Research ArticleBiophysics

Differential Regulation by Cyclic Nucleotides of the CNGA4 and CNGB1b Subunits in Olfactory Cyclic Nucleotide–Gated Channels

Science Signaling  10 Jul 2012:
Vol. 5, Issue 232, pp. ra48
DOI: 10.1126/scisignal.2003110


Olfactory cyclic nucleotide–gated (CNG) ion channels are essential contributors to signal transduction of olfactory sensory neurons. The activity of the channels is controlled by the cyclic nucleotides guanosine 3′,5′-monophosphate (cGMP) and adenosine 3′,5′-monophosphate (cAMP). The olfactory CNG channels are composed of two CNGA2 subunits, one CNGA4 and one CNGB1b subunit, each containing a cyclic nucleotide–binding domain. Using patch-clamp fluorometry, we measured ligand binding and channel activation simultaneously and showed that cGMP activated olfactory CNG channels not only by binding to the two CNGA2 subunits but also by binding to the CNGA4 subunit. In a channel in which the CNGA2 subunits were compromised for ligand binding, cGMP binding to CNGA4 was sufficient to partly activate the channel. In contrast, in heterotetrameric channels, the CNGB1b subunit did not bind cGMP, but channels with this subunit showed activation by cAMP. Thus, the modulatory subunits participate actively in translating ligand binding to activation of heterotetrameric olfactory CNG channels and enable the channels to differentiate between cyclic nucleotides.


Cyclic nucleotide–gated (CNG) ion channels play essential roles in the signal transduction of the olfactory and visual system (14). The channels are activated when the concentration of the intracellular second messenger adenosine 3′,5′-monophosphate (cAMP) or guanosine 3′,5′-monophosphate (cGMP) is increased to the micromolar range, a process that is triggered by the specific signal transduction cascades within the receptor cells.

Architecturally, CNG channels are composed of four subunits that assemble around a centrally located pore (57). Structural data from related channels, including data of the binding domain of a bacterial CNG channel (8), of the C-terminal part of the hyperpolarization-activated cyclic nucleotide–gated pacemaker channel HCN2 (9), of a bacterial voltage-gated K+ channel (10), and of a eukaryotic voltage-gated K+ channel (11), support a tetrameric structure for CNG channels. Seven different genes encoding the CNG channel subunits give rise to cell-specific heterotetramers: CNGA1 and CNGB1a in rod photoreceptor channels (1214), CNGA3 and CNGB3 in cone photoreceptor channels (15), and CNGA2, CNGA4, and CNGB1b in olfactory sensory neuron channels (16). All seven subunit types are homologous to each other, containing six transmembrane helices, a pore region between the fifth and the sixth helix, and the intracellularly located N and C termini (1720). In all subunit types, the C terminus harbors the cyclic nucleotide–binding domain (CNBD). When expressed alone in a heterologous expression system, only three of the seven subunits formed functional channels [CNGA1 (17), CNGA2 (21), and CNGA3 (22)] in the plasma membrane. The properties of these homotetrameric channels, however, differ from those of the respective wild-type channels in the receptor cells. All other subunits, which do not form functional channels as homotetramers, are generally termed “modulatory” subunits.

In olfactory CNG channels, various modulatory effects of the CNGA4 and CNGB1b subunits have been identified by heterologous expression experiments. Both subunits make olfactory channels more sensitive to cAMP; thus, the concentration-activation relationship for cAMP is significantly shifted to lower concentrations (4, 23). At the single-channel level, in the presence of saturating ligand concentration, coexpression of CNGA2 with CNGA4 generates unresolved brief openings (4, 2426) (too short for accurate measurement), and coexpression of CNGA2 with CNGB1b leads to a channel with opening characteristics that are similar to those of the CNGA2 homotetrameric channels, except that the channel exhibits more flickery gating (4). Coexpression of all three subunits generates currents with single-channel properties resembling those of wild-type channels. Additional functional roles of the CNGA4 and CNGB1b subunits have been identified for Ca2+-dependent desensitization (27, 28) of the olfactory channel, of which the CNGB1b subunit seems to have a larger effect (2931).

Despite the growing insight into the function of the CNGA4 and CNGB1b subunits, it remains unclear how these subunits contribute to channel activation. In particular, it is not known whether cyclic nucleotides bind to the CNBDs of these subunits in heterotetrameric olfactory channels. For CNGA4, evidence has been presented that its CNBD can bind cAMP: CNGA2 with the binding domain from CNGA4 formed functional homotetrameric chimeric channels (32, 33); however, a contradictory negative result was also reported (34). Assuming that the CNGA4 binding domain functions in this CNGA2 chimeric subunit, these experiments cannot unequivocally prove that the CNBD of CNGA4 binds cyclic nucleotides when in CNGA4. For the CNBD of CNGB1b, it is unknown whether this domain binds cyclic nucleotides. In a heterotetrameric olfactory channel, not only is it not clear what the affinities of the four binding sites in the channel are, but also how many ligands are required to activate the channel.

For homotetrameric CNGA2 channels, binding of only two ligands is necessary for full channel activation, and binding of two additional ligands only stabilizes the open channel (35). The activation gating was fully described by a sequence of four binding steps, with both positive and negative cooperativity with respect to the equilibrium association constants [C4L-model (35)]. Therefore, it is possible that heterotetrameric olfactory channels, containing two CNGA2 subunits, require ligand binding to only these two CNGA2 subunits to activate the channel and that the other two subunits modulate the channel activity independently of cyclic nucleotide binding. In contrast, Waldeck and co-workers (30) proposed that all subunits of the heterotetrameric channel bind cAMP and contribute to channel activation or stabilization of the open state similar to the C4L-model derived for homotetrameric channels (35).

We addressed the question of whether, like CNGA2, the CNGA4 and CNGB1b subunits are activated by cyclic nucleotides. We analyzed ligand binding in combination with activation gating in channels with subunits modified at their CNBDs. Our results suggest that in heterotetrameric olfactory channels, cGMP binding to the CNGA2 subunits and the CNGA4 subunit, but not to the CNGB1b subunit, is sufficient to fully activate the channel. In contrast, cAMP binds to all three types of subunits, leading to channel activation. This suggests an important role of the CNGB1b subunit in ligand discrimination.


