Functional Role of hCNGB3 in Regulation of Human Cone CNG Channel: Effect of Rod Monochromacy-Associated Mutations in hCNGB3 on Channel Function

Abstract

purpose. The human cone photoreceptor cyclic nucleotide-gated (CNG) channel comprises α- and β-subunits, which are respectively encoded by hCNGA3 and hCNGB3. The purpose was to examine the functional role of hCNGB3 in modulation of human cone CNG channels and to characterize functional consequences of rod monochromacy-associated mutations in hCNGB3 (S435F and D633G).

methods. Macroscopic patch currents were recorded from human embryonic kidney (HEK) 293 cells expressing homomeric (hCNGA3 and hCNGB3) and heteromeric (hCNGA3/hCNGB3, hCNGA3/hCNGB3-S435F, and hCNGA3/hCNGB3-D633G) channels using inside-out patch-clamp technique.

results. Both hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels were activated by cGMP, with half-maximally activating concentration (K _1/2) of 11.1 ± 1.0 and 26.2 ± 1.9 μM, respectively. The hCNGA3 channels appeared to be more sensitive to inhibition by extracellular Ca²⁺ compared with hCNGA3/hCNGB3 channels, when assessed by the degree of outward rectification. Coexpression of either of rod monochromacy-associated mutants of hCNGB3 with hCNGA3 significantly reduced K _1/2 value for cGMP but little affected the sensitivity to extracellular Ca²⁺, compared with wild-type heteromeric channels. The selectivity of hCNGA3, hCNGA3/hCNGB3, hCNGA3/hCNGB3-S435F, and hCNGA3/hCNGB3-D633G channels for monovalent cations were largely similar. Immunoprecipitation experiments showed association of hCNGA3 subunit with both of wild-type and mutant hCNGB3 subunits.

conclusions. The hCNGB3 plays an important modulatory role in the function of human cone CNG channels with respect to cGMP and extracellular Ca²⁺ sensitivities. The rod monochromacy-associated S435F and D633G mutations in hCNGB3 evokes a significant increase in the apparent affinity for cGMP, which should alter cone function and thereby contribute at least partly to pathogenesis of the disease.

Cyclic nucleotide-gated (CNG) channels play an important role in mediating visual and olfactory transductions. In vertebrate rods and cones, CNG channels are thought to be heterotetramers composed of two distinct types of subunit, generically termed α and β. ¹ The native CNG channel in rods comprises the 63-kDa α-subunit and 240-kDa β-subunit, which are respectively encoded by CNGA1 and CNGB1. ^{1 2} When expressed heterologously, the rod α-subunit CNGA1, but not β-subunit CNGB1, forms functional homomeric channels on its own. However, β-subunit coassembles with α-subunit to form the functional heteromeric channels and thereby alters the properties of α-subunit, including the sensitivity to regulation by intracellular cGMP, Ca²⁺-calmodulin, and extracellular Ca²⁺. ^{2 3 4} The native CNG channel in cones is also assumed to function as heterotetramers composed of the 79-kDa α-subunit and 92-kDa β-subunit encoded by CNGA3 and CNGB3, respectively. ^{4 5 6} It has been suggested that β-subunit in cone photoreceptors plays a modulatory role in the function of CNG channels, analogous to β-subunit in rod photoreceptors. ⁷ There is, however, a significant difference in the structure of β-subunits between rods and cones. The rod β-subunit CNGB1 has the glutamic acid-rich part at its amino-terminal half, ⁸ which is reflected as a relatively large molecular weight compared with that of cone β-subunit CNGB3. 

Rod monochromacy is a rare inherited disorder characterized by an absence of functional cone photoreceptors in the retina and typically causes photophobia, nystagmus, low visual acuity in daylight, and a total deficit of color vision. Three genes have been implicated in rod monochromacy; hCNGA3, hCNGB3, and GNAT2. Previous studies have defined approximately 60 distinct disease-causing mutations in hCNGA3, ^{9 10 11} hCNGB3, ^{6 12 13} and GNAT2. ^{14 15} It has also been reported that hCNGA3 and hCNGB3 respectively account for ∼20%–30% ¹⁰ and ∼40%–50% ¹⁴ of the cases while GNAT2 is responsible for only a small percentage. ¹⁴ We previously screened four Japanese individuals with rod monochromacy for mutations in hCNGA3, hCNGB3 and GNAT2. We have not detected any of the previously reported mutations but have found one novel missense mutation in hCNGB3 gene D633G, which resides in cyclic nucleotide-binding domain near the COOH terminus in CNG channel β-subunit (Okada A, IOVS 2001;42:ARVO Abstract 3432). 

In the present study we examined the cGMP sensitivity, ion selectivity, and Ca²⁺ blockade in hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels to elucidate the functional role of hCNGB3 in modulation of human cone CNG channels. Furthermore, we characterized the functional consequences of rod monochromacy-associated S435F ^{6 12} and D633G mutations in hCNGB3 in a similar way. 

