Portal de Periódicos da CAPES

Distinct Structural Determinants of Efficacy and Sensitivity in the Ligand-binding Domain of Cyclic Nucleotide-gated Channels

2004; Elsevier BV; Volume: 279; Issue: 5 Linguagem: Inglês

10.1074/jbc.m310545200

1083-351X

Edgar C. Young, Natalia Krougliak,

Photochromic and Fluorescence Chemistry

Cyclic nucleotide-gated (CNG) channels open in response to direct binding of cyclic nucleotide messengers. Every subunit in a tetrameric CNG channel contains a cytoplasmic ligand-binding domain (BD) that includes a β-roll (flanked by short helices) and a single C-terminal helix called the C-helix that was previously found to control efficacy (maximal open probability) and selectivity for cGMP versus cAMP. We constructed a series of chimeric CNG channel subunits, each containing a distinct BD sequence (chosen from among six phylogenetically divergent isoforms) fused to an invariant non-BD sequence. We assayed these “BD substitution” chimeras as homomeric CNG channels in Xenopus oo-cytes to compare their functions and found that the most efficient activation by both cAMP and cGMP derived from the BD of the catfish CNGA4 olfactory modulatory subunit (fCNGA4). We then tested the effects of replacing subregions of the bovine CNGA1 BD with corresponding fCNGA4 sequence and hence identified parts of the fCNGA4 BD producing efficient activation. For instance, replacing either the “hinge” that connects the roll and C-helix subdomains or the BD sequence N-terminal to the hinge greatly enhanced cAMP efficacy. Replacing the “loop-β8” region (the C-terminal end of the β-roll) improved agonist sensitivity for cGMP selectively over cAMP. Our results thus identify multiple BD elements outside the C-helix that control selective ligand interaction and channel gating steps by distinct mechanisms. This suggests that the purine ring of the cyclic nucleotide may interact with both the β-roll and the C-helix at different points in the mechanism. Cyclic nucleotide-gated (CNG) channels open in response to direct binding of cyclic nucleotide messengers. Every subunit in a tetrameric CNG channel contains a cytoplasmic ligand-binding domain (BD) that includes a β-roll (flanked by short helices) and a single C-terminal helix called the C-helix that was previously found to control efficacy (maximal open probability) and selectivity for cGMP versus cAMP. We constructed a series of chimeric CNG channel subunits, each containing a distinct BD sequence (chosen from among six phylogenetically divergent isoforms) fused to an invariant non-BD sequence. We assayed these “BD substitution” chimeras as homomeric CNG channels in Xenopus oo-cytes to compare their functions and found that the most efficient activation by both cAMP and cGMP derived from the BD of the catfish CNGA4 olfactory modulatory subunit (fCNGA4). We then tested the effects of replacing subregions of the bovine CNGA1 BD with corresponding fCNGA4 sequence and hence identified parts of the fCNGA4 BD producing efficient activation. For instance, replacing either the “hinge” that connects the roll and C-helix subdomains or the BD sequence N-terminal to the hinge greatly enhanced cAMP efficacy. Replacing the “loop-β8” region (the C-terminal end of the β-roll) improved agonist sensitivity for cGMP selectively over cAMP. Our results thus identify multiple BD elements outside the C-helix that control selective ligand interaction and channel gating steps by distinct mechanisms. This suggests that the purine ring of the cyclic nucleotide may interact with both the β-roll and the C-helix at different points in the mechanism. Cyclic nucleotide-gated (CNG) 1The abbreviations used are: CNGcyclic