Cyclic nucleotide-gated (CNG) ion channels mediate cellular responses to sensory stimuli. In vertebrate photoreceptors, CNG channels respond to the light-induced decrease in cGMP by closing an ion-conducting pore that is permeable to cations, including Ca2+ ions. Rod CNG channels are directly inhibited by Ca2+-calmodulin (Ca2+/CaM), but the physiological role of this modulation is unknown. Native rod CNG channels comprise three CNGA1 subunits and one CNGB1 subunit. The single CNGB1 subunit confers several key properties on heteromeric channels, including Ca2+/CaM-dependent modulation. The molecular basis for Ca2+/CaM inhibition of rod CNG channels has been proposed to involve the binding of Ca2+/CaM to a site in the NH2-terminal region of the CNGB1 subunit, which disrupts an interaction between the NH2-terminal region of CNGB1 and the COOH-terminal region of CNGA1. Here, we test this mechanism for Ca2+/CaM-dependent inhibition of CNGA1/CNGB1 channels by simultaneously monitoring protein interactions with fluorescence spectroscopy and channel function with patch-clamp recording. Our results show that Ca2+/CaM binds directly to CNG channels, and that binding is the rate-limiting step for channel inhibition. Further, we show that the NH2- and COOH-terminal regions of CNGB1 and CNGA1 subunits, respectively, are in close proximity, and that Ca2+/CaM binding causes a relative rearrangement or separation of these regions. This motion occurs with the same time course as channel inhibition, consistent with the notion that rearrangement of the NH2- and COOH-terminal regions underlies Ca2+/CaM-dependent inhibition.
Introduction
Cyclic nucleotide-gated (CNG) ion channels mediate the transduction of sensory stimuli into electrical signals (Kaupp and Seifert, 2002; Matulef and Zagotta, 2003). In rod and cone photoreceptors, CNG channels close in response to the decrease in concentration of 5′:3′ cyclic guanosine monophosphate (cGMP) that occurs with light absorption (Yau and Baylor, 1989; Burns and Baylor, 2001). In olfactory sensory neurons, CNG channels open in response to the increase in concentration of 5′:3′ cyclic adenosine monophosphate (cAMP) that occurs with odorant binding (Menini, 1999; Zufall and Munger, 2001). CNG channels are also present in other tissues, including the brain, taste receptors, aorta, liver, and kidney, but their roles in these tissues are not as well understood (Richards and Gordon, 2000).
CNG channels are part of a subfamily of ion channels that are regulated by the direct binding of cyclic nucleotides, including the hyperpolarization-activated cyclic nucleotide-modulated (HCN) channels and the ether-a-go-go (eag) K+ channels (Guy et al., 1991; Robinson and Siegelbaum, 2003). CNG channels are nonselective for cations, but their ion-conducting pore is similar to that of bacterial K+-selective channels whose three-dimensional structure is known (Doyle et al., 1998; Zhou et al., 2001; Jiang et al., 2002; Flynn and Zagotta, 2003). Like their voltage-activated counterparts, CNG channels have six membrane-spanning domains and internal NH2- and COOH-terminal regions; however, they are only weakly activated by voltage (Kaupp et al., 1989; Jan and Jan, 1990). Instead, CNG channels are gated (or opened) by the direct binding of cGMP (Fesenko et al., 1985) or cAMP (Nakamura and Gold, 1987) to a COOH-terminal cyclic nucleotide-binding domain (CNBD). The structure of the CNBD and adjoining C-linker region has recently been solved for a related hyperpolarization-activated cyclic nucleotide-modulated (HCN) channel (Zagotta et al., 2003).
Native rod CNG channels comprise CNGA1 subunits and CNGB1 subunits. These channel subunits are 35% identical on the amino acid level but functionally quite different. CNGA1 subunits form homomeric CNG channels in heterologous systems with some of the properties of native rod channels (Kaupp et al., 1989). CNGB1 subunits do not form functional homomeric channels; however, coexpression of CNGA1 and CNGB1 subunits results in heteromeric channels that recapitulate many of the biophysical and pharmacological properties of native rod channels (Chen et al., 1993, 1994; Korschen et al., 1995). The CNGB1 subunit confers several new properties on the heteromeric channels, including higher fractional activation by cAMP (a partial agonist of the channels), increased permeability of Ca2+ relative to Na+, native-like single-channel behavior, sensitivity to the blocker l-cis-diltiazem, and sensitivity to inhibition by Ca2+-calmodulin (Ca2+/CaM). Unexpectedly, these properties are produced by only a single CNGB1 subunit, as rod channels contain three CNGA1 subunits and only one CNGB1 subunit (Weitz et al., 2002; Zheng et al., 2002; Zhong et al., 2002; Zimmerman, 2002).