Methods

Cloning of Human Cone CNGA3 and CNGB3 Subunit Complementary DNAs

Complementary DNAs for hCNGA3 and hCNGB3 subunits were obtained as the PCR products from the human retinal cDNA (Marathon-Ready cDNA; BD Biosciences Clontech, Palo Alto, CA). The sequences of the primers used were 5′-TGACAAACCGAGAAGATGG-3′ and 5′-ACAGCTGCGGCCACATAC-3′ for the hCNGA3 subunit, and 5′-GGCACAGTCATAAATACAGA-3′ and 5′-AAGAACCAAAGGAAAAGGA-3′ for the hCNGB3 subunit. The enzyme used for the PCR was the Pyrobest DNA polymerase (Takara, Kyoto, Japan). The products were cloned into the pCR3.1 expression vector (Invitrogen Corp., Carlsbad, CA). Two mutations, 1304C>T (S435F) and 1868A>G (D633G), designated hCNGB3-S435F and hCNGB3-D633G, respectively, were each introduced to the hCNGB3 subunit cDNA as previously described, ¹⁶ and these mutant cDNAs were also cloned into pCR3.1. 

Cell Culture and Transfection of Complementary DNAs

HEK293 cells (American Type Culture Collection, Rockville, MD) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS) and were subcultured every 2–3 days. HEK293 cells were plated onto rectangular (5 × 3 mm²) glass coverslips and were transfected with plasmids encoding human cone CNG channels together with a reporter plasmid (green fluorescent protein vector, pQBI25; CPG Inc., Lincoln Park, NJ) using TransIT-LT1 (Mirus Corp., Madison, WI). The amounts of each vector were (μg/dish): 1.5 hCNGA3 and 0.5 GFP or 0.75 hCNGA3, 0.75 hCNGB3 (or hCNGB3-S435F, hCNGB3-D633G) and 0.5 GFP. Patch-clamp experiments were conducted 2–3 days after transfection, on GFP-positive cells. 

Patch-Clamp Technique and Data Analysis

Membrane currents were recorded in an inside-out configuration of the patch-clamp technique ¹⁷ with an EPC-8 patch-clamp amplifier (HEKA Elektronik, Lambrecht, Germany). Borosilicate glass electrodes had a resistance of 2–4 MΩ when filled with the standard high K⁺ solution containing (in mM): 134 KCl, 10 HEPES and 1 EGTA (pH adjusted to 7.2 with KOH). The inside-out membrane patches excised from HEK293 cells were superfused with standard high K⁺ solution without and then with varying concentrations of cGMP in the range between 2 and 500 μM. All patch-clamp recordings were conducted at 24°–26°C. Either square pulses or ramp pulses were used to record macroscopic patch current. The voltage ramp protocol (dV/dt = 0.5 V/s) consisted of three phases: an initial −100 mV descending (hyperpolarizing) phase from a holding potential of 0 mV, a second ascending (depolarizing) phase of 200 mV, and then a third phase returning to the holding potential (see also Fig. 2A ). The current–voltage (I– V) relationship was measured during the second ascending phase and the averages of three to five consecutive I–V relationships are demonstrated in the figures. The activation of hCNGA3, hCNGA3/hCNGB3, hCNGA3/hCNGB3-S435F, and hCNGA3/hCNGB3-D633G channels typically exhibited little if any rundown or runup phenomena during exposure to various concentrations (2–100 μM) of cGMP for a period of at least 20 minutes. In the present study most patch-clamp experiments were completed within approximately 20 minutes after excision to the inside-out mode. 

The concentration of free Ca²⁺ in pipette and bath solutions was calculated using Fabiato and Fabiato’s equations ¹⁸ with a correction by Tsien and Rink. ¹⁹ The free Ca²⁺ concentration in the standard high K⁺ solution was estimated to be approximately 1.0 × 10⁻¹⁰ M (pCa ≈ 10.0). An appropriate amount of Ca²⁺ was added to the standard high K⁺ solution to produce the pipette (extracellular) solution containing low concentration (1.0 μM) of Ca²⁺. The pipette solution containing Ca²⁺ at concentrations of ≥100 μM was made by simply adding Ca²⁺ to the standard high K⁺ solution without EGTA. 

