Proc.Nati.Acad. Sci. USA
Vol.91,pp. 8890-8894,September1994 Neurobiology
Heteromeric olfactory cyclic nucleotide-gated channels: A subunit
increased sensitivity to cAMP
NORMAN DAVIDSON, HENRYA. LESTER, AND KAI
Contributed byNormanDavidson, May20, 1994
ABSTRACT Olfactory receptor neurons respond to odor-ant stimulation witharapid increaseinintracellular cAMP that opens cyclicnucleotide-gated (cng) cation channels. cng chan-nels in rat olfactory neurons are activated by cAMPinthe low micromolar range and are outwardlyrectifying. The cloned rat olfactory cng channel (rOCNC1),however, is much less sen-sitive tocAMPandexhibits very weakrectification. Here we describe thecloningandcharacterizationof a secondratcng channel subunit, denoted rOCNC2. rOCNC2 doesnotform functionalchannels whenexpressed alone. WhenrOCNC1and rOCNC2 are coexpressed, however, an outwardly rectifying cation conductance with cAMP sensitivity near that of the native channel is observed. Insituhybridization withprobes specificforthe twosubunits showsthat theyarecoexpressedin olfactory receptorneurons.Thesedataindicate thatthe native olfactory cng channelis likely to be a heterooligomer ofthe rOCNC1 and rOCNC2 subunits.
One mechanism ofolfactory signal transduction involves a rapid and transient increase in intracellular cAMP. In re-sponse to thebinding ofodorants,guanine nucleotide-binding protein (G protein)-coupled receptors activate a G,-like G protein which increases the enzymatic activity of adenylyl cyclase(reviewedinrefs.1and2).Theresultingincrease in cAMPopenscyclicnucleotide-gated (cng)cation channels in the ciliaand dendrites of olfactory receptor neurons. This produces an influx of monovalent and divalent cations, depolarizingtheneurons(3-7). Theinfluxof Ca2+ activates an outwardCl- conductance (8-10), and Cl- effluxacts to amplify the initial depolarization,
triggeringsensory nerve impulses.
Molecular data support this model of olfactory
transduction. Olfactory receptor neurons express a very large family of G protein-coupled receptors (11). These receptorscouldinteract withGod,aGproteinalsoexpressed inolfactory neurons that is very similarin sequence to
(12). Anadenylyl cyclase (13)andacngchannel(14-16)have also been cloned from olfactory epithelium and character-ized.
The ratolfactory cng channel clone rOCNC1 forms cAMP-activated channels whenheterologously expressedin mam-malian cells. The conductance is characterized by a half-maximallyeffectiveconcentration
and very weakrectificationin the absenceofdivalent cations (14). In contrast, the native conductance in rat olfactory neuronsis much moresensitivetocAMP(EC5oof 2.5 ,uM) and is outwardly rectifying (7). There are severalpossible explanationsfor these functional differences. Forexample,a secondolfactorychannel mightexist that woulddisplaythe cAMP sensitivity and rectification behavior ofthe native channel when heterologously expressed. Alternatively,
an-other channel subunit might modulate the properties of rOCNC1 byformingheterooligomeric channelswith
rOCNC2 cDNA Clones. Primers CN2 [5'-AARYTIGCIG-TIGTNGCNGA-3', corresponding to rOCNC1 aa 500-506 (Lys-Leu-Ala-Val-Val-Ala-Asp)] and CN1
[5'-AT(R)TTI-GCIGTIC(K)IC(K)(R)TTNCC, correspondingto aa535-542 (Gly-Asn-Arg-Arg-Thr-Ala-Asn-Ile)]wereusedfor polymer-asechainreaction (PCR) (17)experiments using oligo(dT)-primed first-strand cDNA synthesized from template rat nasalepithelial RNA. Theannealingtemperature was
and 35 amplification cycles were performed. Restriction analysis ofthe PCR product revealedthat itcontained two sequenceclasses; oneofthesecorrespondedto therOCNC2 sequence. To obtain an rOCNC2-specific hybridization probe, anrOCNC2-specificPCRprimer wasdesignedfrom aa 417-426 and used in combination with vector primers flanking thepolylinker of AZAPII (Stratagene) toPCR am-plify sequences from a rat nasal epithelial AZAPII cDNA library.Theends ofthe1090-bp fragmentthusisolatedwere sequenced, and new rOCNC2-specific PCR primers were designedand used to amplify anon-cross-hybridizing
106phage from thelibrary and8full-length rOCNC2 cloneswereisolated;2 of theseweresequenced (29). Wesequencedthe 5' endsof all8 clones todeterminewhether asecondformof rOCNC2 with a longer N terminus might exist, but we found no evidence for such aform.