CNGA2 subunits promote the incorporation of CNGA4 and CNGB1b subunits into the plasma membrane

Neither CNGA4 nor CNGB1b subunits alone produce functional channels (4, 2325, 36). One possible explanation for this observation is that these subunits are not incorporated in the plasma membrane. For CNGA4, previous results were controversial, with some reporting effective trafficking to the membrane (26, 37) and some not (16). Alternatively, CNGA4 and CNGB1b might be incorporated in the membrane but cannot form functional channels on their own.

We labeled the CNGA4 and CNGB1b subunits by fusing teal fluorescent protein (TFP) (38) to the C terminus and analyzed the fluorescence profile perpendicular to the membrane in Xenopus oocytes expressing various combinations of channel subunits (Fig. 1). The membrane was counterstained from the extracellular side by fluorescently labeled lectin, Alexa Fluor 633 WGA (wheat germ agglutinin; see Materials and Methods). We analyzed the linear unmixed fluorescence images (Fig. 1A) to quantify the fluorescence profiles of the membrane and the tagged subunit (Fig. 1B) and to determine the relative colocalization (Fig. 1C) for each subunit or combination of subunits. Staining of the membrane alone produced a profile with a half-maximum width of ~2 μm, which is much larger than the microscope resolution (0.3 μm) and also much larger than a typical membrane thickness of ~3 nm. This is likely due to the extensive folding that is characteristic of the plasma membrane of Xenopus oocytes (39). Nevertheless, such a profile can be attributed to the plasma membrane, and the amount of overlap in the WGA and TFP profiles represented a quantitative measure of incorporation of the labeled subunits in the plasma membrane (Fig. 1C). To directly compare the TFP and WGA profiles, we normalized the areas under the profiles to unity.

Fig. 1

Control of CNGA4 and CNGB1b incorporation in the plasma membrane. (A) Representative images showing the relative localization of the oocyte plasma membrane (red; labeled lectin Alexa Fluor 633 WGA) and the indicated combinations of expressed subunits in which either the CNGA4 or the CNGB1b subunit was labeled with TFP at the C terminus (see also fig. S1). Autofluorescence contributions were removed by spectral unmixing. (B) Averaged fluorescence profiles (10 to 85 profiles per oocyte, 5 to 10 oocytes for each construct) perpendicular to the plasma membrane. Normalization of the profiles was performed by setting the area under the curves to unity. The teal profiles were obtained from either CNGA4-TFP or CNGB1b-TFP; the red profiles indicate the membrane marker, lectin Alexa Fluor 633 WGA. The shaded areas mark the portion of colocalization. (C) Bar graphs showing the percentage of the TFP-labeled subunits colocalizing with the membrane. For each subunit combination, 5 to 10 measurements were performed. The error bars indicate SEM. *P < 0.05, significant differences based on t test.

When expressed alone, CNGA4-TFP and CNGB1b-TFP showed ~30% colocalization with WGA (Fig. 1C), indicating that at least 70% of the subunits were still intracellular. When CNGA2 and CNGA4-TFP (CNGA2:A4-TFP) were coexpressed, colocalization with WGA increased to ~60%, indicating that the CNGA2 subunits promoted the CNGA4-TFP incorporation into the membrane. Coexpression of CNGA2 with CNGB1b-TFP (CNGA2:B1b-TFP) increased the colocalization with WGA to ~40%, indicating that CNGA2 also aids in incorporating CNGB1b-TFP into the membrane but with a lower efficiency compared to the effect on CNGA4-TFP. However, when CNGB1b-TFP was expressed with CNGA2 and CNGA4 (CNGA2:A4:B1b-TFP), the overlap with WGA was ~54%, representing a degree of incorporation similar to that observed for CNGA2:A4-TFP. Together, these results suggest that CNGA2 effectively promoted CNGA4 incorporation in the plasma membrane, whereas CNGA2 and CNGA4 were required for maximum incorporation of CNGB1b in the oocyte plasma membrane. Notably, although when expressed alone, CNGA4-TFP and CNGB1b-TFP colocalized the least with WGA, at low magnification, each of these produced a ring-shaped confocal image similar to that of CNGA2:A4:B1b-TFP (fig. S1A). Thus, trafficking of the CNGA4-TFP and CNGB1b-TFP subunits through the cell interior to beneath the plasma membrane appeared to occur also in the absence of CNGA2 subunits, and CNGA2 enhanced the incorporation into the plasma membrane.

The CNBDs of CNGA4 and CNGB1b impart an increased apparent affinity for cAMP in heteromultimeric channels

Coexpression of CNGA2 with either CNGA4 or CNGB1b (CNGA2:A4, CNGA2:B1b, RNA ratio 3:1) (16) caused a pronounced shift of the concentration-activation relationship for cAMP to lower concentrations compared to homotetrameric CNGA2 channels (Fig. 2A) (4, 23, 27), which corresponded to a lower median effective concentration (EC50), representing the ligand concentration producing half-maximal current and calculated with Eq. 1 in Materials and Methods. Coexpression of CNGA2 with both CNGA4 and CNGB1b (CNGA2:A4:B1b; RNA ratio 2:1:1) (16) produced an even stronger shift of the concentration-activation relationship for cAMP. The Hill coefficient H (calculated with Eq. 1) was similar in CNGA2 (H = 2.0), CNGA2:A4 (H = 1.9), and CNGA2:A4:B1b (H = 1.9) channels, suggesting that as long as the channel contains at least three CNGA2 subunits or two CNGA2 subunits plus one CNGA4 subunit, the net positive cooperativity was similar. When the channels consisted only of CNGA2 and CNGB1b subunits, H was somewhat smaller (H = 1.6). The shift of the concentration-activation relationships along the concentration axis by the presence of either CNGA4 or CNGB1b along the concentration axis suggested a large effect on the CNGA2 subunits.

