genomic survey of camp and cgmp signalling components in the cyanobacterium synechocystis pcc 6803

Downloaded from www.microbiologyresearch.org by

IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35

Microbiology (2000), 146, 3183–3194 Printed in Great Britain

Genomic survey of cAMP and cGMP signallingcomponents in the cyanobacteriumSynechocystis PCC 6803

Jesu! s A. G. Ochoa de Alda and Jean Houmard

Author for correspondence: Jean Houmard. Tel : 33 144323519. Fax: 33 144323935.e-mail : jhoumard!biologie.ens.fr

Dynamique desMembranes Ve! ge! tales,Complexes Prote! ines-Pigments, CNRS UMR 8543,Ecole Normale Supe! rieure,46 rue d’Ulm 75230 ParisCedex 05, France

Cyanobacteria modulate intracellular levels of cAMP and cGMP in response toenvironmental conditions (light, nutrients and pH). In an attempt to identifycomponents of the cAMP and cGMP signalling pathways in Synechocystis PCC6803, the authors screened its complete genome sequence by usingbioinformatic tools and data from sequence–function studies performed onboth eukaryotic and prokaryotic cAMP/cGMP-dependent proteins. Sll1624 andSlr2100 were tentatively assigned as being two putative cyclic nucleotidephosphodiesterases. Five proteins were identified as having all thedeterminants required to be cyclic nucleotide receptors, two of them beingprobably more specific for cGMP (an element of two-component regulatorysystems – Slr2104 – and a putative cyclic-nucleotide-gated cation channel –Slr1575), the three others being probably more specific for cAMP: (i) a proteinof unidentified function (Slr0842) ; (ii) a putative cyclic-nucleotide-modulatedpermease (Slr0593), previously annotated as a kinase A regulatory subunit;and (iii) a putative transcription factor (CRP-Syn¯Sll1371), which possessescAMP- and DNA-binding determinants homologous to those of the cAMPreceptor protein of Escherichia coli (CRP-Ec). This homology, together with thepresence in Synechocystis of CRP-Ec-like binding sites upstream of crp, cya1,slr1575, and several genes encoding enzymes involved in transport andmetabolism, strongly suggests that CRP-Syn is a global regulator.

Keywords : CRP, cyclic-nucleotide gated channel

INTRODUCTION

Living organisms adapt to a variety of different en-vironmental conditions. After the sensing of environ-mental stimuli, a largely unknown network of signaltransduction pathways allows cells to rapidly respondby changing the activity of target enzymes and}or bymodulating the expression of target genes, thus produc-ing an integrated response. Cyanobacteria are photo-synthetic prokaryotes, the survival of which requiresthe sensing of light, oxygen and nutrient availability.Under standard growth conditions, cyanobacteria con-tain similar intracellular levels of cyclic AMP (cAMP)and cyclic GMP (cGMP) (Herdman & Elmorjani, 1988).Intracellular levels of both nucleotides vary in responseto changes in environmental conditions : light, nutrients

.................................................................................................................................................

Abbreviations: cNMP, cyclic nucleotide monophosphate; CNG, cyclic-nucleotide-gated; CRP, cAMP receptor protein.

and oxygen (Herdman & Elmorjani, 1988; Ohmori,1989; Sakamoto et al., 1991). Therefore, these twocommonly encountered second messengers are likely toplay an important role in cyanobacterial signallingpathways.

Intracellular levels of cAMP and cGMP depend onthe relative rates at which they are synthesized bythe adenylyl- (EC 4\6\1\1) or guanylyl-cyclases (EC4\6\1\2), degraded by the cAMP- (EC 3\1\4\17) orcGMP- (EC 3\1\4\35) phosphodiesterases, and even-tually excreted. The association of cAMP and cGMPwith protein receptors in turn determines the physio-logical response(s). At present, only the cAMP andcGMP biosynthetic enzymes have been identified andcharacterized in cyanobacteria (Kasahara & Ohmori,1999; Katayama & Ohmori, 1997; Ochoa de Alda et al.,2000; Terauchi & Ohmori, 1999).

Cyclic-nucleotide phosphodiesterases play a pivotal rolein cAMP and cGMP signal transduction by regulating

0002-4024 # 2000 SGM 3183


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35

J. A. G. OCHOA DE ALDA and J. HOUMARD

the intracellular levels of cyclic nucleotides. The twoknown classes of phosphodiesterases comprise the sevenfamilies of eukaryotic cyclic-nucleotide phosphodiester-ases (Beavo, 1995) and the proteobacterial cAMP-phosphodiesterases (Macfadyen et al., 1998). AlthoughcAMP-phosphodiesterase activities have been describedin cyanobacteria (Sakamoto et al., 1991), no homologueof the proteobacterial enzymes has been recognized inthe genome of Synechocystis PCC 6803 (Macfadyenet al., 1998). Five hypothetical coding regions in theSynechocystis PCC 6803 genome contain, however, aphosphohydrolase domain, named HD, characteristicof the N-terminal half region of the eukaryotic cAMP-phosphodiesterase catalytic centres (Aravind & Koonin,1998). Mutational analysis of the HD domain of acGMP-specific phosphodiesterase showed that the moreconserved residues within HD domains are involved incatalysis (Turko et al., 1998).

In vertebrates, cAMP and cGMP couple visual andolfactory signals to electrical excitation and Ca#+signalling by modulating cyclic-nucleotide-gated (CNG)channels (Zagotta & Siegelbaum, 1996). Both nucleo-tides can also activate protein kinases, which in turnregulate enzymes and proteins involved in intermediarymetabolism as well as in transcription (Daniel et al.,1998). In Escherichia coli, the real occurrence of cGMP[!10−* mol[(mg protein)−"] is debated (Herdman &Elmorjani, 1988; Vogler & Lengeler, 1987). In thisbacterium, cAMP functions as a cofactor of the cAMP-receptor protein (CRP) rather than as an activator of aprotein kinase. The cAMP-receptor protein CRP, alsoknown as CAP (Catabolite gene Activator Protein), is anallosteric DNA-binding protein that modulates thetranscription of several genes (Gralla & Collado-Vides,1996; Kolb et al., 1993). At present, all the cAMP-dependent responses in bacteria appear to be mediatedthrough the binding of cAMP to its receptor, which inturn regulates directly or indirectly genes involved in pHregulation, sugar metabolism and taxis (Botsford &Harman, 1992).