Channel mRNA Levels. cDNAwasquantitated by ampli-fying with
(-actinprimers (15and 20cycles)andtheamounts ofcDNAineachchannelprimeramplificationwereadjusted onthe basis oftheseresults.Primer sequencesareavailable onrequest. For the channel mRNAamplifications -1ngof cDNAwas amplified for 25, 30, or35cycles. The amplifi-cationproductswereelectrophoresedina
1.5%agarosegel, stained with ethidiumbromide, and
DigoxigeninantisenseRNAprobes weresynthesizedwith an in vitrotranscriptionkit(Ambion, Austin,TX) in the presence of
digoxigenin-UTP(at aratio of 1:3relativeto
probe templatewas a
full-lengthcDNA cloneof3 kb. The17receptorprobe templatewasthe entire coding sequence (984 bp). rOCNC1 and rOCNC2 probe templates were generated from pBluescript (Strata-gene) clones of rOCNC1 and rOCNC2byPCRusing primers 1849
(rOCNC1;corresponding to aa 624-630) and 1666 (rOCNC2;aa518-524), incombinationwithvectorprimers flankingthe transcription promotersatthe 3' endsofthese Abbreviations: cng,cyclic nucleotide-gated; CN,cyclic nucleotide-bindingdomain; HEK, human embryonic kidney; Vm, membrane potential.
*To whomreprintrequestsshould be addressed.
tThe sequence reported in this paper has been deposited in the GenBank data base(accessionno.U12623).
Thepublicationcostsof this articleweredefrayedin partbypagecharge
rOCNC1 NNTEKSNCVK SSPANNENNN PPPSIKANGKDDNRAGSRPQ SVAADDDTSP 50 rOCNC1 ZLQRLAZDMTPRRGRGGFQR IVRLVGVIRD WANKNFZE PRPDSFLKRF 100
rOCNC2 NSQDGKVKTTES TPPAPT1ANEWLPVLDPSGD YYfILN!KV 42
rOCNC1 RGPZLQTVTT NQGDDKGKD GCGKGTKKI ZLFVLDPAGD WYYRLFIA 150
~~~~~~~S2G . .. XK... .L.VLDP.GD .YY.lL.... 150
rOCNC2 FPIKYNLIIV VCRACJPDLQ NISYLVMVTV DYTSDLLYLLDIGVRFIITCF 92
rOCNCl NPVLYNUCLL VARACFSDLQ RNYFVVWLVLDYFSDTVYIA DLIIRLRTGF 200 Consensus P. YN.... V.RACF.DLQ .Y.V.W.VL DY.SD.Y.. D..R..TGF 200
rOCNC2 LKQGILVVDKGKIASR YV8LVTA QLGPIP 142 rOCNC1 LNQCLLVIDP IKLRDNYIXT LQFKLDVASI IPTDLIYFAV GINSPNVRIN 250
Consensus LEQG.LV.D...Y.T ..F.LD.AS. P..Y... G.N.P..R.N 250
rOCNC2 RFLRVPRLN aFORDTRSAYPNAFR L KLYIJVININSCLYFALSR 192 rOClC1 RLLWFARNEW TFDRTRT8 YPNIFRISNL VLYILVIIDN NACIYYVISK 300
Consensus R.LL...R.rF .]FDRTTRT. YPN.FRI..L LYI.V.IXN N.CY .. S. 300 p
rOCNC2 YLGFGRDANVYPDPAQPGFI RLRRQYLYSF YrsTLILT!vGDTPLPDNRZ 242
rOCNC1 SIGFCVDTUVYPNITDPZYG YLARZYIYCLYWSTLTLTTI GETPPPVKDZ 350
Consensus .GFG.D.WV YP....P... L.R.Y.Y.. Y.STL.LTT. G.TP.P.. Z 350 S6
rOCNC2 zYLFKIVQDFL LAVNFATIN GNMNSVIYNMNTADAAFYPD NALVKKYMKL 292
rOCNC1 EYLFVIFDFL IGVLITATIV GNVGSNMISN NATRAZFQAK IDAVKNYNQF 400 Consensus ZYLF..DFL V. FATI. G...S.I.NK N...A.F .VIVK.YK.. 400
rOCNC2 QIVUKRLKRR VIDWYQNLQINM(TNKVAI LQNLPENLRA ZVAVSVNI8T 342 rOCNC1 RNVSKDIAJM VIKNFDYLW NIETVDNRZV LKNLPAKLRA ZIAINVNLNT 450 Consensus . . .1.. VIN....L.. ....Z... L..LP..LRA E. A. . VLST 450