nucleotide-gatedBDligand-binding domainbCNGA1bovine CNGA1fCNGA2catfish CNGA2rCNGA4rat CNGA4fCNGA4catfish CNGA4PBcyclic phosphate-binding.1The abbreviations used are: CNGcyclic nucleotide-gatedBDligand-binding domainbCNGA1bovine CNGA1fCNGA2catfish CNGA2rCNGA4rat CNGA4fCNGA4catfish CNGA4PBcyclic phosphate-binding. channels conduct mono- and divalent cations upon activation by the direct binding of the cytoplasmic messengers cAMP and cGMP. These channels are widespread in the nervous system and in a variety of other tissues, and most notably they are essential signaling components in visual and olfactory transduction, where their activation leads both to changes in membrane potential and to the influx of calcium into the cytoplasm (reviewed in Refs. 1Burns M.E. Baylor D.A. Annu. Rev. Neurosci. 2001; 24: 779-805Google Scholar and 2Zufall F. Firestein S. Shepherd G.M. Annu. Rev. Biophys. Biomol. Struct. 1994; 23: 577-607Google Scholar). Functional CNG channels are tetramers (homomeric or heteromeric) of homologous subunits; vertebrates contain a family of six paralogous CNG channel subunit genes in two phylogenetic subfamilies (3Bradley J. Frings S. Yau K.W. Reed R. Science. 2001; 294: 2095-2096Google Scholar), CNGA and CNGB. Distinct combinations of these paralogues are expressed in each tissue type to produce CNG channels whose response parameters are presumably adapted to the physiological requirements of the tissue; these parameters include sensitivity and efficacy (maximal ligandgated open probability) for cAMP and cGMP and selectivity for one agonist over the other. Thus, CNG channels hold promise as targets for tissue-specific pharmacological regulation of signaling through cyclic nucleotide-dependent, electrical, and calcium-dependent pathways. It is therefore important to understand how CNG channel function is determined by the sequences of individual subunits and more generally how structural elements in the channel work together to control the quantitative properties of the ligand gating mechanism.All CNG channel subunits have a common modular architecture (reviewed in Refs. 4Zagotta W.N. Siegelbaum S.A. Annu. Rev. Neurosci. 1996; 19: 235-263Google Scholar and 5Biel M. Zong X. Ludwig A. Sautter A. Hofmann F. Rev. Physiol. Biochem. Pharmacol. 1999; 135: 151-171Google Scholar) incorporating recognized structural motifs, namely a 6TM domain joined at its cytoplasmic C-terminal end to a conserved “C-linker” region of ∼80 residues, followed by a cyclic nucleotide-binding domain (BD). The 6TM domain is homologous to that of the voltage-gated potassium channel family, with six membrane-spanning segments (S1-S6) and a re-entrant “P-loop” between S5 and S6 that lines the aqueous pore. The BD is homologous to those of cAMP- and cGMP-dependent protein kinases, cAMP-dependent G-protein exchange factors, and the bacterial catabolite activator protein. BD sequences are also found (associated with the 6TM domain and C-linker regions) in several ion channel families with homology to CNG channels, such as the HCN “pacemaker” and the EAG channels. Useful models for the CNG channel BD are based on known three-dimensional structures of homologous BDs (6Weber I.T. Steitz T.A. J. Mol. Biol. 1987; 198: 311-326Google Scholar, 7Su Y. Dostmann W.R. Herberg F.W. Durick K. Xuong N.H. Ten Eyck L. Taylor S.S. Varughese K.I. Science. 1995; 269: 807-813Google Scholar, 8Rehmann H. Prakash B. Wolf E. Rueppel A. De Rooij J. Bos J.L. Wittinghofer A. Nat. Struct. Biol. 2003; 10: 26-32Google Scholar, 9Zagotta W.N. Olivier N.B. Black K.D. Young E.C. Olson R. Gouaux E. Nature. 