The allosteric effect promoted by cyclic nucleotides onCRP, kinases and the aforementioned channels isexerted through a similar cyclic nucleotide mono-phosphate (cNMP)-binding domain (Zagotta & Sie-gelbaum, 1996). The homodimeric three-dimensionalstructure of the E. coli cAMP-receptor protein (CRP-Ec)has been determined (Weber & Steitz, 1987). Thestructural data combined with site-directed mutagenesisstudies showed the importance of the residues that areconserved among the cyclic-nucleotide-binding domainsof CRPs, kinases and cyclic-nucleotide-gated channels(Varnum et al., 1995; Woodford et al., 1989; Zagotta &Siegelbaum, 1996). All the well-characterized cyclic-nucleotide-binding domains can accommodate bothcGMP and cAMP, but the degree of activation that theyconfer upon the output domain depends on the boundnucleotide. Both cAMP and cGMP bind to the E. colicAMP-receptor protein but cGMP does not activatetranscription (Ebright et al., 1985). The olfactory cyclic-

nucleotide-gated channel is fully activated by cGMP andcAMP. In contrast, for the photoreceptor channels,cAMP acts as a partial agonist, producing only afraction of the current induced by cGMP (Zagotta &Siegelbaum, 1996).

The protein GAF domains (encountered in cGMP-specific phosphodiesterases, Adenylyl cyclases andFormate hydrogen lyases) which may allostericallyregulate catalytic activities via ligand binding representanother type of cyclic nucleotide receptor (Aravind &Ponting, 1997). GAF domains participate in the ar-chitecture of many signalling proteins (phytochromes,ethylene receptors and members of two-componentregulatory systems). The GAF domain of cGMP-stim-ulated phosphodiesterases binds cGMP. Structure–function studies of this GAF domain performedby alanine mutagenesis have identified a motif[N(K}R)XnD] necessary for full cGMP-binding activity(McAllister-Lucas et al., 1995; Turko et al., 1996).

Apart from the adenylyl cyclase Cya1 (Slr1991), whichhas just been shown to be involved in the regulation ofcell motility in Synechocystis PCC 6803 (Terauchi &Ohmori, 1999), cGMP and cAMP signalling pathwaysin cyanobacteria remain sketchy. Here we report on theidentification of putative components of the cGMP andcAMP cyanobacterial signalling pathways by makinguse of : (i) the complete genome sequence of Syne-chocystis PCC 6803 (Terauchi & Ohmori, 1999) ; (ii)recently available programs and databases (Table 1) ;and (iii) the data obtained by sequence–function studiesfor cAMP}cGMP-dependent proteins. Our study leadsin particular to the presumptive identification of aprokaryotic cyclic-nucleotide-gated channel protein.Hypotheses are presented that could serve as goodstarting-points for functional genomic studies.

RESULTS AND DISCUSSION

Identification of potential cAMP- andcGMP-phosphodiesterases

Using sequence similarity search programs ( and-) no proteobacterial cAMP-phosphodiesterasehomologues could be recognized in the SynechocystisPCC 6803 genome. Such homologues could howeverexist in other cyanobacteria because the fusion of thetwo overlapping ORFs, PID g139932 and g139969,of the unicellular cyanobacterium Synechococcus PCC6301 (Kumano et al., 1983) would result in an ORF 30%identical (47% similar) to E. coli and Haemophilusinfluenzae phosphodiesterases. Recently, Aravind &Koonin (1998) reported that Slr1885, Slr2100, Sll1624and Slr0104 contained a phosphohydrolase domainnamed HD, which is related to the catalytic region ofeukaryotic cyclic-nucleotide phosphodiesterases. Weobserved that the HD-domains of ORFs Slr2100 andSll1624 were similar (30% identity, 40% similarity),and that each HD domain was fused to a different N-

3184


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35

cAMP and cGMP signalling components in Synechocystis

Table 1. Web sites and programs used to identify components of the cAMP and cGMP signalling pathways inSynechocystis PCC 6083.................................................................................................................................................................................................................................................................................................................

Parameters are shown in the footnotes.

Tool or content Website Reference

Annotated analysis of the Synechocystis PCC 6803 genome

Cyanobase http:}}www.kazusa.or.jp :8080

}cyano}Nakamura et al. (1998)

GeneQuiz http:}}columba.ebi.ac.uk:8765}ext-

genequiz}}genomes}cy9709

}index.html

Scharf et al. (1994)

GenBank http:}}www.ncbi.nlm.nih.gov Benson et al. (1999)

Sequence similarity search tool

* and -† search http:}}www.ncbi.nlm.nih.gov

}}Altschul & Koonin (1998) ;

Altschul et al. (1997)

Protein structure prediction

Secondary structure prediction http:}}www.embl-heidelberg.de}predictprotein

}predictprotein.html

Rost & Sander (1994)

Coiled coils prediction‡ http:}}www.isrec.isb-

sib.ch}software}COILSjform.html

Lupas (1996)

Prediction of transmembrane helices and topology of proteins

Neural network (htm and top) http:}}www.embl-heidelberg.de}predictprotein

}predictprotein.html

Rost et al. (1996)

Amino acid distribution ()§ http:}}www.enzim.hu}hmmtop

}submit.html

Tusnady & Simon (1998)

Protein structure data bank

Structural information of proteins http:}}www.rcsb.org} Berman (1999)

Domain architecture

http:}}coot.embl-heidelberg.de}SMART} Schultz et al. (1998)

Alignment

s http:}}www2.ebi.ac.uk}clustalw} Thompson et al. (1994)

*,†Matrix, Blosum 62; gap existence cost, 11; per residue gap cost, 1 ; lambda ratio, 0±85.

‡Matrix, MTIDK; weighting, no; window width, all.

§ Size of pseudocount vector, 1000; length of helices, from 17 to 25; lengths of tails, from 1 to 5; lengths of loops, "1; iterations, 1.

sGap opening 10; gap extension 0±005; gap distance 0±05.

terminal response regulator domain through a coiled-coil structural domain (Fig. 1). Interestingly, this modu-lar organization is that of the cAMP-specific phos-phodiesterase of Dictyostelium discoideum (RegA), forwhich the receiver domain has been shown to modulatethe phosphodiesterase activity (Thomasson et al., 1998).These similarities in sequence and domain organizationsuggest that Slr2100 and Sll1624 could be cyanobacterialcGMP- and cAMP-phosphodiesterases and that thecyclic nucleotide levels are likely controlled throughtwo-component systems. In the filamentous cyano-bacterium Spirulina platensis, such a regulation indeedcontrols the adenylyl cyclase CyaC and, thus, theintracellular levels of cAMP (Kasahara & Ohmori,1999).

