CNGA3 achromatopsia-associated mutation potentiates the phosphoinositide sensitivity of cone photoreceptor CNG channels by altering intersubunit interactions

Abstract

Cyclic nucleotide-gated (CNG) channels are critical for sensory transduction in retinal photoreceptors and olfactory receptor cells; their activity is modulated by phosphoinositides (PIP_n) such as phosphatidylinositol 4,5-bisphosphate (PIP₂) and phosphatidylinositol 3,4,5-trisphosphate (PIP₃). An achromatopsia-associated mutation in cone photoreceptor CNGA3, L633P, is located in a carboxyl (COOH)-terminal leucine zipper domain shown previously to be important for channel assembly and PIP_n regulation. We determined the functional consequences of this mutation using electrophysiological recordings of patches excised from cells expressing wild-type and mutant CNG channel subunits. CNGA3-L633P subunits formed functional channels with or without CNGB3, producing an increase in apparent cGMP affinity. Surprisingly, L633P dramatically potentiated PIP_n inhibition of apparent cGMP affinity for these channels. The impact of L633P on PIP_n sensitivity depended on an intact amino (NH₂) terminal PIP_n regulation module. These observations led us to hypothesize that L633P enhances PIP_n inhibition by altering the coupling between NH₂- and COOH-terminal regions of CNGA3. A recombinant COOH-terminal fragment partially restored normal PIP_n sensitivity to channels with COOH-terminal truncation, but L633P prevented this effect. Furthermore, coimmunoprecipitation of channel fragments, and thermodynamic linkage analysis, also provided evidence for NH₂-COOH interactions. Finally, tandem dimers of CNGA3 subunits that specify the arrangement of subunits containing L633P and other mutations indicated that the putative interdomain interaction occurs between channel subunits (intersubunit) rather than exclusively within the same subunit (intrasubunit). Collectively, these studies support a model in which intersubunit interactions control the sensitivity of cone CNG channels to regulation by phosphoinositides. Aberrant channel regulation may contribute to disease progression in patients with the L633P mutation.

cyclic nucleotide-gated (CNG) channels of retinal cone photoreceptors are crucial to color and high-acuity vision (5, 29). CNG channels are nonselective cation channels, the opening of which is controlled by intracellular cGMP in photoreceptors (10, 28). Phototransduction is initiated by light activation of opsins, which leads to lower cGMP levels and closure of the CNG channels. Thus, light signals are translated into electrical responses through CNG channels, resulting in a hyperpolarization of retinal photoreceptors. Each protein subunit of CNG channels consists of six transmembrane domains (S1–S6), a P-loop between S5 and S6, and cytoplasmic amino (NH₂)- and carboxyl (COOH)-terminal regions. The COOH-terminal region of each subunit is composed of a COOH-linker region, a cyclic nucleotide-binding domain (CNBD), and the post-CNBD region. Binding of cGMP to the CNBD induces a conformational change that is allosterically coupled to channel opening via gating rearrangements of the C-linker, the inner helix (S6), the pore helix, and the channel selectivity filter (7, 12, 60). Native cone CNG channels are tetrameric proteins formed by CNGA3 and CNGB3 subunits (2, 14, 65). The stoichiometry of cone CNG channel subunits is thought to be 2:2 (42, but also see 8), in contrast to the 3:1 ratio of CNGA1 to CNGB1 subunits in rod CNG channels (51, 69, 66). One structural determinant contributing to assembly of CNG channels is the carboxyl-terminal leucine zipper (CLZ) domain within the post-CNBD region. The CLZ domains are α-helices composed of heptad repeats of periodic hydrophobic residues. Previous studies have shown that the CLZ domain can assemble via coiled-coil interactions and thus may help control CNG channel subunit arrangement (51, 68, 69).

CNG channels do not desensitize to continuous cyclic nucleotide exposure, but they are regulated by various mechanisms, including via Ca²⁺-calmodulin (CaM) in olfactory and visual sensory cells (16, 25, 47, 57). This feedback regulation is thought to contribute, for example, to the ability of cone photoreceptors to adapt to a wide range of background light intensities (9). For cone CNG channels, the Ca²⁺-CaM mediated decrease in apparent ligand affinity is only modest (23, 40) and cannot account for the considerable Ca²⁺-dependent inhibition of native cone channels (21). A recent study suggests that another Ca²⁺-binding protein, CNG-modulin, may represent the authentic Ca²⁺-dependent modulator of cone CNG channels in striped bass (46). However, other mechanisms for regulation of CNG channels have been revealed and might interact with the aforementioned Ca²⁺-dependent modulators to adjust cone CNG channel activity.

Phosphoinositides (PIP_n), a family of acidic phospholipids within cell membranes, have been demonstrated to be modulators of many types of ion channels, including CNG channels (13, 53). Receptor-mediated changes in the concentration of PIP_n species, particularly phosphatidylinositol 4,5-bisphosphate (PIP₂) and phosphatidylinositol 3,4,5-bisphosphate (PIP₃), underlie many forms of ion channel regulation. For most channels, these phosphoinositides serve to support or promote channel activity. For photoreceptor and olfactory CNG channels, however, phosphoinositides primarily inhibit channel activity (3, 64). In photoreceptors, PIP_n levels are controlled by light, intracellular Ca²⁺, paracrine signals, and circadian oscillators (6, 19, 22, 31). Thus, PIP_n inhibition of CNG channels has the potential to explain several key aspects of photoreceptor physiology. We have found previously that cone CNGA3 + CNGB3 channels are inhibited by PIP₂ and PIP₃, showing a rightward shift in the cGMP dose-response relationship (a decrease in the apparent cGMP affinity) after PIP_n application (4, 7a). In addition, we have recently defined two structural components that are essential for PIP_n regulation of cone CNG channels; these are located within the cytoplasmic NH₂- and the COOH-terminal regions of CNGA3, respectively (7a).

Mutations in the genes encoding both CNGA3 and CNGB3 subunits of cone photoreceptor CNG channels have been linked to debilitating eye diseases in humans, including complete and incomplete achromatopsia, cone dystrophy, and inherited macular degeneration (1, 39, 45, 52, 54, 63). These pathologies are associated with cone dysfunction and progressive photoreceptor cell death, producing loss of central, high-acuity vision and day blindness (52). Human CNG channelopathies highlight the essential role of these proteins for photoreceptor health. Electrophysiological and biochemical characterization of CNGA3 and CNGB3 mutations has revealed both loss- and gain-of function defects (1, 32, 41, 48), suggesting that tight control of CNG channel activity is critical for the normal function of cone photoreceptors. While the basic functional effects for many disease-associated mutations have been defined, the potential impact of disease-associated mutations on cone CNG channel regulation has remained unknown.

Here, we report the functional consequences of a previously uncharacterized achromatopsia-associated mutation: L633P in human CNGA3 (17). This leucine residue is located in the part of the protein after the CNBD, within the COOH-terminal CLZ domain described above. The fundamental activation properties of cone CNG channels were only modestly altered by L633P. However, L633P conspicuously enhanced the PIP₂ or PIP₃ inhibition of cGMP-dependent channel gating, an effect that required the PIP_n regulation module located within the NH₂-terminal region of CNGA3. On the basis of in vitro coimmunoprecipitation studies and thermodynamic linkage analysis, we propose that L633P alters the interaction between NH₂- and COOH-terminal regions of CNGA3 and therefore potentiates the NH₂-terminal component of PIP_n regulation within CNGA3 subunits.¹

MATERIALS AND METHODS

Molecular biology.