Fig. 2

Concentration-activation relationships in heterotetramers with cAMP. Each data point represents the mean of measurements in 5 to 12 oocytes. The continuous curves are best fits with Eq. 1. The cartoons symbolize the four subunits (square), their cyclic nucleotide binding site (circle), and the respective ligand (gray circle). The colors are as follows: CNGA2, yellow; CNGA4, blue; and CNGB1b, orange. The voltage was set to +100 mV. (A) Effect of cAMP on wild-type channels. The calculated values of the curves are as follows: CNGA2 (red curve): EC50 = 49.1 μM cAMP, H = 2.0; CNGA2:B1b (orange curve): EC50 = 21.1 μM cAMP, H = 1.6; CNGA2:A4 (blue curve): EC50 = 11.1 μM cAMP, H = 1.9; and CNGA2:A4:B1b (black curve): EC50 = 4.6 μM cAMP, H = 1.9. (B) Effect of the CNGA4 subunit with defective CNBD on the concentration-activation relationship (n = 5 to 11 measurements obtained from at least four oocytes). The curves for CNGA2 (red squares) and CNGA2:A4 (blue squares) correspond to the respective curves in (A). The continuous curve is the best fit for CNGA2:A4(R430E) with Eq. 2 (blue circles). The parameters are as follows: EC50,h = 35.2 μM cAMP, Hh = 1.9; EC50,l = 656.9 μM cAMP, Hl = 1.5, a = 0.43. In the cartoon, the red X in the CNBD symbolizes disabled binding. (C) Effect of the CNGB1b subunit with defective CNBD on the concentration-activation relationship (n = 5 to 14). The curves for CNGA2 (red squares) and CNGA2:B1b (orange squares) correspond to the respective curves in (A). The continuous curve is the best fit for CNGA2:B1b(R657E) with Eq. 1 (orange circles). The parameters are EC50 = 192.0 μM cAMP and H = 1.2.

The small errors associated with the measurements argue for specific formation of the respective heterotetrameric channels according to the injected RNA ratios (16). Moreover, similar to a previous report (4, 40), single-channel currents of channels with different subunit combinations (fig. S2) showed the characteristic properties for each corresponding heterotetrameric channel (fig. S2), which further confirms a consistent channel composition.

In principle, the increased apparent affinity for cAMP in the heterotetramers, compared to the homotetramers, could result from cAMP binding to the CNBDs of CNGA4 and CNGB1b and evoking a conformational change in these subunits or from allosteric effects of the CNGA4 and CNGB1b subunits on the binding of cAMP to the CNGA2 subunits. To distinguish between these two possibilities, we replaced a specific arginine with a glutamic acid, CNGA4(R430E) (24, 25) and CNGB1b(R657E) (4, 23), in the binding sites of CNGA4 and CNGB1b, respectively. In CNGA1 channels, this mutation dramatically decreases the apparent affinity for cGMP (41).

CNGA2:A4(R430E) channels produced a concentration-activation relationship shifted significantly to higher concentrations compared to that of CNGA2:A4, and two well-resolved components could be identified (Fig. 2B). The data could not be fitted with a single Hill equation (Eq. 1), but required the sum of two Hill equations (Eq. 2). For the high-affinity component, the EC50,h value of the CNGA2:A4(R430E) channels for cAMP was 35.2 μM, which was similar to the EC50 value of CNGA2 channels (49.1 μM), suggesting that the high-affinity component was caused by the CNGA2 subunits and was relatively independent of the compromised CNGA4 subunit. For the low-affinity component, the EC50,l was 656.9 μM, suggesting that Arg430 in the CNBD of the CNGA4 subunit was involved in ligand binding and control of channel activation. The robust effect of this single point mutation in the CNBD of CNGA4 suggested that the binding of cAMP to CNGA4 contributes to the effect of CNGA4 on channel activation. However, it did not rule out a role for the CNGB1b subunit in CNGA2:A4:B1b channels.

CNGA2:B1b(R657E) channels also showed a lower apparent affinity (EC50 = 192 μM) compared to CNGA2:B1b channels (EC50 = 21.1 μM) and CNGA2 homotetrameric channels (EC50 = 49.1 μM), suggesting that cAMP bound to the CNGB1b subunit (Fig. 2C). In addition, the Hill coefficient, which was already smaller for CNGA2:B1b than for CNGA2 channels (Fig. 2A), was further reduced for CNGA2:B1b(R657E) compared to CNGA2:B1b channels (the mutated channel had a Hill coefficient of 1.2 and the wild-type heterotetramer had a coefficient of 1.6). This suggested that in heterotetrameric channels both the CNGA2 subunits and the CNGB1b subunit bound cAMP and that this binding may promote positive cooperativity among the subunits.

In heterotetrameric channels, only CNGA4, and not CNGB1b, alters channel activity in response to cGMP

Olfactory heterotetrameric channels respond to both cAMP and cGMP (1). Therefore, to further unravel the action of cyclic nucleotides on heterotetrameric channels, we evaluated the effect of cGMP on channels of various subunit compositions (Fig. 3). The low concentrations of cGMP required to activate the channels (Fig. 3A) allowed us to study channel activation and ligand binding in parallel by means of confocal patch-clamp fluorometry with a fluorescence-labeled cGMP (fcGMP) (35).

Fig. 3

Effect of cGMP and fcGMP on heterotetrameric olfactory channels. (A) Concentration-activation relationships for the wild-type heteromultimers measured at +100 mV. The calculated values are as follows: CNGA2:B1b (orange): EC50 = 1.9 μM cGMP, H = 2.0; CNGA2:A4 (blue): EC50 = 1.28 μM cGMP, H = 1.73; CNGA2:A4:B1b (black): EC50 = 0.76 μM cGMP, H = 2.0; and CNGA2 (red): EC50 = 1.74 μM cGMP, H = 1.9. The curves for CNGA2 and CNGA2:B1b are indistinguishable. (B) Concentration-activation relationships of the CNGA2:A4:B1b channels activated by either cGMP (black) or fcGMP (green). The currents were recorded at +10 mV. Data points, obtained from 5 to 12 patches, were fitted with Eq. 1. The calculated values are EC50 = 1.3 μM, H = 1.7 for cGMP and EC50 = 1.8 μM, H = 1.6 for fcGMP. (C) Concentration-activation relationship for CNGA2(R538E) with cGMP. The data points, means of five to nine experiments, were fitted with Eq. 1. The calculated values are EC50 = 606.7 μM cGMP and H = 2.7. CNGA2 channels are represented by red symbols and the CNGA2(R538E) channels by black symbols. In the cartoon, the R538E mutation in the binding domain of CNGA2 is indicated by red lines.