To measure the selectivity for monovalent cations, 134 mM KCl in the standard high K⁺ solution perfusing the cytoplasmic side of the membrane was entirely substituted by an equimolar concentration of NaCl, LiCl, RbCl, or CsCl. In each of these solutions, pH was adjusted to 7.2 with the hydroxide form of the monovalent cation. Reversal potentials (V _rev) of homomeric and heteromeric channels were measured under symmetrical bi-ionic conditions with K⁺ in the pipette, and the permeability of test cation C⁺ relative to potassium (P _C/P _K) was calculated using the Goldman–Hodgkin–Katz equation:  

\[\frac{P_{\mathrm{C}}}{P_{\mathrm{K}}}\ {=}\ \frac{{[}\mathrm{K}^{{+}}{]}_{\mathrm{o}}}{{[}\mathrm{C}^{{+}}{]}_{\mathrm{i}}}\mathrm{exp}\left(\frac{{-}F{\Delta}V_{\mathrm{rev}}}{RT}\right)\]

where ΔV _rev is the difference in the reversal potentials observed when K⁺ on the cytoplasmic membrane surface was entirely substituted with an equimolar concentration of C⁺; [K⁺]_o is the activity of K⁺ at the external side, and [C⁺]_i is the activity of the cation C⁺ at the cytoplasmic side of the membrane; R, T, and F have their usual meanings. Ionic activities were calculated from concentrations using tables of activity coefficient. ²⁰ To test the Cl⁻ permeability, KCl in the standard high K⁺ solution on the cytoplasmic side was partially replaced with an equilimolar concentration of potassium aspartate. 

Current and voltage signals were stored on a digital audiotape using a PCM data recorder (RD-120TE; TEAC, Tokyo, Japan) and later played back through a low-pass 3 kHz filter (48 dB per octave, E-3201A; NF, Tokyo, Japan) for computer analysis. Current records were digitized at a sampling frequency of 1 kHz using a 100 kHz A/D board (ADX-98; Canopus, Kobe, Japan) installed in a personal computer (PC98RL; NEC, Tokyo, Japan) and were then analyzed using in-house programs. 

Averaged data were given as mean values ± SEM. Statistical comparisons were made using Student’s t-test, and differences were considered significant at P < 0.05. 

Immunoprecipitation for Examination of Assembly of Human Cone CNGA3 and CNGB3 Subunits

The hCNGA3 subunit cDNA was cloned into the HindIII-XbaI site of pcDNA3.1/Myc-His(+) plasmid (Invitrogen). The hCNGB3, hCNGB3-S435F, and hCNGB3-D633G subunit cDNAs were cloned into the KpnI site of pFLAG-CMV plasmid (Sigma Chemical Co., St. Louis, MO). They were transfected in combination into HEK293 cells, and after 2 days the cells were lysed in a buffer containing 1% NP-40, 50 mM Tris-Cl (pH 8.0), 150 mM NaCl, 10 μM leupeptin, 1 mM pepstatin, and 1 mM PMSF and were then centrifuged. The supernatant was subjected to immunoprecipitation using the Anti-FLAG M2 affinity gel (Sigma) according to the manufacturer’s protocol. The precipitate was subjected to SDS-PAGE and the separated proteins were blotted to a PVDF membrane (Millipore, Bedford, MO). Anti-Myc monoclonal antibody conjugated with alkaline-phosphatase (Invitrogen) was used for detection of the hCNGA3 subunit. The Prestained SDS-PAGE Standards (broad range, Bio-Rad Laboratory, Hercules, CA) were used as the size marker. 

Results

Activation of hCNGA3 Homomeric and hCNGA3/hCNGB3 Heteromeric Channels by cGMP

The cGMP-dependent activation of homomeric (either hCNGA3 or hCNGB3) and heteromeric hCNGA3/hCNGB3 channels heterologously expressed in HEK293 cells was characterized. After excision to the inside-out mode, square voltage pulses of 200 ms duration were applied from a holding potential of 0 mV to various test potentials between +100 and −100 mV in 10 mV steps. Figure 1A demonstrates a representative example for the activation of the hCNGA3 homomeric channels by cGMP (100 μM) in the bath (cytoplasmic side) solution. In the absence of cGMP (Control), membrane currents during 200 ms test pulses of various amplitudes were of small amplitude and time-independent (Fig. 1A 1a and 1d) , suggesting that activation of endogenous ion channels was practically absent in this HEK293 cell under the present recording conditions. Application of 100 μM cGMP to the bath evoked macroscopic patch current that exhibited little time-dependence during test pulses to various potentials, in most of patches examined (≥90%, Fig. 1A 1b and 1d ). This current activation rapidly returned to baseline after removal of cGMP (data not shown). The hCNGA3 homomeric channel current was determined as cGMP-activated component, which was obtained by digitally subtracting current trace in the absence of cGMP from that in its presence at each test potential. The hCNGA3 homomeric channel current thus isolated exhibited little, if any, time-dependence (Fig. 1A 1c) , an essentially liner I–V relationship and a reversal potential of around 0 mV in the presence of symmetrical concentrations of K⁺ inside and outside the membrane patch (Fig. 1A 1e) . On the other hand, application of 100 μM cGMP consistently failed to evoke any appreciable membrane current in inside-out patches excised from HEK293 cells expressing hCNGB3 alone (n = 30 cells, data not shown), consistent with the previous reports for vertebrate rod and cone CNG channel β subunits. ^{7 8 21}  

The hCNGA3/hCNGB3 heteromeric channel current (Fig. 1B 1c) , determined as cGMP-activated current by similar digital subtraction of corresponding current traces, was largely time-independent, with the exception of small amplitude of current decay observed at the initial part of strong depolarizations (≥ +80 mV). The I–V relationship for the hCNGA3/hCNGB3 heteromeric channel current intersected the voltage axis at around 0 mV (i.e., reversal potential of ∼ 0 mV) and exhibited a practically linear conductance over the test potential range (Fig. 1B 1e) . 