Human CNGB3 was cloned as previously described (40). Human CNGA3 was a gift of K.-W. Yau (Johns Hopkins University, Baltimore, MD). For heterologous expression in Xenopus oocytes, CNGA3 and CNGB3 coding sequences were subcloned into pGEMHE, where they are flanked by the Xenopus β-globin gene 5′- and 3′-untranslated regions. The procedures for making point mutations and generating amino-terminal fusions of enhanced green fluorescent protein (eGFP) with CNGA3 were described previously (40). All mutations were confirmed by DNA sequencing. For expression of channel fragments as GST-tagged proteins, CNGA3 sequences were amplified and inserted into pGEX-5X2; coding sequences for these GST fusion proteins were subsequently subcloned into pGEMHE. For concatenated cDNA constructs, the stop codon for the leading subunit and the start codon for the trailing subunit were replaced by a short linker sequence (IAGGGGGRARLPA), combining the coding sequences for the two subunits into a single open reading frame (42). For expression in Xenopus oocytes, cRNA was synthesized in vitro from linearized channel-subunit cDNAs using an upstream T-7 promoter and the mMessage mMachine kit (Ambion, Austin, TX).

Patch-clamp electrophysiology.

Xenopus laevis oocytes were isolated as previously described (42) and injected with ∼20 ng of cRNA. The animal use protocols were consistent with the recommendations of the American Veterinary Medical Association and were approved by the Institutional Animal Care and Use Committee (IACUC) of Washington State University. Patch-clamp experiments were performed using an Axopatch 200B amplifier (Axon Instruments, Union City, CA) in the inside-out configuration. Initial pipette resistances were 0.4–0.8 MΩ. Intracellular and extracellular solutions contained 130 mM NaCl, 0.2 mM EDTA, and 3 mM HEPES (pH 7.2). Cyclic nucleotides were added to intracellular solution as needed. Macroscopic patch current density was calculated by estimating the patch area A (μm²) from the initial pipette resistance R (MΩ) based on the equation A = 12.6(1/R + 0.018) (50). Intracellular solutions were changed using an RSC-160 rapid solution changer (Molecular Kinetics, Indianapolis, IN). Solutions containing PIP₃ or PIP₂ analogs, diC8-PI (3,4,5) P₃ or diC8-PI (4,5) P₂ (Echelon Biosciences, Salt Lake City, UT), were prepared with FVPP (a phosphatase inhibitor cocktail) as previously described (4). Recordings were made at 20 to 22°C.

Protein biochemistry.

After 4 days of incubation at 16°C, oocytes that were injected with cRNA were homogenized in oocyte lysis buffer (20 mM Tris·HCl, 2 mM NaPO₄, 150 mM NaCl, 1 mM EDTA, 0.5% vol/vol Triton X-100, 0.5% vol/vol NP-40, 0.5% vol/vol CHAPS, pH 7.4). The homogenization buffer was supplemented with 1 protease inhibitor cocktail tablet (Roche Applied Science, Indianapolis, IN) per 10 ml buffer. We used 20–40 oocytes per sample group and 20 μl lysis buffer per oocyte. The homogenate was incubated on ice for 1 h and then centrifuged two to three times at 20,000 g for 12 min at 4°C to separate the supernatant from insoluble material and the floating fat layer. Soluble lysates were precleared using 10 μl of a 50% slurry of control agarose beads (Thermo Scientific, Waltham, MA). After 1 h of incubation with rocking, the samples with control beads were centrifuged at 2,000 g for 3 min and the supernatants were removed. The supernatants again were centrifuged at 20,000 g for 10 min and the supernatants were collected again. Then we applied 15 μl 50% slurry of anti-GFP beads (Vector Laboratories, Burlingame, CA) to the supernatant. After 2 h of incubation with rocking, the anti-GFP beads were gently pelleted and washed four times with lysis buffer. Protein complexes that interacted with anti-GFP beads were eluted using 2 × NuPAGE sample buffer (Invitrogen, Carlsbad, CA). Protein samples and the control oocyte lysates were then separated under reducing conditions using SDS-PAGE in 4–12% Bis-Tris gels. After gel electrophoresis, proteins were transferred to a nitrocellulose filter by electroblotting with the NuPAGE transfer buffer system (Invitrogen). The transferred proteins were detected via Western blotting using a mouse anti-GST monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) overnight at a dilution of 1:1,000 in TTBS solution composed of 1% milk, 20 mM Tris·HCl (pH 7.5), 500 mM NaCl, and 0.05% Tween 20. The blotted proteins from control oocyte lysates also were probed using a 1:2,500 diluted mouse anti-GFP antibody (Clontech, Palo Alto, CA) and the same anti-GST antibody. Anti-actin antibodies (Chemicon International, Temecula, CA) were used to ensure that the same amounts of input proteins were used. Chemiluminescent detection with horseradish peroxidase-conjugated anti-mouse IgG secondary antibodies was performed using the Super-Signal West Dura Extended Duration detection kit (Pierce).

Data analysis.

Data were acquired and analyzed using Pulse (HEKA Elektronik, Lambrecht, Germany), Igor (Wavemetrics, Portland, OR), and SigmaPlot/SigmaStat (Systat Software, San Jose, CA). Currents in the absence of cyclic nucleotide were subtracted. For channel activation by cyclic nucleotides, dose-response data were fitted with the Hill equation: I/I_max = [cNMP]^h/(K_1/2^h + [cNMP]^h), where I is the current amplitude, I_max is the maximum current elicited by saturating concentration of ligand, [cNMP] is the ligand concentration, K_1/2 is the concentration of cNMP producing half-maximal current, and h is the Hill coefficient. To confirm the formation of heteromeric CNG channels, block by 25 μM l-cis-diltiazem (Sigma-Aldrich, St. Louis, MO) in 1 mM cGMP at +80 mV was measured. The Gibbs free energy of the overall reaction of CNG channel opening, ΔG, was calculated using the equation ΔG = −RT ln (1/K_1/2^h) (7, 30), where R is the gas constant, 1.987 cal·K⁻¹·mol⁻¹, and T is the temperature in degrees Kelvin. We calculated the phosphoinositide-induced change in free energy difference for the channel opening reaction, ΔΔG, using the following: ΔΔG = ΔG_PIPn − ΔG_cont = RT ln {[K_1/2,PIPn^h(PIPn)]/[K_1/2,cont^h(cont)]}, where cont (control) and PIP_n indicate the values before and after PIP₂ or PIP₃ application. The interaction energy ΔΔΔ_intG for the thermodynamic linkage analysis was calculated as follows: ΔΔΔ_intG = (ΔΔG_PA-LP,L633P − ΔΔG_PA-LP) − (ΔΔG_L633P − ΔΔG_WT), where ∣ΔΔΔ_intG∣ = 0 indicates no interaction and purely additive effects. In our study, ΔΔΔ_intG > 1 kcal/mol was considered significantly interacted or coupled (30, 62). Data parameters are expressed as means ± SE of n experiments unless otherwise indicated. Statistical significance was determined by using Student's t-test or Mann-Whitney U-test.