2003; 425: 200-205Google Scholar) and delineate a “roll subdomain” and “C-helix subdomain” connected by a conserved proline. The roll subdomain consists of an eight-stranded β-roll flanked by two short helices, called the “A-helix” and the “B-helix.” The β-roll itself contains (between its sixth and seventh β-strands) a short “P-helix” and “PB-loop” that constitute a conserved phosphate-binding cassette (10Diller T.C. Xuong Madhusudan, N.H. Taylor S.S. Structure. 2001; 9: 73-82Google Scholar). The C-helix subdomain consists of a single long helix; one particular C-helix residue has been shown to control agonist selectivity in bovine CNGA1, presumably by forming a specific contact with the purine ring of the ligand (11Varnum M.D. Black K.D. Zagotta W.N. Neuron. 1995; 15: 619-625Google Scholar).We reasoned that phylogenetically divergent CNG channel subunit isoforms should have accumulated mutations in many functionally important parts of the BD in the course of adaptation to different physiological settings, and analysis of these mutations could be useful in investigating the structural determinants of channel activation. We previously assessed the importance of sequence polymorphism in the BD by constructing a series of chimeric CNG channel subunits in which each chimera contained a distinct BD sequence, but all chimeras shared identical sequence outside the BD (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar). These “BD substitution” chimeras enabled direct functional comparisons between BD sequences. We have now expanded the previous set of three BD substitution chimeras to test BDs from a phylogenetically diverse range of CNG channel isoforms. This revealed a broad spectrum of response properties deriving from BD sequence polymorphism and led to the identification of one particular sequence (from catfish CNGA4) that produced extremely efficient activation properties. This BD sequence was then dissected to identify specific subsequences in which polymorphism strongly influences efficacy or sensitivity. Our analysis shows that multiple BD regions (outside the previously studied C-helix ligand contact) contribute to highly efficient ligand gating; these regions moreover control distinct sets of molecular interactions during the activation process.EXPERIMENTAL PROCEDURESMolecular Subcloning—Subcloning in the oocyte expression vector pGEM-HE and chimera construction and mutagenesis by PCR were done as described (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar, 13Paoletti P. Young E.C. Siegelbaum S.A. J. Gen. Physiol. 1999; 113: 17-34Google Scholar) using the following gene sequences: CNGA1 from cow (14Kaupp U.B. Niidome T. Tanabe T. Terada S. Bonigk W. Stuhmer W. Cook N.J. Kangawa K. Matsuo H. Hirose T. et al.Nature. 1989; 342: 762-766Google Scholar), CNGA2 from catfish (15Goulding E.H. Ngai J. Kramer R.H. Colicos S. Axel R. Siegelbaum S.A. Chess A. Neuron. 1992; 8: 45-58Google Scholar), CNGA4 from rat (16Bradley J. Li J. Davidson N. Lester H.A. Zinn K. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 8890-8894Google Scholar, 17Liman E.R. Buck L.B. Neuron. 1994; 13: 611-621Google Scholar), CNGA4 from catfish (18Young E.C. Yao H. Vosshall L.B. Sun Z.-P. Krougliak N. Tibbs G.R. Siegelbaum S.A. Biophys. J. 2002; 82: 275AGoogle Scholar) (GenBank™ accession number AF522297), TAX-4 from C. elegans (19Komatsu H. Mori I. Rhee J.S. Akaike N. Ohshima Y. Neuron. 1996; 17: 707-718Google Scholar), and TAX-2 from Caenorhabditis elegans (20Coburn C.M. Bargmann C.I. Neuron. 