we never observed -gal-positive knobs of Gucy2d +/– mice respond to either of these two chemicals with either a Ca2+ rise (n = 16, Fig. 5I) or an increase in action potential frequency (n = 11, data not shown). On a finer scale we did observe some heterogeneity among GC-D cells in Gucy2d +/– mice. This observation, first suggested in the knob patch clamp recordings (Fig. 4), was clearly apparent when cells were characterized by Ca2+ imaging: whereas the peptide mixture stimulated virtually all GC-D neurons, the individual peptides stimulated subsets of GC-D neurons that only partially overlap (Fig. 5J). Therefore, GC-D neurons can be divided into at least three functional classes based on peptide recognition profiles: guanylin-sensitive/uroguanylin-insensitive (7/26, 27%), uroguanylin-sensitive/guanylin-insensitive (7/26, 27%), and guanylin- and uroguanylin-sensitive (12/26, 46%) (Fig. 5J and SI Fig. 9). Imaging GC-D cell function at cellular resolution in Gucy2d –/– mice revealed that the recognition of single peptide ligands is abolished as a whole (Fig. 5I), indicating that GC-D plays an essential role in mediating peptide responsivity in all three functional classes of GC-D neurons.
|
Discussion |
|---|
|
TopAbstractResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
Our analyses of Gucy2d +/– and –/– mice, together with measurements performed in Cnga2 and Cnga3 null mice, provide several lines of evidence that the GC-D cells employ a cGMP-dependent second messenger cascade for chemosensory transduction. Furthermore, our data show that not only is this cascade responsible for action potential generation and intracellular Ca2+ elevation in these cells but that modulation of GC-D by peptide ligands alters these signals. For example, uroguanylin and guanylin responses are unaffected by the perturbation of the canonical OSN cAMP signaling cascade either by inhibition (Fig. 2C) or activation (Fig. 5F) of adenylyl cyclase or by deletion of CNGA2, the functionally critical subunit of the olfactory cAMP-gated channel (35) (Fig. 2B). Uroguanylin- and guanylin-dependent responses are absent in Gucy2d –/– and Cnga3 –/– mice (Figs. 2D, 3B, 3C, 4D, and 5I), whereas in Gucy2d +/– mice they are reduced by the CNG channel blocker, L-cis-diltiazem (Fig. 2C) and mimicked by the PDE inhibitors IBMX and EHNA (Fig. 5 E and F). 8-bromo-cGMP stimulates GC-D neurons in both Gucy2d +/– and –/– mice (Figs. 4 E and F and 5 F and G) because this cGMP analog bypasses the guanylyl cyclase GC-D in the signaling cascade to gate the CNG channel in these cells directly. Thus, a second cyclic nucleotide-based signaling system, which depends on elevation of cGMP, not cAMP, is used for chemodetection by the MOE.
The excitatory cGMP cascade mediating sensory responses in GC-D neurons differs critically from that of vertebrate retinal photoreceptors because stimulation of those sensory cells by light leads to the G protein activation of a PDE, hydrolysis of cGMP, closing of CNG channels, and a graded membrane hyperpolarization (31). In contrast, stimulation of GC-D neurons leads to membrane depolarization and an increase in action potential frequency (Fig. 4). Interestingly, photoreceptors of the lizard parietal eye also show cGMP-dependent depolarizing responses to sensory stimulation, although this pathway depends on the G protein-dependent inhibition of PDE activity (36). The modulation of GC-D neuronal firing, rather than graded changes in membrane voltage, represents the critical signal processed in the necklace glomeruli of the OB. We cannot distinguish between a model in which direct modulation of GC-D is responsible for information flow and an alternative scenario in which a separate G protein-coupled receptor and downstream effectors impinge on the GC-D pathway. Regardless, our studies demonstrate that the GC-D protein is an essential component for signaling in these cells and that peptide ligands modulate physiological responses to produce signals detected in the target tissue.
A surprising finding was that the GC-D cells are highly sensitive detectors of two peptide hormones, uroguanylin and guanylin, both of which are known to elevate cGMP levels in several tissues (27). Both peptides are secreted from intestine and influence salt and water transport in intestine and kidney (37). GC-D cells are also excited by dilute urine from mature males and females. Uroguanylin is present in relatively high concentrations in urine (27) and may also be present in feces. Guanylin is degraded by chymotrypsin in different parts of the kidney (37), but its presence in urine, at least at low concentrations, cannot be ruled out. Notably, a third member of this peptide family, the heat-stable Escherichia coli enterotoxin STp, was ineffective as a chemostimulus for GC-D cells. This latter finding might argue against a possible function of GC-D cells as detectors of pathogenic bacteria, unlike a previously characterized olfactory pathway in Caenorhabditis elegans (38).
The mechanisms used for olfactory coding by the GC-D cells also differ sharply from those used by canonical OSNs or vomeronasal sensory neurons. Most notably, our results indicate that a mixture consisting of only two peptide ligands can activate virtually all GC-D neurons, and stimulation with a single peptide activates almost 75% of the cells (Fig. 5 I and J). This remarkable uniformity is consistent with a model in which GC-D itself functions as the peptide receptor. Other receptor guanylyl cyclases do function as peptide receptors: GC-A and GC-B exhibit differential sensitivities to natriuretic atriopeptins, including atrial natriuretic peptide and urodilatin, whereas GC-C responds to guanylin, uroguanylin, and the enterotoxin STp (24). However, the differential stimulus tuning of GC-D neurons suggests that at least two receptors or receptor variants are expressed across this subpopulation (one activated by uroguanylin only, the other activated by guanylin only). Because receptor guanylyl cyclases commonly function as dimers (24), GC-D could form a common heterodimeric receptor partner to other receptor guanylyl cyclases or similar proteins that are differentially expressed across the population. This possibility seems doubtful, however, because homomeric receptor guanylyl cyclases would likely also be expressed in GC-D neurons, conferring sensitivity to STp, urodilatin, or other peptides that do not stimulate these cells, or maintaining uroguanylin and guanylin sensitivity in Gucy2d –/– mice. Alternative splicing of Gucy2d might account for the differential selectivity of GC-D receptors. However, we have found no evidence for this possibility (data not shown).