Rod channels, cone channels, and olfactory channels are all inhibited by extrinsic Ca2+/CaM (Hsu and Molday, 1993; Chen and Yau, 1994; Hsu and Molday, 1994; Gordon et al., 1995; Korschen et al., 1995; Bauer, 1996; Hackos and Korenbrot, 1997; Peng et al., 2003). Ca2+/CaM-dependent inhibition of native olfactory channels is a major determinant of olfactory adaptation (Kurahashi and Menini, 1997), and in cone CNG channels, Ca2+/CaM (along with an unknown factor) may form a feedback loop that contributes to light adaptation (Hackos and Korenbrot, 1997; Rebrik and Korenbrot, 1998). In rod CNG channels, the role for Ca2+/CaM-dependent inhibition is unknown. In rod CNG channels, a site was identified in the NH2-terminal region of the CNGB1 subunit as important for direct Ca2+/CaM binding and functional modification (Grunwald et al., 1998; Weitz et al., 1998; Trudeau and Zagotta, 2002b). Fusion proteins containing this site bound Ca2+/CaM, and deletion of this site rendered channels insensitive to modulation by Ca2+/CaM. In addition, the NH2-terminal region of CNGB1 subunits and the distal COOH-terminal region of CNGA1 subunits formed a specific interaction in a biochemical interaction assay (Trudeau and Zagotta, 2002a). This interaction required an intact Ca2+/CaM-binding site and was disrupted in the presence of Ca2+/CaM (Trudeau and Zagotta, 2002b). Based on these results, it has been proposed that Ca2+/CaM modulation of rod channels involves Ca2+/CaM binding to a site on the NH2-terminal region, which disrupts an interaction between the NH2- and COOH-terminal regions, and causes channel inhibition. A similar but distinct mechanism has been proposed for Ca2+/CaM modulation of homomeric olfactory channels (Varnum and Zagotta, 1997; Trudeau and Zagotta, 2003; Zheng et al., 2003).
In this paper, we have sought to test this mechanism for Ca2+/CaM inhibition using fluorescence methods that offer distinct advantages over previous biochemical and electrophysiological approaches. We find that Ca2+/CaM binding to CNG channels is the rate-limiting step for inhibition. We show that Ca2+/CaM binds in close proximity to the putative Ca2+/CaM-binding site in the NH2-terminal region of CNGB1. Finally, we show that the NH2- and COOH-terminal regions of the channel subunits are in close proximity in intact channels, and that these regions undergo a rearrangement with the same time course as Ca2+/CaM-dependent inhibition. These results provide further support for the proposal that Ca2+/CaM inhibition involves a Ca2+/CaM-dependent disruption of an interaction between the NH2- and COOH-terminal regions.
Materials And Methods
Molecular Biology and Oocyte Expression
The CNGA1 clone used here was previously described (Gordon and Zagotta, 1995; Trudeau and Zagotta, 2002a) and was identical to the original clone (Kaupp et al., 1989), except for a K2E change and an eight–amino acid FLAG epitope on the COOH-terminal end. The CNGB1 clone was a gift from R. Molday (University of British Columbia, Vancouver, BC, Canada) (Korschen et al., 1995) and contained an I2V change that did not affect any of the properties studied here. The enhanced cyan fluorescent protein (eCFP) and enhanced yellow fluorescent protein (eYFP) clones were a gift from R.Y. Tsien (University of California, San Diego, CA) (Miyawaki et al., 1997). Fluorescent proteins were genetically attached to CNGA1 and CNGB1 subunits using an overlapping PCR strategy and confirmed with DNA sequencing. In CNGA1-eYFP, eYFP was fused immediately after the FLAG epitope tag at the COOH terminus. In eCFP-CNGB1, eCFP was fused before amino acid 677 of CNGB1 and was immediately upstream of the CaM-binding site (Trudeau and Zagotta, 2002b). eCFPΔCaM-CNGB1 had a deletion mutation of the Ca2+/CaM site (which eliminated an additional 25 amino acids), and the eCFP was fused before amino acid 702 of the CNGB1 subunit. Channel cDNAs were subcloned into the pGEMHE vector (a gift of E. Liman, University of Southern California, Los Angeles, CA) for heterologous expression. RNA was transcribed with the mESSAGE mACHINE kit (Ambion) and injected with a micropipette into Xenopus oocytes. Oocytes were prepared as described elsewhere (Gordon et al., 1995) and incubated with shaking for 3–10 d at 16°C.