rOClC2 LSRVQIFQNCXASLLNLVLILQPQTYNPG NYCKGDIG RNKYIRNNGQ 392
rOCNC1 LKKVRIFQDCZAGLLVZLVL ILRPQVNSPG DYICRNDIG tUKYIYIKI 500
Consensus L..V.IFQ.C KA.LL.ZLVLKL.PQ. SPC .Y.CRNKDIG .DKYII.KG. 500
rOCNC2 LAVVADDGVT QYAVI4&GLY ZI8IINIKI GUKSGNRRTA NIKNLCYSDL 442
rOCNC1 LAVVADDWVT QYAOLLNAGSC NINILNIK ONS URTA NJIRSLCYDL 550
Consensus LAVVADDGVT QYA.L.AG.. FGIISI.NIK C... CRRTA NI.SLGYSDL 550 rOCNC2 FCLSKNDLRN VLSNYPQAQA VNKIKLNNILLDEN WVNAZAAZIALQEA 492
rOCNC1 rCLSXDDIMN AVTZYPDANR VLKKRGRZIL MKGLLDDNI VAANNNVD-- 598 Consensus TCLSK.DL.3 ....ZP... V.ZN.GRCIL .... LD.N.aa... 600
rOCNC2 TKRLK.GLDQ QLDDLQTKFARLLAILISSA LIIAYRINRLKNQTIZWPNP 542
rOCICi VQNKLNQLNTNDIDTLYTRFARLLAZYTGAQ QKLQRITVL&--TX---- 641 Consensus ....L..L.. ..D.L.T.FA RLLAZ. RI. .LZ.T... 650 rOCNC2 KDNGADDKA PGNGTSIDG GACQAGCPSGIZ 575
rOCNC1 --IIQNMCDDYL8DGIIIT-- ---PZPTAAR 664
Consensus . .P.P. ..Z 683
FIG. 1. Aligned amino acid sequences of rOCNC1 and rOCNC2 proteins.Inthealignment, a dash indicates the absence of an amino acid,andadot in the consensus line indicates an amino acid that is not conserved between the sequences. S1-S6are putative trans-membranedomains, and P is the putative pore region. The cyclic nucleotide-binding (CN) domain is defined by homology to the sequences of otherproteins that bindcAMP and cGMP (20, 21). genes. These probes did not cross-hybridize to the other channelsequencebyNorthernblot analysis using full-length sense-strand in vitro transcripts of rOCNC1 and rOCNC2 cDNA clones(J.B.,unpublishedresults). In situ hybridiza-tionwasperformedasdescribed (18). Sections weremounted
Proc. Natl. Acad. Sci. USA 91 (1994) 8891 in glycerol andphotographed undercoverslips withaZeiss Axioplan microscope.
Electrophysiolgy.The pCIS expression vector was used and transfections into humanembryonic kidney (HEK) 293 cellswereperformedasdescribed(19). Patch-clamp record-ingsweremade1-3daysafter transfection. Electrodeswere fabricated from borosilicateglassand hadresistances of1-3 Mfl. Cyclic nucleotides wereappliedbyafast microperfu-sion system that allowed solution changes within 100ms. Divalent cation-free solution was 140 mM NaCl/5 mM KCI/10 mM Na Hepes/0.5 mM Na EDTA/0.5 mM Na EGTA, pH 7.6. Solution withdivalentswas140mMNaCl/5 mM KCl/10mM NaHepes/2 mM
CaCl2/1mMMgCl2,pH 7.6. Single-channel recordings weremade from excised in-side-outpatches withsymmetrical divalentcation-free solu-tions. Datawerefiltered at2kHz,digitizedat16-bit resolu-tionand 22-kHzfrequencyforstorage onmagnetictape,and redigitized and analyzed at 4 kHz with AXOTAPE 2.0/ FETCHAN 6.0(Axon Instruments, FosterCity, CA).