RESULTS

We examined the functional effects of the disease-associated mutation L633P, which is located in the carboxyl-terminal leucine zipper (CLZ) domain of CNGA3 subunits, on the basic properties of channels formed by cone photoreceptor CNGA3 (A3) and CNGB3 (B3) subunits. A3 subunits can assemble to form functional homomeric (A3-only) channels; B3 subunits alone cannot form functional channels. However, the expression of A3 plus B3 subunits together generates heteromeric channels that essentially recapitulate the properties of native channels found in cone photoreceptor outer segments. Previous studies (68) have shown that CLZ-defective CNGA3 mutants have a decreased capacity to form heteromeric channels with CNGB1, suggesting that the CLZ domain is important for selective heteromeric assembly of CNG channels. L633P introduces a structurally distinct and conformationally rigid amino acid proline, which is commonly found as disruptors of protein secondary-structural components (such as α-helices) (34, 38). Therefore, we hypothesized that L633P could exert a profound disruption of the CLZ domain, potentially altering channel gating and regulation. Our approach to studying this mutation was to express homomeric and heteromeric wild-type or L633P-containing channels in Xenopus oocytes, followed by examining the properties of these channels in excised, inside-out membrane patches using electrophysiological recordings and application of the channel activators cGMP or cAMP to the cytoplasmic face of the patch. We found that the average maximal current densities (current/patch area) produced by 1 mM cGMP (Fig. 1A and Table 1) were significantly decreased by L633P: for heteromeric channels, current densities were 146.6 ± 49.6 pA/μm² for A3 + B3 and 30.0 ± 7.2 pA/μm² for A3-L633P + B3 (P < 0.05); for homomeric channels, current densities were 587.0 ± 244.5 pA/μm² for wild-type A3 and 111.2 ± 33.3 pA/μm² for A3-L633P (P < 0.05). Decreased current density for L633P-containing channels indicates that the mutation impaired but did not prevent functional expression of CNG channels.

Fig. 1.

Summary of effects of L633P mutation on basic properties of homomeric and heteromeric cone cyclic nucleotide-gated (CNG) channels. A: representative currents elicited by saturating 1 mM cGMP (black) and 5 μM cGMP (gray) for heteromeric A3 + B3 channels composed of A3 wild-type or A3-L633P. Currents were elicited by voltage steps from a holding potential of 0 mV to + 80 mV, then to −80 mV and 0 mV. Leak currents in the absence of cyclic nucleotide were subtracted. B: representative cGMP dose-response relationships for activation of heteromeric A3 + B3 and A3 L633P + B3 channels, fitted with the Hill equation. C: K_1/2,cGMP values for homomeric A3 channels and heteromeric A3 + B3 channels with or without the L633P mutation are shown using a box plot. The horizontal lines within the boxes indicate the median values; the bottom and top of the boxes show the 25th and 75th percentiles of the data, respectively; the ends of whiskers show the 5th and 95th percentiles of the data. D and E: outward current rectification (I_max,cGMP + 80 mV/−80 mV; D) and relative cAMP efficacy (I_max,cAMP/I_max,cGMP; E) of A3 and A3 + B3 channels were not significantly altered by L633P. F: l-cis-diltiazem inhibition of maximal cGMP current was attenuated by L633P. The ratio of cGMP current in the presence and absence of 25 μM l-cis-diltiazem (I_{L-cis-diltiazem}/I) was 0.42 ± 0.05 (n = 8) for wild-type A3 + B3 channels and 0.62 ± 0.03 (n = 8) for A3-L633P + B3 channels. *P < 0.05.

Table 1.

Effects of PIP_n application on gating parameters of CNGA3 or CNGA3 + CNGB3 channels

		Control		PIP2		PIP3
	Imax, A	K1/2, cGMP, μM	h	K1/2, cGMP, μM	h	K1/2, cGMP, μM	h
A3	1.4 ± 0.5 × 10−8 (9)	8.6 ± 0.5 (9)	1.9 ± 0.0	9.1 ± 0.5 (4)	1.8 ± 0.1	7.7 ± 0.7 (5)	2.0 ± 0.0
A3 L633P	3.0 ± 1.0 × 10−9 (13)	5.9 ± 0.3 (13)	1.9 ± 0.0	12.5 ± 0.9 (8)*	1.7 ± 0.0	13.4 ± 0.7 (5)*	1.9 ± 0.1
A3+B3	3.9 ± 1.7 × 10−9 (9)	13.5 ± 0.9 (9)	2.0 ± 0.0	23.2 ± 3.0 (5)*	1.6 ± 0.1	18.8 ± 2.0 (4)*	1.8 ± 0.2
A3 L633P+B3	2.0 ± 0.6 × 10−9 (12)	8.1 ± 0.4 (12)	1.9 ± 0.0	20.6 ± 1.8 (7)*	1.6 ± 0.1	19.9 ± 3.1 (5)*	1.6 ± 0.1
A3 613X	1.9 ± 0.7 × 10−10 (10)	13.2 ± 0.1 (10)	1.6 ± 0.1	44.4 ± 5.1 (4)*	1.3 ± 0.1	47.1 ± 3.2 (6)*	1.4 ± 0.1
A3 624X	7.3 ± 1.7 × 10−10 (8)	6.8 ± 0.4 (8)	2.0 ± 0.0	15.6 ± 3.9 (3)*	1.5 ± 0.0	14.9 ± 0.5 (5)*	1.5 ± 0.0
A3 635X	2.8 ± 1.0 × 10−9 (4)	7.1 ± 0.5 (4)	1.8 ± 0.0	ND	ND	6.6 ± 0.2 (4)	1.7 ± 0.1
A3 635X, L633P	1.4 ± 0.5 × 10−9 (5)	6.7 ± 0.2 (5)	2.0 ± 0.0	ND	ND	11.7 ± 0.7 (5)	1.6 ± 0.1
A3 E627R	1.1 ± 0.6 × 10−9 (9)	8.8 ± 0.5 (9)	2.0 ± 0.0	8.9 ± 1.1 (4)	2.0 ± 0.0	10.1 ± 1.2 (5)	1.8 ± 0.1
A3 E627,628R	1.9 ± 0.5 × 10−10 (9)	11.3 ± 0.7 (9)	1.8 ± 0.0	17.0 ± 2.4 (5)*	1.5 ± 0.1	15.7 ± 1.4 (4)*	1.9 ± 0.0
A3 E631R	2.2 ± 0.8 × 10−9 (4)	7.7 ± 0.4 (4)	1.9 ± 0.1	ND	ND	7.9 ± 0.4 (4)	1.9 ± 0.1
A3 N7R-A	2.7 ± 0.4 × 10−11 (4)	8.9 ± 1.2 (4)	1.6 ± 0.2	ND	ND	8.7 ± 0.8 (4)	1.8 ± 0.1
A3 N7R-A,L633P	3.5 ± 0.5 × 10−11 (11)	6.3 ± 0.3 (11)	1.8 ± 0.1	6.6 ± 0.4 (5)	1.5 ± 0.1	6.9 ± 0.3 (6)	1.9 ± 0.1
A3 N7RA,L633P+B3	1.1 ± 0.3 × 10−10 (8)	9.5 ± 0.4 (8)	1.6 ± 0.0	9.3 ± 1.2 (4)	1.6 ± 0.0	9.5 ± 1.7 (4)	1.6 ± 0.1
A3 PA-LP	1.5 ± 0.6 × 10−9 (10)	6.1 ± 0.3 (10)	1.8 ± 0.1	11.0 ± 0.4 (6)*	1.6 ± 0.0	9.8 ± 1.4 (4)*	1.8 ± 0.1
A3 PA-LP, L633P	5.0 ± 1.3 × 10−10 (8)	4.5 ± 0.3 (8)	1.8 ± 0.0	10.6 ± 0.9 (8)*	1.7 ± 0.0	ND	ND
A3//A3	6.8 ± 3.4 × 10−9 (4)	8.2 ± 0.5 (4)	2.0 ± 0.0	8.7 ± 0.9 (4)	2.0 ± 0.0	ND	ND
A3//A3 L633P	2.4 ± 0.5 × 10−9 (4)	5.3 ± 0.3 (4)	1.6 ± 0.0	12.3 ± 1.3 (4)*	1.6 ± 0.0	ND	ND
A3 N7R-A//A3 L633P	5.8 ± 1.0 × 10−11 (4)	7.9 ± 0.4 (4)	1.7 ± 0.0	8.1 ± 0.4 (4)	1.7 ± 0.0	ND	ND
A3//A3 N7R-A,L633P	4.8 ± 0.5 × 10−11 (4)	6.3 ± 0.3 (4)	1.8 ± 0.1	12.6 ± 0.8 (4)*	1.7 ± 0.0	ND	ND

