Abstract
The highly specialized metabotropic glutamate receptor type 6 (mGluR6) is postsynaptically localized and expressed only in the dendrites of ON bipolar cells. Upon activation of mGluR6 by glutamate released from photoreceptors, a nonselective cation channel is inhibited, causing these cells to hyperpolarize. Mutations in this gene have been implicated in the development of congenital stationary night blindness type 1 (CSNB1). We investigated five known mGluR6 point mutants that lead to CSNB1 to determine the molecular mechanism of each phenotype. In agreement with other studies, four mutants demonstrated trafficking impairment. However, mGluR6 E775K (E781K in humans) suggested no trafficking or signaling deficiencies measured by our initial assays. Most importantly, our results indicate a switch in G-protein coupling, in which E775K loses Go coupling but retains coupling to Gi, which may explain the phenotype.
- mGluR, metabotropic glutamate receptor
- BBS, bungarotoxin binding site
- CSNB, congenital stationary night blindness
- SCG, superior cervical ganglion
- HEK, human embryonic kidney
- PTX, pertussis toxin
- m3, muscarinic type 3 receptor
- AM, acetoxymethyl ester
- GFP, green fluorescent protein
- i3, third intracellular loop
- GLU, glutamate.
The metabotropic glutamate receptors (mGluRs) are classified into three subfamilies based on their molecular identity, pharmacology, and G-protein coupling profile (Schoepp, 2001). Group III mGluRs (4, 6, 7, and 8) couple exclusively to Gi/o proteins and negatively regulate adenylyl cyclase (Prézeau et al., 1994). mGluR6 is a highly specialized G-protein-coupled receptor that is exclusively expressed in the postsynaptic dendritic region of retinal ON bipolar cells (Nakajima et al., 1993; Vardi et al., 2000; Quraishi et al., 2007). Upon activation, mGluR6 initiates a signaling cascade that ultimately results in the inhibition of a nonselective cation channel, leading to hyperpolarization of the ON bipolar cells (Nawy, 1999; Masland, 2001). Recent studies suggest that the molecular identity of the regulated cation channel is probably transient receptor potential channel melastatin 1 (TRPM1) (Bellone et al., 2008; Nakajima et al., 2009; Shen et al., 2009), and a few reports suggest that regulation is mediated by Gαo rather than the Gαi subunits or by Gβγ (Nawy, 1999; Dhingra et al., 2000, 2004). Our laboratory and others have demonstrated strong coupling of mGluR6 to Gαo in reconstitution experiments (Tian and Kammermeier, 2006), and antibodies directed toward Gαo were able to block the effect of glutamate on the channel in retinal slice recordings (Nawy, 1999).
Genetic defects in the GRM6 gene, which encodes mGluR6, can lead to congenital stationary night blindness (CSNB), characterized by myopia and mild to severe impairment of night vision. In humans, mGluR6 signaling abnormalities, caused by autosomal recessive mutations, lead to complete loss of night vision, CSNB1 (Dryja et al., 2005; Zeitz et al., 2005, 2007), along with the inability to adapt to light-dark cycles and reduced responsiveness to light stimuli in animal models (Takao et al., 2000; Pinto et al., 2007). CSNB2, which is characterized by less severe phenotypes, resolves from mutations occurring elsewhere in the ON visual pathway (Chang et al., 2006; Zeitz et al., 2006). Genetic studies in humans with CSNB1 and deletion of the GMR6 gene in mouse models result in the absence of the electroretinogram b-wave (Dhingra et al., 2000, 2002; Dryja et al., 2005; Zeitz et al., 2005; Pinto et al., 2007), which reflects the depolarization of ON bipolar cells after exposure of photoreceptors to light. Interestingly, the absence of the b-wave was also observed in mice lacking Gαo in bipolar cells (Dhingra et al., 2000, 2002), which correlates well with the hypothesis that mGluR6 strongly couples to Gαo to facilitate the modulation of the cation channel.
To date, there are at least five mGluR6 point mutations that lead to CSNB in humans (P46L, G58R, G150S, C522Y, and E781K). Recent work suggests that each of these mutants exhibits impaired plasma membrane expression in HEK293 cells (Zeitz et al., 2005). Based on the location of these mutations, we anticipated that these mutants would have impaired glutamate responses due to decreased affinity for glutamate or compromised G-protein coupling. Our results confirm that four of these molecules fail to express on the plasma membrane in neurons. The mGluR6 E775K (E781K in humans) expressed and functioned normally in our preliminary assays. However, calcium mobilization data suggest that E775K couples predominantly to Gαi rather than to Gαo, which may explain the CSNB phenotype in humans.