The analysis of the Synechocystis PCC 6803 genomeperformed by Mizuno et al. (1996) indicated that slr2100was clustered with elements of two-component systems

(slr2098, slr2099 and slr2104), all these genes possiblybeing transcribed as a single unit. The correspondingpolypeptides contain both the transmitter (sensoryhistidine kinase) and the receiver (response regulator)domains, likely forming a specific multi-step phospho-relay system (Appleby et al., 1996; Kotani & Tabata,1998; Mizuno et al., 1996). By using a program aimed atdetecting already identified protein domains (),we observed that ORFs Slr2098 and Slr2104 contain PASsignalling motifs in the sensor region (Fig. 1), motifsknown to regulate light- and O

#-stimulated signalling

pathways in a variety of organisms (Zhulin & Taylor,1998). PAS domains have also been related to theperception of blue light (Pellequer et al., 1998; Zhulin etal., 1997). Interestingly, in Synechocystis PCC 6803 bluelight induces changes in the intracellular levels of cAMP(Terauchi & Ohmori, 1998). Altogether, data indicatethat Slr2100 might participate in the blue light signaltransduction pathway.

3185


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


(a)

(b)

.................................................................................................................................................................................................................................................................................................................

Fig. 1. (a) Physical map of the genes surrounding slr2100, and domain architecture of the two-component regulatoryelements (black arrows) clustered with this putative cNMP phosphodiesterase. Two hypothetical proteins (cross-hatchedarrows) and the cell division protein FtsY homologue (dotted arrow) are located just downstream of the putativephosphodiesterase. (b) Comparison of the domain architecture of the putative cNMP phosphodiesterases Slr2100 andSll1624 with that of the cAMP-phosphodiesterase RegA of Dictyostelium discoideum.

Cyclic nucleotide receptor proteins

Protein domain analysis of the Synechocystis genome,using the program, indicated that single ormultiple GAF domains are present in 28 ORFs of theSynechocystis genome (12 of which also containing aPAS domain). server provides a alignment of GAF domains that permits comparison ofthe Synechocystis GAF domains with those of someeukaryotic phosphodiesterases. The GAF domain ofORF Slr2104 contains all the determinants (Asn258,Lys259, Asp267) for cGMP-binding. That some of theseresidues are missing in other GAF domains of Syne-chocystis may indicate : (i) that these sites do notparticipate in cGMP binding; (ii) that the substitutedresidues can functionally replace those found in GAFdomains of eukaryotic phosphodiesterases ; or (iii) thatthis GAF domain serves some function other than cGMPbinding.

Based on protein domain sequence analyses (), 12ORFs of the Synechocystis genome could have cNMP-binding domains (Fig. 2). All of them contain a highlyconserved stretch of approximately 120 amino acids thatis homologous to the cNMP-binding domains of otherproteins, including the cAMP- and cGMP-dependentprotein kinases, the cyclic-nucleotide-gated channels

and the cAMP receptor protein (CRP). In five of theseORFs the cNMP-binding domain is joined to putativetransmembrane helices (Slr0593, Slr1529, Slr1575,Slr0510 and Sll1180), whereas in another five it isattached to a C-terminal domain which containshelix–turn–helix (HTH) DNA-binding motifs (Sll1371,Sll1924, Sll1169, Slr0449 and Slr1423).

Protein secondary structure predictions showed that theoverall structure of the cNMP-binding domain of CRP-Ec was conserved in the putative corresponding domainsof the 12 ORFs (Fig. 2b), suggesting a similar tertiarystructure. In CRP-Ec, the αC helix (conserved in the12 putative cNMP-binding domains) is required forthe induction, via the binding of the nucleotide,of the conformational change that subsequently permitssequence-specific DNA-binding (Heyduk et al., 1992). Incyclic-nucleotide-gated channels this helix stabilizes theligand in the open-state form of the channel (Tibbs et al.,1998). The strongest interactions between the ligand andCRP-Ec are achieved through non-covalent bonds: (i)between Arg82, Ser83 and the cyclic phosphate; (ii)between Gly71, Glu72 and the ribose; and (iii) betweenthe N-6 amino group of the cAMP purine ring, Thr127of one subunit and Ser128 of the other subunit of thedimer (Weber & Steitz, 1987). The residues that bridge

3186


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


(a)

(b)

.................................................................................................................................................................................................................................................................................................................

Fig. 2. (a) Domain architecture of 12 ORFs from Synechocystis that contain a putative cNMP-binding domain (cNMP-BD,white boxes), as detected by SMART. The helix–turn–helix motifs (H-T-H, dotted box), the σ54-interacting domain (σ54), theferredoxin-like domain (Fd) and the ABC transporter signature (ABC) were found using PSI-BLAST. The DUF2 domain (boxfilled with diamonds) was identified by SMART. Transmembrane helices (hatched barrels) and topology of Slr0593, Slr1529and HlyB were predicted using HMMTOP (for references see Table 1). Putative transmembrane domains of Slr1575 andSlr0510 were predicted using PHDhtm, and their topology inferred from the homology of Slr1575 with the bacterial K+

channel KcsA (Doyle et al., 1998). (b) Amino acid sequence comparison of the 12 putative cNMP-binding domains ofSynechocystis aligned to the cAMP-binding domain of E. coli cAMP receptor protein (CRP-Ec). Dots on top of thesequence indicate the residues close to cAMP in the cAMP–CRP complex. The residues involved in the strongestinteractions between CRP-Ec and the ribose, cyclic phosphate and purine ring of cAMP are highlighted above the dots.Numbers at the left indicate the position of the first residue in the corresponding sequence. Numbers in parenthesesrepresent amino acids that are not shown but that were considered in the CLUSTAL W alignment. Predicted secondarystructures (PHDsec) of the alignment, with an expected accuracy higher than 82%, are shown: H denotes an α-helix and Ea β-strand. Lines below the sequence highlight the secondary-structure motifs (α-helixC and β-strands 2–8) deduced fromthe three-dimensional structure of cAMP receptor protein (CRP; Weber & Steitz, 1987).

the cyclic phosphate and the ribose are conservedin cNMP-dependent protein kinases and in cyclic-nucleotide-gated channels. Site-directed mutagenesis ofthese conserved residues showed that they are indeedessential for the binding of cyclic nucleotides (Moore etal., 1992; Tibbs et al., 1998; Woodford et al., 1989). Ouralignment showed that these residues were conserved inSll1371 (CRP-Syn), Slr1575, Slr0593 and Slr0842 (Fig. 2).Therefore, among the 12 selected ORFs containingputative cNMP-binding domains, at least 4 polypeptideslikely possess a functional cNMP-binding domain.