Values are means ± SE. The numbers in parentheses after the values indicate number of patches (n); K_{1/2, cGMP} was determined using the Hill equation. h, Hill coefficient; I_max, maximum current; ND, not determined; CNG, cyclic nucleotide-gated; K_{1/2, cGMP}, concentration of nucleoside 3′,5′-cyclic monophosphate producing half-maximal current; PIP₂, phosphatidylinositol (4,5)-biphosphate; PIP₃, phosphatidylinositol (3,4,5)-trisphosphate.

^*Significant difference between Hill parameters before and after phosphoinositide (PIP_n) treatment (P < 0.05).

Next we determined the effect of L633P on channel activation properties. Native photoreceptor CNG channels operate under low cGMP concentrations in the dark, and these levels fall in response to light stimulation of the cells (43). Therefore, alterations in CNG channel ligand sensitivity can have profound consequences for channel activity under physiological conditions. We found that L633P significantly increased the current activated by a subsaturating concentration of cGMP (5 μM) relative to the current activated by a saturating concentration of cGMP (1 mM) (Fig. 1A); this low concentration of cGMP approximates the intracellular cGMP levels within vertebrate photoreceptors in the dark. The L633P mutation increased the apparent cGMP affinity (decreased K_1/2,cGMP) for heteromeric CNGA3 + CNGB3 channels (Fig. 1, B and C). Similarly, L633P increased the apparent cGMP affinity for homomeric CNGA3 channels (Fig. 1C).

As L633P may influence channel assembly, we sought to confirm that the mutation did not prevent the formation of heteromeric channels. To accomplish this, we investigated three features that are used to functionally distinguish between heteromeric versus homomeric CNG channels. First, we tested the outward rectification of currents mediated by L633P-containing channels. Heteromeric A3 + B3 channels characteristically produce less outwardly rectifying currents (outward current at +80 mV versus inward current at −80 mV) in saturating concentrations of cGMP compared with homomeric A3-only channels (42). L633P had no significant effect on current rectification for either homomeric (Fig. 1D) or heteromeric channels (Fig. 1, A and D), suggesting that CNGA3 subunits containing the mutation were able to assemble into heteromeric channels with CNGB3 subunits. Second, we examined the ratio of current elicited by a saturating concentration of cAMP relative to that produced by saturating cGMP (I_max,cAMP/I_max,cGMP) for L633P-containing channels. For cone photoreceptor CNG channels, cAMP is a partial agonist; heteromeric A3 + B3 channels are characterized by a greater I_max,cAMP/I_max,cGMP ratio compared with homomeric A3-only channels (42). The I_max,cAMP/I_max,cGMP ratio was not significantly altered by L633P (Fig. 1E), again indicating that A3-L633P was able to form functional heteromeric channels. Finally, we examined l-cis-diltiazem inhibition of cGMP-induced currents. l-cis-diltiazem block represents another reporter for the formation of functional heteromeric cone CNG channels, as channel block is minimal in the absence of CNGB3 (I_{L-cis-diltiazem}/I = 0.94 ± 0.02 for homomeric A3 channels) (42). Heteromeric A3-L633P + B3 channels showed decreased l-cis-diltiazem inhibition in 1 mM cGMP compared with wild-type heteromeric channels (Fig. 1F). However, diltiazem inhibition for A3-L633P + B3 channels was significantly greater than homomeric (wild-type or L633P) channels (P = 0.001). Together, these results show that L633P altered the gating properties of both homomeric and heteromeric channels, producing increased sensitivity to cGMP. In addition, the results suggest a slight impairment of heteromeric channel formation, presumably due to disruption of the CLZ domain.

Since the L633P mutation is located in a region of CNGA3 that we have previously shown to be important for regulation by phosphoinositides, we predicted that L633P might alter the PIP_n sensitivity of cone CNG channels. To test this prediction, we applied 1 μM diC8-PI (3,4,5) P₃ (PIP₃) or 10 μM diC8-PI (4,5) P₂ (PIP₂) to A3-L633P + B3 channels. The rational for using these concentrations of PIP_n analogs was based on the following. First, generic membrane levels of free PIP₂ in cells are thought to be ∼10 μM, and PIP₃ levels are expected to be at least tenfold lower (15, 24, 59). Second, we have shown previously that 10 μM PIP₂ and 1 μM PIP₃ induce near maximal changes in K_1/2,cGMP for activation of wild-type A3 + B3 channels; PIP_n concentration-response relationships suggest that the approximate apparent affinities of A3 + B3 channels for natural PIP₂, diC8-PIP₂, and diC8-PIP₃ are 2.8, 3.4, and 0.6 μM, respectively (7a). For regulation of wild-type A3 + B3 channels, 1 μM PIP₃ or 10 μM PIP₂ produced similar shifts in cGMP sensitivity, causing a 47.2 ± 4.2% (n = 4) or 60.7 ± 12.1% (n = 4) increase in K_1/2,cGMP for channel activation, respectively (Fig. 2A). For A3-L633P + B3 channels, we found that PIP₃ or PIP₂ generated a much greater increase in K_1/2,cGMP: 153.3 ± 22.2% (n = 5) after PIP₃ and 136.8 ± 15.6% (n = 5) after PIP₂ (Fig. 2B and Table 1). Next, we examined whether L633P alters the PIP_n sensitivity of homomeric A3-only channels. Wild-type homomeric A3 channels are insensitive to PIP₃ or PIP₂ regarding apparent ligand affinity (K_1/2,cGMP) but exhibit a more than twofold increase in cAMP efficacy (I_max,cAMP/I_max,cGMP ratio) after PIP_n application. This PIP_n-induced increase in cAMP efficacy is dependent on positively charged arginines located within the CLZ domain of CNGA3 (7a). Similar to heteromeric channels, we found that homomeric A3-L633P channels exhibited a large increase in K_1/2,cGMP after PIP₂ or PIP₃ (Fig. 3, A and B, and Table 1). Interestingly, the PIP_n-induced increase in I_max,cAMP/I_,cGMP for homomeric channels was eliminated by L633P (Fig. 3, C and D). These results demonstrate that L633P dramatically alters PIP_n regulation of both heteromeric and homomeric cone CNG channels.

Fig. 2.

L633P significantly enhances phosphoinositide (PIP_n) sensitivity of CNGA3 + CNGB3 channels. A and B: representative cGMP dose-response relationships before (black) and after (gray) 10 μM phosphatidylinositol (4,5)-bisphosphate (PIP₂) (diC8-PIP₂) application for activation of heteromeric A3 + B3 channels, fitted with the Hill equation. L633P enhanced the rightward shift of K_1/2,cGMP after PIP₂. The Hill parameters were as follows: for A3 + B3 channels, before PIP₂: K_1/2,cGMP = 13.8 μM, h = 2.1; after PIP₂: K_1/2,cGMP = 23.9 μM, h = 1.8; for A3-L633P + B3 channels, before PIP₂: K_1/2 = 6.6 μM, h = 1.8; after PIP₂: K_1/2 = 16.4 μM, h = 1.7. C: summary illustrating the K_1/2,cGMP values before and after 10 μM PIP₂ (diC8-PIP₂) or 1 μM PIP₃ (diC8-PIP₃) for heteromeric A3 + B3 or A3 L633P + B3 channels. PIP₃, phosphatidylinositol 3,4,5-trisphosphate; WT, wild type.

