A second isoform of the ferredoxin:NADPoxidoreductase generated by an in-frameinitiation of translationJean-Claude Thomas*, Bettina Ughy†‡, Bernard Lagoutte‡, and Ghada Ajlani‡§

*Departement de Biologie, Ecole Normale Superieure, F-75230 Paris, France; †Plant Biology Institute, Biological Research Center, H-6701, Szeged, Hungary;and ‡Departement de Biologie Joliot-Curie, Centre National de la Recherche Scientifique, Commissariat a l’Energie Atomique, F-91191 Gif-sur-Yvette, France

Edited by Robert Haselkorn, University of Chicago, Chicago, IL, and approved October 10, 2006 (received for review September 4, 2006)

Ferredoxin:NADP oxidoreductases (FNRs) constitute a family offlavoenzymes that catalyze the exchange of reducing equivalentsbetween one-electron carriers and the two-electron-carryingNADP(H). The main role of FNRs in cyanobacteria and leaf plastidsis to provide the NADPH for photoautotrophic metabolism. In rootplastids, a distinct FNR isoform is found that has been postulatedto function in the opposite direction, providing electrons fornitrogen assimilation at the expense of NADPH generated byheterotrophic metabolism. A multiple gene family encodes FNRisoenzymes in plants, whereas there is only one FNR gene (petH)in cyanobacteria. Nevertheless, we detected two FNR isoforms inthe cyanobacterium Synechocystis sp. strain PCC6803. One of them(FNRS �34 kDa) is similar in size to the plastid FNR and specificallyaccumulates under heterotrophic conditions, whereas the otherone (FNRL �46 kDa) contains an extra N-terminal domain thatallows its association with the phycobilisome. Site-directed mu-tants allowed us to conclude that the smaller isoform, FNRS, isproduced from an internal ribosome entry site within the petHORF. Thus we have uncovered a mechanism by which two isoformsare produced from a single gene, which is, to our knowledge, novelin photosynthetic bacteria. Our results strongly suggest that FNRL

is an NADP� reductase, whereas FNRS is an NADPH oxidase.

cyanobacteria � phycobilisome � internal ribosome entry

Cyanobacteria and chloroplasts, the eukaryotic organellesderived from cyanobacteria, are defined by their ability to

carry out the oxygenic photosynthesis required for their photo-autotrophic growth. Although some cyanobacterial strains cangrow chemoheterotrophically in the dark at the expense ofexternal carbohydrates, their chemoheterotrophy is usually re-stricted to the mobilization of reserves during dark periods or tospecialized cells like heterocysts.

Ferredoxin:NADP oxidoreductases (FNRs) catalyze the ex-change of reducing equivalents between ferredoxin (or fla-vodoxin) and NADP(H). In cyanobacteria and photosyntheticplastids (chloroplasts) the main role of the FNR is to catalyze thefinal step of photosynthetic electron transport, providingNADPH for CO2 assimilation and other reductive metabolism.In nonphotosynthetic root plastids a genetically distinct FNRisoform is postulated to function in the opposite direction,providing electrons for nitrogen assimilation at the expense ofNADPH generated by heterotrophic metabolism (1, 2). It hasbeen suggested, for both cyanobacteria and plastids, that FNRcould participate in respiration and cyclic electron transfer as anNADPH dehydrogenase (3–6).

Despite the similarity in both biochemical properties andprotein structure for the cyanobacterial and plastid FNR (7), theunique petH gene in most phycobilisome (PBS)-containingcyanobacteria, except Gloeobacter violaceus (8) and Synechococ-cus spp. strains OS-A and OS-B� (9), encodes a �46-kDa FNRthat contains an N-terminal domain, of �80 aa, whose sequenceis similar to PBS rod-linker polypeptides (10). The PBS is a largeand abundant bilin–protein complex that harvests light for

photosynthesis. Nondiazotrophic cyanobacteria respond to thelack of combined nitrogen by slowing down photosynthesis andanabolic reactions, while catabolizing reserves to survive pro-longed periods of starvation (11). During nitrogen starvation,PBS degradation supplies amino acids for the synthesis ofessential proteins, while also reducing light harvesting for pho-tosynthesis (12).