Materials and Methods
Superior Cervical Ganglion Neuron Isolation, cDNA Injection, and Molecular Methods.
Detailed descriptions of the isolation and injection procedures have been described previously (Ikeda, 1997). In brief, the superior cervical ganglion (SCG) neurons were dissected from adult rats and incubated in Earle's balanced salt solution (Invitrogen, Carlsbad, CA) containing 0.6 mg/ml trypsin (Worthington Biochemicals, Freehold, NJ) and 0.8 mg/ml collagenase D (Roche Diagnostics, Indianapolis, IN) for 1 h at 35°C. Cells were centrifuged twice, transferred to minimum essential medium (Thermo Fisher Scientific, Waltham, MA), plated, and placed in an incubator at 37°C until cDNA injection. Injection of cDNA was performed with an Eppendorf 5247 microinjector and Injectman NI2 micromanipulator (Eppendorf North America, New York, NY) 4 to 6 h after cell isolation. Plasmids were stored at −20°C as a 1 μg/μl stock solution in buffer containing 10 mM Tris and 1 mM EDTA, pH 8. mGluR6 insert was subcloned into pCDNA3.1(−) (Invitrogen). All receptor cDNAs were injected at 0.1 μg/μl. Neurons were coinjected with “enhanced” green fluorescent protein cDNA (0.005 μg/μl, pEGFPN1; Clontech, Mountain View, CA) or other fluorescent marker if necessary for identification of injected cells. Point mutants (P40L, G52R, G144S, C516Y, and E775K) were generated using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA), and sequence was verified. Bungarotoxin binding site (BBS)-mGluR6 was constructed by subcloning the bungarotoxin binding site sequence (5′-TGGAGATACTACGAGAGCTCCCTGGAGCCCTACCCTGAC-3′) between the 22nd and 23rd residues of the original mGluR6 clone using the overlapping extension PCR method. Gq chimeras were also constructed using the QuikChange strategy. After injection, cells were incubated overnight at 37°C, and experiments were performed the next day.
Electrophysiology and Data Analysis.
Patch-clamp recordings were made using 8250 glass (Garner Glass, Claremont, CA). Pipette resistances were 1 to 3 MΩ yielding uncompensated series resistances of 1 to 5 MΩ. Series resistance compensation of 70 to 80% was used in all recordings. Data were recorded using an Axopatch 200B patch-clamp amplifier from Molecular Devices (Sunnyvale, CA). Voltage protocol generation and data acquisition were performed using custom data acquisition software (generously donated by Stephen R. Ikeda, National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD) on a Macintosh G3 computer with an Instrutech ITC18 data acquisition board (HEKA Elektronik, Lambrecht/Pfalz, Germany). Currents were sampled at 0.5 to 5 kHz low pass-filtered at 3 to 5 kHz using the 4-pole Bessel filter in the patch-clamp amplifier, digitized, and stored on the computer for later analysis. All patch-clamp experiments were performed at 21–24°C (room temperature). Data analysis was performed using Igor software (WaveMetrics, Lake Oswego, OR). For calcium current recordings in SCG, the external (bath) solution contained 145 mM tetraethylammonium methanesulfonate, 10 mM HEPES, 15 mM glucose, 10 mM CaCl2, and 300 nM tetrodotoxin, pH 7.4, with osmolality of 320 mOsmol/kg. The internal (pipette) solution contained 120 mM N-methyl-d-glucamine methanesulfonate, 20 mM tetraethylammonium, 11 mM EGTA, 10 mM HEPES, 10 mM sucrose, 1 mM CaCl2, 4 mM MgATP, 0.3 mM Na2GTP, and 14 mM Tris-creatine phosphate, pH 7.2, with osmolality of 300 mOsmol/kg. Pertussis toxin (PTX) (List Biological Lab. Inc., Campbell, CA) was applied at 500 ng/ml media overnight at 37°C.
PTX-Insensitive Reconstitution Recordings.