Patch-clamp Electrophysiology and Fluorescence Imaging
Ionic currents through CNG channels expressed in oocytes were recorded in the excised, inside-out patch-clamp configuration (Hamill et al., 1981) with an Axopatch 200B patch-clamp amplifier (Axon Instruments, Inc.). Data were digitized with an ITC-16 (Instrutech) and recorded and analyzed with the Pulse software package (Instrutech) and Igor software running on a Pentium III computer. The patch pipette (external) solution contained 130 mM NaCl, 0.2 mM EDTA, 3 mM HEPES, pH 7.2 (with 500 μM niflumic acid to block endogenous Cl− channels). The Ca2+-free bath (internal) solution contained 130 mM NaCl, 0.2 mM EDTA, 3 mM HEPES, pH 7.2, and 50 μM cGMP (Sigma-Aldrich) to activate CNG channels. In solutions with internal Ca2+ ions, (Ca2+-only and Ca2+/CaM), 2 mM NTA replaced EDTA and 50 μM total Ca2+ was added to achieve a free Ca2+ concentration of 1 μM, as determined with WinMaxC (Bers et al., 1994). CaM (Calbiochem) or CaM conjugated to the fluorescent dye Alexa-488 (CaM-488) (Molecular Probes) was added to Ca2+-containing solutions at a concentration of 250 nM. Internal solutions were applied to the cytoplasmic face of a membrane patch with an RSC-200 solution changer (Molecular Kinetics).
For patch-clamp fluorometry (PCF) experiments, fluorescent signals were recorded by imaging the patch pipette tip with a cooled CCD camera (Princeton Instruments) while the ionic current was simultaneously recorded with a patch-clamp. Fluorescence was observed with a 40× oil-immersion objective (NA 1.3) on a Nikon Diaphot inverted microscope. Fluorophores were excited at the appropriate wavelength using a monochrometer (Cairn) with a xenon lamp light source, and the appropriate excitation filter and dichroic mirror configuration (for eCFP, exciter: 440 ± 10 nm, dichroic: 455 nm; for eYFP or CaM-488, exciter: 470 ± 20 nm, dichroic: 510 nm; Chroma Technology Corp.). Emission spectra were recorded with 10-nm bandpass emission filters (Chroma Technology Corp.) set into a coupled pair of filter wheels (Sutter Instrument Co.). Fluorescence data were acquired and analyzed with the MetaMorph software package (Universal Imaging Corp.).
After a membrane patch was excised, ionic currents were recorded every 10 s with a voltage pulse from −60 to 60 mV (from a holding voltage of 0 mV) in a subsaturating (50 μM) concentration of cGMP. Often there was a characteristic increase in current associated with dephosphorylation after patch excision (Gordon et al., 1992; Molokanova et al., 1997). Experiments were conducted after the current reached a steady level. Ca2+/CaM or Ca2+/CaM-488 was then perfused for a given amount of time while the current was recorded at 10-s intervals. Then CaM (or CaM-488) was removed and replaced with Ca2+-only solution, containing 1 μM Ca2+, and the current was recorded. In the Ca2+-only solution, the previous inhibition by Ca2+/CaM was maintained, and the currents were stable. An emission spectra of nine wavelengths was then determined while the membrane was held at 0 mV. In this way, the ionic current and fluorescent signals were recorded after the same cumulative time in Ca2+/CaM (or Ca2+/CaM-488). This method also minimized solution artifacts, as the spectra were always determined in the presence of the same internal solution (containing 1 μM Ca2+). The time course of channel inhibition by Ca2+/CaM was determined using the cumulative time the patch was exposed to the modifier. To washout Ca2+/CaM (or Ca2+/CaM-488), patches were exposed to Ca2+-free solution (containing 0.2 mM EDTA) for the indicated amount of time, during which time the Ca2+/CaM-dependent inhibition was alleviated. As above, solutions were switched from Ca2+-free solution to Ca2+-only solution for current and fluorescence measurements. The cumulative time refers to the time spent in the Ca2+-free solution.
Over the course of an experiment, the eCFP fluorescence decreased by ∼5–10% due to bleaching. The bleaching time course was measured from patches with eCFP-only containing channels subunits and corrected in the time courses of the eCFP fluorescence in Fig. 3 (C and F).
Confocal Microscopy
Emission spectra from whole oocytes expressing fluorescently labeled channel subunits were collected using a confocal microscope (Leica TCS) with laser excitation. The microscope objective was 5× with 0.15 NA. The measurement of spectra was enabled by acousto-optical tunable filters with an emission window of 5 nm. Fluorescence resonance energy transfer (FRET) was measured from the stimulated emission of the acceptor fluorophore (eYFP) with excitation of the donor fluorophore (eCFP). We took a ratiometric approach to solve for the spectral component due to FRET (Selvin, 1995). The experimental spectra with excitation at 458 nm (Fig. 4 A, red) had two components, one at 480 nm due to eCFP and one at 530 nm due to eYFP. The eCFP component of the experimental trace was removed by subtracting a scaled eCFP spectra (Fig. 4 A, cyan) from a control experiment with eCFP-CNGB1/CNGA1 channels. This yielded the extracted spectra (Fig. 4 A, green) due to eYFP emission (F458), which contained two components: the emission due to FRET (F458FRET) and the emission due to the small but finite excitation of eYFP directly by the 458 laser (F458direct). F458FRET was obtained by dividing the extracted spectra F458 by the YFP emission spectra with direct excitation at 488 nm (F488; Fig. 4 A, black trace) (Selvin, 1995). This ratio, Ratio A, contained two components.