1996; 17: 695-706Google Scholar). The C-terminal regions of these genes were aligned using Clustal W (see Fig. 1C) to define homologous positions; then the BD sequence was defined as bCNGA1 Leu485-Ala614 and homologous sequences in the other subunits. Similarly, the putative ligand contact position is defined as bCNGA1 Asp604 and homologous positions in other BDs. New X chimeras were constructed by substitution of the BD in the chimera ROON-S2 previously studied (21Tibbs G.R. Goulding E.H. Siegelbaum S.A. Nature. 1997; 386: 612-615Google Scholar, 22Tibbs G.R. Liu D.T. Leypold B.G. Siegelbaum S.A. J. Biol. Chem. 1998; 273: 4497-4505Google Scholar); new RO chimeras were constructed by BD substitution in the chimera RO133 previously studied (13Paoletti P. Young E.C. Siegelbaum S.A. J. Gen. Physiol. 1999; 113: 17-34Google Scholar, 21Tibbs G.R. Goulding E.H. Siegelbaum S.A. Nature. 1997; 386: 612-615Google Scholar, 23Goulding E.H. Tibbs G.R. Liu D. Siegelbaum S.A. Nature. 1993; 364: 61-64Google Scholar, 24Liu D.T. Tibbs G.R. Siegelbaum S.A. Neuron. 1996; 16: 983-990Google Scholar, 25Liu D.T. Tibbs G.R. Paoletti P. Siegelbaum S.A. Neuron. 1998; 21: 235-248Google Scholar). X-bA1 and RO-bA1 in this work are synonymous with ROON-S2 and RO133, respectively. All sequences subjected to PCR were dideoxy-sequenced.Patch Clamp Recording of Channel Currents and Data Analysis— The procedures were essentially performed as described (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar). In brief, Xenopus oocytes were injected with 0.25-25 ng of RNA; 1-5 days later, inside-out patches were obtained with electrodes of resistance 1-5 MΩ (coated with Sylgard for single-channel recording). The pipette and bath solutions both contained 67 mm KCl, 30 mm NaCl, 10 mm HEPES, 10 mm EGTA, 1 mm EDTA, pH 7.2 with KOH. Na-cAMP or Na-cGMP were included in the bath solution by iso-osmolar replacement of NaCl and applied by gravity perfusion. Patch clamp equipment, software, and data acquisition and analysis were as described (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar).Macroscopic currents elicited by cyclic nucleotide were recorded at -100 mV (filtered at 4 kHz, digitized at 1 kHz) after the steady-state current level was reached and were corrected by subtraction of leak currents recorded without agonist. Time stationarity of dose-response curves (i.e. completion of spontaneous run-up or run-down (26Molokanova E. Maddox F. Luetje C.W. Kramer R.H. J. Neurosci. 1999; 19: 4786-4795Google Scholar)) was verified as described (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar). For each curve, response current I at agonist concentration [A] was fitted (with weighting by 1/S.D.) with the Hill equation, I = Imax/(1 + (K1/2/[A])h), where K1/2 is the concentration of A eliciting half-maximal activation, h is the Hill coefficient, and Imax is the maximal current amplitude. Dose responses collected for cAMP and cGMP in the same patch were used to evaluate selectivity ratios (K1/2,cGMP/K1/2,cAMP and Imax,cAMP/Imax,cGMP). Ratios significantly less than or greater than unity indicate selectivity for cGMP or cAMP, respectively.Single-channel currents in steady-state agonist concentrations were recorded at -80 mV (filtered at 4 kHz, digitized at 20 kHz), and open probability (Popen) was evaluated from all points current amplitude histograms as described (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar). Maximal conductance and two prominent subconductances from proton block were characteristic of previously studied X chimeras (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar, 22Tibbs G.R. Liu D.T. Leypold B.G. Siegelbaum S.A. J. Biol. Chem. 1998; 273: 4497-4505Google Scholar) and the RO chimera, RO-bA1 (24Liu D.T. Tibbs G.R. Siegelbaum S.A. Neuron. 1996; 16: 983-990Google Scholar, 25Liu D.T. Tibbs G.R. Paoletti P. Siegelbaum S.A. Neuron. 