Fig. 3.

Effects of L633P on PIP_n sensitivity of homomeric CNGA3 channels. A: representative cGMP dose-response relationships before (black) and after (gray) 1 μM PIP₃ application for activation of homomeric A3 channels, fitted with the Hill equation. L633P enhanced the rightward shift of K_1/2 cGMP after PIP₃. The Hill parameters were as follows: for A3 wild-type channels, before PIP₃: K_1/2,cGMP = 12.8 μM, h = 2; after PIP₃: K_1/2,cGMP = 10.5 μM, h = 2; for A3 L633P channels, before PIP₃: K_1/2,cGMP = 6.9 μM, h = 1.7; after PIP₃: K_1/2,cGMP = 15.4 μM, h = 1.6. B: summary of the K_1/2,cGMP values before and after 10 μM PIP₂ (n = 4) or 1 μM PIP₃ (n = 5) for homomeric A3 and A3-L633P channels. C: representative currents elicited by saturating 1 mM cGMP (black) and 10 mM cAMP (gray), for homomeric channels composed of A3 wild-type or A3-L633P. Arrowhead indicates that application of 1 μM PIP₃ increased the saturating cAMP current for wild-type channels. L633P attenuated the increase of cAMP current by PIP₃. D: summary illustrating both PIP₃- and PIP₂-dependent changes in I_{max,cAMP/cGMP} for A3 and A3-L633P homomeric channels. *P < 0.05.

Given the location of L633P, we sought to characterize features surrounding this residue that may help control PIP_n regulation of the channel. Thus, we introduced targeted deletions and mutations within the region immediately following the CNBD of CNGA3 (Fig. 4A). We have shown previously that truncations of the distal COOH-terminal region after 613 (A3–613X), but not after 635 (A3–635X), unmasked the increase in K_1/2,cGMP after PIP_n application for homomeric channels (Fig. 4B and Table 1). On the basis of these experiments, we hypothesized that the A3 COOH-terminal region between amino acids 613 and 635 has an attenuating effect on the NH₂-terminal PIP_n regulation module. Deletion of the A3 COOH-terminal region after 613 relieves this inhibition and activates the NH₂-terminal regulatory module, which then is able to elicit a considerable increase in K_1/2,cGMP after PIP_n application. Similar to A3–613X, A3–624X exhibited sensitivity to PIP₃, manifested as a decrease in apparent cGMP affinity (Fig. 4B). The difference in PIP_n sensitivity for A3 624X versus 635X channels suggests that these residues bracket a structural element essential for tuning channel regulation by PIP_n. In addition, we found that combining 635X with L633P generated homomeric channels that were sensitive to PIP₃ (Fig. 4B), indicating that the regulatory features still present in 635X channels remain sensitive to disturbance by L633P. Furthermore, mutations of select charged residues in this region produced only a modest increase in PIP_n sensitivity compared with wild-type channels (Fig. 4C): for A3 channels having single mutations E627R or E631R, the percent increase of K_1/2,cGMP after 1 μM PIP₃ was 16.6 ± 9.2% (n = 5) and 3.2 ± 5.7% (n = 4), respectively; for A3 channels with combined E627R and E628R mutations, the percent increase in K_1/2,cGMP after PIP₃ was 37.2 ± 15.3% (n = 4; P < 0.05 compared with wild-type channels). Together, these observations indicate that the COOH-terminal region between 624 and 635 may be critical for influencing PIP_n sensitivity of CNGA3 channels.

Fig. 4.

CNGA3 COOH-terminal region surrounding L633P is critical for control of PIP_n sensitivity. A: schematic diagram showing cyclic nucleotide-binding domain (CNBD) and COOH-terminal leucine zipper (CLZ) domain within the human CNGA3 COOH-terminal region. Sequence alignments represent human CLZ domain (hA3: 626–672) and nearby upstream sequences (hA3: 613–625) together with homologous regions of canine, mouse, and bovine CNGA3. The leucine residue L633 is highlighted in bold. B: representative cGMP dose-response curves before (black) and after (gray) PIP₃ application for activation of homomeric A3 channels with COOH-terminal regions deleted, fitted with the Hill equation. C: summary illustrating the percent increase in K_1/2,cGMP after PIP₃ for channels with COOH-terminal truncations and mutations. *P < 0.05 for A3–635X vs. A3–635X, L633P channels.

CNGA3 subunits contain two distinct modules supporting regulation by phosphoinositides: one each in the NH₂- and COOH-terminal cytoplasmic domains. On the basis of the following rationale, we hypothesized that the enhancement of PIP_n sensitivity by L633P depends on the NH₂-terminal PIP_n regulation module. First, L633P attenuated the component of PIP_n sensitivity that is mediated by the COOH-terminal regulation module, which for homomeric CNGA3-only channels is manifested as an increase in cAMP efficacy (Fig. 3, C and D). Second, L633P was able to potentiate PIP_n inhibition in the context of the 635X truncation (Fig. 4C and Table 1); 635X deletes the COOH-terminal regulation module of CNGA3. To test our hypothesis, we used mutations to cripple the NH₂-terminal PIP_n regulation module while leaving the COOH-terminal PIP_n regulation module intact. As demonstrated previously, neutralizing the positively charged arginines in this region (N7R-A: R72A, R75A, R81A, R82A, R86A, R101A, R103A) eliminated the PIP_n sensitivity of A3–613X channels (Fig. 5B) and attenuated binding of an NH₂-terminal domain fragment of CNGA3 to PIP₃-linked agarose beads (7a). Here, we tested whether N7R-A mutations inhibited the PIP_n sensitivity of A3-L633P channels similar to A3–613X channels. As expected, N7R-A abolished the PIP_n-induced increase in K_1/2,cGMP for both homomeric and heteromeric L633P channels (Fig. 5B); the K_1/2,cGMP was not significantly altered by PIP₂ application (P = 0.38 and P = 0.89, respectively, for homomeric and heteromeric channels). These results suggest that similar to COOH-terminal truncations, L633P exerts a disinhibition effect on the NH₂-terminal PIP_n regulation site. Silencing the NH₂-terminal PIP_n interaction site removes the augmentation of PIP_n sensitivity caused by L633P. Furthermore, the data suggest that a potential interaction occurs between NH₂- and COOH-terminal cytoplasmic regions of CNGA3 subunits.

Fig. 5.

Enhancement of PIP_n sensitivity by L633P depends on the NH₂-terminal PIP_n regulation site of CNGA3. A: cartoon showing the PIP_n regulation site within the NH₂-terminal region of CNGA3; neutralization of the seven arginines within this region (N7R-A) is sufficient to prevent PIP_n regulation of the channel. B: representative cGMP dose-response relationships before (black) and after (gray) 10 μM PIP₂ for activation of homomeric A3–613X (Δ613–694), A3 N7R-A,613X, A3 N7R-A,L633P channels, and heteromeric A3 N7R-A, L633P + B3 channels.