A good stoichiometric balance between the α and βγ subunits was necessary to perform the reconstitution experiments. As described previously, the triple pulse protocol that produces the “pre” and the “post” currents is an effective way to detect an imbalance (Kammermeier et al., 2003). Excess of Gα subunits will result in a “pre” current that is slightly higher than the “post” current due to α-GDP serving as a “sponge” for Gβγ. By contrast, surplus of Gβγ will result in a “pre” current that is already inhibited (Herlitze et al., 1996; Ikeda, 1996; Ikeda and Dunlap, 1999). To overcome this obstacle, we only analyzed cells that had a post/pre ratio between 1 and 1.3. Ratios between these values are indicative a good stoichiometric G-protein subunit balance (Kammermeier et al., 2003). The α subunit plasmids were injected at 5 to 6 ng/μl and Gβ1 and Gγ2 plasmids were injected at 8 to 10 ng/μl final concentration.
Confocal Microscopy.
Injected SCG neurons (in glass coverslips) with the desired plasmid along with “enhanced” green fluorescent protein cDNA to confirm expression were incubated overnight at 37°C. Cells (16–18 h after injection) were transferred into a perfusion chamber and washed with PBS, followed by treatment with 1 μM α-bungarotoxin Alexa Fluor 647 conjugate (Invitrogen) for 1 min. Cells were washed again with PBS, and imaging was performed using an inverted Nikon microscope through a 40× oil immersion objective lens. Confocal images were obtained by using the EZ-C1 3.60 software program (Nikon, Melville, NY). Cells were excited at 488 and 633 nm, and signal was detected at 530 ± 30 and 688 ± 20 nm for GFP and Alexa Fluor 647, respectively.
Digital Imaging of Intracellular Ca2+ in Muscarinic Type 3 Stable HEK293 Cells.
Muscarinic type 3 (m3) HEK293 cells, obtained from Dr. Trevor Shuttleworth (University of Rochester, Rochester, NY) were loaded with 2 mM Ca2+-sensitive dye fura-2 AM at 37°C for 15 to 20 min. Transfected cells were loaded with fura-2 AM and then transferred into a perfusion chamber. Cells were perfused in HEPES-buffered physiological saline containing 137 mM NaCl, 0.56 mM MgCl2, 4.7 mM KCl, 1 mM Na2HPO4, 10 mM HEPES, 5.5 mM glucose, and 1.26 mM CaCl2, pH 7.4. Imaging was performed using an inverted Nikon microscope (Nikon, Tokyo, Japan) through a 40× oil immersion objective lens (numerical aperture, 1.3). Fura-2 AM loaded cells were excited alternately with light at 340 and 380 nm by using a monochromator-based illumination system (TILL Photonics, Pleasanton, CA), and the emission at 510 nm was captured by using a digital frame transfer charge-coupled device camera. Images were captured every 2 s with an exposure of 40 ms and 4 × 4 binning. Analysis was performed by TILL Vision software. Any signal below 0.1 ratio was designated as background noise. The receptors and the Gq chimeras were transfected at 1 μg/35-mm dish, β1 and γ2 plasmids were transfected at 0.5 μg/35 mm dish, and finally dSRed-nuc was transfected at 0.2 to 0.3 μg/35-mm dish (for the supplementary experiment) and cherry for the Ca2+ mobilization experiments.
Results
Modulation N-Type (CaV2.2) Calcium Channels in SCGs by mGluR6 and Five CSNB Point Mutants.
Our initial hypothesis regarding the CSNB point mutants evolved from the idea that these receptors have a reduced or diminished ability to respond to glutamate. Thus, we examined modulation of CaV2.2 channels in SCG neurons by each point mutant to examine receptor activity. The triple-pulse voltage protocol, shown in Fig. 1A (inset), was used to assess channel activity and consists of two test pulses to +10 mV from a −80 mV holding potential, separated by a strong depolarizing step to +80 mV. The +80-mV step partially reverses the well characterized, voltage-dependent, Gβγ-mediated inhibition (Elmslie et al., 1990; Ikeda, 1991, 1996).
To determine the effects of five CSNB point mutations on mGluR6 signaling, the corresponding mutants were made in the rat mGluR6 sequence, and each was expressed in SCG neurons. Figure 1A shows a typical time course of an SCG neuron expressing mGluR6, illustrating the inhibition of calcium currents during glutamate application. No effect was observed in uninjected or GFP-only injected cells (data not shown).