1998; 21: 235-248Google Scholar). Pmax was determined as Popen from continuous stretches of data (>30 s) in 10 μm cAMP for X-fA4, 3 mm cAMP or cGMP for RO-fA4, 30 mm cAMP for RO-rA4, 3 mm cGMP, or 30 mm cAMP for RO-fA4 Asp. To normalize dose responses in terms of Popen for Figs. 3, 4, and 6, data and Hill fits for both agonists were multiplied by Pmax/Imax of the agonist giving higher Imax.Fig. 3Chimeras with disfavored opening reveal efficacy differences.A, design and single-channel recordings of RO chimera homomers containing fCNGA4 or rCNGA4 BDs. Note that except for the P-loop and BD, sequence is identical to that of intact bCNGA1 (thin black lines). Recording excerpts (collected as in Fig. 2A) have Popen values of 0.964 (RO-fA4) and 0.436 (RO-rA4). B, points plot cAMP (solid symbols) and cGMP (open symbols) activation data from representative patches of RO-fA4 (circles) and RO-rA4 (triangles), normalized using respective Pmax,cAMP values of 0.974 and 0.468 (means from single-channel measurements); the lines show Hill fits.View Large Image Figure ViewerDownload (PPT)Fig. 4fCNGA4 BD with a cGMP-selective ligand contact mutation maintains efficient cAMP activation.A, points plot representative patch data for cAMP (solid symbols) and cGMP (open symbols) activation of RO-fA4 Asp (circles) and RO-bA1 (triangles), normalized with respective single-channel mean Pmax,cGMP values of 0.99 and 0.94. The lines show Hill fits for cAMP (solid lines) or cGMP (dashed lines) activation of RO-fA4 Asp and RO-bA1 (black) and of RO-fA4 (gray, curves from Fig. 3B). B, recordings from a single RO-fA4 Asp homomer. Popen for excerpts: 0.997 in cGMP and 0.609 in cAMP.View Large Image Figure ViewerDownload (PPT)Fig. 6fCNGA4 subregion replacements with differential effects on efficacy, binding, and selectivity.A, replacements of the A-helix + β-roll and hinge show different effects on sensitivity. Points plot representative data for activation by cAMP (solid symbols) and cGMP (open symbols) of RO-bA1{1-3} (circles) and RO-bA1{4} (triangles), normalized with arbitrary Pmax,cGMP = 0.99. The lines show Hill fits for cAMP (solid lines) or cGMP (dashed lines) activation of RO-bA1{1-3} and RO-bA1{4} (black lines) and of RO-bA1 (gray lines, curves from Fig. 4A). B, loop-β8 replacement shows cGMP selectivity. Plots of RO-bA1{4} and RO-bA1 data are as in A; inverted triangles plot dose-response data for activation of RO-bA1{3} by cAMP (solid symbols) and cGMP (open symbols), normalized as in A. C, hinge (red) or A-helix + β-roll and loop-β8 (yellow) in the putative three-dimensional structure of the BD, represented by the cAMP-liganded HCN2 BD (RASTER3D diagram); a star marks C-helix ligand contact.View Large Image Figure ViewerDownload (PPT)Pmax,cAMP values in Table III were determined as Pmax,cAMP = Imax,cAMP/Imax,cGMP measured at -100 mV in macroscopic current patches. This assumes that open channel conductance is the same in saturating cAMP and cGMP and that Pmax,cGMP is near unity. If the latter assumption is omitted, Imax,cAMP/Imax,cGMP is an upper limit on Pmax,cAMP. However, both assumptions were validated for RO-bA1 previously (13Paoletti P. Young E.C. Siegelbaum S.A. J. Gen. Physiol. 1999; 113: 17-34Google Scholar, 21Tibbs G.R. Goulding E.H. Siegelbaum S.A. Nature. 