The concentration-activation relationships for CNGA2:A4:B1b channels with either cGMP or fcGMP were nearly identical, with only a slight decrease in the apparent affinity for fcGMP (EC50 1.75 μM for fcGMP versus 1.28 μM for cGMP) and nearly identical Hill coefficients (1.6 for fcGMP and 1.7 for cGMP) (Fig. 3B). The effect of either the labeled or the unlabeled cGMP was fully and rapidly reversible (fig. S3). Hence, we considered fcGMP an appropriate fluorescent ligand not only for the analysis of homotetrameric CNGA2 (35) but also for the analysis of the heterotetrameric CNGA2:A4:B1b channels (Fig. 3B).

The concentration-activation relationship for CNGA2:B1b channels was indistinguishable from that for CNGA2 channels, whereas that for channels that included the CNGA4 subunit (CNGA2:B1b:A4 and CNGA2:A4) was shifted to lower concentrations (Fig. 3A). We also generated a CNGA2 subunit with a decreased apparent affinity for cGMP by mutating Arg538 to Glu in the CNBD, a mutation that is analogous to those used to compromise cAMP binding to CNGA4 and CNGB1b. For cGMP, the EC50 value of CNGA2(R538E) was 606.7 μM, and the channels were not activated by concentrations up to 30 μM cGMP (Fig. 3C). This mutant and the differential properties of CNGA4 and CNGB1b for cGMP regulation allowed us to analyze the contribution of the cyclic nucleotide binding to the CNGA4 subunit on the CNGA2 subunits, independent of cyclic nucleotide–mediated effects induced by the CNGB1b subunit and independent of the direct effects of cGMP on the CNGA2 channels.

CNGA4 and CNGB1b exhibit differential channel regulation in response to cGMP

Because the experiments with the TFP-fused subunits did not exclude that a fraction of CNGA4 or CNGB1b when expressed individually reached the membrane, we first tested whether the expression of CNGA4 or CNGB1b, when expressed individually, produced fcGMP binding in patches. The binding studies showed that fcGMP did not bind to CNGA4 or CNGB1b when expressed alone in the oocytes (Fig. 4A).

Fig. 4

Binding of fcGMP to the different subunits in olfactory heterotetrameric channels. (A) Quantification of fcGMP binding to inside-out patches from oocytes expressing either CNGA4 or CNGB1b subunits or CNGA2:A4:B1b channels. The nonspecific fluorescence, measured in patches from water-injected oocytes, has been subtracted. The question marks in the cartoons of the channels indicate that it is unknown whether fcGMP binds to these subunits in the heterotetrameric channel. (B) Quantification of fcGMP binding to CNGA2(R538E), CNGA2(R538E):B1b, and CNGA2(R538E):A4 channels. Red lines in the cartoons represent the R538E mutation in the CNBD. (C) Bar graph showing the current amplitude I at 10 μM fcGMP corresponding to the binding experiments in (B), normalized to the current amplitude Imax recorded at 3 mM cGMP.

We next coexpressed the cGMP-compromised mutant CNGA2(R538E) with either CNGA4 or CNGB1b and measured binding in the presence of 10 μM fcGMP, a ligand concentration at which no binding to the CNGA2(R538E) subunits was expected (Fig. 4B). Consistent with the differential action of cGMP on CNGA2:A4 and CNGA2:B1b channels (Fig. 3A), CNGA2(R538E):A4 showed robust ligand binding, whereas CNGA2(R538E):B1b and the homotetrameric channel CNGA2(R538E) did not.

To demonstrate that the lack of binding observed for CNGA2(R538E) and CNGA2(R538E):B1b was not due to low expression, we measured the maximum current elicited in the presence of the saturating concentration of 3 mM cGMP to activate the mutated CNGA2(R538E) subunits. All currents were in the range of nanoamperes and thus in the range of currents obtained with heterotetrameric CNGA2:A4:B1b channels (fig. S4A). However, currents measured in the presence of 10 μM fcGMP showed that only CNGA2(R538E):A4 channels were activated at the selected concentration (Fig. 4C). Thus, only the CNGA2(R538E):A4 showed detectable binding to the CNGA4 subunit, and this binding was also translated into channel activation. The amplitude of this current was less than 10% of the maximum current elicited by saturating cGMP (Fig. 4C). Channels with the mutated CNGA2 subunits lacking the CNGA4 subunit were essentially inactive, producing ≤1% of the maximum current. These results suggested that the CNGA4 subunit bound fcGMP and that this binding was sufficient to trigger opening of the channel. However, binding of fcGMP to CNGA4 required the presence of CNGA2. In contrast, CNGB1b reached the plasma membrane with CNGA2 (Fig. 1) but did not bind fcGMP.

To confirm that cGMP binding to the CNGA4 subunit of a heterotetrameric channel was sufficient for channel activation, we introduced the mutation T539M into the CNBD of CNGA2. Although this mutation only moderately decreased the apparent affinity for cGMP (Fig. 5A), it caused a larger decrease in the apparent affinity for fcGMP. Correspondingly, the binding to the homotetrameric CNGA2(T539M) channels in the presence of 10 μM fcGMP, monitored by fluorescence imaging, was negligible (Fig. 5B). Hence, when CNGA4 and CNGB1b are coexpressed with CNGA2(T539M), application of 10 μM fcGMP should lead to a specific binding to the modulatory subunits only. The CNGA2(T539M):A4 channels exhibited considerable binding, whereas no binding was observed to the CNGA2(T539M):B1b channels (Fig. 5B). To verify the specificity of binding only to CNGA4 in these heterotetrameric channels, we mutated the CNBD of CNGA4 [CNGA4(R430E)] and coexpressed this construct with CNGA2(T539M). No binding of fcGMP was detected (Fig. 5B), suggesting that the observed binding with CNGA2(T539M):A4 was indeed caused by CNGA4 only.