We usedPCR toamplifycngchannel-relatedsequencesfrom primary olfactory epithelial cDNA. We identifiedan ampli-fication product whose sequence is closely related to rOCNC1. This product was used to isolate a full-length cDNA clone, denotedrOCNC2. The predicted amino acid sequence of rOCNC2 comprises 575 residues and is 51% identical to rOCNC1 (Fig. 1). Hydropathyplots of the two sequences are almost superimposable, indicating similar transmembranetopology. Each proteincontains six putative transmembrane regions (S1-S6), and a pore(P) region ho-mologoustothe Pregion of voltage-gated channels.
The CN domain is highly conserved (77% identity). A threonineresidue whichinfluencesselectivity for cGMPover cAMP (22) is present in both sequences (residue 539 in rOCNC1). However, there are three adjacent nonconserva-tive differences Ser-Lys-Met (aa 532-534) in rOCNC1 vs. Asn-Met-Ser (aa 424-426) in rOCNC2 withinthe most con-served part of the CN domain (20, 21). The S4 region of rOCNC2, whichcorrespondstothe putativevoltage sensor ofvoltage-gated channels, is likelytobearone or two more positive charges than the corresponding sequence from rOCNC1 (two ifHis254ofrOCNC1 is uncharged). The two sequencesalso differ at aposition in the P region(Glu3M2in rOCNC1, which corresponds positionally to Asp234 in rOCNC2)thatinfluencesinteractions betweendivalent
MQZ.> X0 ow ato os n> cW0 0 a)MM Mzm
ooIoo+-M U Ml 50G.
:29322C(2 12- --i4 159 "
specificfor the 3' untranslatedregionsof thetwomRNAswereusedtoamplify
oligo(dT)-primedcDNAmade from RNAisolated from variousrattissues. Total RNAwasusedexceptwhereindicated. This isanegativeimageof the ethidiumbromide-stainedagarosegel. The rOCNC1 PCR
productis 122bp.Lanes:M,markers(sizesinbases indicatedatright);NepA, nasalepithelium
poly(A)+mRNA];Lu,lung;Ov,ovary; Nel andNe2,nasalepithelium(twoisolates);
Co,cortex;Ce,cerebellum; He, heart;OBI and OB2,
8892 Neurobiology: Bradleyetal.
ions and the channelpore(23-25).Finally, rOCNC2 is109 aa shorter than rOCNC1 at its N terminus.
Weused quantitative reverse transcription-PCR to assay the tissue-specific expression patterns of the rOCNC1 and rOCNC2 mRNAs. A 30-cycle amplification is shown in Fig. 2; both mRNAs are highly enriched in the olfactory epithe-lium, although low-level expression can be detected in brain and olfactory bulb. In 35-cycle amplifications, bothof the PCRproducts can be detectedwhencDNA from wholebrain, cortex,cerebellum, and olfactory bulb is used, but notwith cDNA from any of the other tissues (J.B., unpublished results).
To examine the cell-specific expression patterns of the mRNAs, we performed in situ hybridization of digoxigenin-labeled rOCNC1 and rOCNC2 probes to sections of rat olfactory epithelium. We also used probes recognizing mRNAsencoding
Golf(12) and the I7 olfactory receptor (11) to visualize olfactory neurons and to provide controls for nonspecific background hybridization.
GolfmRNA is ex-pressed athighlevelsthroughoutthe neuronallayer (Fig. 3B), whereasI7 mRNA, like other olfactory receptormRNAs(26, 27), isexpressedonly in a small subset ofneurons(Fig. 3A). BothrOCNC1 (Fig. 3C)and rOCNC2(Fig. 3D)mRNAs are expressedin theolfactoryneuronallayer.There issignificant heterogeneity in rOCNC2 expression among individual neu-rons, whereas rOCNC1 is more homogenously expressed.
AN -r_ 0
z."V- + 60mV 1.0 0.8 0.6 0.4 10 100 cAMP,
jgMV-+ 60 mV rOCNCI 0.2 -0.0 0.1 (1:1) 10 cGMP,JAM B +60 +60 -60
10k-I50 . 5,10,20_ [cAMP],,uM 20 10 5
1_0, 10,5 50 100 200
72 5 10 20 SO 400 pA 200 ms rOCNC1 200
pAL200 ms rOCNC1/rOCNC2(1:1) t " 4 4 ;^ B..