The first component (Ratio A0) was determined with an additional control experiment in which eYFP spectra were measured from CNGA1 channels with only eYFP attached (Fig. 4 B) with excitation at 458 nm (F458; Fig. 4 B, red trace) or 488 nm (F488; Fig. 4 B, black trace), which yielded Ratio A0.
The FRET component was calculated by subtracting Ratio A0 from Ratio A, which yields a value that is proportional to FRET efficiency.
Results
Ca2+/CaM Binding Determines the Rate of CNG Channel Inhibition
To study the mechanism of Ca2+/CaM modulation of rod CNG channels, we used the PCF technique in which fluorescence intensity and ionic current are recorded simultaneously from an inside-out patch of membrane containing fluorescently labeled channels (Zheng and Zagotta, 2000, 2003). PCF provides time-resolved measurements of protein rearrangements in intact functioning channels in the plasma membrane. To directly measure Ca2+/CaM binding we used a CaM molecule that was conjugated to the fluorescent dye Alexa-488 (CaM-488). Fig. 1 A shows that Ca2+/CaM-488 inhibited heteromeric CNGA1/CNGB1 channels expressed in Xenopus oocytes. CNG currents were recorded in the inside-out configuration of the patch-clamp technique with voltage steps from −60 to 60 mV in the presence of a subsaturating concentration of cGMP (50 μM) and were leak subtracted (Fig. 1 A). The patch was then exposed to Ca2+/CaM-488 for 3 min followed by a wash in a Ca2+-only solution to remove unbound CaM-488 (but not bound CaM-488). The current was stable in the Ca2+-only solution and was inhibited about twofold (Fig. 1 A, center). After washout of Ca2+/CaM-488 in Ca2+-free solution, the current recovered to its initial value (Fig. 1 A, right).
To directly measure the binding of Ca2+/CaM-488 we monitored the fluorescence intensity in the membrane patch (Fig. 1 B) while we simultaneously measured the ionic current. The fluorescence intensity at 530 nm was measured in response to excitation at 470 nm. Before addition of Ca2+/CaM-488 there was no detectable fluorescent signal coming from the pipette or patch membrane (Fig. 1 B, left). After addition of Ca2+/CaM-488, followed by washing in a Ca2+-only solution to remove the unbound CaM-488, we observed a large increase in the fluorescence at the pipette tip. No significant fluorescence was detected in patches from uninjected oocytes (not depicted), suggesting that Ca2+/CaM-488 was associating directly with the channels in the membrane patch (Fig. 1 B, center). Upon washout of the Ca2+/CaM-488 in Ca2+-free solution and recovery of the current, the fluorescence of the patch was likewise eliminated (Fig. 1 B, right), indicating that Ca2+/CaM-488 has dissociated from the channels.
To determine if the Ca2+/CaM-488 bound to the membrane was indeed the same Ca2+/CaM-488 that was modulating the channels, we compared the time courses of the changes in ionic current and fluorescence intensity recorded simultaneously (Fig. 2). The time course of the current decrease by Ca2+/CaM-488 is similar to that seen previously for Ca2+/CaM (Grunwald et al., 1998; Weitz et al., 1998; Trudeau and Zagotta, 2002b) (Fig. 2 B). The time course of current inhibition and recovery (Fig. 2 B) was mirrored precisely by the time course of fluorescence intensity increase and decrease (Fig. 2 C). Single exponential fits to these data revealed no significant difference between the time constant of current inhibition (τ = 14.7 ± 2.0 s, n = 3) and the time constant for fluorescence increase (τ = 18.5 ± 2.3 s, n = 3). Similarly there was no significant difference between the time constant of current recovery (τ = 68 ± 19 s, n = 3) and the time constant for fluorescence decrease (τ = 51 ± 20 s, n = 3). This similarity between the ionic current and fluorescence intensity time courses suggests that the Ca2+/CaM-488 bound to the membrane was indeed the same Ca2+/CaM-488 that was modulating the channels. The single exponential time course of the fluorescence change is consistent with a single Ca2+/CaM-488 binding to the intact channel, as expected from the 3:1 CNGA1:CNGB1 stoichiometry of the channel. Finally, the similarity between the current and fluorescence time courses indicates that the binding of Ca2+/CaM is the rate-limiting step for channel inhibition.