Fig. 5

Activation of olfactory heterotetrameric channels by CNGA4. (A) Concentration-activation relationship for CNGA2 channels containing the mutation T539M in the CNBD. The data points, obtained from 5 to 10 patches, were fitted with Eq. 1. The calculated values for CNGA2(T539M) (blue curve) are EC50 = 10.4 μM cGMP, H = 2.3. The curve for CNGA2 (red) corresponds to the respective curve in Fig. 3A. In the cartoons, the T539M mutation in the CNBD is indicated by the blue lines. (B) Bar graph showing the fluorescence intensity in the presence of 10 μM fcGMP measured in patches containing CNGA2(T539M), CNGA2(T539M):A4, CNGA2(T539M):B1b, or CNGA2(T539M):A4(R430E) channels. The nonspecific fluorescence of 1.73 a.u., measured in water-injected oocytes, has been subtracted. (C) Bar graph showing the current amplitude I corresponding to the binding experiments in (B) at 10 μM fcGMP, with respect to the current amplitude Imax recorded at saturating cAMP concentrations.

To determine whether the presence of wild-type CNGA4 was sufficient to mediate activation in response to 10 μM fcGMP, we also measured the current produced by the constructs shown in Fig. 5, B and C. The CNGA2(T539M):A4 channels generated a robust current that was ~12% of the maximum amplitude detected with a saturating concentration of cAMP. The other channels [CNGA2(T539M), CNGA2(T539M):B1b, and CNGA2(T539M):CNGA4(R430E)] did not produce significant current (amplitude ≤1%) in the presence of fcGMP. The fact that CNGA2(T539M):A4 generated current is consistent with the observed binding in these mutant channels and also with the results for the CNGA2(R538E):A4 channels. To ensure similar amounts of expression, we also measured the maximum current elicited by the saturating concentrations of cAMP and found that all channel combinations produced currents in the range of nanoamperes (fig. S4B).

These results demonstrated that the CNGA4 subunit, when assembled with CNGA2 subunits, not only binds fcGMP but can also activate the heterotetrameric channels without requiring binding of fcGMP to the CNGA2 subunits.

The apparent affinity of the CNGA4 subunit for cGMP is smaller than that of the CNGA2 subunit

Because CNGA2(T539M) subunits are not activated by fcGMP below 30 μM and CNGA4-containing channels are, and assuming that the concentration-activation relationship exhibits a clear separable component at low concentrations of fcGMP, we can determine the apparent affinity of the CNGA4 subunit in heterotetrameric CNGA2(T539M):A4 channels. Indeed, the concentration-activation relationship was composed of a component of low apparent affinity, which is most likely caused by the CNGA2(T539M) subunits, and a component of high apparent affinity, which is most likely caused by the CNGA4 subunit only (Fig. 6A). The high-affinity component (calculated with Eq. 2) had an EC50,h value of 6.8 μM and a Hill coefficient (H) of 1.3. These data quantify the action of the CNGA4 subunit in the presence of nonactivated CNGA2 subunits, conditions that are different from those in the heterotetrameric channels.

Fig. 6

Role of the CNGA4 subunit in determining the apparent affinity in heterotetrameric channels. (A) Log-log plot of the concentration-activation relationship of CNGA2(T539M):A4 channels. The plot suggests the superimposition of a component with high apparent affinity and a component with low apparent affinity. Each data point is the mean of five to seven patches. The curve represents the fit of the data points with Eq. 2, with EC50,h = 6.8 μM fcGMP, Hh = 1.3, a = 0.12 and EC50,l = 52.2 μM fcGMP, Hl = 3.1. The proposed ligand binding to the channels is shown by the cartoons, where the high-affinity state characterizes the binding to CNGA4 only. (B) Concentration-activation relationships of CNGA2(T539M):A4 (blue), CNGA2(T539M):A4(R430E) (violet), and CNGA2(T539M) (red) channels with cGMP. The voltage was set to +100 mV. Each data point is the mean obtained from 8 to 10 patches. The curves represent fits with Eq. 1: CNGA2(T539M):A4: EC50 = 5.1 μM, H = 1.6; CNGA2(T539M):A4(R430E): EC50 = 10.4 μM, H = 2.4; and CNGA2(T539M): EC50 = 10.4 μM, H = 2.3. (C) Functional characteristics of a CNGA2 channel with the CNBD of CNGA4. The cartoon symbolizes the four CNGA2 subunits (squares) with CNGA4 binding sites (blue circles) and the respective ligand (gray circles). Inset: Saturating concentrations of cGMP and cAMP activate CNGA2(CNBD:A4) channels. The voltage was stepped from 0 to +100 mV. The concentration-activation relationship for CNGA2(CNBD:A4) with cGMP was measured at +100 mV. The data points, means of five to seven measurements, were fitted with Eq. 1, yielding EC50 = 3.3 μM cGMP and H = 1.4.

To test whether CNGA4 contributes to channel activation exclusively by the binding of a ligand or, in addition, by an allosteric effect independent of the binding, we measured the concentration-activation relationships for CNGA2(T539M):A4, CNGA2(T539M):A4(R430E), and CNGA2(T539M) channels with cGMP (Fig. 6B). Only the incorporation of wild-type CNGA4 increased the apparent affinity compared to that of the homotetrameric CNGA2(T539M) channels, similar to the results with heterotetrameric CNGA2:A4 and homotetrameric CNGA2 channels (Fig. 2A). However, when the binding of cGMP to CNGA4 was disabled, the effect of CNGA4 was abolished, suggesting that CNGA4 exerts its effect on the activation of CNGA2 subunits by ligand binding only.