_'...:X '1 ' 'S 'S./;; fi -z; ,$5 .S N
BpC D C C C.) E 0 z -60 mV -0.2
FIG. 3. Expression of rOCNC1and rOCNC2 mRNAs within the
neuronallayerofthe olfactory epithelium. We performed in situ
hybridization to20-Mmhorizontalsections ofratolfactory epithe-lium, usingdigoxigenin-labeledantisense RNAsasprobes.A-Dare
high-magnification photographs (x788),taken with Nomarskioptics, of thesameregionof theepithelium.Theolfactoryciliaareatthetop,
and below themarethesupportingcelllayer (S),the neuronallayers (N), and the basal cell layer (B),asindicated in B. Cellsexpressing aparticular mRNA are visualized asdark disks. (A) 17 olfactory
receptorprobe. One I7-expressingneuronisobservedinthis pho-tograph. The 17probe, which is about twiceaslargeasthe channel probes, alsoserves as acontrol fornonspecific hybridization. (B) Golf
probe. Notethe uniformhybridizationwithin the neuronallayers. (C and D) rOCNC1 and rOCNC2 probes, respectively. Widespread hybridization in the neuronal layers is observed with these two probes. Note the heterogeneityinrOCNC2mRNAexpression.Acell
expressing rOCNC2mRNA athigh levelsis indicatedbyablack
arrowhead, anda cell lackingthe mRNAorexpressing itat low
levels, byawhite arrowhead. C andDareadjacent sections;Aand
Baresections within50,uMofCand D. Examination of theepithelial sections at low magnification shows that rOCNC1 and rOCNC2
mRNAsarepresent in allregionsof theepitheliumthatcontainG.If mRNA. (Bar=5,uM.)
FIG. 4. cngconductancesproduced by rOCNC1 expression and
by rOCNC1/rOCNC2 coexpression. Recordingswereobtained from inside-out patches excised from HEK 293 cells transfected with
rOCNC1plasmidorwith a1:1 mixture ofrOCNC1 and rOCNC2 plasmids. (A) Normalized dose-response relations for cAMP (Left) andcGMP(Right)at Vm = +60 mV.Leftis derived from
macro-scopiccurrent data for the two cells shownin B. Right displays cGMPresults fromtwoother cells. Currentsweremeasuredatthe
end of the 800-msvoltage pulse.Thesymbolsrepresent theaverage responses;the smooth linesaredescribedby the Hill equations with best-fitting values. (B) Macroscopic currents recorded from two patches (Left, rOCNC1; Right,rOCNCl/rOCNC2).Vmwasstepped
from 0 to +60or-60mVfor 800ms. The patches werekept in
symmetrical divalent cation-free solutions,andthebathcontained cAMPatthe concentrationsindicated.Toptracesshow thevoltage commands for eachepisodeinatrial.Episodesshownareaveraged
from threetosixtrials,each takenduringaseparateseries of cAMP
applicationsatascending concentrations. Leakcurrentshave been
subtracted. Filtercorner frequency, 2 kHz. (C) Current-voltage relation in thepresenceof extracellular divalent cations. Currents
wereactivated with cAMPat50 MM forrOCNC1/rOCNC2 patches
(n = 3), andat200pM for rOCNC1 patches (n = 4). Both were
normalized to currents at Vm of +60 mV. Leak currents were
subtracted. Error bars indicate SEM.
Proc. Natl. Acad. Sci. USA 91 (1994) 8893 Table 1. Properties ofcyclicnucleotide-activated conductances observed in HEK 293 cellsexpressingtherOCNC1 and
rOCNC1/rOCNC2channels,comparedwith thosepreviously measuredfor theratnativeolfactorycngchannels (7)
Channel type Vm, mV EC5o, /JM Hillcoefficient EC50, uM Hillcoefficient EC50(cGMP) rOCNC1 -60 48 ±4.6(n= 11) 2.8±0.3(n = 11) 1.6 ±0.2(n =7) 2.1 +0.2(n =7) 30
+60 47 ±3.5(n= 11) 2.6 ±0.3(n = 11) 1.4 ± 0.1(n =7) 2.4+0.4(n =7) 34
rOCNC1/rOCNC2 -60 10.8 ±1.7(n=8) 1.8 +0.2(n =8) 2.7±0.5(n = 8) 1.8 ±0.1(n= 8) 4.0 +60 7.3 ±0.8(n= 8) 1.9 +0.1(n = 8) 2.9±0.4(n =8) 1.6 + 0.1(n= 8) 2.5
Native channel +50 2.5 1.8 1.0 1.3 2.5
Results are given as mean ± SEM.