In general, ligand discrimination of cNMP-bindingdomains is achieved by the residues analogous to Ser83and Thr127 of CRP-Ec (Altenhofen et al., 1991; Lee et

al., 1994; Shabb et al., 1990; Varnum et al., 1995). TheSer83 residue (or a conservative replacement of it such asThr) allows the binding of both cGMP and cAMP. Sincein CRP-Syn, Slr1575, Slr0593 and Slr0842 the positionanalogous to Ser83 of CRP-Ec is occupied by either a Seror a Thr residue, this position probably does not allowdiscrimination between cAMP and cGMP. In CRP-Ec,residues Thr127 and Ser128 determine nucleotide se-lectivity and participate in the intersubunit commun-ication. Amino acid substitutions that introduce ahydrophobic amino acid side chain at positions 127 or128 decrease CRP-Ec discrimination between cAMP andcGMP (Lee et al., 1994). Mutation T127L results in anactivation of transcription upon the binding of cGMPinstead of cAMP (Moore et al., 1996). According to

3187


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


these data, CRP-Syn, Slr1575 and Slr0593 would beactivated by cGMP, since the positions equivalent toThr127 in CRP-Ec are occupied by either Leu or Alaresidues (Fig. 2). However, reconstruction of the puta-tive cNMP-binding domain using the CRP-Ec structureas a template (Fig. 3) showed that : (i) the substitution ofSer128 by Asn (in CRP-Syn and Slr0842) preserved theinteractions with positions N-6 and N-7 of the adenine;and (ii) the loss of specificity that would result from thesubstitution of Thr127 by Leu or Ala (as in CRP-Syn andSlr0593) could be overcome by the substitution of Ser62by an Asn, which could create a new hydrogen bondwith the N1 of the adenine (Fig. 3). Thus we proposethat the cNMP-binding domains of CRP-Syn, Slr0593and Slr0842 would be activated preferentially by cAMP(Fig. 3). This deductive method cannot be used to inferthe specificity of Slr1575 because the aforementionedcritical positions are occupied by non-polar amino acids.

Characterization of the proteins that could bind cyclicnucleotides

Slr2104 is a member of two-component regulatory systems.Searches for characteristic architecture domains by the program indicated that Slr2104 contains a sen-sory histidine kinase N-terminal domain joined to aresponse regulator domain followed by a phospho-acceptor domain of histidine kinases (Fig. 1). Thus,Slr2104 integrates a multi-step phospho-relay. Thesensor domain (input domain) of the histidine kinase isformed by a putative membrane domain, a cGMP-binding GAF domain and a PAS domain. Strikingly, thisORF is clustered with the putative phosphodiesteraseSlr2100. Hence, two-component regulatory systems arelikely involved in the control, amplification and inte-gration of cyclic nucleotide signals.

Slr1575 is a putative cyclic-nucleotide-gated channel (CNG).ORF Slr1575 possesses structural elements clearly rela-ted to voltage-gated and cyclic-nucleotide-gated chan-nels (Fig. 4). In such channels, the major determinants inchannel gating have been mapped in the cytoplasmiclinker (C-linker) that connects the last transmembranesegments of the channel and the cNMP-binding domain(Zong et al., 1998). The alignment of the cyclic-nucleotide-gated bovine cone photoreceptor channel(CNG3) with Slr1575 (Fig. 4a) showed that two (I439and D481) of the three amino acids in the C-linker thatdetermine the high specificity of the cone photoreceptorchannel for cGMP were conserved, the Asp494 of CNG3being replaced by a Gln residue. This alignment stronglysuggests that Slr1575 would be activated preferentiallyby cGMP.

ORF Slr1575 contains a sequence similar to the pore-forming region of ion channels and a stretch ofhydrophobic and basic amino acids similar to thetransmembrane voltage-sensing S4 domain of voltage-activated and cyclic-nucleotide-gated channels (Fig. 4b,c). Putative voltage-sensing and pore-forming sequencesform the N-terminus of Slr1575. We deduced its putative

topology by comparison with the well-known structureof the bacterial K+ channel (KcsA) (Doyle et al., 1998).In the pore-forming sequence, the GXG motif boxed inFig. 4(c) is of particular interest since it determinescation selectivity in ion channels (Heginbotham et al.,1992; Kerr & Sansom, 1995). The pore of Slr1575 wouldresemble the non-selective cation pore of the P2X2purine receptor. Our comparative analysis leads us toconclude that Slr1575 would be a non-selective cationchannel gated by cyclic nucleotides that would belong tothe S4 ion channel superfamily (Heginbotham et al.,1992).

The subunit composition of cyclic-nucleotide-gatedchannels, heteromeric or homomeric, modulates bothcyclic nucleotide sensitivity (Bradley et al., 1994) andcation permeation (Dzeja et al., 1999). Although Slr0510does not contain the determinants for cyclic nucleotidebinding (Fig. 1), this hypothetical protein is homologousto Slr1575 in both sequence (33% identity and 51%similarity) and domain architecture (Fig. 2). By analogywith eukaryotic cyclic-nucleotide-gated channels, thespecificity for cations and cyclic nucleotides of Slr1575homomeric channels might be modulated by hetero-merization with Slr0510.

Slr0593 is a putative cyclic-nucleotide-modulated permease.In both Cyanobase and Genequiz databases, Slr0593 isannotated as a cAMP protein kinase regulatory chainbecause the cNMP-binding domains of these proteinsshare a high degree of homology (30% identity and 48%similarity). Our sequence analysis indicates, however,that of the five domains that define such proteins (Tayloret al., 1990), two were not conserved, namely the dimer-interaction site and the peptide-inhibitory site. Twoprograms aimed at predicting transmembrane helicesand protein topology (PHDhtm and ) predicteight transmembrane helices for Slr0593, the cNMP-binding domain being located on the cytosolic side,between the third and fourth helices (Fig. 2). The threeC-terminal membrane helices of Slr0593 show 27%identity (43% similarity) with three membrane helicesof the amino acid permease ROCC of Bacillus subtilis(PID g730600) (Glaser et al., 1993). The five C-terminalmembrane helices of Slr0593 show 26% identity (42%similarity) with the phenylacetic acid permease PhaJ ofPseudomonas putida (PID g3253206) (Olivera et al.,1998). Although the specificity of the permease cannotbe inferred from sequence comparison, our resultsindicate that Slr0593 is likely to be a permease, theactivity of which would be modulated by cyclic nu-cleotides.

Slr0842 is a putative signal transducer. In ORF Slr0842, thecNMP-binding domain is joined to a DUF2 domain (Fig.2). This domain was first recognized in the N-terminalregion of Acetobacter xylinum enzymes that control theturnover of c-di-GMP:diguanylate cyclase and phos-phodiesterase A (Tal et al., 1998). It is possible thatDUF2 domains mediate one of these activities. Using aprogram for detecting characteristic architecture doma-ins (), we have found DUF2 domains in 13 ORFs

3188


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


.................................................................................................................................................................................................................................................................................................................