Replacement of the CNBD of CNGA2 with that of CNGA4 produces functional channels

When the CNBD of CNGA2 was replaced with the CNBD of CNGA4 and channel function was studied, previous results were controversial; functional channels were either observed (32, 33) or not (34). We created a similar chimeric subunit and expressed this CNGA2(CNBD:A4) channel construct alone in oocytes. These chimeric CNGA2(CNBD:A4) channels produced current with either cGMP or cAMP (inset in Fig. 6C), supporting the notion that the CNBD of CNGA4 is, in principle, functional. This functionality has been shown previously to require the C-linker of CNGA2 (37), which was also present in our construct. The concentration-activation relationship for cGMP could be fitted with a single Hill equation (Eq. 1) (Fig. 6C). Compared to homotetrameric CNGA2 channels, the EC50 value for cGMP was somewhat larger (3.3 μM for the chimeric channel versus 1.91 μM for the wild-type channel), and the Hill coefficient was smaller (1.4 versus 2.0), suggesting that the net positive cooperativity of these artificial subunits was less effective than in wild-type homotetrameric CNGA2 channels.

The concentration-binding and concentration-activation relationships for CNGA2:A4:B1b channels with fcGMP intersect

Our data so far suggested that cGMP activated CNGA2:A4:B1b channels by binding to the CNGA2 subunits and the CNGA4 subunit, both with an apparent affinity in the low micromolar range, and that the CNGB1b subunit did not bind cGMP. However, the apparent affinities for the CNGA2 and CNGA4 subunits in the context of the heterotetrameric CNGA2:A4:B1b channels may be different. Indeed, even the “simplest” of these tetrameric channels, the homotetrameric CNGA2 channels, show a pronounced and complex cooperativity of the identical subunits, resulting in functional differences between them (35).

To gain more information about the activation of CNGA2:A4:B1b channels, we monitored the binding of fcGMP and binding-induced activation by confocal patch-clamp fluorometry over a wide concentration range and plotted the concentration-binding and the concentration-activation relationship in a log-log plot (Fig. 7). This analysis revealed that both relationships required two components (calculated with Eqs. 2 and 3; for parameters see the legend to Fig. 7). The low-affinity component dominated in both ligand binding and channel activation. The larger Hill coefficient for activation (1.9) than for binding (1.3) suggested that part of the cooperativity was not associated with the binding reaction per se but with the subsequent channel opening. Additionally, the parameters for the component of high-affinity binding did not describe a single binding or activation process, but presumably reflect an initial gating mechanism triggering the larger low-affinity component. Substantial activation of the channel occurred only after the high-affinity binding event (relative binding <0.1), suggesting either partial activation after the first ligand has bound or, alternatively, a pronounced positive cooperativity for the subsequent binding steps. In the range of high-affinity binding, ligand binding relatively exceeded channel activation, suggesting that more than one ligand was necessary to maximally open the channel. In the range of low-affinity binding, activation relatively exceeded ligand binding, suggesting that not all ligands that could bind to the channel were required for channel activation. The intersection of the relationships confirms that at least three ligands bind to the channel and are involved in activation. This is because the binding of maximally two ligands would cause a concentration-binding relationship exclusively left of or equal to the concentration-activation relationship.

Fig. 7

Steady-state activation and fcGMP binding in CNGA2:A4:B1b channels. Log-log plot of the specific binding of fcGMP (green) and current activation (black). The actual fluorescence F and current amplitude I were normalized with respect to the values at the saturating concentration of 30 μM fcGMP (Fmax, Imax). Each data point is the mean obtained from 6 to 12 patches. The data points of activation and ligand binding were fitted with Eqs. 2 and 3, respectively. The calculated values are EC50,l = 1.8 μM, EC50,h = 0.03 μM, Hl = 1.9, Hh = 1.9, a = 0.97; BC50,l = 2.8 μM, BC50,h = 0.05 μM, Jl = 1.3, Jh = 1.4, and b = 0.95. The confocal images show ligand binding to the same patch at three different fcGMP concentrations.

Although these results are consistent with an activation process of CNGA2:A4:B1b channels involving ligand binding to three subunits, two CNGA2 subunits, and one CNGA4 subunit, it remains unknown which of the subunits mediates the low- and which mediates the high-affinity component.


The cyclic nucleotides cAMP and cGMP differentially bind to and control the activity of many proteins, including CNG and HCN channels. In rod and cone photoreceptors, CNG channels distinguish well between cGMP and cAMP, whereas in olfactory sensory neurons, the channels respond equally well to both ligands. Despite this, cAMP is the main second messenger for the olfactory signal transduction, although cGMP is also available for channel regulation (1).

Here, we address the differential effects of cAMP and cGMP on the activation of the heterotetrameric olfactory CNG channels. We provide evidence that cAMP activates these channels by binding to both CNGA4 and CNGB1b subunits in addition to binding to the CNGA2 subunits. We also provide a detailed analysis of the action of cGMP, and its fluorescent derivative fcGMP, in CNGA2:A4:B1b channels. In contrast to cAMP, cGMP binds and activates the CNGA4 subunit but not the CNGB1b subunit. Moreover, in heterotetrameric channels containing three CNGA2 subunits with compromised binding sites, the CNGA4 subunit alone binds fcGMP, resulting in opening of the channel. When expressed with CNGA2(T539M) subunits, CNGA4 without a functional CNBD failed to alter the channel activity in response to cGMP. This suggested that the effects of CNGA4 depended on binding cyclic nucleotides and were not due to a ligand-independent allosteric influence on the CNGA2 subunits. For the CNGB1b subunit, the situation differed: Coexpression with compromised CNGA2 subunits resulted in nonfunctional channels that also did not bind fcGMP, indicating that CNGB1b does not contribute to activation of the channel by cGMP.