These patterns suggest that some olfactory neurons may expressonlytherOCNC1 channel.Thiscouldexplainearlier observationsthat the cngconductance ofasubpopulationof neurons exhibited alower sensitivity to cAMP(3). In situ hybridization to brain sections shows that both channel subunit mRNAs arealsoexpressed insubsets ofneuronsin theolfactory bulb, cerebellum,andcortex(J.B., unpublished results).
To determine whether rOCNC2 could function as a cng channel, we transientlytransfected HEK 293 cells with an rOCNC2 expression vector. Tento40% of the transfected cells exhibited bright surface staining with an rOCNC2-specificantiserum(J.B., unpublished results), butwecould notdetect any cyclic nucleotide-activated conductances in excised inside-out patches from these cells.
To evaluatewhetherrOCNC2 couldalter theproperties of therOCNC1 channel, wetransfectedHEK293 cells with the rOCNC1 expression vector alone orwith amixture of the rOCNC1 and rOCNC2 plasmids at a 1:1 molar ratio. At a membrane potential (Vm) of +60 mV, an inside-out patch from anrOCNCl-expressing cell displayed acyclic nucle-otide-activated conductance with an EC50 for cAMP of64 ,uM.Acellexpressingbothsubunits,in contrast, hadanEC50 for cAMP of 6.3,uMat+60 mV(Fig.4ALeft; macroscopic current tracesfrom thesetwopatchesareshown inFig. 4B). Whileexpression of rOCNC2increased the apparentaffinity of the channelforcAMP,ithadtheopposite effect forcGMP. TheEC50 for cGMP was 1.5 ,uMfor acellexpressing only rOCNC1 butwas2.8,uM foracellexpressingboth subunits (Fig. 4A Right). The data obtained from 34 patches are summarized in Table 1.
Table 1 also shows that the Hill coefficients for the rOCNC1/rOCNC2channelapproximatethose observedfor the nativechannel (7)andaresignificantlylower than for the rOCNC1 channel. We do not know whether the lower Hill coefficients of the
rOCNC1/rOCNC2channelare due to a reducedcooperativity atthe molecular levelor tothe pres-enceofaheterogeneouspopulationof channels in thepatch thathavedifferent subunitstoichiometries. Incells cotrans-fected with rOCNC2 andrOCNC1plasmidsatratios of3:1or 6:1, the averageEC50for cAMPwas6.9± 0.8at+60 mV(n = 8). Thus, increasing the relative proportion of rOCNC2 doesnotincrease thesensitivitytocAMPbeyondthe values observedwitha1:1ratio. TheEC50values forbothcAMP and cGMP thatwemeasuredfor therOCNC1/rOCNC2channel in HEK 293 cellsare
=3times those observed for the native ratchannel (7)(Table 1). This may be due to differences in experimental conditions, in posttranslational modification, orininteractions with cell-specific modulatory factors. For example, a recent study demonstrated that the cyclic-nucleotidesensitivityof therOCNC1and native rat channels canbedramatically alteredby direct interaction withCa2+/ calmodulin (28).
The differences in apparent agonist affinity between the heterooligomericandhomooligomeric channels may be par-tiallydue tothethree adjacent nonconservative amino acid substitutionswithin thehighlyconserved CN domain(Fig.1).
In the determined three-dimensional structures of CN do-mains from another protein (21), these three positions are within aloop that forms part of thecyclicnucleotide-binding pocket.
TherOCNC1/rOCNC2 channel shows outward rectifica-tion insymmetrical divalent-freesolutions.Thekineticbasis forthisrectificationisa currentrelaxation (timeconstantof 30-50ms) to smaller or larger amplitudes followingastep from zero to negative or positive membrane potentials, respectively (Fig. 4B Right). The relaxation was not observed for cells transfected with rOCNC1 alone(Fig.4BLeft).The increased voltagedependence of the heterooligomeric chan-nel could be due to thelarger numberofpositive charges in the S4 domain of rOCNC2 relative torOCNC1.
In the presenceofextracellular divalentcations, both the homooligomeric and heterooligomeric channels are out-wardlyrectifying, but the divalents have a smaller effect on the heterooligomeric channel. In the absence of divalent cations, theratio ofcurrentamplitudesat -60and +60 mV (I-60/I+60)was0.84forrOCNC1(seeFig. 4B), and this ratio decreasedby8-fold,to0.11, with theinclusion of 2 mMCa2+ and 1 mM Mg2+ in the extracellular solution (Fig. 4C). Divalent cations had a smaller effect on the rOCNC1/ rOCNC2channel, decreasingI-60/I+60 by only a factor of 3 (from0.62 to0.2). These data suggest that the two channels could have different Ca2+ permeabilities. Permeability to Ca2+ is likely to be functionally important since a major componentof thedepolarizingcurrentinolfactoryneuronsis carried byaCa2+-activatedCl- conductance (8-10).