Fig. 3. Models for the cAMP-binding sites of CRP-Syn, Slr1575, Slr0593 and Slr0842, by analogy with the cAMP receptorprotein (CRP-Ec) structure. Key amino acids were deduced from the alignment shown in Fig. 1. Spatial structures weremodelled by substituting the corresponding residues in the three-dimensional structure of CRP-Ec (PDB entry 3GAP).Residues contacting the ribose and cyclic phosphate are not shown because they are fully conserved.

of Synechocystis PCC 6803 (Slr0842, Sll0267, Slr1588,Slr1102, Sll0821, Slr1103, Slr0359, Slr1104, Sll1895,Slr1305, Slr1593, Slr2077 and Slr1692). They may bepreceded by input sensory signalling domains likeresponse regulators, GAF domains, FHA (Forkhead-associated domain) domains and}or PAS domains.Hence, DUF2 domains can probably be regulated bydifferent signals. We conclude that ORF Slr0842 mightbe a cyclic-nucleotide-modulated signal transducer of apathway yet to be determined.

The cAMP receptor protein (CRP-Syn, sll1371) is a tran-scription factor. The homology of CRP-Syn with the E.coli cAMP receptor protein (CRP-Ec) extends to thehelix–turn–helix DNA-binding domain (Fig. 5). Thecrystal structure of the CRP-DNA complex enabled theidentification of the residues of CRP-Ec that interactwith the DNA (Schultz et al., 1991), and the compilationof several CRP-DNA binding sites led to the iden-tification of a 22 bp palindromic consensus site ofsequence AAATGTGATCT*AGATCACATTT (the

most conserved bases are underlined and the base pairsthat interact with the protein are doubly underlined)(Berg & von Hippel, 1988). Interactions of CRP-Ec withthe consensus DNA-binding sequence involve specificbase contacts with Arg180, Glu181 and Arg185, thedeterminants of specificity (Kolb et al., 1993). These andother less selective contacts involved in the specificrecognition of the consensus DNA binding site by E. colicAMP receptor protein are conserved in the Syne-chocystis PCC 6803 protein (Fig. 5a). This suggests thatCRP-Syn could recognize the same DNA sequence asCRP-Ec. Upstream of the CRP-Syn coding region wefound DNA sequences showing a high degree ofsimilarity with the promoter region of the E. coli crp,including the two CRP binding sites (Fig. 5b). Theseidentities reinforce the hypothesis that CRP-Syn binds tothe same DNA sequence as CRP-Ec, and suggest thatcrp is autoregulated as it is in E. coli (Hanamura &Aiba, 1991). While this paper was under review,Yoshimura et al. (2000) reported that sll1371 (referred toas CRP-Syn in this article) not only binds cAMP but

3189


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


(a)

(b)

(c)

.................................................................................................................................................................................................................................................................................................................

Fig. 4. Features common to Slr1575 and cation channels. (a) PSI-BLAST sequence alignment of Slr1575 with the C-linker(doubled underlined) and cNMP-binding domain (underlined) regions of the bovine cone photoreceptor cyclic-nucleotide-gated channel (CNG3). Boxes contain the determinants for the specificity of CNG3 for cGMP (Zong et al.,1998). (b) Alignment of the voltage-sensing S4 region of the shaker K+ channel, CNG3 (Finn et al., 1996), and Slr1575.Basic residues are underlined. (c) Alignment of the pore-forming region of the shaker K+ channel, CNG3 (Zong et al.,1998), the P2X2 purinoreceptor (Kerr & Sansom, 1995) and Slr1575. The putative determinants for cation selectivity (Kerr& Sansom, 1995) are boxed. Numbers at the left indicate the position of the first residue in the corresponding sequence.Sequence identity and similarity are represented by (*) and (:), respectively. GenBank accession codes (PID) are shown tothe right of the alignment.

also, when complexed with cAMP, can alter theelectrophoretic mobility of an oligonucleotide thatcontains the E. coli CRP binding site. Slr0593 (referredto as a cyclic nucleotide permease in this article) alsobound cAMP while sll1924 could not. These data are inagreement with the hypotheses proposed in this article.

We then searched for CRP-binding sites in the upstreamregion of the Synechocystis PCC 6803 genes that arehomologues of E. coli CRP-regulated genes (Gralla &Collado-Vides, 1996). As in E. coli, we found putativeCRP-binding sites upstream of the adenylyl cyclase gene(cya1) and of the acetolactate synthase small subunitgene (ilvN¯ sll0065). Further screening of the wholegenome of Synechocystis using a program aimed atrecognizing short highly similar DNA sequences (Find-patterns, GCG package) revealed putative CRP-bindingsites upstream of the genes encoding: (i) the ribose-5-phosphate isomerase (rpiA¯ slr0194) involved in thepentose phosphate pathway; (ii) the UDP-N-acetyl-muramoylalanyl--glutamyl-2,6-diaminopimelate--alanyl--alanine ligase (murF¯ slr1351), implicated inthe synthesis of the cell envelope; (iii) the above-mentioned putative cyclic-nucleotide-gated-channel

slr1575 ; (iv) the high-affinity branched-chain amino acidtransport permease protein (livH¯ slr1200) ; (v) aputative methyl-accepting chemotaxis protein I (cheD¯ sll0041) involved in bacterial taxis ; (vi) the AT103protein (AT103¯ sll1874), a putative phytochrome-regulated gene; (vii) subunit 3 of the NADH dehydro-genase of the respiratory electron transport (ndhC¯slr1279) ; (viii) the homologue of the response regulatorNarL (sll1708) involved in the adaptation of E. coli toanaerobic respiration; (ix) the uncharacterized histidinekinase Slr1805; and (x) the permease protein PstA (pstAor phoT) implicated in phosphate transport. Thediversity of the putative CRP-Syn-regulated genes sug-gests the existence in Synechocystis PCC 6803 of aCRP-Syn-regulated modulon.

Concluding remarks

After standard search for identification by homology,about half of the putative ORFs of every fully sequencedgenome are annotated as hypothetical proteins. Morethorough analyses, coupled with the physiological andstructure–function data that are available, make itpossible to refine the assignments and}or propose

3190


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


(a)

(b)

(c)

.................................................................................................................................................................................................................................................................................................................