Our analysis of the localization of the TFP-labeled subunits (Fig. 1) showed that CNGA2 promoted the incorporation of the CNGA4 subunit in the plasma membrane and that the combination of CNGA2 plus CNGA4 promoted the incorporation of the CNGB1b subunit. Although we cannot unequivocally state whether the CNGA4 subunits can reach the membrane alone, the fluorescence data showed that if these subunits reach the membrane, they do not bind fcGMP. Only the presence of CNGA2 subunits enabled the CNGA4 subunits to bind cyclic nucleotides and to translate this binding, in combination with the CNGA2 subunits, into channel activation. Furthermore, ligand binding to the CNGA2 subunits was not required for CNGA4 to mediate partial activation of the heterotetrameric channels in response to fcGMP. These results, together with the fact that the CNGA2(CNBD-A4) chimera formed functional channels, suggest a specific interaction between CNGA2 and CNGA4 for which, as previously reported (37), the C-linker of CNGA2 plays an important role.

In contrast to olfactory neurons, where the CNGB1b subunit is needed for successfully reaching the cilia (42, 43), in the oocyte system, CNGB1b was not necessary for targeting the channels to the membrane, because CNGA2:A4 channels reached the membrane as effectively as did CNGA2:A4:B1b channels. The different experimental systems might explain these results.

The CNGB1b subunit differed from the CNGA4 subunit in that it influenced channel activity only in the presence of cAMP and not in the presence of cGMP. When CNGB1b was coexpressed with compromised CNGA2 subunits [CNGA2(R538E) or CNGA2(T539M)], neither ligand binding nor channel activation was observed. Thus, the CNGB1b subunit apparently could distinguish between cGMP and cAMP. It remains open at present by which molecular mechanism the CNGB1b subunit discriminates between cAMP and cGMP, in particular whether this discriminative ability is an intrinsic property of the CNGB1b subunit or controlled by the other subunits.

Our experiments may also provide insight into the sequence of ligand binding to the various subunits. The EC50 values for CNGA4 coexpressed with CNGA2 with compromised binding sites [CNGA2(T539M):A4] (6.8 μM with fcGMP) and for the homotetrameric CNGA2 channels containing the CNBD of CNGA4 [CNGA2(CNBD:A4)] (3.3 μM with cGMP) were similar. Assuming that these low micromolar values are representative also of the CNGA4 subunit in the context with wild-type CNGA2 subunits, CNGA4 binds and activates the channel at only moderately higher concentrations than those observed for the heterotetrameric CNGA2:A4:B1b channel (EC50 = 1.8 μM with fcGMP). Conversely, the fact that the EC50 value for CNGA2:A4:B1b channels was lower than that for the CNGA4 subunit coexpressed with compromised CNGA2 suggested that at least one CNGA2 subunit in the CNGA2:A4:B1b channels becomes activated at lower concentrations than those required to activate the CNGA4 subunit. A previously reported detailed analysis of the gating of homotetrameric CNGA2 channels supports this conclusion: The first, the third, and the fourth binding step have an equilibrium association constant (KA) of 1.01 × 106 M−1 (35), resulting in a concentration of half-maximum binding of 1/(1.01 × 106 M−1) = 0.99 μM. However, this analysis also showed that the equilibrium association constant of the second binding step is more than three orders of magnitude smaller than that of the three other binding steps (KA = 2.94 × 102 M−1), resulting in a concentration of half-maximum binding of 1/(2.94 × 102 M−1) = 3.4 mM. This shows a pronounced complex cooperativity among the subunits; thus, it is impossible, on the basis of our data, to assign the CNGA4 subunit a number in the sequence of binding steps. It is also possible that the sequence of binding is not always the same. Nevertheless, the lower apparent affinity of the CNGA4 subunit (EC50 = 6.8 μM) compared to that of the CNGA2 subunits (EC50 = 1.0 μM) in homotetrameric channels suggests that in heterotetrameric channels the first binding is governed by the CNGA2 subunit and not by the CNGA4 subunit. This conclusion differs from that in a previous report proposing CNGA4-CNGA2-CNGA2-CNGB1b as the sequence of binding (30).

Materials and Methods

Molecular biology

Rat olfactory channel subunits CNGA2 (accession number AF126808), CNGA4 (accession number U12623), and CNGB1b (accession number AF068572), as well as all genetically modified subunit variants were subcloned into vector pGEMHEnew (44). DNA constructs for chimeras, for subunits tagged with TFP (sequence from pmTFP1-N, Allele Biotechnology and Pharmaceuticals Inc.), and for subunits modified at the CNBD were constructed by recombinant polymerase chain reaction (PCR) techniques. The deduced amino acid composition of chimera CNGA2(CNBD:A4) was as follows: CNGA2(1–461)-CNGA4(354–475)-CNGA2(584–664).

The TFP-fusion constructs encoded the full-length CNGA4 or CNGB1b subunit coupled to the full-length mTFP1 protein, as follows: CNGA4(1–575)-TFP(1–236) and CNGB1b(1–858)-TFP(1–236). Capped complementary RNA (cRNA) was obtained after linearization with Not I followed by in vitro transcription with T7 RNA polymerase.

Preparation of oocytes and cRNA injection

Oocytes were obtained under anesthesia (0.3% aminobenzoic acid ethyl ester) from adult females of Xenopus laevis. The oocytes were prepared by incubation for 60 to 80 min with collagenase (1.2 mg/ml; type I, Roche Diagnostics). Within 2 to 7 hours after isolation and defolliculation, the oocytes were injected with 40 to 70 nl of the desired cRNA solution, as previously described (45). The oocytes were incubated at 18°C in Barth medium for up to 6 days after injection.


cGMP and cAMP were obtained from Sigma. 8-DY547-AET-cGMP (fcGMP) was prepared as reported previously (35). All chemicals were of analytical grade.

Current recording

The patch pipettes were pulled from borosilicate glass tubing (outer diameter, 2.0 mm; inner diameter, 1.0 mm; resistance, 0.7 to 1.4 megohms). The angle between the pipette and the chamber bottom was about 30°. The patch pipettes, containing inside-out macropatches, were positioned in front of a multibarrel application system implemented in the experimental chamber. The different experimental solutions with the different concentrations of cGMP, cAMP, or fcGMP were administered with a multibarrel application system to the cytosolic face of the patch. All experiments were performed with the following solution in the bath and the pipette containing 150 mM KCl, 1 mM EGTA, and 5 mM Hepes (pH 7.4) with KOH. Each patch was initially exposed to a solution containing no cyclic nucleotide and then to a solution containing the saturating concentration of either 100 μM cGMP, 1 mM cAMP, or 3 mM cGMP. Recording was performed with an Axopatch 200B amplifier (Molecular Devices LLC). The currents were measured at +10 mV or, if mentioned, at +100 mV.