In Fig. 1B, the glutamate concentration-response curve for the wild-type and each point mutant is shown. Four of the mutants, P40L, G52R, G144S, and C516Y, showed no response to glutamate. By contrast, cells expressing mGluR6 E775K showed responses similar to those of the wild-type mGluR6. We noticed a slight rightward shift in the EC50 values of mGluR6 E775K concentration-response curve (EC50 for mGluR6 was 8 μM and 20 μM GLU for the mutant). However, at 100 μM glutamate applications, the responses were identical (45 ± 5 and 44 ± 2% current inhibition for mGluR6 and E775K, respectively). These data suggest that four of the mGluR6 CSNB point mutants are nonfunctional. By contrast, mGluR6 E775K seems to function similarly to the wild type.
Four CSNB Point Mutants Lack Plasma Membrane Expression.
To determine why the nonfunctional mutant receptors lacked responses, the plasma membrane expression of each was examined. Zeitz et al. (2007) reported that all of the CSNB point mutants are retained in the endoplasmic reticulum when overexpressed in HEK293 cells. To evaluate plasma membrane expression in SCG neurons, wild-type mGluR6 and each CSNB mutant were tagged with a N-terminal, extracellular BBS (Sekine-Aizawa and Huganir, 2004). The plasma membrane expression of each receptor was then examined by application of a fluorescently tagged α-bungarotoxin to SCG neurons expressing each receptor. Figure 2 illustrates representative neurons with the indicated plasmid along with phospho-enhanced green fluorescent protein. Cells expressing GFP alone showed no membrane fluorescence after treatment with α-bungarotoxin-Alexa Fluor 647 (shown in red). Furthermore, no detectable membrane labeling was evident in cells expressing P40L, G52R, G144S, or C516Y. However, cells expressing the wild-type or mGluR6 E775K mutant demonstrated comparable surface membrane labeling (Fig. 2). To verify that the bungarotoxin binding site tag did not interfere with receptor function, electrophysiological experiments were performed by monitoring calcium-channel modulation by 100 μM glutamate in cells expressing mGluR6 or BBS-mGluR6. Similar responses were observed in SCG neurons expressing each receptor (Fig. 2). Calcium currents were inhibited 43 ± 4 (n = 14) and 48 ± 5% (n = 8) in cells expressing mGluR6 and BBS-mGluR6, respectively.
E775K Does Not Activate Gαo.
Although most of our labeling data correlated with the previous report (Zeitz et al., 2007), one CSNB mutant, mGluR6 E775K, was able to traffic to the plasma membrane and function similarly to the wild-type mGluR6 in SCG neurons. Because wild-type mGluR6 seems to act in the retina through GαoA stimulation (Dhingra et al., 2002), it is possible that the E775K mutant may produce CSNB due to an inability to activate GαoA rather than lack of expression or due to trafficking defects. Because the E775K residue resides in the third intracellular loop (i3), a change in the G-protein coupling seemed possible (Francesconi and Duvoisin, 1998). Therefore, the G-protein coupling of mGluR6 E775K was examined more closely.
As with wild-type mGluR6, E775K responses were abolished by overnight PTX treatment, indicating that the mutant coupled exclusively to Gi/o proteins (see Fig. 3). Calcium currents were inhibited 45 ± 4% (n = 6) by 100 μM GLU and 6 ± 1% (n = 5) with overnight PTX treatment when mGluR6 was expressed in SCG neurons. Similar responses were obtained (42 ± 3%, n = 11, and 3 ± 1%, n = 7) when mGluR6 E775K was expressed.
Next, the possibility that the E775K mutation changes from a predominantly Go-coupled to a predominantly Gi-coupled receptor was tested. Therefore, coupling of mGluR6 E775K to specific G-proteins was examined by reconstitution using PTX-insensitive Gi/o protein mutations of the cysteine residue near the end of the C terminus (351/2 position, the site for ADP ribosylation) to glycine, rendering the G-proteins insensitive to PTX, as described previously (Bahia et al., 1998). SCG neurons expressing either mGluR6 wild-type or E775K were treated with PTX to inactivate endogenously expressed Gαi/o proteins, and calcium current modulation was examined when Gβ1 and Gγ2 were coexpressed along with each Gαi/oCG PTX-insensitive protein (Fig. 3).