Ca2+/CaM Binds in Close Proximity to a Putative Ca2+/CaM-binding Site in the CNGB1 Subunit
The previous experiment suggests that Ca2+/CaM binds to a constituent of the membrane patch and is associated with inhibition of CNG channels. Here, using FRET, we show that Ca2+/CaM binds directly to intact CNG channels. FRET is the transfer of energy from an excited donor fluorophore to an acceptor fluorophore when the molecules are within a permissive distance (<80 Å) (see Lakowicz, 1999). Ca2+/CaM-488 has spectral properties similar to eYFP, and thus can be used as the acceptor in a FRET pair, with eCFP as the donor, to monitor proximity (Heim and Tsien, 1996). We genetically attached eCFP to the NH2-terminal region of the CNGB1 channel subunit at a site immediately adjacent to the putative Ca2+/CaM-binding site (eCFP-CNGB1) (Fig. 3 A). We then monitored FRET between eCFP on CNGB1 and Ca2+/CaM-488 by measuring the quenching of eCFP fluorescence (which was excited at 440 nm) by the nearby Alexa-488 dye.
Using PCF, we simultaneously measured the current and eCFP fluorescence intensity during the application and removal of Ca2+/CaM-488 to inside-out patches containing CNGA1/eCFP-CNGB1 channels. As before, the indicated solution was applied for various amounts of time, followed by a wash in Ca2+-only solution, at which point the current and fluorescence were measured. We found that the eCFP (donor) emission decreased (quenched) in the presence of Ca2+/CaM-488 (acceptor) (Fig. 3 C). Furthermore, the time course of eCFP quenching (τ = 24.6 ± 6.8 s, n = 4) precisely matched the time course of ionic current decay (τ = 23.1 ± 9.2 s, n = 4) (Fig. 3, B and C). During the washout of CaM-488 in Ca2+-free solution, current recovery again had a time course (τ = 61 ± 6 s, n = 4) similar to the time course of recovery of eCFP fluorescence (dequenching) (τ = 129 ± 20 s, n = 4). We note that while similar, the eCFP recovery was about twofold slower than the current recovery. The reason for this small difference is not clear, given the matching recovery time courses of current and fluorescence in Figs. 2 and 6. The occurrence of FRET in these experiments (Fig. 3 C) indicates that Ca2+/CaM binds within close proximity (<80 Å) to the Ca2+/CaM-binding site in the NH2-terminal region of CNGB1, suggesting a direct association between Ca2+/CaM and CNGB1 subunits in intact channels.
As a control experiment, we deleted the putative Ca2+/CaM-binding region from the NH2-terminal region of CNGB1 (eCFP-CNGB1ΔCaM) and coexpressed these subunits with CNGA1 subunits (Fig. 3 D). The resultant heteromeric channels were insensitive to Ca2+/CaM (Fig. 3 E). Likewise, the eCFP fluorophore was not quenched when Ca2+/CaM-488 was introduced (Fig. 3 F). These results confirm that this region is necessary for binding of Ca2+/CaM to CNGB1 and inhibition of CNGA1/CNGB1 channels by Ca2+/CaM.
NH2- and COOH-terminal Regions Are in Close Proximity
What are the molecular regions of CNG channels that underlie channel inhibition by Ca2+/CaM? Previously, using polypeptide fragments from the channels, we showed that a physical interaction existed between the NH2- and COOH-terminal regions of CNGB1 and CNGA1 subunits, respectively (Trudeau and Zagotta, 2002a,b). We sought to study this interaction in intact CNG channels with FRET spectroscopy. Fluorescent proteins were fused next to the cytoplasmic regions of the CNG channel subunits that were previously shown to interact. Specifically, eCFP was attached proximal to the NH2-terminal Ca2+/CaM-binding region of CNGB1 and eYFP was attached to the distal COOH-terminal region of CNGA1 (Fig. 4).
First, we measured FRET from these CNGA1-eYFP/eCFP-CNGB1 channels in whole oocytes by measuring the stimulated emission of eYFP due to excitation of a nearby eCFP. The emission spectra were measured just from the surface membrane using a confocal microscope. Excitation of eCFP at 458 nm produced a complex emission spectrum with a small peak at ∼480 nm, corresponding to eCFP, and a larger peak at ∼530 nm, corresponding to eYFP (Fig. 4 A, red trace). To determine the spectral component due to just eYFP, we first subtracted off a scaled spectra from an oocyte expressing only eCFP-fused channel subunits (Fig. 4 A, dashed cyan trace). This yielded just the eYFP component of the spectra (F458; Fig. 4 A, green trace). To determine the eYFP component due to FRET, we took a ratiometric approach (materials and methods; Selvin, 1995) in which F458 is normalized by the eYFP emission after direct excitation of eYFP at 488 nm (F488; Fig. 4 A, black trace). The F458/F488 ratio (Ratio A) contained a FRET component as well as a component due to direct excitation of eYFP by the 458-nm laser. The direct component was determined by subtracting the F458/F488 ratio (Ratio A0) in a control experiment in which CNG channels contained only eYFP (Fig. 4 B). The quantity Ratio A − Ratio A0 is directly proportional to FRET.