Guided by the observations described above, we hypothesized that L633P alters the coupling between NH₂- and COOH-terminal regions of CNGA3 during channel activation and regulation. We used thermodynamic linkage analysis to investigate possible coupling between the NH₂- and COOH-terminal domains. One previously characterized CNGA3 change, P99L, A105P (PA-LP) (18), shows an increase both in initial cGMP affinity and in PIP_n sensitivity (Table 1), similar to L633P channels. In contrast to L633P, PA-LP is located within the NH₂-terminal region of A3 (Fig. 6A). If interdomain NH₂-COOH (N-C) coupling occurs, then the NH₂-terminal PA-LP and COOH-terminal L633P mutations might alter channel gating and tune PIP_n sensitivity of the channel in a convergent manner. By constituting channels with a combination of mutations, we can compare PIP_n regulation of channels having single changes (PA-LP in the NH₂-terminal region or L633P in the COOH-terminal region) to those with combined mutations (A3 PA-LP, L633P). The initial K_1/2,cGMP for A3 PA-LP, L633P channels was significantly lower than that of A3 PA-LP, but it was not statistically different from that of A3 L633P (P = 0.204). Similarly, we found the PIP₂-induced increase in K_1/2,cGMP for A3 PA-LP, L633P was not statistically different from that of L633P (P = 0.751) (Fig. 6C). The changes in Gibbs free energy (ΔΔG) for channel opening reactions caused by PIP_n were calculated (Fig. 6D). The ΔΔG for the combined mutations did not exhibit additive consequences when compared with the individual mutants (P > 0.05), suggesting that the effects of L633P and PA-LP on PIP_n sensitivity of A3 channels overlap. Alternatively, if the effects of L633P and PA-LP on PIP_n sensitivity were independent, we would have expected the PIP_n-induced ΔΔG of the mutant combination to represent an approximate summation of the ΔΔG for the individual mutants. On the basis of the thermodynamic cycle shown in Fig. 6B, the interaction energy ΔΔΔ_intG was −1.80 ± 0.44 kcal/mol, differing substantially from zero. These results are consistent with interdomain coupling between NH₂- and COOH-terminal regions of CNGA3.

Fig. 6.

Thermodynamic linkage analysis is consistent with coupling between NH₂- and COOH-terminal regions of CNGA3. A: cartoon showing approximate locations of NH₂-terminal P99L, A105P (PA-LP) mutations and COOH-terminal L633P mutation in CNGA3. Box represents PIP_n regulation module within the NH₂-terminal region. B: scheme illustrating the thermodynamic cycle. C: summary of K_1/2,cGMP before and after application of 10 μM PIP₂ for wild-type, individual-mutant, and combined-mutant channels. The PIP₂-induced percent increase in K_1/2,cGMP for A3 PA-LP, L633P was 140.7 ± 0.16% (n = 8) compared with 133.8 ± 13.4% for A3-L633P (n = 8) and 63.2 ± 5.0% for A3 PA-LP (n = 10). D: the differences (ΔΔG) are shown for the Gibbs free energy change (ΔG) of the channel opening reactions caused by PIP₂ for A3 wild-type, A3 PA-LP, A3 L633P, and A3 PA-LP, L633P channels (ns, not significantly different).

On the basis of the above results, our working hypothesis was that L633P disrupts a physical interaction between NH₂- and COOH-terminal cytoplasmic domains that helps control PIP_n sensitivity. To test this idea, we expressed a soluble fragment of the A3 post-CNBD region (A3C) in combination with functional A3ΔC subunits having this region truncated (613X) (Fig. 7A). We predicted that if the A3C fragment could directly interact with the truncated channel, then it would alter the response of A3ΔC channels to PIP_n, attenuating the PIP_n-induced increase in K_1/2,cGMP and partially restoring normal PIP_n sensitivity. In combination with A3ΔC, we expressed COOH-terminal fragments representing all or part of the post-CNBD region: full-length (aa 611–694), proximal (aa 611–654), and distal (aa 655–694) fragments. We found that both the full-length fragment and the proximal fragment, but not the distal fragment, significantly attenuated the increase in K_1/2,cGMP for A3ΔC channels after application of PIP₂ (Fig. 7B). Partial rather than full restoration of normal PIP_n sensitivity may reflect low affinity and a low effective concentration for the soluble fragment compared with the situation where this domain is physically attached to the channel (high effective concentration). In contrast, coexpression of the A3C 611–694 fragment having the L633P mutation with A3ΔC channels produced PIP₂ sensitivity that was not significantly different from A3ΔC channels alone (P = 0.961). Again, these data are consistent with a scenario where the proximal COOH-terminal region after the CNBD is necessary for coupling with the NH₂-terminal region, and L633P can disrupt this coupling.

Fig. 7.

Coexpression of soluble CNGA3 COOH-terminal fragments supports interdomain NH₂-COOH (N-C) interactions regulating PIP_n sensitivity. A: cartoon showing the constructs used for coexpression experiments. CNGA3ΔC is equivalent to 613X, and CNGA3 ΔN, ΔC combines NH₂-terminal Δ2–107 and COOH-terminal Δ613–694 truncations. The NH₂-terminal PIP_n regulation site is shown as a black box. eGFP, enhanced green fluorescent protein. B: summary showing effects of PIP₂ on A3ΔC channels coexpressed with soluble A3 COOH-terminal fragments. Coexpression of A3 COOH-terminal fragments representing the region after the CNBD (aa 611–694) with A3ΔC channels significantly attenuated the increase in K_1/2,cGMP after PIP₂ application. However, addition of L633P to this fragment prevented this effect. The proximal A3 COOH-terminal fragment (611–654) exerted a similar effect to the full-length fragment while the more distal A3 COOH-terminal fragment (655–694) failed to attenuate the PIP₂-induced increase in K_1/2,cGMP for A3 ΔC channels; *P < 0.05 compared with A3ΔC channels alone. C: immunoblots are shown for coimmunoprecipitation of GST-tagged COOH-terminal fragments with eGFP-tagged A3ΔC or A3ΔNΔC channels. A3 C(611–694) was able to bind to A3ΔC (2 + 4), but not to A3ΔN,ΔC (3 + 4) in vitro. L633P prevented binding of A3 C(611–694) to A3ΔC channels (2 + 5). Both of the smaller A3 COOH-terminal fragments showed weak binding (2 + 6, 2 + 7). Approximate molecular weight markers are indicated to the right of the blots. The slower gel migration of the L633P-containing fragment (5) likely reflects a less compact structure compared with the wild-type fragment. Lower-molecular-weight bands for fragments 6 and 7 represent minor proteolytic products that retain the GST tag. Note that the top molecular weight bands shown in the bottom blot represent glycosylated CNGA3 subunits.

The putative interaction between NH₂- and COOH-terminal domains of CNGA3 was further investigated using in vitro coimmunoprecipitation studies (Fig. 7C). We coexpressed enhanced GFP-tagged A3–613X subunits (eGFP-A3ΔC) with soluble GST-tagged A3 post-CNBD fragments (GST-A3C) in oocytes. Use of A3ΔC subunits is expected to prevent the homotypic interactions between intact CLZ domains previously described (69). We isolated A3ΔC channels from oocyte lysates using anti-eGFP antibodies linked to agarose beads, followed by immunoblotting the pulled-down protein complexes using anti-GST antibodies. We found that A3C (611–694) was able to bind to A3ΔC channels; the shorter COOH-terminal fragments A3C (611–654) and A3C (655–694) also exhibited (weak) interactions with A3ΔC channels in vitro. We next asked whether coimmunoprecipitation of A3C with A3ΔC requires an intact NH₂-terminal domain. Thus, we coexpressed channel subunits having both NH₂- and COOH-terminal regions deleted (eGFP-A3ΔN, ΔC) together with GST-A3C fragments. We found the NH₂-terminal deletion eliminated the binding of the COOH-terminal fragment to the channel. Finally, we observed that the L633P mutation prevented binding of GST-A3C (611–694) fragments to eGFP-A3ΔC channels, in agreement with the electrophysiological data described above. Taken together, these observations support interactions between the NH₂- and COOH-terminal regions of CNGA3 that help control PIP_n sensitivity.