Consistent with previous results (Tian and Kammermeier, 2006), reconstitution with GαoACG resulted in a strong coupling to mGluR6 wild-type (32 ± 1% inhibition, n = 7). However, no coupling was detected with E775K mutant (2 ± 1% inhibition n = 6). Surprisingly, no coupling was observed with Gi1CG, Gi2CG, or Gi3CG either (3 ± 2, 3 ± 2, and 4 ± 2% inhibition, respectively) (Fig. 4), despite a strong PTX-sensitive signal when mGluR6 E775K was expressed alone. Thus, although the glutamic acid-lysine mutation seemed to abolish signaling through GαoA, it is possible that the reconstitution assay is not sufficiently sensitive, or that the cysteine-glycine mutation may selectively disrupt coupling to the mutant receptor. Therefore, an alternative approach to investigate mGluR6 E775K signaling was necessary.
E775K Couples Predominantly to Gi Proteins.
As an alternative strategy to investigate mGluR6 E775K signaling, Ca2+ mobilization was examined in m3 HEK293 stable cell lines expressing GqGi/o chimeric proteins in which the extreme C-terminal amino acids were exchanged to allow activation by mGluR6 (Blahos et al., 1998). Gq chimeras have been widely used to study Gi/o-coupled G-protein-coupled receptor signaling (Kowal et al., 2002, 2003; Zhang et al., 2003; Walker et al., 2005).
To this end, GqGoA(9) and GqGi1/2(9) chimeras were generated (the last nine residues for Gi1 and Gi2 are identical) and used to perform Ca2+ fura-2 AM imaging experiments in m3 HEK293 stable cell lines. M3 receptors couple well to native Gq, which would serve as an internal control when 300 nM carbachol was applied. These cells, were heterologously transfected with each receptor (mGluR6 or E775K), the Gq chimera, and Gβ1γ2 constructs, along with cherry (a red fluorescent protein), to identify expressing cells.
Figure 5A shows representative Ca2+ traces illustrating Ca2+ signals when each agonist was applied. In cells transfected with the receptor alone, no signal was detected upon glutamate application (mGluR6 151 cells analyzed; E775K 149 cells). These results suggested that neither mGluR6 nor mGluR6 E775K could induce an intracellular Ca2+ signal via natively expressing G proteins. When the wild-type receptor was coexpressed with GqGoA(9) and Gβ1γ2, a strong Ca2+ response was detected in 31 of 208 cells analyzed. By contrast, only 1 of 82 cells expressing mGluR6 E775K produced a detectable response, and this response was weak (Fig. 5, A and B). These results correlated well with the electrophysiological data (Fig. 3). When GqGi1/2(9) was expressed, some signaling was detected with mGluR6 (3 of 88 cells). However, this signal was weaker and less frequent than when GqGoA(9) was expressed. Interestingly, mGluR6 E775K was able to signal in response to glutamate in 15 of 460 GqGi1/2(9)-expressing cells (Fig. 5, A and B). These data confirm that mGluR6 couples strongly to Gαo and weakly to Gαi while demonstrating that E775K mutant receptor predominantly activates Gαi. Furthermore, because this mutation causes CSNB in humans, effectively resulting in loss of function, these data support the conclusion that mGluR6 in retinal ON bipolar cells produces its primary signal via Gαo activation rather then via Gαi or Gβγ.
Discussion
We report here that a disease causing mGluR6 mutant (E775K) functions and expresses similar to the wild-type receptor in SCG neurons. Further investigation of the signaling cascade revealed that mGluR6 E775K is incapable of activating Gαo and primarily couples to Gαi proteins. Overall, the data presented here suggest that mGluR6 primarily functions through Gαo proteins in retinal ON bipolar cells and an inability of the receptor to activate Gαo results in loss of function, leading to CSNB.
Three of the CSNB mGluR6 point mutants examined here, P40L, G52R, and G144S, are located in the Venus flytrap domain near the agonist binding site (Bessis et al., 2000; Rosemond et al., 2002). The Cys516 residue is located in the cysteine-rich region of the receptor, which separates the Venus flytrap domain from the seven-transmembrane region and is believed to be important in the intramolecular signal transmission (Rondard et al., 2006). Therefore, the mechanism of loss of function for each seemed apparent. Surprisingly, each of these mutants lacked function not because of a predicted deficit in ligand binding or intramolecular signal transduction but because of a loss of plasma membrane expression. These results are consistent with those of Zeitz et al. (2007) in HEK293 cells and support the notion that mGluR6 trafficking is tightly regulated such that even small structural perturbations can severely alter receptor trafficking.