For CNGA1-eYFP/eCFP-CNGB1 channels, the Ratio A − Ratio A0 was significantly greater than zero, indicating FRET (Fig. 5). This suggests that the NH2- and COOH-terminal regions of CNGB1 and CNGA1, respectively, are within close proximity in intact, heteromeric CNG channels. We have also previously shown that the Ca2+/CaM-binding site is necessary for the biochemical interaction with COOH-terminal fragments of CNGA1 subunits (Trudeau and Zagotta, 2002b). Deletion of the Ca2+/CaM region from CNGB1 (eCFPΔCaM-CNGB1) caused a significant (P < 0.05, Student's t test) reduction in FRET of 29% (Fig. 5), consistent with the NH2- and COOH-terminal regions being further apart or reoriented in the absence of the Ca2+/CaM-binding domain. While reduced, some FRET still remained in channels with the eCFPΔCaM-CNGB1 deletion, presumably because the fluorophores were still within a permissive distance for FRET. It is possible that deletion of the CaM-binding site in eCFPΔCaM-CNGB1 simply shortened the NH2-terminal region and increased the distance between the CNGB1 NH2-terminal region and the CNGA1 COOH-terminal region (and the attached fluorophores), reducing FRET. However, our results with Ca2+/CaM (shown below) are more consistent with a mechanism where deletion of the Ca2+/CaM-binding site disrupted an interaction between the CNGB1 NH2-terminal region and the CNGA1 COOH-terminal region.
Molecular Rearrangements Between NH2- and COOH-terminal Regions Underlie Ca2+/CaM-dependent Inhibition
The NH2-terminal region of CNGB1 and the COOH-terminal region of CNGA1 are in close proximity, and we wished to determine if these regions undergo molecular rearrangements during modulation by Ca2+/CaM. To test this, we expressed CNGA1-eYFP/eCFP-CNGB1 channels (Fig. 6 A) and then introduced Ca2+/CaM to the internal face of membrane patches while simultaneously measuring the ionic current and fluorescence spectra using PCF. In the presence of Ca2+/CaM, the channels were inhibited to the same extent and with the same time course (Fig. 6 B) as channels that do not contain fluorescent proteins (Fig. 2 B), indicating that the fluorophores did not alter the Ca2+/CaM-dependent inhibition of the channels.
We measured emission spectra from these fluorescent channels with 440-nm excitation during the application and washout of Ca2+/CaM and plotted the ratio of the eYFP emission (F530) to the eCFP emission (F480). The F530/F480 ratio is related to FRET efficiency, increasing as FRET increases and decreasing as FRET decreases. The initial F530/F480 ratio (1.8 ± 0.15, n = 4) decreased, with cumulative time in Ca2+/CaM, to a lower F530/F480 ratio (1.54 ± 0.09, n = 4), indicating a decrease in FRET (Fig. 6 C). This suggests that the NH2- and COOH-terminal regions rearrange and perhaps separate in the presence of Ca2+/CaM. The time course of the current decay (τ = 22.6 ± 7.1 s, n = 5) and recovery (τ = 35.2 ± 10.3 s, n = 5) was also closely matched by the time course of F530/F480 decrease (τ = 40.7 ± 12.1 s, n = 5) and increase (τ = 32.8 ± 6.1 s, n = 5). The similar time courses suggest that a molecular rearrangement between the NH2- and COOH-terminal regions is involved in the Ca2+/CaM inhibition of CNGA1/CNGB1 channels.
In a control experiment in which the Ca2+/CaM-binding region was deleted from the CNGB1 subunit (Fig. 6 D), we did not detect inhibition of channels by Ca2+/CaM (Fig. 6 E). We also did not detect a change in FRET in these channels (Fig. 6 F), as the initial F530/F480 ratio (1.39 ± 0.08, n = 3) was the same as the F530/F480 ratio after a 4-min cumulative exposure to Ca2+/CaM (1.36 ± 0.05, n = 3). This result is consistent with the notion that the Ca2+/CaM-binding region is necessary for the rearrangement between the NH2- and COOH-terminal regions of CNGB1 and CNGA1. The F530/F480 ratio in intact channels in the presence of Ca2+/CaM (1.54 ± 0.09, n = 4) (Fig. 6 C) was similar to that in channels with the Ca2+/CaM site deleted (1.39 ± 0.08, n = 3) (Fig. 6 F), indicating that Ca2+/CaM caused a rearrangement in the channels that was equivalent to removal of the Ca2+/CaM-binding site.
Discussion
Using fluorescence spectroscopy, we have made a number of novel findings regarding inhibition of rod CNG channels by Ca2+/CaM. A major advantage in using fluorescence microscopy over previous biochemical investigations is that we could examine intact channels that are membrane embedded and therefore in an environment approximating their native one. Here we have shown that (a) Ca2+/CaM binding is the rate-limiting step for channel inhibition, (b) Ca2+/CaM binds in close proximity to the NH2-terminal Ca2+/CaM-binding site of the CNGB1 subunit in intact, functioning heteromeric channels, (c) the NH2-terminal region of the CNGB1 subunit is in close proximity to the COOH-terminal region of CNGA1 subunits, and (d) the NH2-terminal region of the CNGB1 subunit and the distal COOH-terminal region of the CNGA1 subunits undergo a relative rearrangement or separation as a consequence of Ca2+/CaM binding, and that this movement is coupled to channel inhibition by Ca2+/CaM.