In the experiment represented in Figure 2 , I–V relationships determined using voltage ramp protocol were compared with the current magnitudes at steady state obtained during square pulse protocol at various test potentials (see Fig. 1 ). Figure 2A illustrates macroscopic patch currents recorded in an inside-out membrane patch excised from the HEK293 cell expressing hCNGA3/hCNGB3 heteromers (the same patch as shown in Fig. 1B ) when the voltage ramp protocol (dV/dt = ± 0.5 V/s) was applied (upper row) in the absence and presence of 100 μM cGMP (middle row). The cGMP-activated macroscopic patch current was determined as difference current (lower row) and the current trace associated with the ascending phases of this voltage ramp protocol is illustrated in Figure 2B (as solid line). The steady state magnitude of the current measured near the end of 200 ms square pulse is also plotted (filled triangles) on the same I–V axis. It is clear that the I–V relationships obtained by these two pulse protocols were almost superimposable. Difference currents derived from voltage ramp protocols were used throughout the remainder of the study to measure the hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric currents, since the full I–V relationships could be quickly obtained using voltage ramp protocols. 

Differences in Apparent Affinity for cGMP in hCNGA3 Homomeric and hCNGA3/hCNGB3 Heteromeric Channels

Figures 3A and 3B , respectively, illustrate representative examples for hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels activated by various concentrations (2–500 μM) of cGMP. To quantitatively assess the stimulatory effect of cGMP at each concentration, the slope conductance was measured at around the reversal potential (∼0 mV) by fitting the I– V curve to a straight line by linear regression, then normalized with reference to its maximal value obtained at 100 μM cGMP. Figure 3C illustrates the mean concentration–response relationships for the activation of hCNGA3 homomeric (filled circles) and hCNGA3/hCNGB3 heteromeric (open circles) channels by cGMP. The data were reasonably well described by a Hill equation with the following parameters: K _1/2 of 11.1 ± 1.0 μM and n _H of 1.83 ± 0.19 (n = 13) for hCNGA3; K _1/2 of 26.2 ± 1.9 μM and n _H of 1.82 ± 0.11 (n = 16) for hCNGA3/hCNGB3. The K _1/2 value for hCNGA3 homomeric channels is significantly (P < 0.001) smaller than that for hCNGA3/hCNGB3 heteromeric channels, thus showing that the apparent affinity for cGMP is reduced when hCNGA3 is associated with hCNGB3. Moreover, immunoprecipitation experiments also demonstrated that the Myc-tagged hCNGA3 subunit is associated with the coexpressed FLAG-tagged hCNGB3 subunit (Fig. 7) . 

Effects of Rod Monochromacy-Associated Mutations in hCNGB3 on Apparent Affinity for cGMP in Heteromeric Channels

The cGMP sensitivity of the heteromeric channels composed of hCNGA3 and rod monochromacy-associated mutants of hCNGB3 (hCNGB3-D633G and hCNGB3-S435F) was characterized to clarify the functional consequences of these two mutations in the hCNGB3. Application of cGMP concentration-dependently increased the macroscopic patch currents that exhibited an essentially linear I–V relationship with a reversal potential of ∼ 0 mV in HEK293 cells expressing hCNGA3/hCNGB3-D633G (Fig. 4A) and hCNGA3/hCNGB3-S435F (Fig. 4B) heteromeric channels. Figures 4C and 4D , respectively, represent the concentration–response relationships for the activation of hCNGA3/hCNGB3-D633G and hCNGA3/hCNGB3-S435F heteromeric channels by cGMP. The solid curves through the data points represent least-squares fit of the Hill equation, yielding K _1/2 of 11.9 ± 1.0 μM and n _H of 1.87 ± 0.15 for hCNGA3/hCNGB3-D633G (n = 11; Fig. 4C ), and K _1/2 of 12.0 ± 2.1 μM and n _H of 1.85 ± 0.12 for hCNGA3/hCNGB3-S435F (n = 10; Fig. 4D ). The K _1/2 value for the activation of both hCNGA3/hCNGB3-D633G and hCNGA3/hCNGB3-S435F is thus similar to that for hCNGA3 homomers (11.1 ± 1.0 μM) but is significantly smaller than that for hCNGA3/hCNGB3 heteromers (26.2 ± 1.9 μM, P < 0.001). Thus, the apparent affinity for cGMP is not appreciably affected by association of the hCNGA3 subunit with these rod monochromacy-associated mutants of hCNGB3 subunits. Immunoprecipitation experiments showed that both mutants of hCNGB3 (hCNGB3-D633G and hCNGB3-S435F) were associated with the hCNGA3 subunit (Fig. 7) . It should also be noted that neither mutant of the hCNGB3 subunit appears to form functional homomeric channels when expressed alone, since no appreciable current was evoked by addition of 100 μM cGMP. 