Previous experiments analyzing odor-induced activity in the OB or the effects of genetic ablations suggested an involvement of the necklace glomeruli in the detection of suckling pheromones (18, 39–41). However, Gucy2d –/– mice exhibited no obvious defects in suckling effectiveness (SI Fig. 7), nor did they show any obvious mating defects. Although we cannot yet determine the exact biological function of this olfactory subsystem, the chemostimuli identified here suggest an association with the detection of information related to metabolic status, specifically to regulatory mechanisms that mediate the integration of salt and water balance. It will be interesting to see whether higher neuronal projections from the necklace glomeruli link the GC-D cells with hypothalamic nuclei involved in the control of extracellular fluid volume, Na+ balance, and osmolarity. In any case, our work identifies guanylin and uroguanylin as mammalian semiochemicals that are recognized by a unique olfactory detection system.
|
Materials and Methods |
|---|
|
TopAbstractResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
Gene Targeting. Portions of Gucy2d (19) were targeted for replacement with an IRES-Mapt-lacZ reporter (25) in mouse embryonic stem cells. Neo-positive clones undergoing homologous recombination were injected into blastocysts, and chimeric animals were obtained. The selection cassette was removed by cre-dependent recombination (42). The relevant Institutional Animal Care and Use Committees approved all procedures.
Immuno- and X-Gal Histochemistry. Frozen sections of MOE or OB from adult Gucy2d +/– and –/– mice and Cnga3 –/– mice (30) were prepared as described previously (25) and then incubated with primary antisera. Immunoreactivity was visualized with Cy2- and Cy3-labeled secondary antibodies and confocal microscopy. X-Gal histochemistry was performed as described previously (26).
Local Field Potential Recording. The submerged EOG technique was used (12, 25) to record local field potentials from the endoturbinates of adult Gucy2d +/– and –/–, Cnga2-null (26), Cnga3 –/–, and wild-type mice. Most stimuli and drugs were focally ejected by using multibarrel stimulation pipettes (33); SQ22536 and L-cis-diltiazem were applied by bath flow. The spatial distribution of guanylin- or uroguanylin-evoked potentials fell within a broad central zone of the MOE, consistent with previous GC-D in situ hybridizations (16). Epithelial hot spots were often found at the dorsal rim of each endoturbinate.
Live Cell Imaging and Single-Knob Electrophysiology.
Intact MOE from the nasal septum of adult mice was mounted en face (32). Live dendritic knobs expressing the reporter were visualized by using the fluorogenic -gal substrate fluorescein digalactoside and imaged with infrared differential interference contrast and fluorescence videomicroscopy. Action potential-driven capacitive currents (32–34) were recorded from identified knobs by using patch pipettes.
Simultaneous Imaging of -Gal and Ca2+.
The epithelial preparation was loaded with the Ca2+ indicator fluo-4/AM (33, 34). Changes in intracellular Ca2+ concentration were imaged in single knobs by using a confocal laser system (33, 34). Simultaneous visualization of -gal was achieved by coloading the cells with the fluorogenic substrate resorufin galactoside. No spillover of fluo-4 fluorescence onto the resorufin channel (or vice versa) was observed. Urine was freshly collected from adult C57BL/6 mice. We observed no differences in male or female urine on GC-D cell responses. Data are expressed as means ± SD if not otherwise stated.
|
Acknowledgements |
|---|
|
TopAbstractResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
|
Footnotes |
|---|
Abbreviations: CNG, cyclic nucleotide-gated; EHNA, erythro-9-[2-hydroxy-3-nonyl]-adenine; EOG, electroolfactogram; G, guanylin; IBMX, 3-isobutyl-1-methylxanthine; IHC, immunohistochemistry; MOE, main olfactory epithelium; OB, olfactory bulb; OSN, olfactory sensory neuron; PDE, phosphodiesterase; TH, tyrosine hydroxylase; UG, uroguanylin.
||To whom correspondence may be addressed. E-mail: frank.zufall{at}uks.eu
**To whom correspondence may be addressed at: Department of Anatomy and Neurobiology, University of Maryland School of Medicine, 20 Penn Street, S251, Baltimore, MD 21201. E-mail: smung001{at}umaryland.edu
Author contributions: T.L.-Z., D.L.G., R.R.R., F.Z., and S.D.M. designed research; T.L.-Z., R.E.C., and S.D.M. performed research; T.L.-Z., R.E.C., F.Z., and S.D.M. analyzed data; T.L-Z., S.M., M.B., D.L.G., R.R.R., and S.D.M. contributed new reagents/analytic tools; and F.Z. and S.D.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0704965104/DC1.
© 2007 by The National Academy of Sciences of the USA
|
References |
|---|
|
TopAbstractResultsDiscussionMaterials and MethodsAcknowledgementsReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg What's this?
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
F/F peak Ca2+ responses in Gucy2d +/– (F) or –/– mice (G) to stimulation with 60 mM KCl, 100 µM 8-bromo-cGMP, 50 µM EHNA, 100 µM IBMX, or 20 µM forskolin. The number of experiments is indicated above each bar. (H) Example of Ca2+ knob responses to UG (1 µM), G (1 µM), UG/G mixture (1 µM each), and urine (1:100). No responses were observed in GC-D cells from Gucy2d –/– mice. (I) Percentage of