The proposed mechanism for Ca2+/CaM-dependent inhibition of rod CNG channels is presented in Fig. 7. It postulates that rod CNG channels have an intersubunit interaction between the NH2-terminal region (including a Ca2+/CaM-binding site) of CNGB1 subunits and the distal COOH-terminal region of CNGA1 subunits (Fig. 7 A). Ca2+/CaM binds directly to the CNGB1 subunit, and this is the rate-limiting step in channel inhibition. Upon binding, Ca2+/CaM causes a rearrangement or separation of the NH2- and COOH-terminal regions of CNGB1 and CNGA1 subunits, respectively (Fig. 7 B). We propose that this rearrangement underlies channel inhibition. Since only a single CNGB1 subunit is present in the CNG channel tetramer, a single Ca2+/CaM molecule appears to be sufficient for channel inhibition.
FRET is an excellent technique for the measurement of the dynamics of rearrangements within a protein or between proteins. What, if anything, can we say about absolute distances? FRET efficiency, which is steeply dependent on distance, is also dependent on κ2, a value that depends on the relative orientation of fluorophores (Lakowicz, 1999). If the fluorophores are highly mobile, κ2 can be estimated based on the average of the possible orientations of the fluorophores, and the resulting distance calculated with reasonable accuracy (dos Remedios and Moens, 1995). However, restricted mobility of the fluorophores renders absolute distance calculations meaningless. Absolute efficiency also requires stoichiometric numbers of fluorophores (Lakowicz, 1999). In our experiments, we either have one eCFP to three eYFP's or, for CaM-488, two Alexa-488 fluorophores to one eCFP (the molar ratio of Alexa-488 to CaM is usually 2:1). Thus, we cannot make estimates regarding absolute distances from our results.
Like rod channels, olfactory CNG channels are also inhibited by Ca2+/CaM (Chen and Yau, 1994; Liu et al., 1994; Bradley et al., 2001; Munger et al., 2001). Ca2+/CaM-dependent inhibition of olfactory channels decreases the apparent affinity for cAMP by ∼10-fold, which is much larger than the decrease in apparent cGMP affinity in rod channels, which is about twofold (Hsu and Molday, 1993; Chen and Yau, 1994; Liu et al., 1994; Gordon et al., 1995; Grunwald et al., 1998; Weitz et al., 1998). The mechanism of inhibition in homomeric olfactory channels involves the disruption of an interaction between the NH2-terminal region and COOH-terminal region of CNGA2 subunits (Varnum and Zagotta, 1997; Zheng et al., 2003). This interaction occurs between subunits, rather than within a subunit, and the NH2- and COOH-terminal regions undergo a relative rearrangement with respect to one another (Zheng and Zagotta, 2003). In a broad sense, the mechanisms of inhibition in olfactory and rod channels are similar, but in detail they are quite different. The Ca2+/CaM-binding site in CNGA2 subunits is a 1-8-14 type of binding motif, whereas the binding site in the NH2-terminal region of CNGB1 subunits is unconventional, but has some similarities to IQ motifs. The NH2-terminal region of CNGA2 channels has an autoexcitatory effect on channel opening, and deletion of the Ca2+/CaM-binding site in this region reduces the apparent affinity of the channels for cAMP by 10-fold (Liu et al., 1994; Varnum and Zagotta, 1997). In contrast, the NH2-terminal Ca2+/CaM-binding site in CNGB1 subunits does not appear to have an excitatory effect on channel opening, as deletion of this region does not change the apparent affinity of channels for cyclic nucleotides (Trudeau and Zagotta, 2002b). In both rod and homomeric olfactory channels there is an interaction between the NH2-terminal region of the channels and the COOH-terminal region, although the specific regions that interact are not related. The autoexcitatory effect of the NH2-terminal region of CNGA2 subunits is mediated through an interaction with the COOH-terminal CNBD and the C-linker region of olfactory subunits (Varnum and Zagotta, 1997). In rod channels, the NH2-terminal region of CNGB1 subunits interacts with a region distal to the CNBD in CNGA1 subunits (Trudeau and Zagotta, 2002a). In both rod and olfactory channels, Ca2+/CaM disrupted NH2- and COOH-terminal interactions, leading to inhibition. In homomeric olfactory channels, Ca2+/CaM disrupted the autoexcitatory effect of the NH2-terminal region, which can completely account for the change in the apparent affinity for cAMP. In contrast, in rod channels, Ca2+/CaM disrupted an interaction that is not autoexcitatory, but rather, Ca2+/CaM may directly inhibit the channels.