Fig. 5. (a) CLUSTAL W sequence alignment of the cAMP receptor protein (CRP-Syn and CRP-Ec) sequences involved in DNA-binding. Contacts involved in specific recognition of DNA by CRP-Ec (Schultz et al., 1991) are indicated above itssequence: specific base contacts (y), charged interactions with the DNA backbone phosphate (), side chain H bonds(+) and backbone amide H bonds (r). Numbers at the left indicate the position of the first residue in the correspondingsequence. Sequence identity and similarity are represented by (*) and (:), respectively. Numbers in parentheses representamino acids not shown but considered for the alignment. (b) Comparison of the E. coli crp promoter region Pcrp with aregion located upstream of the gene encoding CRP-Syn. The CRP-Ec binding sites I (repressing site) and II (activating site),®35 and ®10 sequences and the ATG are underlined (Hanamura & Aiba, 1991). The CRP-dependent transcription startpoint is boxed. Sequences of Synechocystis PCC 6803 matching the CRP-Ec consensus DNA-binding site are doubleunderlined. (c) Putative CRP-binding sites located upstream of some Synechocystis PCC 6803 coding regions. The topsequence corresponds to the E. coli CRP consensus binding site, with the 10 most conserved nucleotides of thepalindrome (Berg & von Hippel, 1988) boxed, and those involved in specific base contacts with CRP-Ec (Schultz et al.,1991) are marked with a triangle (y). CRP-binding sites controlling cyaA and ilvB expression in E. coli are included toallow comparison with the Synechocystis genes cya1 and ilvN, respectively. Putative translation start points areunderlined. Identity scores (at the right) correspond to the number of nucleotides matching those of the residuesinvolved in specific base contacts (y) with respect to the number of residues matching the 10 most conserved nucleotidesof the palindrome (boxed). Residues identical to the consensus are doubly underlined.

working hypotheses (Pallen, 1999). Using this approachin conjunction with targeted gene inactivation, we haverecently shown that Synechocystis Cya2, a proteinpreviously identified as an adenylyl cyclase (Kaneko etal., 1996), is responsible for cGMP synthesis in thisorganism, thus identifying the first prokaryotic guanylylcyclase (Ochoa de Alda et al., 2000). Although phos-phodiesterase activities have been measured in cyano-bacteria (Herdman & Elmorjani, 1988), includingSynechocystis PCC 6803, no obvious homologues ofproteobacterial enzymes have been previously found;we tentatively propose Sll1624 and Slr2100 as beingcyanobacterial phosphodiesterases.

Among 40 ORFs showing a high overall homology with

cyclic nucleotide receptors, we identified 5 that containthe determinants required for ligand binding. Oneimportant conclusion derived from the presence ofcGMP and cAMP at similar levels in Synechocystis PCC6803 (Herdman & Elmorjani, 1988), and from thepresence of the aforementioned cNMP-binding do-mains, is that one cyclic nucleotide could function as aninhibitor or as an agonist of the activity induced by theother, both of them being able to bind to the cNMP-binding domain but only one of the two inducing fullactivation.

Our sequence analysis study of the genome of Syne-chocystis PCC 6803 revealed the previously unrecog-nized presence of eukaryotic-type cGMP}cAMP signal-

3191


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


ling components in a prokaryote, such as the cyclic-nucleotide-gated channels and the cGMP-binding GAFdomain. The presence of the genes encoding eukaryotic-type cGMP}cAMP signalling components that aredetected, together with that of a putative CRP modulon,suggests that the complexity of the cAMP}cGMPsignalling pathways in cyanobacteria is greater than inother bacteria, in agreement with the striking mor-phological and physiological diversity found in thecyanobacteria. This complexity could be even higher insome filamentous cyanobacteria able to carry outcellular differentiation, because Anabaena PCC 7120 forexample has five different nucleotide cyclases (Katayama& Ohmori, 1997).

Targeted inactivation of the ORFs that we have char-acterized would be a first approach to validate ourhypotheses. In parallel, site-directed mutagenesis of thehypothetical residues essential for the function of theidentified domains would give more information aboutthe signal transduction pathways in which they par-ticipate. Substitution of the gene encoding slr2100 by amutated gene, for example, should remove the phos-phodiesterase activity and affect the phosphorelay(s) inwhich cyclic nucleotides are involved.

ACKNOWLEDGEMENTS

We are grateful to M.-I. Tapia for careful reading of themanuscript and to J.-P. Roux for computer assistance. Wewish to thank A.-L. Etienne for her encouraging support. Thiswork was supported by grants from the Centre National de laRecherche Scientifique to the UMR8543 and to J. A. G. Ochoade Alda, and from the Ecole Normale supe! rieure (Paris).

REFERENCES

Altenhofen, W., Ludwig, J., Eismann, E., Kraus, W., Bonigk, W. &Kaupp, U. B. (1991). Control of ligand specificity in cyclicnucleotide-gated channels from rod photoreceptors and olfactoryepithelium. Proc Natl Acad Sci U SA 88, 9868–9872.

Altschul, S. F. & Koonin, E. V. (1998). Iterated profile searches with- : a tool for discovery in protein databases. TrendsBiochem Sci 23, 444–447.

Altschul, S. F., Madden, T. L., Schaffer, A. A., Zhang, J., Zhang, Z.,Miller, W. & Lipman, D. J. (1997). Gapped and - : anew generation of protein database search programs. NucleicAcids Res 25, 3389–3402.

Appleby, J. L., Parkinson, J. S. & Bourret, R. B. (1996). Signaltransduction via the multi-step phosphorelay : not necessarily aroad less travelled. Cell 86, 845–848.

Aravind, L. & Koonin, E. V. (1998). The HD domain defines anew superfamily of metal-dependent phosphohydrolases. TrendsBiochem Sci 23, 469–472.

Aravind, L. & Ponting, C. P. (1997). The GAF domain: anevolutionary link between diverse phototransducing proteins.Trends Biochem Sci 22, 458–459.

Beavo, J. A. (1995). Cyclic nucleotide phosphodiesterases : func-tional implications of multiple isoforms. Physiol Rev 75, 725–748.

Benson, D. A., Boguski, M. S., Lipman, D. J., Ostell, J., Ouellette,B. F., Rapp, B. A. & Wheeler, D. L. (1999). GenBank. Nucleic AcidsRes 27, 12–17.

Berg, O. G. & von Hippel, P. H. (1988). Selection of DNA bindingsites by regulatory proteins. II. The binding specificity of cyclicAMP receptor protein to recognition sites. J Mol Biol 200,709–723.

Berman, H. M. (1999). The past and future of structure databases.Curr Opin Biotechnol 10, 76–80.

Botsford, J. L. & Harman, J. G. (1992). Cyclic AMP in prokaryotes.Microbiol Rev 56, 100–122.

Bradley, J., Li, J., Davidson, N., Lester, H. A. & Zinn, K. (1994).Heteromeric olfactory cyclic nucleotide-gated channels : a subunitthat confers increased sensitivity to cAMP. Proc Natl Acad SciU SA 91, 8890–8894.