Interactions between NH₂- and COOH-terminal regions of the channel may occur via contacts within the same subunit (intrasubunit) or between channel subunits (intersubunit). On the basis of our model suggesting that L633P separates or uncouples the A3 COOH-terminal region from the NH₂-terminal region, resulting in unmasking of PIP_n sensitivity that depends on an NH₂-terminal regulation module, we designed concatenated cDNA constructs to produce tandem dimers of A3 subunits containing the L633P mutation. Tandem linkage of A3 subunits by itself did not alter PIP_n regulation of the channels: the apparent cGMP affinity of A3//A3 dimers remained insensitive to 10 μM PIP₂ (Fig. 8A), while A3//A3 L633P channels showed a 133.1 ± 20.2% (n = 4) increase in K_1/2,cGMP (Fig. 8B). Since charge neutralizations within the NH₂-terminal PIP_n-regulation module (N7R-A) antagonized the PIP_n sensitivity of L633P channels, we generated tandem dimers of A3 subunits with N7R-A and L633P mutations either in alternating subunits (A3 N7R-A//A3 L633P) or within the same subunit (A3//A3 N7R-A, L633P). If the N-C interactions occur within the same subunit (intrasubunit), then A3 N7R-A//A3 L633P channels are expected to be PIP_n sensitive; N7R-A within one subunit would not affect the L633P-activated PIP_n regulation module within the NH₂-terminal region of the adjacent subunit. Conversely, if the N-C interaction occurs between subunits (intersubunit), then A3 N7R-A//A3 L633P channels are expected to be PIP_n insensitive, because the NH₂-terminal PIP_n regulation site within one subunit that was unmasked by L633P of the adjacent subunit would be silenced by N7R-A mutations. Channels formed by A3 N7R-A//A3 L633P dimers were found to be insensitive to PIP₂ regarding apparent cGMP affinity (Fig. 8C), showing only a 3.3 ± 3.2% increase in K_1/2,cGMP (n = 4; P = 0.701 compared with untreated channels). We next tested A3//A3 N7R-A, L633P dimers. Since channels formed by either wild-type A3 or A3 N7R-A, L633P subunits expressed alone as homomultimers are PIP_n insensitive, PIP_n sensitivity observed when combining these subunits together in the same channel would provide evidence for an intersubunit interaction. Channels formed by A3//A3 N7R-A, L633P dimers exhibited a 102.6 ± 22.1% (n = 4) increase in K_1/2,cGMP (Fig. 8D), indicating that L633P within one subunit was able to activate the intact PIP_n-regulation module within the NH₂-terminal region of an adjacent subunit, even though the PIP_n regulation module within the NH₂-terminal region of the same subunit had been silenced. These results, summarized in Fig. 8E, are consistent with intersubunit coupling between NH₂- and COOH-terminal regions that controls the PIP_n sensitivity of CNGA3 channels.

Fig. 8.

Tandem CNGA3 heterodimers containing L633P reveal intersubunit N-C terminal interaction controlling PIP_n sensitivity. Four scenarios illustrate how intersubunit N-C terminal interactions control PIP_n regulation of CNGA3 subunits, respectively. These PIP_n regulation modules can be silenced (red) or activated (green) owing to changes in intersubunit interactions. Representative cGMP dose-response curves before and after 10 μM PIP₂ (gray) are shown for activation of each channel made up of these dimers. A: tandem linkage of wild-type A3 subunits. B: channels formed by tandem A3//A3 L633P dimers. The Hill parameters for representative A3//A3 L633P channels were as follows: before PIP₂: K_1/2,cGMP = 5.72 μM, h = 1.6; after PIP₂: K_1/2,cGMP = 13.3 μM, h = 1.6. C: A3 N7R-A//A3 L633P tandem dimers. D: channels formed by A3//A3 N7R-A, L633P tandem dimers. The Hill parameters for representative A3//A3 N7R-A, L633P channels were as follows: before PIP₂: K_1/2,cGMP = 6.58 μM, h = 1.6; after PIP₂: K_1/2,cGMP = 13.1 μM, h = 1.8. E: summary showing the percent increase in K_1/2,cGMP after 10 μM PIP₂ for tandem dimer constructs, as in A–D, or after coexpression of the respective constituent monomers (n = 4). *P < 0.05 compared with A3//A3 dimer channels.

The interpretation of tandem dimer experiments requires some caution since these approaches can be sensitive to potential artifacts due to aberrant assembly (35). Potential artifacts include exclusion of the second tandem dimer subunit from the channel tetramer, or possible linker proteolysis. For A3//A3 N7R-A, L633P dimers, since each individual subunit of the dimer alone generates PIP_n-insensitive channels, neither of these potential assembly artifacts can explain the formation of PIP_n-sensitive channels. In support of our interpretation of intersubunit N-C interactions, we found that coexpression of the respective PIP_n-insensitive A3 wild-type and A3 N7R-A, L633P monomer constituents also produced channels that reconstituted PIP_n inhibition of apparent cGMP affinity, exhibiting a 62.2 ± 12.5% (n = 4) increase in K_1/2,cGMP (Fig. 8E). Since A3 wild-type subunits produce functional channels more efficiently than A3 N7R-A, L633P subunits (Table 1), a 1:5 ratio of the respective cRNAs was injected for coexpression experiments. Assuming a binominal distribution of channel species via free assembly of the monomer constituents, an average of approximately one favorable intersubunit N-C interaction per channel is expected. In contrast to PIP_n-insensitive A3 N7R-A//A3 L633P dimers, coexpression of the respective A3 N7R-A and A3 L633P monomer constituents produced some channels that were PIP_n sensitive, exhibiting a 48.3 ± 9.2% increase in K_1/2,cGMP (n = 4) (Fig. 8E). This difference in PIP_n sensitivity observed between A3 N7R-A//A3 L633P dimers and coexpression A3-N7R-A and A3-L633P monomers again suggests that the tandem dimer configuration successfully constrains the subunit arrangement for CNGA3 tetramers. Taken together, these results validate our interpretation of the tandem dimer experiments described above and are consistent with intersubunit interactions controlling PIP_n sensitivity of CNGA3 channels.

DISCUSSION

Characterization of the achromatopsia-associated mutation L633P in human CNGA3 has provided mechanistic insight into cone CNG channel regulation by PIP_n and the potentiation of PIP_n sensitivity by L633P. We propose a model where L633P disrupts an interaction between NH₂- and COOH-terminal cytoplasmic domains of adjacent A3 subunits, which serves to unmask and activate the PIP_n interaction module located within the NH₂-terminal domain of A3. The results suggest that in addition to a role in assembly of CNG channel subunits, the CLZ domain of CNGA3 also may help control PIP_n regulation of the channel by participating in intersubunit interactions. Similarly, assembly of A3 with B3 subunits likely alters some of these N-C interactions (by substitution), tuning the channel's PIP_n sensitivity. L633P still can potentiate the PIP_n sensitivity of heteromeric A3 + B3 channels, possibly due to effects on A3-A3 interactions within the tetramer.