The rOCNC1/rOCNC2 channel also differs from the rOCNC1 channel in itssingle-channel properties. The
open-rOCNCl rOCNCl/rOCNC2 i---.-z--- 0 2
I-0----+60mV MpA 0 ms s, ..
-I-0 2 -60mV --02
FIG. 5. Single-channel properties of homooligomeric rOCNC1 channel (the patch contained hundreds of channels, thus 0.1 ,uM cAMP was used) (Left) and heterooligomeric rOCNC1/rOCNC2 channel(the patch contained one channel, so 2pMcAMP was used) (Right). Vmwas +60mV (Upper) or -60 mV (Lower).01and 02 represent theconductance from the openings of one and two chan-nelswithin therOCNC1patch, respectively. C represents the closed
state.The dotted line represents the currentamplitude corresponding
to asingle-channelconductance of -48pS. Consecutive sweeps are shown.
8894 Neurobiology: Bradleyet al. rOCNC2 hRCNC2 rOCNC1_ bOCNCI mRCNC1 -hRCNC1 'bRCNC1
FIG. 6. Oneunrooted parsimonioustreeofcngchannel coding-region DNAsequences, calculatedon aVAXcomputerusing the Phylogeny Inference Package ofprograms(30). Prefixes:h, human;
r,rat;m,mouse;b,bovine; f, catfish.
ings of the rOCNC1 channel are stable, last for tens of milliseconds, and haveamaximalconductance of=48 pSat +60 mV (Fig. 5 Left). The rOCNC1/rOCNC2 channel is
flickeryat+60mV, making it difficulttoaccuratelymeasure
single-channel conductances. At -60 mV, the flickering is further accentuated(Fig. 5 Right).Thisreduces the effective single-channel conductanceand contributestooutward
rec-tification. Flickery opening behaviorhas alsobeenobserved forthe nativechannel(7).
Wehaveidentifiedasecondsubunit of theratolfactorycng
channel, rOCNC2. This subunit doesnotformafunctional channelbyitself butheterooligomerizeswith thepreviously identifiedolfactory channel, rOCNC1,toproduceachannel whoseelectrophysiological behaviordiffersfrom that ofthe homooligomeric rOCNC1channel.
Additionalsubunits thatmodulate channelpropertieshave been identified forK+ channelsand for variousligand-gated channels. A modulatory subunit for the human rod cng
channel, hRCNC2,has alsobeen described(19).This subunit doesnotfunctionas anion channel when expressed alone,
butcanformfunctionalheterooligomerswiththepreviously characterizedrodcngchannel,hRCNC1.Comparedwith the hRCNC1 channel, the hRCNC1/hRCNC2 channel is more
similartothenativephotoreceptorchannel in itssensitivity todrug blockade and in its single-channel properties. The openings of botholfactoryand retinalheterooligomeric chan-nelsareflickery, while theopeningsof bothhomooligomeric
channelsarestable. LikerOCNC2,hRCNC2isshorteratits Nterminus than thefirst rod channel subunit.
The data described above indicate that the native cng
channels inolfactoryneuronsandinretinalrodsare similar inmany respects. An evolutionary tree of the cngchannel
sequences (Fig. 6) suggests that therOCNC2 and hRCNC2 subunits arose from a common ancestral protein after the rOCNC1 and hRCNC1 subunits diverged. Thus, the olfac-toryand retinalchannels mayhaveonce sharedacommon
The rOCNC2 subunit confers several propertiesthat are
characteristic of the native channel fromolfactory neurons
butdifferfromthoseofthe homooligomericrOCNC1 chan-nel.These includeagonistsensitivity, rectification,and sin-gle-channel behavior. Theratio of theEC50values for cAMP
versuscGMPis34for thehomooligomeric rOCNC1 channel
and 2.5for theheterooligomeric rOCNC1/rOCNC2 channel (Table 1).The valueof 2.5 is equaltothatobservedforthe
native rat channel from olfactory neurons. Our results, together with the studiesonthe retinal channel (19),provide molecular and functional evidence for the presence of heterooligomericcngchannels in vivo.