Optical localization of TFP-labeled subunits

To identify the TFP signal originating from CNGA4-TFP and CNGB1b-TFP subunits localized in the plasma membrane, we labeled the plasma membrane with a fluorescence-labeled lectin from the extracellular side (Alexa Fluor 633 WGA; Invitrogen Life Technologies GmbH). Fluorescence was quantified by confocal microscopy (LSM 710; Zeiss). The oocytes were incubated for 5 min in a solution containing lectin (~5.0 μg/ml). The TFP signal was separated from the overlapping spectra of autofluorescence arising from the oocytes and of Alexa Fluor 633 WGA by spectral unmixing with a software routine provided by the microscope manufacturer (46).

TFP and WGA fluorescence was simultaneously excited by the 458-nm line of an argon laser and a 543-nm helium-neon laser, respectively. The lasers were projected through a main beam splitter (MBS) 458/543 and a 40×/1.2 C-Apochromat water-immersion objective onto the sample. The emitted light was collected through the objective and spatially filtered by a 37-μm pinhole [1 arbitrary unit (a.u.)], resulting in a z slice of 1 μm, and recorded with the spectra-resolving Quasar detector of the system. A 23-channel spectrum ranging from 465 to 688 nm was acquired for each image point.

The recorded spectra are a linear combination of the spectra of the different fluorophores present in the sample, convoluted by the transmission properties of the microscope system. The relative contributions of the different fluorophores were extracted by means of reference spectra acquired separately with an identical microscope configuration. Reference spectra were obtained from an Alexa Fluor 633 WGA solution, uninjected oocytes, and oocytes expressing TFP-labeled CNGA4. The TFP spectra were offline-corrected for the autofluorescence contribution of the oocytes.

To quantify colocalization, we averaged a total of 10 to 85 profiles per oocyte. TFP and Alexa Fluor 633 WGA signals were normalized with respect to the integral of their profiles. The integral of the overlap between the two profiles was used as a quantitative measure of colocalization, thus TFP membrane localization.

Confocal patch-clamp fluorometry

Ligand binding to inside-out patches was measured by a combination of patch-clamp fluorometry (47, 48) with confocal microscopy (35, 49, 50). Using fcGMP, we simultaneously measured ligand binding and binding-induced activation in CNGA2 channels (35). fcGMP produces an intensive green fluorescence when excited at 543 nm. To separate the fluorescence of the bound fcGMP from the fluorescence generated by the free fcGMP in the bath solution, we used an additional red dye, DY647 (Dyomics). The rationale to use the red dye and the whole method of confocal patch-clamp fluorometry was described previously (35). fcGMP was preferred over the fluorescent analog fcAMP (47) to avoid problems with the optical measurements arising from the lower potency of fcAMP.

Data acquisition and analysis

Measurements were controlled and data were recorded with the ISO2 and ISO3 soft- and hardware (MFK) running on a PC. The sampling rate was either 2 or 5 kHz, and the filter was set to 1 kHz. Curves were fitted to the data with the nonlinear approximation algorithm implemented in the Origin 8.1G software (OriginLab Corporation).

Concentration-activation relationships were obtained by relating the actual current amplitude, I, to the maximum current amplitude, Imax, and plotting this ratio as a function of the ligand concentration, x. The data points were fitted with the Hill equationI/Imax=1/[1+(EC50/x)H](1)where EC50 is the ligand concentration of half-maximum effect and H is the Hill coefficient for activation.

When the concentration-activation relationship showed two components, the data points were fitted to the sum of two Hill equations according toI/Imax=a/[1+(EC50,h/x)Hh]+(1a)/[1+(EC50,l/x)Hl](2)where EC50,h, EC50,l, Hh, and Hl are the ligand concentration of half-maximum current and the Hill coefficient for the high- and low-affinity component, respectively. a and (1 − a) denote the contribution of the high- and low-affinity component, respectively.

A two-component concentration-binding relationship was fitted with a respective equation,B/Bmax=b/[1+(BC50,h/x)Jh]+(1b)/[1+(BC50,l/x)Jl](3)where BC50,h, BC50,l, Jh, and Jl are the ligand concentration of half-maximum binding and the Hill coefficient for the high- and low-affinity component, respectively. Accordingly, b and (1 − b) denote the contribution of the high- and low-affinity component, respectively.

Statistical data are given as means ± SEM.

Supplementary Materials

Fig. S1. Spectrally coded confocal images of Xenopus oocytes injected with RNA encoding TFP-labeled subunits.

Fig. S2. Representative single-channel currents of CNGA2, CNGA2:A4, CNGA2:B1b, and CNGA2:A4:B1b channels.

Fig. S3. Rapid and reversible activation by fcGMP or cGMP of heterotetrameric olfactory channels.

Fig. S4. Maximum current generated by the different CNGA2(R538E)- and CNGA2(T539M)-containing channels.

References and Notes

Acknowledgments: We are indebted to U. B. Kaupp (Bonn) for providing the complementary DNA encoding rat CNGA2, CNGA4, and CNGB1b subunits. We also thank K. Schoknecht, S. Bernhardt, A. Kolchmeier, and B. Tietsch for excellent technical assistance. Funding: This work was supported by grants of the Deutsche Forschungsgemeinschaft to K.B. Author contributions: V.N., electrophysiological and fluorometrical measurements, manuscript preparation; T.Z., molecular biology; N.W. and L.R., electrophysiological measurements; R. Schmauder, TFP-data analysis; J.K., experimental design; W.B., molecular biology; R. Seifert, experimental design; C.B., software development; F.S., synthesis of fluorescent ligand; and K.B., project planning, experimental design, and manuscript writing. Competing interests: F.S. is R&D head of BIOLOG Life Science Institute.
View Abstract

Navigate This Article