Selectivity of Homomeric and Heteromeric Channels for Monovalent Cations

The permeability of homomeric and heteromeric channels for monovalent cations was examined. The permeability for the test cations relative to K⁺ was determined by measuring the shift in reversal potentials when K⁺ on the cytoplasmic side of the membrane was entirely substituted by an equimolar concentration of Na⁺, Li⁺, Rb⁺, and Cs⁺. Figure 5 illustrates representative I–V relationships for hCNGA3 (A), hCNGA3/hCNGB3 (B), hCNGA3/hCNGB3-D633G (C), and hCNGA3/hCNGB3-S435F (D) channels activated by 100 μM cGMP under the various bi-ionic solutions. The relative permeability ratios P _C/P _K, derived from the measured reversal potential using the Goldman–Hodgkin–Katz equation (see Methods), are given in Table 1 . The relative conductance ratios G _C/G _K, obtained from the measurement of outward current at +80 mV in bi-ionic conditions, are also listed in Table 1 . The relative permeability and conductance ratios of hCNGA3/hCNGB3 heteromeric channels are largely similar if not identical to those of hCNGA3 homomeric channels. In addition, these parameters of heteromeric channels formed with hCNGA3 and either of rod monochromacy-associated mutants of hCNGB3 are also basically comparable to those of wild-type heteromeric channels. 

When the Cl⁻ concentration in the medium perfusing the cytoplasmic membrane surface was reduced from 134 mM to 67 or 34 mM by replacing KCl by the same concentration of potassium aspartate, the reversal potentials for homomeric (hCNGA3) and heteromeric (hCNGA3/hCNGB3, hCNGA3/hCNGB3-D633G and hCNGA3/hCNGB3-S435F) channels were not appreciably affected (data not shown). This result indicates that these homomeric and heteromeric channels are virtually impermeable to Cl⁻ ions. 

Inhibition of Homomeric and Heteromeric CNG Channels by Extracellular Ca²⁺

It has been suggested that extracellular Ca²⁺ electrostatically binds to negatively charged residues (e.g., glutamate at position 340 in bovine retinal rods) at pore region of the CNG channel α-subunit and thereby inhibits the flux of monovalent cations through the channel. ^{22 23 24 25} The inhibitory effects of extracellular Ca²⁺ on hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels activated by 100 μM cGMP were assessed. Macroscopic patch currents were recorded in an inside-out configuration with pipette solutions containing various concentrations of Ca²⁺. Figures 6A and 6B , respectively, illustrate I–V relationships for hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels in the presence of Ca²⁺ at concentrations of 1 × 10^-10 M (filled circle), 1 × 10⁻⁶ M (open square), 1 × 10⁻⁴ M (filled square), and 1.8 × 10⁻³ M (open circle). The I–V relationship for hCNGA3 homomeric channels is essentially linear in the presence of extremely low concentrations of Ca²⁺ (1 × 10⁻¹⁰ M) but exhibits a marked outward rectification with Ca²⁺ at concentrations of 1 × 10⁻⁶ M or more (Fig. 6A) . This characteristic I–V relationship is assumed to arise from voltage dependent blockade of channel currents by extracellular Ca²⁺; inward current at negative potentials is more sensitive to inhibition compared with outward current at positive potentials. ^{24 25} On the other hand, the I–V relationship for hCNGA3/hCNGB3 heteromeric channels with high extracellular Ca²⁺ concentrations is also outwardly rectified but exhibits some relaxations at potentials of ≤ ∼ −50 mV (Fig. 6B) , which has been ascribed to relief of Ca²⁺ blockade at these strongly hyperpolarized potentials. ⁸ The degree of Ca²⁺ blockade was assessed by normalizing the current amplitude at −80 mV with reference to that at +80 mV in I–V relationships, recorded in the presence of 1.8 mM Ca²⁺. The current ratio (expressed as %) averages 5.2 ± 2.0% (n = 4) for hCNGA3 and 29.7 ± 3.1% for hCNGA3/hCNGB3 (n = 4, P < 0.01, Fig. 6E ). 

The sensitivity of heteromeric channels produced by coexpression of hCNGA3 with hCNGB3 mutants (S435F and D633G) to blockade by extracellular Ca²⁺ was tested in a similar way. The current ratios at +80 and −80 mV recorded in the presence of 1.8 mM Ca²⁺ for hCNGA3/hCNGB3-S435F (Fig. 6C) and hCNGA3/hCNGB3-D633G (Fig. 6D) heteromers were 25.1 ± 3.6% (n = 7) and 28.3 ± 6.1% (n = 8), respectively (Fig. 6E) . Thus, when coexpressed with hCNGA3, both of these mutants of hCNGB3 produced the heteromeric channels that exhibit the sensitivity to extracellular Ca²⁺ similar to wild-type hCNGA3/hCNGB3 heteromeric channels. The association of these mutants of hCNGB3 with wild-type hCNGA3 was confirmed by immunoprecipitation experiments (Fig. 7) . 