The native olfactory CNG channels are made up of three different CNG channel subunits, CNGA2, CNGA4, and CNGB1b, an alternatively spliced form of CNGB1 (Dhallan et al., 1990; Bradley et al., 1994; Liman and Buck, 1994; Sautter et al., 1998; Bonigk et al., 1999) with a stoichiometry of 2:1:1 (Zheng and Zagotta, 2004). All three subunit types are necessary to reconstitute the fast Ca2+/CaM-dependent inhibition seen in native olfactory channels (Sautter et al., 1998; Bonigk et al., 1999; Bradley et al., 2001; Munger et al., 2001). Recently it has been shown that the rapid Ca2+/CaM modulation of these heteromeric olfactory channels involves the Ca2+/CaM-binding site in the NH2-terminal region of CNGB1b, identical to the site on CNGB1, and perhaps another site on the CNGA4 subunit (Bradley et al., 2004). Therefore the mechanism for the modulation in heteromeric olfactory channels may have many similarities to the mechanism described here.
Cone channels are also inhibited by Ca2+/CaM. Like rod channels, Ca2+/CaM reduces the apparent affinity of cone channels for cGMP by about twofold (Hackos and Korenbrot, 1997; Rebrik and Korenbrot, 1998; Peng et al., 2003). CNGA3 and CNGB3 subunits are required to recapitulate the properties of native cone channels (Weyand et al., 1994; Gerstner et al., 2000), and both CNGA3 and CNGB3 subunits are required for functional modification by Ca2+/CaM (Peng et al., 2003). The mechanism of Ca2+/CaM-dependent inhibition in cone channels involves two Ca2+/CaM-binding sites in CNGB3 subunits, one on the NH2-terminal region and one on the COOH-terminal region. Deletion of both sites was necessary to remove Ca2+/CaM inhibition (Peng et al., 2003). Thus, the mechanism of cone channel inhibition appears quite distinct from that in rod or olfactory channels.
A role for Ca2+/CaM-dependent modulation of CNG channels in photoreceptor function is not firmly established. Ca2+ ions, CaM, and CNG channels have been theorized to form a feedback system that may play a role in range extension and adaptation in photoreceptors (Hsu and Molday, 1993). This view, however, has been challenged on various grounds. The effect of Ca2+/CaM on CNG channels is relatively modest, compared with the large range over which adaptation occurs (Koutalos et al., 1995; Nakatani et al., 1995). The Ca2+ levels in native rods falls from 500 nM in the dark to 50 nM in bright light (Gray-Keller and Detwiler, 1994; Sagoo and Lagnado, 1996), whereas the half maximal concentration for Ca2+/CaM association with channels is 50 nM Ca2+ (Hsu and Molday, 1993). This means that at low Ca2+ levels in rods, Ca2+/CaM may only partially dissociate from CNG channels, resulting in a smaller dynamic range of inhibition in native channels than that seen in heterologous systems. Nevertheless, a physiological role for Ca2+/CaM-dependent inhibition of rod channels cannot be ruled out. In rods, an unidentified endogenous factor, which appears distinct from CaM, inhibits CNG channels in a Ca2+-dependent manner (Gordon et al., 1995). In native cone channels there is also evidence for an inhibitory cytoplasmic factor that dissociates from channels in zero Ca2+ (Rebrik and Korenbrot, 1998). These unidentified factors could be the physiological inhibitors or regulators of native rod and cone CNG channels. In native olfactory channels, however, the role of Ca2+/CaM inhibition is much more clear, where inhibition accounts for perhaps all the negative feedback and adaptation to odorants (Kurahashi and Menini, 1997).
Acknowledgments
We thank the following people for the generous gift of DNA constructs: R. Molday for CNGB1, R.Y. Tsien for eCFP and eYFP, and E. Liman for pGEMHE. We are grateful to Heidi Utsugi, Kevin Black, Gay Sheridan, and Shellee Cunnington for technical assistance, Greg Martin and Paulette Brunner of the Keck Imaging Center for help with confocal microscopy, and to members of the Zagotta lab for insightful discussions. We thank Michael Varnum and Ariela Gordon-Shagg for comments on an earlier version of the manuscript.
This work was supported by the Howard Hughes Medical Institute and a grant from the National Eye Institute (EY10329) to W.N. Zagotta.
David C. Gadsby served as editor.
M.C. Trudeau's present address is Department of Physiology, University of Maryland School of Medicine, Baltimore, MD 21201.
Abbreviations used in this paper: CaM, calmodulin; cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; CNBD, cyclic nucleotide-binding domain; CNG, cyclic nucleotide-gated; eCFP, enhanced cyan fluorescent protein; eYFP, enhanced yellow fluorescent protein; FRET, fluorescence resonance energy transfer; PCF, patch-clamp fluorometry.