The CLZ domain of CNG channels has been proposed to play an important role in channel assembly. Previous studies by Zhong et al. (68) have shown that mutation of hydrophobic heptad-repeat residues within the CLZ domain of CNGA3 decreased cGMP-activated current density in HEK 293T cells. In addition, L633A mutation attenuated l-cis-diltiazem inhibition and the I_max,cAMP/I_max,cGMP ratio for CNGA3 + CNGB1 channels, indicating that the CLZ domain is critical for selective assembly of heteromeric channels (68). Consistent with these studies, we have observed that the current density for L633P-containing channels is similarly decreased and that formation of heteromeric CNGA3 + CNGB3 channels is slightly impaired by L633P when channels are expressed in oocytes. However, we also observed changes in the gating properties of L633P channels that were not reported with CLZ deletions or other mutations (68). This is not surprising because radically disrupting the α-helical structure of the CLZ domain may have a different effect on channel gating compared with deleting the CLZ domain. Furthermore, we found that mutating a cluster of hydrophobic residues within the distal part of the CLZ domain (L662A, L665A, and V669A) similarly potentiated the PIP_n sensitivity of the channel; the K_1/2,cGMP was shifted from 5.67 ± 0.21 μM to 10.90 ± 0.93 μM by PIP₂ (92.4 ± 11.2% increase; n = 4). Together, these data suggest that the structural integrity of the entire CLZ domain might be necessary for proper PIP_n regulation of the channel.

We have used multiple approaches that together suggest N-C interactions help control PIP_n regulation of CNGA3 channels. However, we have not yet localized specific residues necessary for this interaction. We investigated whether possible salt bridges between critical subdomains might contribute to interactions between NH₂- and COOH-terminal regions. However, several charge reversal mutations within COOH-terminal region between 624 and 635 (E627R, E628R, E631R) (Fig. 4C), as well as charge reversals within the region surrounding P99 and A105 in the NH₂-terminal domain (R101E, R103E and K108E), produced only subtle effects on initial K_1/2,cGMP and PIP_n sensitivity of A3 channels (for A3 R101E, R103E, K108E channels, the percent increase of K_1/2,cGMP after 1 μM PIP₃ was 28.8 ± 4.0%). Besides salt bridges, other types of protein-protein interfaces, such as hydrophobic interactions, may play a role. Alternatively, indirect interactions mediated by other regions of the channel or by unknown channel-associated proteins may support N-C coupling in CNGA3 channels.

Intersubunit N-C interactions also are characteristic of other CNG channel types. In rod CNG channels, the NH₂-terminal region of CNGB1 has been shown to interact with the CLZ domain of CNGA1; Ca²⁺-CaM binding to the NH₂-terminal calmodulin-binding module within CNGB1 disrupts this intersubunit interaction (56). A recent study showed that coexpression of CNGA1-ΔCLZ with CNGB1-ΔN subunits generated channels with a mixture of channel subunit stoichiometries (51). Interestingly, coexpression of CNGA1-ΔCLZ with intact CNGB1 subunits basically eliminated all functional channel expression in Xenopus oocytes, whereas coexpression of CNGA1-ΔCLZ with CNGB1-ΔN subunits restored expression, consistent with an intersubunit interaction between CNGA1 and the NH₂-terminal region of CNGB1 (51). In addition, intersubunit disulfide bonds can be formed between endogenous cysteines within NH₂-terminal and C-linker regions of CNGA1 (49). Furthermore, the NH₂-terminal GARP domain of CNGB1 was found to act as gating inhibitor of rod channels via direct protein-protein interactions with CNGA1 (36). In CNGA2 channels, Ca²⁺-CaM can bind to an NH₂-terminal site, which disrupts an autoexcitatory interdomain interaction between the NH₂- and COOH-terminal regions of adjacent CNGA2 subunits and downregulates channel activity (61, 67). Furthermore, N-C interactions have been observed in other homologous CNBD-containing channels. In human ether-á-go-go-related gene (hERG) potassium channels, intersubunit interactions between the NH₂-terminal “EAG domain” and the COOH-terminal CNB-homology domain have been shown to regulate the slow deactivation of the channel (20, 37). CNG channels belong to the superfamily of voltage-gated cation channels (27). Plausible N-C interactions are consistent with the fundamental structural organization of voltage-gated sodium and calcium channels having N-C linkages between pseudo subunits, and with the successful generation of functional channels using concatenated cDNA constructs for tandem potassium and CNG channel subunits (42, 58). Furthermore, at a different structural level, the potential for intersubunit coupling is suggested by the structure of voltage-gated potassium channels, which show the offset voltage-sensor domain (S1–S4) of one subunit in close proximity to the pore domain and contiguous COOH-terminal module of the adjacent subunit (33). Together with the intersubunit interactions of CNGA3 channels described in this paper, we propose that N-C interdomain interactions may be a general feature for the family of CNBD-containing ion channels.

In addition to influencing N-C coupling, L633P mutation might exert an effect on the CNBD of CNGA3. Previous work has demonstrated that specific residues within the αC helix of the CNBD are critical for ligand selectivity of CNG channels and hyperpolarization-activated cyclic nucleotide-regulated (HCN) channels (11, 60, 70). In HCN2 channels, the αC helix undergoes a conformational rearrangement and is stabilized after binding of the ligand to the CNBD (55). Because of the close proximity of the CNBD and CLZ domains, disruption of the CLZ domain by L633P or other mutations may have an effect on the αC helix of the CNBD. The PIP_n-dependent increase in cAMP efficacy for homomeric A3 channels was eliminated by L633P. It is possible that PIP_n could have an effect on the CLZ domain (directly or indirectly) and therefore influence the αC helix of CNBD and ultimately alter the coupling of cGMP or cAMP binding to channel gating. L633P may abolish either a potential PIP_n interaction with the COOH-terminal region of A3 or the ability of the COOH-terminal region to communicate the PIP_n-binding event elsewhere in the channel to the CNBD.

Overall, our results show that the achromatopsia-associated mutation L633P produces a gain-of-function change for gating of CNG channels, increasing the apparent affinity for cGMP. However, L633P produces a loss-of-function effect regarding functional expression level (current density). In addition, L633P results in abnormal channel regulation by PIP_n. Phosphoinositide metabolism is very active in retinal photoreceptors. For example, light activation of the insulin receptor/phosphoinositide 3-kinase (PI3K)/Akt pathway generates PIP₃ in photoreceptors (31, 44). Downregulation of PI3K/Akt pathway in cone photoreceptor results in cone degeneration (26). Moreover, circadian phase-dependent regulation of cone CNG channels by somatostatin is mediated by phospholipase C (PLC), which cleaves PIP₂ within the membrane (6). Therefore, light and/or circadian-dependent modulation of photoreceptor CNG channels may be mediated in part by phosphoinositides. Enhanced PIP_n sensitivity of the cone channel due to L633P could potentially impair the ability of the channels to appropriately adjust ligand sensitivity in response to light or circadian inputs.

GRANTS

This research was supported by National Institutes of Health National Eye Institute Grant EY-12836 (to M. D. Varnum).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

G.D. performed the experiments; G.D. and M.D.V. analyzed the data; G.D. and M.D.V. interpreted the results of the experiments; G.D. and M.D.V. prepared the figures; G.D. drafted the manuscript; G.D. and M.D.V. edited and revised the manuscript; G.D. and M.D.V. approved the final version of the manuscript; M.D.V. conception and design of the research.