Daniel, P. B., Walker, W. H. & Habener, J. F. (1998). Cyclic AMPsignalling and gene regulation. Annu Rev Nutr 18, 353–383.

Doyle, D. A., Morais Cabral, J., Pfuetzner, R. A., Kuo, A., Gulbis,J. M., Cohen, S. L., Chait, B. T. & MacKinnon, R. (1998). The struc-ture of the potassium channel : molecular basis of K+ conductionand selectivity. Science 280, 69–77.

Dzeja, C., Hagen, V., Kaupp, U. B. & Frings, S. (1999). Ca#+

permeation in cyclic nucleotide-gated channels. EMBO J 18,131–144.

Ebright, R. H., Le Grice, S. F., Miller, J. P. & Krakow, J. S. (1985).Analogues of cyclic AMP that elicit the biochemically definedconformational change in catabolite gene activator protein (CAP)but do not stimulate binding to DNA. J Mol Biol 182, 91–107.

Finn, J. T., Grunwald, M. E. & Yau, K. W. (1996). Cyclic nucleotide-gated ion channels : an extended family with diverse functions.Annu Rev Physiol 58, 395–426.

Glaser, P., Kunst, F., Arnaud, M. & 14 other authors (1993).Bacillus subtilis genome project : cloning and sequencing of the97 kb region from 325 degrees to 333 degrees. Mol Microbiol 10,371–384.

Gralla, J. D. & Collado-Vides, J. (1996). Organization and functionof transcription regulatory elements. In Escherichia coli andSalmonella : Cellular and Molecular Biology, pp. 1232–1245.Edited by F. C. Neidhardt and others. Washington, DC:American Society for Microbiology.

Hanamura, A. & Aiba, H. (1991). Molecular mechanismof negativeautoregulation of Escherichia coli crp gene. Nucleic Acids Res 19,4413–4419.

Heginbotham, L., Abramson, T. & MacKinnon, R. (1992). Afunctional connection between the pores of distantly related ionchannels as revealed by mutant K+ channels. Science 258,1152–1155.

Herdman, M. & Elmorjani, K. (1988). Cyclic nucleotides. MethodsEnzymol 167, 584–591.

Heyduk, E., Heyduk, T. & Lee, J. C. (1992). Intersubunit comm-unications inEscherichia coli cyclicAMP receptor protein: studiesof the ligand binding domain. Biochemistry 31, 3682–3688.

Kaneko, T., Sato, S., Kotani, H. & 21 other authors (1996).Sequence analysis of the genome of the unicellular cyanob-acterium Synechocystis sp. strain PCC 6803. II. Sequence dete-rmination of the entire genome and assignment of potentialprotein-coding regions. DNA Res 3, 109–136.

Kasahara, M. & Ohmori, M. (1999). Activation of a cyanobacterialadenylate cyclase, CyaC, by autophosphorylation and a subs-equent phosphotransfer reaction. J Biol Chem 274, 15167–15172.

Katayama, M. & Ohmori, M. (1997). Isolation and characterizationof multiple adenylate cyclase genes from the cyanobacteriumAnabaena sp. strain PCC 7120. J Bacteriol 179, 3588–3593.

3192


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


Kerr, I. D. & Sansom, M. S. (1995). Cation selectivity in ionchannels. Nature 373, 112.

Kolb, A., Busby, S., Buc, H., Garges, S. & Adhya, S. (1993).Transcriptional regulation by cAMP and its receptor protein.Annu Rev Biochem 62, 749–795.

Kotani, H. & Tabata, S. (1998). Lessons from sequencing of theunicellular cyanobacterium Synechocystis sp. PCC 6803. AnnuRev Plant Physiol Plant Mol Biol 49, 151–171.

Kumano, M., Tomioka, N. & Sugiura, M. (1983). The completenucleotide sequence of a 23S rRNA gene from a blue-green alga,Anacystis nidulans. Gene 24, 219–225.

Lee, E. J., Glasgow, J., Leu, S. F., Belduz, A. O. & Harman, J. G.(1994). Mutagenesis of the cyclic AMP receptor protein ofEscherichia coli : targeting positions 83, 127 and 128 of the cyclicnucleotide binding pocket. Nucleic Acids Res 22, 2894–2901.

Lupas, A. (1996). Coiled coils : new structures and new functions.Trends Biochem Sci 21, 375–382.

McAllister-Lucas, L. M., Haik, T. L., Colbran, J. L., Sonnenburg,W. K., Seger, D., Turko, I. V., Beavo, J. A., Francis, S. H. & Corbin,J. D. (1995). An essential aspartic acid at each of two allostericcGMP-binding sites of a cGMP-specific phosphodiesterase. J BiolChem 270, 30671–30679.

Macfadyen, L. P., Ma, C. & Redfield, R. J. (1998). A 3«,5«-cyclicAMP (cAMP) phosphodiesterase modulates cAMP levels andoptimizes competence in Haemophilus influenzae Rd. J Bacteriol180, 4401–4405.

Mizuno, T., Kaneko, T. & Tabata, S. (1996). Compilation of allgenes encoding bacterial two-component signal transducers in thegenome of the cyanobacterium, Synechocystis sp. strain PCC6803. DNA Res 3, 407–414.

Moore, J., Kantorow, M., Vanderzwaag, D. & McKenney, K.(1992). Escherichia coli cyclic AMP receptor protein mutantsprovide evidence for ligand contacts important in activation. JBacteriol 174, 8030–8035.

Moore, J. L., Gorshkova, I. I., Brown, J. W., McKenney, K. H. &Schwarz, F. P. (1996). Effect of cAMP binding site mutationson the interaction of cAMP receptor protein with cyclic nu-cleoside monophosphate ligands and DNA. J Biol Chem 271,21273–21278.

Nakamura, Y., Kaneko, T., Hirosawa, M., Miyajima, N. & Tabata,S. (1998). CyanoBase, a www database containing the completenucleotide sequence of the genome of Synechocystis sp. strainPCC 6803. Nucleic Acids Res 26, 63–67.

Ochoa de Alda, J. A. G., Ajlani, G. & Houmard, J. (2000). TheSynechocystis PCC 6803 cya2, a prokaryotic gene that encodes aguanylyl cyclase. J Bacteriol 182, 3839–3842.

Ohmori, M. (1989). cAMP in Anabaena cylindrica : rapid changesin cellular levels in response to changes in extracellular envi-ronments. Plant Cell Physiol 30, 911–914.