1997; 386: 612-615Google Scholar) and for RO-fA4 Asp in this study and so likely hold for RO chimeras derived from these original two chimeras. In addition, Pmax,cGMP > 0.95 was verified directly in single-channel recordings for at least two patches each (>3 mm cGMP) of RO-bA1{4}, RO-bA1{4.1}, RO-bA1{4.2}, and RO-bA1{1-3}. We assumed that a fully liganded channel has a single closed and a single open state, so the free energy of opening in saturating cAMP is ΔGsat,cAMP = -RT ln [Pmax,cAMP/(1 - Pmax,cAMP)]. The change in ΔGsat,cAMP associated with converting RO-bA1 to a subregion chimera was calculated by subtracting the mean ΔGsat,cAMP for RO-bA1 from the mean ΔGsat,cAMP for the new chimera, that is, ΔΔGsat = ΔGsat,cAMP[new] - ΔGsat,cAMP[RO-bA1]. Unless otherwise noted, the means are reported ± S.D. with the sample size (n), and unpaired t test was used to assess the significance in population differences.Table IIIActivation properties of subregion replacement RO chimeras The values are the means ± S.D. of measurements from individual patches (numbers of patches in parentheses). ND, not determined.ChimerabCNGA1 sequence replacedK1/2Efficacy for cAMPcAMPcGMPPmax,cAMPΔGsat,cAMPΔΔGsatμmkJ/molkJ/molRO-bA12140 ± 410 (4)55 ± 27 (18)0.025 ± 0.12 (17)+9.4 ± 1.3RO-fA4 AspVal487—Asn613264 ± 62 (7)0.82 ± 0.19 (9)0.71 ± 0.18 (10)—2.5 ± 2.1—11.9 ± 2.5RO-bA1 {1}Val487—Arg5122160 ± 730 (2)23 ± 13 (8)0.082 ± 0.097 (9)+6.9 ± 2.4—2.5 ± 2.7RO-bA1 {2}Ile530—Tyr541949.5 ± 2.1 (2)9.1 ± 1.8 (7)0.058 ± 0.023 (7)+7.06 ± 0.93—2.3 ± 1.6RO-bA1 {3}Ser553—Cys5731910 ± 870 (4)17.2 ± 4.5 (9)0.042 ± 0.23 (10)+8.1 ± 1.6—1.3 ± 2.0RO-bA1 {4}Asp577—Met5923550 ± 760 (6)13.5 ± 4.8 (10)0.47 ± 0.14 (12)+0.4 ± 1.5—9.0 ± 1.9RO-bA1 {5}Lys598—Asn613ND185 ± 52 (3)0.0028 ± 0.0019 (4)+14.9 ± 1.5+5.5 ± 2.0RO-bA1 {4.1}Asp577—Tyr586ND16.5 ± 3.9 (6)0.154 ± 0.061 (5)+4.3 ± 1.1—5.0 ± 1.6RO-bA1 {4.2}Asp588—Met5922130 ± 320 (4)22 ± 17 (10)0.164 ± 0.074 (9)+4.4 ± 1.7—5.0 ± 2.2RO-bA1 {1—3}Val487—Cys573586 ± 63 (8)1.48 ± 0.27 (9)0.424 ± 0.098 (11)+0.8 ± 1.0—8.6 ± 1.6RO-bA1R/fA4C AspAsp588—Asn613ND72 ± 25 (6)0.0171 ± 0.0099 (7)+10.5 ± 1.7+1.1 ± 2.1RO-fA4R/bA1CVal487—Tyr586284 ± 71 (6)0.361 ± 0.069 (6)0.825 ± 0.054 (9)—4.0 ± 1.0—13.3 ± 1.6 Open table in a new tab RESULTSAgonist Sensitivities of Diverse BD Sequences Directly Compared Using BD Substitution Chimeras—Our previously studied “X chimeras” (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar) all share identical sequence outside the BD region; differences between X chimeras can thus be attributed to BD sequence differences. The invariant non-BD sequence (X) consists of the bovine CNGA1 (bCNGA1) sequence with two sequence regions replaced by corresponding residues from catfish CNGA2 (fCNGA2) for technical advantages (Fig. 1A). Replacement of the P-loop increases ion conductance through the open channel (23Goulding E.H. Tibbs G.R. Liu D. Siegelbaum S.A. Nature. 1993; 364: 61-64Google Scholar), facilitating current detection in both macroscopic current and single-channel recordings. Replacement of the “N-S2” region favors the intrinsic opening transition, enhancing the response to any agonist (21Tibbs G.R. Goulding E.H. Siegelbaum S.A. Nature. 1997; 386: 612-615Google Scholar, 27Goulding E.H. Tibbs G.R. Siegelbaum S.A. Nature. 1994; 372: 369-374Google Scholar). These two replacements increase the chances of observing cyclic nucleotide-activated currents from X chimeras containing BDs of unknown functionality that might work poorly in activation.The previous study tested BDs from bCNGA1, fCNGA2, and rat CNGA4 (rCNGA4); this study included BDs from more phylogenetically diverse isoforms, namely CNGA4 recently cloned from catfish olfactory epithelium (fCNGA4) and TAX-4 and TAX-2 from the nematode C. elegans (Fig. 1, B and C). Whereas bCNGA1, fCNGA2, and TAX-4 are classed as “conventional” type because they can form functional homomeric CNG channels, rCNGA4, fCNGA4, and TAX-2 are classed as “modulatory” type because they cannot form functional homomeric CNG channels but do coassemble with conventional subunits in heteromers. Nonetheless, X chimeras derived from