Purified PBS from several cyanobacterial strains containsignificant amounts of the 46-kDa FNR, one to two moleculesper PBS (10, 13, 14). It was proposed that the FNR binds to thePBS rods via its linker domain to fulfill functions in both cyclicelectron transport and respiration, in close proximity to themembrane (6, 15). A �35-kDa FNR (similar in size to that ofplastids) has been purified from several cyanobacterial strains,Spirulina sp. (16), Anabaena cylindrica (17), and Synechocystis sp.strain PCC6803 (18). This result was attributed to proteolyticdegradation of the N-terminal domain of the 46-kDa FNR(6, 10).

In this article, the use of a highly specific antibody togetherwith genetic and physiological studies reveals that two FNRisoforms are produced by differential translation initiation of thepetH ORF in Synechocystis sp. PCC6803 (hereafter called Syn-echocystis). One of these isoforms accumulates under conditionsof heterotrophic metabolism. This phenomenon is shown tooccur only in cyanobacteria capable of heterotrophy.

ResultsFNR Undergoes Proteolysis in Rodless PBS Mutants. In an attempt toclarify the FNR localization issue we examined phycobiliprotein(PBP) complexes purified from PBS-deficient mutants: CB(�cpcC1C2), in which the lack of two rod linkers (LR

33 and LR30)

results in PBS containing only one phycocyanin (PC) hexamerper rod instead of three in the WT (19), CK (�cpcBAC1C2)where the PBS is restricted to the allophycocyanin (APC)-containing core (B.U. and G.A., unpublished work), and AB(�apcAB) producing PC only and assembling uncoupled rods(20) (Fig. 1A). PBS from CB and WT contained similar amountsof FNR, whereas none was visible in PBS from CK. Rodspurified from AB contained a major contamination in the 45- to50-kDa range that prevented observation of the FNR (Fig. 1B).Immunological analysis with an FNR-specific antibody con-firmed the absence of FNR in purified cores from CK andrevealed its presence in rods from AB (Fig. 1B�). We probed

Author contributions: J.-C.T. and G.A. designed research; J.-C.T., B.U., and G.A. performedresearch; B.L. contributed new reagents/analytic tools; J.-C.T., B.U., and G.A. analyzed data;and G.A. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS direct submission.

Abbreviations: APC, allophycocyanin; FNR, ferredoxin:NADP(H) oxidoreductase; IRES,internal ribosome entry site; PBS, phycobilisome; PBP, phycobiliprotein; PC, phycocyanin.

§To whom correspondence should be addressed. E-mail: gajlani@cea.fr.

© 2006 by The National Academy of Sciences of the USA

18368–18373 � PNAS � November 28, 2006 � vol. 103 � no. 48 www.pnas.org�cgi�doi�10.1073�pnas.0607718103

Dow

nloa

ded

by g

uest

on

July

23,

202

1

total cell extracts from the WT and four PBS-mutants: CB, CK,AB, and PAL (PC�, �apcAB, �apcE), a mutant totally devoidof PBPs (21). In WT, CB, and AB extracts two isoforms weredetected at 46 and 34 kDa, which we designated FNRL andFNRS, respectively (Fig. 1C�). In CK and PAL cell extracts FNRLseemed to undergo proteolysis (Fig. 1C�) as has been observed(10). It should be noted that cell extracts from AB, PAL, and CKcontain relatively high amounts of FNRS, a point that is ad-dressed in Discussion. The fragments detected in extracts fromrodless mutants were reminiscent of the profile of the Anabaenasp. FNR produced in Escherichia coli (22); by the use of aproteinase-deficient mutant, Martinez-Julvez et al. (22) elimi-nated the intermediary products and obtained two polypeptidesof 49 and 36 kDa. Those results and our results suggest thatFNRL undergoes proteolysis when not bound to the PBS,whereas FNRS appears not be a proteolytic product of FNRL.

FNRS Accumulates When Photosynthesis Is Slowed Down. To deter-mine the origin of FNRS in Synechocystis, we tested the effect ofnitrogen starvation (known to induce PBS degradation) on theFNR in cell extracts from WT and two mutants in which PBSwere not degraded under nitrogen starvation conditions: N1lacking nblA1, a gene that is induced and necessary for PBStrimming under nitrogen starvation (23), and M55 deficient inndhB, encoding a subunit of the NDH-I complex (24). The latter

mutant exhibits attenuated PBS degradation when nitrogen-starved, probably because of deficient respiration (J.-C.T. andG.A., unpublished work). Under the nitrogen-starvation condi-tions used in our experiment, leading to the loss of two rod-linkers, LR