Discussion

The present study characterized the ligand (cGMP) sensitivity, ion selectivity, and Ca²⁺ blockade in hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels heterologously expressed in HEK293 cells. Association of hCNGA3 with hCNGB3 subunits significantly reduced an apparent affinity for cGMP (Fig. 3) and attenuated the Ca²⁺ blockade of inward current at negative membrane potentials (Fig. 6) but produced small effects on selectivity for monovalent cations (Fig. 5 and Table 1 ). The present study further characterized the functional consequences of the rod monochromacy-associated mutations in hCNGB3 (S435F and D633G). Coexpression of either of rod-monochromacy-associated mutants of hCNGB3 with hCNGA3 produced functional heteromeric channels with a higher apparent affinity for cGMP (Fig. 4) compared with wild-type heteromeric channels. The selectivity for monovalent cations (Fig. 5) and sensitivity to inhibition by extracellular Ca²⁺ (Fig. 6) in these two mutant heteromeric channels were largely similar to those in wild-type heteromeric channels. 

It has previously been demonstrated that coexpression of bovine rod CNGA1 and CNGB1 subunits in Xenopus oocytes results in a reduction in the apparent affinity for cGMP. ^{3 4} The present result also confirmed that, as previously reported, ²⁶ the apparent affinity for cGMP in hCNGA3/hCNGB3 heteromeric channels is significantly lower than that for hCNGA3 homomeric channels (Fig. 3) . Since cGMP is the physiological ligand for rod and cone photoreceptor CNG channels, this raises the possibility that the functional role of β-subunits is to substantially reduce the affinity for cGMP and thereby to allow the channels to be more susceptible to changes in intracellular cGMP levels after reception of photons. However, the heteromeric channels composed of hCNGA3 and rod monochromacy-associated mutants of hCNGB3 (hCNGB3-D633G and hCNGB3-S435F) exhibited a higher apparent affinity for cGMP compared with wild-type heteromeric channels (Fig. 4) . This functional alteration in cone CNG channels associated with D633G or S435F mutation in hCNGB3 should cause the channels to be reluctant to the decrease in intracellular cGMP level after reception of photons, leading to their being always open, which should be responsible at least partly for the pathogenesis of rod monochromacy. Peng et al. recently reported a similar increase in the apparent affinity for cGMP in the rod monochromacy-associated S435F mutation in hCNGB3 subunits. ²⁶ Serine at position 435 resides in the transmembrane segment S6 (indicated by 1, Fig. 8 ) and is evolutionally conserved within the β-subunits but not in the α-subunits. It has been suggested that amino acid residue with smaller side chain at position 435, five amino acids COOH terminus to the gating hinge (conserved as glycine, marked with * in Fig. 8 ), is conserved as alanine, glycine, or serine ²⁷ and is required to attain the stable closed conformation of the channels. It is likely that phenylalanine with bulky aromatic side chain at this position destabilizes the closed state of the channel and is thereby responsible for the increase in the cGMP sensitivity. ²⁶ When coexpressed with hCNGA3, the S435A mutant of hCNGB3 subunit produceed functional heteromeric channels with the apparent affinity for cGMP of 19.6 ± 3.0 μM (n = 3, data not shown, authors’ unpublished observations, 2003), a value relatively similar to that for wild-type hCNGA3/hCNGB3 heteromeric channels (26.2 ± 1.9 μM, n = 16). 

The present study also detected an increase in the apparent affinity for cGMP in heteromeric cone CNG channels containing the rod monochromacy-associated D633G mutation in hCNGB3 subunits (Fig. 4C) . Moreover, the apparent affinity for cGMP (24.8±1.7 μM, n = 4) in hCNGA3/hCNGB3-D633T heteromeric channels was comparable to that for wild-type hCNGA3/hCNGB3 heteromeric channels (data not shown, authors’ unpublished observations, 2003). Since both aspartate and threonine, but not glycine, have polar side chains, it seems likely that the polarity of amino acid in the cGMP-binding domain (indicated by 2, Fig. 8 ) is important for decreasing the sensitivity to cGMP. 