Note Added in Proof. Similar data on therOCNC2channelsubunit havebeenobtainedby E.Liman and L. Buck, and a paper describing theirresults is in press in Neuron.
Wethank T.-Y. Chen and K.-W. Yauforhelpful discussions,HEK 293 cells, and an rOCNC1 clone; W. Zagotta for providing the sequences ofthe CN1andCN2 primers;Genentech forthe pCIS plasmid; M. Quick and Y. Uezono of Caltech for experimental contributionstothe early phase of this project; andL.BuckandE. Liman for communicating data before publication. J.B. was sup-portedbyaNationalInstitutesof Healthgraduatetraininggrant and an Achievement Award for College Scientists. This work was supportedbygrantsfrom theNational InstituteofMentalHealth and theNational Institutes of Health toK.Z., H.A.L., andN.D.
1. Reed,R. R. (1992) Neuron 8, 205-209.
2. Lancet, D. &Ben-Arie,N.(1993) Curr.Biol. 3, 668-674. 3. Nakamura, T. & Gold, G.H. (1987) Nature (London) 325,
4. Kurahashi,T.(1990) J.Physiol. (London) 430,355-371. 5. Firestein,S.,Darrow, B.&Shepherd, G. M.(1991) Neuron 6,
6. Zufall, F.,Firestein, S. & Shepherd, G.M. (1991) J. Neurosci. 11,3573-3580.
7. Frings, S., Lynch, J. W. & Lindemann, B. (1992) J. Gen. Physiol. 100, 45-67.
8. Kleene, S.J. (1993)Neuron11,123-132.
9. Kurahashi, T. & Yau, K.-W. (1993) Nature (London) 363, 71-74.
10. Lowe, G. & Gold,G.(1993)Nature(London)366,283-286. 11. Buck,L. &Axel,R.(1991)Cell65,175-187.
12. Jones,D. T.&Reed, R. R.(1989) Science 244, 790-795. 13. Bakalyar, H. A. &Reed, R. R. (1990) Science 250, 1403-1406. 14. DhaLlan, R.S., Yau, K.-W., Schrader,K. A. & Reed,R. R.
(1990)Nature(London) 347, 184-187.
15. Ludwig,J.,Margalit, T., Eismann,E.,Lancet,D. & Kaupp, U. B.(1990)FEBS Lett.270,24-29.
16. Goulding,E.H.,Ngai, J., Kramer,R. H.,Colicos, S.,Axel, R., Siegelbaum, S.A. &Chess,A. (1992)Neuron 8, 45-58. 17. Saiki,R.K.,Gelfand,D.H.,Stoffel, S.,Scharf,S.J., Higuchi,
R., Horn, G. T.,Mullis,K. B.&Erlich,H. A.(1988)Science 239,487-491.
18. Schaeren-Wiemers,N.&Gerfm-Moser,A.(1993) Histochem-istry 100, 431-440.
19. Chen,T.-Y., Peng, Y.-W.,Dhallan,R.S., Ahamed,B.,Reed, R. R.&Yau,K.-W.(1993)Nature(London)362, 764-767. 20. Kaupp,U. B.(1991) TrendsNeuroSci. 14,150-157.
21. Shabb, J. B. & Corbin, J. D. (1992) J. Biol. Chem. 267, 5723-5726.
22. Altenhofen, W., Ludwig, J.,Eismann, E., Kraus, W., Bonigk, W. &Kaupp, U. B. (1991)Proc. Natl. Acad. Sci. USA 88, 9868-9872.
23. Heginbotham, L., Abramson, T. & MacKinnon, R. (1992) Science258, 1152-1155.
24. Root,M. J. &MacKinnon,R.(1993)Neuron11,459-466. 25. Eismann, E., Muller, F., Heinemann,S. H. & Kaupp, U. B.
(1994)Proc. Nati.Acad. Sci. USA 91,1109-1113.
26. Ressler, K.J., Sullivan, S. L. &Buck, L.B.(1993) Cell 73, 597-610.
27. Vassar,R.,Ngai,J. & Axel,R.(1993)Cell 74, 309-318. 28. Chen, T.-Y. & Yau, K.-W. (1994) Nature (London) 368,
29. Bradley, J., Uezono, Y.,Davidson, N., Lester,H. A.&Zinn, K. (1992)Soc. Neurosci.Abstr. 18,596.