BDs of either conventional or modulatory subunits were equally capable of expression alone in Xenopus oocytes, forming channels that responded to cyclic nucleotide when assayed in excised inside-out membrane patches using voltage-clamp recording. Thus, like the previously studied rCNGA4 BD, the BDs of the modulatory subunits fCNGA4 and TAX-2 can support channel activation in a homomeric channel without relying on any residues of a conventional subunit BD.Sensitivity of the X chimeras to low agonist concentrations was quantified by K1/2, the concentration eliciting half-maximal activation (Table I). The chimera X-fA4, containing the fCNGA4 BD, stands out because of its extreme sensitivity to cAMP. Single-channel records (Fig. 2A) show that 10 μm cAMP is sufficient to increase the open probability (Popen) of X-fA4 to >0.99 (only rare brief channel closures were detected). X-fA4 is also sensitive to micromolar cGMP, but this property is not unique, appearing also in X-bA1, X-TAX4, and X-TAX2. However, these latter chimeras have K1/2,cAMP more than an order of magnitude higher than K1/2,cGMP. Thus, the fCNGA4 BD in X-fA4 is unique in that its high sensitivity applies similarly to both agonists.Table IActivation properties of BD substitution CNG channel chimeras (“X-chimera” series) The numbers of patches are given in parentheses for the means ± S.D. of values from individual patches; the values without number of patches are ratios of means.ChimeraBD sourceK1/2Selectivity ratiosSynonym and reference for previous datacAMPcGMPSensitivity (K1/2,cGMP/K1/2,cAMP)Efficacy (Imax,cAMP/Imax,cGMP)μmX-fA4catfish CNGA41.01 ± 0.17 (11)0.73 ± 0.10 (11)0.721 ± 0.034 (11)0.991 ± 0.020 (11)X-rA4rat CNGA481 ± 85 (9)120 ± 140 (9)1.45 ± 0.14 (9)1.024 ± 0.019 (9)X-β (12)X-fA2catfish CNGA2643 ± 29 (6)261 ± 12 (6)0.407 ± 0.0261.38 ± 0.11 (5)X-α (12)X-bA1bovine CNGA1432 ± 99 (4)1.8 ± 1.0 (10)0.0042 ± 0.00250.98ROON-S2 (21,22)X-TAX4C. elegans TAX-4680 ± 37 (2)3.70 ± 0.59 (2)0.00546 ± 0.00057 (2)0.743 ± 0.039 (2)X-TAX2C. elegans TAX-274 ± 33 (6)2.0 ± 1.3 (6)0.0286 ± 0.0049 (6)0.926 ± 0.044 (6) Open table in a new tab Fig. 2The fCNGA4 BD mediates efficient activation with cGMP selectivity.A, excerpted recordings of a single homomeric X-fA4 channel in an inside-out patch held at -80 mV, with steady-state cAMP concentrations indicated. Dashes mark closed (C) and open (O) channel currents. B, macroscopic currents of X-fA4 or X-rA4R/fA4C homomers (line styles are as described in the legend Fig. 1A) in inside-out patches at -80 mV, in response to perfused agonists (open bars; dotted outlines indicate agonist washout). Gaps in traces are arbitrary time intervals. C, relative currents of X-fA4 (circles) and X-rA4R/fA4C (triangles) activated by cAMP (solid symbols) and cGMP (open symbols), in the patches shown in B at -100 mV. The lines are fits of the Hill equation. D, each connected pair of points plots K1/2,cAMP (solid symbols) and K1/2,cGMP (open symbols) obtained from an individual patch from a distinct oocyte expressing either X-fA4 (circles), X-rA4R/fA4C (triangles), or X-rA4 (inverted triangles). Point pairs are horizontally displaced for visual clarity. The data for X-rA4 are from Ref. 12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar.View Large Image Figure ViewerDownload (PPT)A given subsaturating concentration of cGMP reliably elicited larger X-fA4 currents than did the same concentration of cAMP (Fig. 2B, upper traces). This cGMP selectivity was unexpected because the fCNGA4 BD has a methionine residue in its putative C-helix ligand contact position (11Varnum M.D. Black K.D. Zagotta W.N. Neuron. 