The present investigation confirmed that I– V relationships for both hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels are essentially linear with extracellular Ca²⁺ of extremely low levels (∼10⁻¹⁰ M) but exhibit a marked outward rectification by increasing the concentrations to micromolar levels or more (Fig. 6) . In addition, the degree of outward rectification evoked by millimolar concentration (1.8 mM) of Ca²⁺, assessed by comparing the current amplitudes at −80 and +80 mV, was more pronounced in hCNGA3 homomeric channels than in hCNGA3/hCNGB3 heteromeric channels (Fig. 5) . An extensive site-directed mutagenesis study has demonstrated that negatively charged residue glutamate in the P (pore) loop, which is highly conserved among CNGA1–CNGA3 subunits (but not in CNGB subunits), constitutes a cation-binding site important for blockade by external Ca²⁺. ^{22 23 24 25} It is therefore reasonable to assume that the high-affinity Ca²⁺ binding-site is produced by a set of four glutamate residues in homomeric hCNGA3 channels, leading to a marked inhibition of the channel current especially at negative membrane potentials, which may account for the difference in the current ratios (at −80 and +80 mV) observed between hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels (Fig. 6) . Furthermore, the present result demonstrated that rod monochromacy-associated S435F and D633G mutations in hCNGB3 subunit do not largely affect the sensitivity to block by extracellular Ca²⁺ in heteromeric channels (Fig. 6) . It is likely that serine and aspartate at positions 435 and 633, respectively, in the hCNGB3 subunits are functionally independent of Ca²⁺ binding to the negatively charged residue in the pore region. 

In summary, the present study provided the functional evidence to suggest that coexpression of hCNGA3 with rod monochromacy-associated mutants of hCNGB3 (hCNGB3-S435F and hCNGB3-D633G) results in a significant increase in the apparent affinity for cGMP, which should alter cone function and thereby contribute at least partly to the pathogenesis of the disease. 

 Supported in part by Grants-in-Aid for Scientific Research from Japan Society of the Promotion of Science (14571667 to SY and 14770953 to SO).

 Submitted for publication October 2, 2003; revised January 1, 2004; accepted January 12, 2004.

 Disclosure: A. Okada, None; H. Ueyama, None; F. Toyoda, None; S. Oda, None; W.-G. Ding, None; S. Tanabe, None; S. Yamade, None; H. Matsuura, None; I. Ohkubo, None; K. Kani, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 2.

 

Measurement of hCNGA3/hCNGB3 heteromeric channel current using square pulse and voltage ramp protocols. (A) Voltage ramp protocol at dV/dt of ± 0.5 V/s (upper row). Macroscopic patch currents during the voltage ramp protocol in the absence (control) and presence of 100 μM cGMP (middle row). Difference current obtained by digital subtraction of current trace in the absence of cGMP from that in its presence shown in middle row (lower row). These data were obtained from the inside-out membrane patch represented in Figure 1B . (B) I–V relationship determined during the ascending portion of voltage ramp protocol (from −100 to +100 mV, solid line) superimposed with the current level at each test potential (from −100 to +100 mV in 10 mV steps) determined by square pulse protocols (identical with the data demonstrated in Fig. 1B 1e ).

Figure 1.

 

Inside-out macropatch recordings from HEK293 cells expressing hCNGA3 homomeric (A) and hCNGA3/hCNGB3 heteromeric (B) channels. (a and b) Macroscopic patch currents evoked by 200 ms voltage-clamp pulses to potential levels between +100 and −100 mV in 10 mV steps applied from a holding potential of 0 mV in the absence (a) and presence (b) of 100 μM cGMP. A schematic diagram of the voltage-clamp protocol is given above the control traces in A. (c) cGMP-activated patch currents obtained by digital subtraction of the current traces recorded in the absence of cGMP from those in its presence. (d) I–V relationships for macroscopic patch currents recorded in the absence (open circles) and presence (filled circles) of cGMP (100 μM), shown in a and b, respectively. (e) I–V relationships for cGMP-activated macroscopic patch currents shown in c. The current level was measured near the end of 200 ms clamp steps (d, e).

Figure 3.

 

Concentration-dependent activation of hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels by intracellular cGMP. (A, B) Macroscopic patch current activated by various concentrations (2, 5, 10, 50, 100, and 500 μM) of cGMP in HEK293 cells expressing hCNGA3 homomeric (A) and hCNGA3/hCNGB3 heteromeric (B) channels, obtained by digital subtraction of current traces in the absence of cGMP from that in its presence. (C) Concentration–response relationships for the activation of hCNGA3 homomeric (filled circles) and hCNGA3/hCNGB3 heteromeric (open circles) channels. Slope conductance of macroscopic patch current activated by each concentration of cGMP was measured near the reversal potential (∼ 0 mV) and then normalized with reference to the maximum response elicited by a saturating concentration (100 μM) of cGMP. Data points represent the means ± SEM. The curves were drawn by a least-squares fit of the Hill equation, I = 1/(1 + (K _1/2/[cGMP])^nH), where I is the slope conductance of macroscopic current activated by each concentration of cGMP normalized with reference to that at 100 μM cGMP, K _1/2 is the concentration of cGMP causing a half-maximal activation, and n _H is the Hill coefficient.

Figure 7.

 

The Myc-tagged hCNGA3 subunit associated with the coexpressed FLAG-tagged hCNGB3, hCNGB3-S435F, and hCNGB3-D633G subunits.

Figure 4.