The Glu775 residue is located in the third intracellular loop of the receptor. The second (i2) and the third (i3) of mGluRs participate in G-protein contact and activation. It is interesting that the i3 loop of all mGluRs contains a conserved region, and single point mutations in this region have been known to impair receptor/G-protein coupling (Francesconi and Duvoisin, 1998).
According to the functional data presented here, E775K produced responses similar to the wild-type receptor at 100 μM GLU with only a slight shift in potency (Fig. 1). Membrane surface labeling indicated that the mutant receptor is present on the plasma membrane in SCGs after nuclear cDNA injection (Fig. 2) and in HEK293 cells when transfected (Supplemental Fig. 1). However, the observation that an mGluR6 point mutation that resulted in a nearly normal functioning receptor was puzzling.
Wild-type mGluR6 couples to Go proteins in ON bipolar cells (Nawy, 1999) and in reconstitution studies in SCG neurons (Tian and Kammermeier, 2006). Furthermore, the primary signal transduction induced by mGluR6 in retinal ON bipolar cells, the inhibition of a nonselective cation conductance, presumably transient receptor potential channel melastatin 1 (Shen et al., 2009), seems to require the activation of Gαo because disruption of neither Gαi nor Gβγ can occlude this effect (Nawy, 1999; Dhingra et al., 2002). Therefore, a change in G-protein coupling preference of mGluR6 from Gαo to Gαi could result in CSNB even if the mutant receptor is otherwise functional. This hypothesis was tested for mGluR6 E775K with two strategies: first, by reconstitution of signaling using PTX-insensitive mutant Gα proteins in SCG neurons, and second, by examining intracellular calcium signaling with fura-2 AM in HEK293 cells expressing GαqGαi/o chimeric proteins.
Unfortunately, reconstitution of signaling with the mutant receptor in SCG neurons with PTX-insensitive G proteins could not produce a complete answer. The data suggested a loss of E775K coupling to GoACG. However, when the GiCG mutants were used, no coupling was observed. These negative results may have resulted from a weak expression of the G-protein mutant in SCGs, low sensitivity of the assay (channel modulation), or the possibility that mGluR6 E775K coupled weakly to more than one Gi protein but not robustly to any one in isolation.
To this end, a more sensitive assay, Ca2+-imaging experiments with fura-2 AM, was used. We constructed GqGαi/o chimeras in which the last nine residues were swapped with that of the GoA, Gi1, and Gi2 (the last nine residues of Gi1 and Gi2 are identical, so this construct was called GqGi1/2). Similar constructs have been used to study mGluR2 and mGluR4 signaling in Chinese hamster ovary cells (Kowal et al., 2003). In accordance with our electrophysiological data, mGluR6 was able to strongly couple to GqGoA with less robust coupling to GqGi1/2. However, mGluR6 E775K failed to couple to GqGoA but elicited substantial Ca2+ signals when expressed with GqGi1/2. It is important to note that during these experiments, several plasmid constructs were required (receptor, chimera, Gβ1, Gγ2, and cherry), so the probability of getting all of the components in any one cell is low. Therefore, the low number of responding cells in each experiment was expected.
In conclusion, we demonstrate here an explanation for the CSNB phenotype (for E781K mutant in humans) as an alternative to those proposed previously. These results indicate that mGluR6 E775K functions and expresses like the wild-type receptor, but its ability to activate GαoA has been diminished. The mutant receptor retains the capability to activate Gi proteins in heterologous and reconstitution assays, but that may not be sufficient to produce effective signaling in retinal ON bipolar cells.
Acknowledgments
We thank D. I. Yule (University of Rochester) for use of equipment and reagents and Trevor Shuttleworth (University of Rochester) for m3 HEK293 cells.
Footnotes
-
↵ The online version of this article (available at http://molpharm.aspetjournals.org) contains supplemental material.
-
Article, publication date, and citation information can be found at http://molpharm.aspetjournals.org.
doi:10.1124/mol.109.058628
-
ABBREVIATIONS:
-
↵1 Current affiliation: Department of Physiology and Cellular Biophysics, Columbia University, New York, New York.
- Received June 16, 2009.
- Accepted August 6, 2009.
- © 2009 The American Society for Pharmacology and Experimental Therapeutics