1995; 15: 619-625Google Scholar), where other cGMP-selective X chimeras (X-bA1, X-TAX4, and X-TAX-2) have aspartate. In fact, in the rCNGA4 BD, a methionine ligand contact imparts cAMP selectivity (28Pagès F. Ildefonse M. Ragno M. Crouzy S. Bennett N. Biophys. J. 2000; 78: 1227-1239Google Scholar, 29Shapiro M.S. Zagotta W.N. Biophys. J. 2000; 78: 2307-2320Google Scholar), as confirmed in X-rA4 (12Young E.C. Sciubba D.M. Siegelbaum S.A. J. Gen. Physiol. 2001; 118: 523-546Google Scholar). The cGMP selectivity of X-fA4 might thus suggest that the methionine in the fCNGA4 C-helix cannot form the ligand interaction required for cAMP selectivity, perhaps because of a different structure in this region. We disproved this possibility by constructing a new X chimera, named X-rA4R/fA4C, which contains the roll subdomain of rCNGA4 (indicated by subscript R) and the C-helix of fCNGA4 (indicated by subscript C); this chimera shows cAMP selectivity at low agonist concentrations (Fig. 2, B and C). Fig. 2D summarizes K1/2 measurements for cAMP and cGMP in individual patches of X-fA4, X-rA4, and X-rA4R/fA4C, which all have the methionine ligand contact. Even though the absolute K1/2 values observed for a chimera varied from patch to patch, the K1/2 selectivity of X-fA4 was always in favor of cGMP, whereas both X-rA4 and X-rA4R/fA4C always favored cAMP. This shows that the fCNGA4 C-helix has cAMP-selective determinants (presumably including the ligand contact methionine) similar to those of the rCNGA4 C-helix. We propose that the intact fCNGA4 BD exhibits cGMP selectivity because the fCNGA4 roll subdomain contains some cGMP-selective elements, whose energetic contributions to gating properties outweigh those of the cAMP-selective C-helix.Efficacies Compared Using a New Series of BD Substitution Chimeras—Comparisons of X chimeras clearly show differences in BD sensitivity, but the K1/2 parameter in isolation is poorly informative of microscopic physical processes such as ligand binding or channel opening (30Colquhoun D. Br. J. Pharmacol. 1998; 125: 924-947Google Scholar). It is more valuable to compare the efficacy (maximal open probability, Pmax) in saturating agonist concentrations where every channel should have uniform (i.e. maximal) BD occupancy. Unfortunately, several X chimeras have extremely high efficacy (e.g. X-fA4 in Fig. 2A), and differences in their respective equilibrium constants for channel opening at saturating agonist concentration would not be easily detectable, because the numerical difference in their Pmax values would be too small. We predicted that BD substitution chimeras using a design similar to the X chimeras but containing a different N-S2 sequence that disfavored intrinsic channel opening might exhibit efficacies that were significantly less than unity; efficacy differences between chimeras would then be more readily apparent. We therefore incorporated the BDs from modulatory subunits (rCNGA4, fCNGA4, and TAX-2) into a new series of chimeras, called “RO chimeras,” in which the N-S2 sequence was that of intact bCNGA1 (Fig. 3A). The RO chimeras still contain the P-loop of fCNGA2 to facilitate single-channel recording, but all of the sequence N-terminal to the P-loop matches exactly that of bCNGA1. All of the RO chimeras were found to be capable of forming functional homomeric CNG channels, repeating our success with the X chimeras. Table II summarizes the properties of the new RO chimeras and includes also RO-bA1 containing the bCNGA1 BD, which was called RO133 in previous studies (13Paoletti P. Young E.C. Siegelbaum S.A. J. Gen. Physiol. 1999; 113: 17-34Google Scholar, 21Tibbs G.R. Goulding E.H. Siegelbaum S.A. Nature. 1997; 386: 612-615Google Scholar, 23Goulding E.H. Tibb