 

Effect of two mutations in hCNGB3 associated with rod monochromacy on the apparent affinity for cGMP in heteromeric CNG channels. Macropatch currents activated by various concentrations of cGMP, recorded from HEK293 cells expressing hCNGA3/hCNGB3-D633G (A) and hCNGA3/hCNGB3-S435F (B) heteromeric channels. Concentration–response relationships for the activation of hCNGA3/hCNGB3-D633G (C, filled squares) and hCNGA3/hCNGB3-S435F (D, open squares) heteromeric channels. The data points represent the slope conductance of macroscopic current activated by each concentration of cGMP at around 0 mV normalized with reference to the maximal value evoked by 100 μM cGMP. The solid line shows a least-squares fit of the Hill equation. For reference, concentration–response relationships for hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels (see Fig. 3C ) are respectively represented by dotted and dashed curves.

Figure 5.

 

Selectivity of homomeric and heteromeric CNG channels for monovalent cations. I–V relationships for macroscopic patch currents activated by 100 μM cGMP under symmetrical bi-ionic conditions with K⁺ on the extracellular membrane surface (in the pipette) and different monovalent cations on the cytoplasmic membrane surface (in the bath). Panels A, B, C, and D represent the current traces recorded from hCNGA3, hCNGA3/hCNGB3, hCNGA3/hCNGB3-D633G, and hCNGA3/hCNGB3-S435F channels, respectively.

Table 1.

 

Relative Permeability and Conductance Ratios of Wild-Type and Mutant Channels for Monovalent Cations

	Relative Permeability Ratio (P C/P K)									Relative Conductance Ratio (G C/G K)
		K+	Na+	Li+	Rb+	Cs+	K+	Na+	Li+	Rb+	Cs+
hCNGA3	1.0	0.96 ± 0.01	0.76 ± 0.02	0.84 ± 0.02	0.51 ± 0.03	1.0	1.70 ± 0.06	0.57 ± 0.01	0.38 ± 0.02	0.27 ± 0.01
hCNGA3/hCNGB3	1.0	1.00 ± 0.01	0.88 ± 0.03	0.83 ± 0.02	0.61 ± 0.05	1.0	1.12 ± 0.06	0.57 ± 0.01	0.58 ± 0.02	0.40 ± 0.03
hCNGA3/hCNGB3-D633G	1.0	0.97 ± 0.01	0.75 ± 0.08	0.87 ± 0.04	0.51 ± 0.11	1.0	1.43 ± 0.09	0.56 ± 0.01	0.51 ± 0.04	0.28 ± 0.06
hCNGA3/hCNGB3-S435F	1.0	0.98 ± 0.02	0.77 ± 0.04	0.77 ± 0.03	0.46 ± 0.03	1.0	1.84 ± 0.04	0.60 ± 0.08	0.36 ± 0.04	0.23 ± 0.06

 Data are mean values ± SEM from 4–8 patches.

Figure 6.

 

Blocking effects of extracellular Ca²⁺ on hCNGA3 homomeric and hCNGA3/hCNGB3 heteromeric channels. Macroscopic patch currents were recorded in an inside-out configuration with pipette solutions (extracellular side) containing various concentrations of Ca²⁺. I–V relationship obtained at each test concentration of Ca²⁺ was normalized with reference to the current amplitude at +100 mV before averaging. (A, B) I–V relationships for hCNGA3 homomeric (A) and hCNGA3/hCNGB3 heteromeric channels (B) in the presence of Ca²⁺ at concentrations of 1 × 10⁻¹⁰ (filled circle), 1 × 10⁻⁶ (open square), 1 × 10⁻⁴ (filled square) and 1.8 × 10⁻³ M (open circle). (C, D) I–V relationships for hCNGA3/hCNGB3-S435F (C) and hCNGA3/hCNGB3-D633G (D) heteromeric channels in the presence of extracellular Ca²⁺ at concentrations of 1 × 10⁻¹⁰ (filled circle) and 1.8 × 10⁻³ M (open circle). (E) Current ratios obtained by normalizing the amplitude at −80 mV with reference to that at +80 mV for hCNGA3, hCNGA3/hCNGB3, hCNGA3/hCNGB3-S435F, and hCNGA3/hCNGB3-D633G channels. *P < 0.05 and **P < 0.01 when current ratio for hCNGA3/hCNGB3, hCNGA3/hCNGB3-S435F, or hCNGA3/hCNGB3-D633G is compared with that for hCNGA3. Measurements were conducted in 4–8 cells in each channel.

Figure 8.

 

Comparison of α- and β-subunits at primary structure at sixth transmembrane domain (S6) and cGMP-binding domain. The data are adopted from the following references: human rod CNG α, ²⁸ bovine cone CNG α, ^{29 30} chicken cone CNG α, ³¹ human cone CNG α, ^{5 32} murine cone CNG β, ²¹ canine cone CNG β, ³³ and human cone CNG β ⁶ subunits. * denotes amino acid at the gating hinge. ²⁷

The authors thank Masashi Suzaki at Shiga University of Medical Science (Central Research Laboratory) for his technical help.