Olivera, E. R., Minambres, B., Garcia, B., Muniz, C., Moreno,M. A., Ferrandez, A., Diaz, E., Garcia, J. L. & Luengo, J. M. (1998).Molecular characterization of the phenylacetic acid catabolicpathway in Pseudomonas putida U: the phenylacetyl-CoAcatabolon. Proc Natl Acad Sci U SA 95, 6419–6424.

Pallen, M. J. (1999). Microbial genomes. Mol Microbiol 32,907–912.

Pellequer, J. L., Wager-Smith, K. A., Kay, S. A. & Getzoff, E. D.(1998). Photoactive yellow protein: a structural prototype for thethree-dimensional fold of the PAS domain superfamily. Proc NatlAcad Sci U SA 95, 5884–5890.

Rost, B. & Sander, C. (1994). Combining evolutionary information

and neural networks to predict protein secondary structure.Proteins 19, 55–72.

Rost, B., Fariselli, P. & Casadio, R. (1996). Topology prediction forhelical transmembrane proteins at 86% accuracy. Protein Sci 5,1704–1718.

Sakamoto, T., Murata, N. & Ohmori, M. (1991). The concentrationof cyclic AMP and adenylate cyclase activity in cyanobacteria.Plant Cell Physiol 32, 581–584.

Scharf, M., Schneider, R., Casari, G., Bork, P., Valencia, A.,Ouzounis, C. & Sander, C. (1994). GeneQuiz : a workbench forsequence analysis. ISMB 2, 348–353.

Schultz, S. C., Shields, G. C. & Steitz, T. A. (1991). Crystal structureof a CAP–DNA complex: the DNA is bent by 90 degrees. Science253, 1001–1007.

Schultz, J., Milpetz, F., Bork, P. & Ponting, C. P. (1998). , asimple modular architecture research tool : identification ofsignalling domains. Proc Natl Acad Sci U SA 95, 5857–5864.

Shabb, J. B., Ng, L. & Corbin, J. D. (1990). One amino acid changeproduces a high affinity cGMP-binding site in cAMP-dependentprotein kinase. J Biol Chem 265, 16031–16034.

Tal, R., Wong, H. C., Calhoon, R. & 11 other authors (1998). Threecdg operons control cellular turnover of cyclic di-GMP inAcetobacter xylinum : genetic organization and occurrence ofconserved domains in isoenzymes. J Bacteriol 180, 4416–4425.

Taylor, S. S., Buechler, J. A. & Yonemoto, W. (1990). cAMP-dependent protein kinase : framework for a diverse family ofregulatory enzymes. Annu Rev Biochem 59, 971–1005.

Terauchi, K. & Ohmori, M. (1998). An adenylate cyclase, CyaD,mediates the signal of blue light in the cyanobacterium Syne-chocystis sp. PCC 6803. Plant Cell Physiol 39, s153.

Terauchi, K. & Ohmori, M. (1999). An adenylate cyclase, Cya1,regulates cell motility in the cyanobacterium Synechocystis sp.PCC 6803. Plant Cell Physiol 40, 248–251.

Thomasson, P. A., Traynor, D., Cavet, G., Chang, W. T., Harwood,A. J. & Kay, R. R. (1998). An intersection of the cAMP}PKA andtwo-component signal transduction systems in Dictyostelium.EMBO J 17, 2838–2845.

Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994). :improving the sensitivity of progressive multiple sequence alig-nment through sequence weighting, position-specific gap pena-lties and weight matrix choice. Nucleic Acids Res 22, 4673–4680.

Tibbs, G. R., Liu, D. T., Leypold, B. G. & Siegelbaum, S. A. (1998).A state-independent interaction between ligand and a conservedarginine residue in cyclic nucleotide-gated channels reveals afunctional polarity of the cyclic nucleotide binding site. J BiolChem 273, 4497–4505.

Turko, I. V., Haik, T. L., McAllister-Lucas, L. M., Burns, F., Francis,S. H. & Corbin, J. D. (1996). Identification of key amino acids in aconserved cGMP-binding site of cGMP-binding phosphodie-sterases. A putative NKXnD motif for cGMP binding. J BiolChem 271, 22240–22244.

Turko, I. V., Francis, S. H. & Corbin, J. D. (1998). Potential roles ofconserved amino acids in the catalytic domain of the cGMP-binding cGMP-specific phosphodiesterase. J Biol Chem 273,6460–6466.

Tusnady, G. E. & Simon, I. (1998). Principles governing amino acidcomposition of integral membrane proteins : application totopology prediction. J Mol Biol 283, 489–506.

Varnum, M. D., Black, K. D. & Zagotta, W. N. (1995). Molecularmechanism for ligand discrimination of cyclic nucleotide-gatedchannels. Neuron 15, 619–625.

3193


IP: 54.224.135.207

On: Sat, 09 Apr 2016 16:43:35


Vogler, A. P. & Lengeler, J. W. (1987). Indirect role of adenylatecyclase and cyclic AMP in chemotaxis to phosphotransferasesystem carbohydrates in Escherichia coli K-12. J Bacteriol 169,593–599.

Weber, I. T. & Steitz, T. A. (1987). Structure of a complex ofcatabolite gene activator protein and cyclic AMP refined at 2±5 A/resolution. J Mol Biol 198, 311–326.

Woodford, T. A., Correll, L. A., McKnight, G. S. & Corbin, J. D.(1989). Expression and characterization of mutant forms of thetype I regulatory subunit of cAMP-dependent protein kinase. Theeffect of defective cAMP binding on holoenzyme activation. J BiolChem 264, 13321–13328.

Yoshimura, H., Hisabori, T., Yanagisawa, S. & Ohmori, M. (2000).Identification and characterization of a novel cAMP receptorprotein in the cyanobacterium Synechocystis sp. PCC 6803. J BiolChem 275, 6241–6245.

Zagotta, W. N. & Siegelbaum, S. A. (1996). Structure and functionof cyclic nucleotide-gated channels. Annu Rev Neurosci 19,235–263.

Zhulin, I. B. & Taylor, B. L. (1998). Correlation of PAS domainswith electron transport-associated proteins in completely sequ-enced microbial genomes. Mol Microbiol 29, 1522–1523.

Zhulin, I. B., Taylor, B. L. & Dixon, R. (1997). PAS domain S-boxesin Archaea, Bacteria and sensors for oxygen and redox. TrendsBiochem Sci 22, 331–333.

Zong, X., Zucker, H., Hofmann, F. & Biel, M. (1998). Three aminoacids in the C-linker are major determinants of gating in cyclicnucleotide-gated channels. EMBO J 17, 353–362.

.................................................................................................................................................

Received 7 February 2000; revised 14 August 2000; accepted 30 August2000.

3194

genomic survey of camp and cgmp signalling components in the cyanobacterium synechocystis pcc 6803

Documents