33 and LR30, and their associated PC in the WT, FNRL

did not undergo proteolysis in the WT nor in either nonbleachingmutant. FNRS accumulates in all three strains, although to aslightly lesser extent in the mutants (Fig. 2). We also probed totalextracts from the WT grown under different growth environ-ments: photoheterotrophic (light, 3-(3,4-dichlorophenyl)-1-1dimethylurea, glucose), chemoheterotrophic (dark, glucose),mixotrophic (dim-light, glucose), and iron starvation. Again,although no intermediary proteolysis of FNRL was observed, therelative amount of FNRS increased significantly in cells grownunder heterotrophic and iron-starvation conditions comparedwith cells grown under photoautotrophic conditions (Fig. 3).

FNRS Derives from a Second Translation Initiation Site. The uniquepetH gene in Synechocystis encodes a polypeptide of 413 aa (46kDa) and was shown to be transcribed into a single mRNA (25),indicating that FNRS is generated posttranscriptionally. A pu-tative translation start is found at Met-113, which would producean FNRS with the right size (34 kDa) and an N-terminal

Fig. 1. Characterization of the FNR in purified PBP complexes and total extractsfrom Synechocystis. (A) Representation of PBSs in the WT, a mutant bearing onePC hexamer per rod (CB), a PC-deficient mutant (CK), and PC-hexamer structuresin the APC-deficient mutant (AB). PC hexamers are represented as dark graycylinders and APC-containing cores are represented as three light-gray circles. (B)Polypeptide composition of purified PBP complexes. (B�) Western blot analysis ofthe FNR in the same PBS samples separated on an identical gel loaded with fourtimes less sample per well. (C) Coomassie-stained PAGE of total cell extracts fromthe above strains plus extracts from PAL, totally devoid of PBP. (C�) Western blotanalysis of the FNR in the same extracts separated on an identical gel loaded withhalf as much sample per well. The identities of the PBS subunits are labeled on theleft and approximate masses are labeled in kDa in the middle. LX, linker polypep-tide located at position X, where X can be R (rod), RC (rod core), C (core), or CM(core membrane); linkers with identical location are distinguished by a super-script indicating their mass.

Fig. 2. Characterization of the FNR during nitrogen starvation (for 0, 16, and34 h) in two mutants unable to degrade PBS, N1 and M55, compared with theWT. (A) Coomassie-stained PAGE of total extracts. (B) Western blot analysis ofthe FNR in the same extracts separated on an identical gel loaded with half asmuch sample per well.

Fig. 3. Immunodetection of the FNR in cell extracts of Synechocystis grownunder: PA, photoautotrophic; PH, photoheterotrophic; CH, chemoheterotro-phic; MX, mixotrophic; �Fe, iron starvation conditions.

Thomas et al. PNAS � November 28, 2006 � vol. 103 � no. 48 � 18369

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

23,

202

1

sequence matching the one found by Matsuo et al. (18) for anFNR that copurifies with the NDH-1 complex in Synechocystis.Although this may sound curious, many examples of internalribosome entry sites (IRESs) within coding regions have beendescribed in various organisms, although rarely in prokaryotes(26–28). To test this hypothesis, we introduced missense andframe-shift mutations in petH by a procedure similar to thatdescribed in ref. 6. Each construct (described in Methods and Fig.8, which is published as supporting information on the PNASweb site) was used to transform WT Synechocystis where doublerecombination led to its integration into the chromosome. Thefully segregated mutant carried the modified petH gene and anantibiotic-resistance cassette. The missense mutation in MI6changed the putative start methionine into an isoleucine. Frame-shift mutations, created by single base deletion or insertion,caused premature translation stops upstream and downstream ofMet-113 in FS1 and FS2, respectively (Fig. 4A). PCR, restrictionanalysis, and DNA sequencing confirmed the genotypes of themutants (data not shown).

As shown in Fig. 4B, cell extracts from mutant MI6 containedonly FNRL, consistent with FNRS originating from either anIRES at Met-113 or from a specific cleavage of FNRL justupstream of this residue. In cell extracts from FS1 and FS2(where translation initiated at Met-1 terminates before or justafter Met-113, respectively) FNRS was present at the WT FNRLlevel, whereas FNRL was absent (Fig. 4B, FS1 and FS2). There-fore, FNRS must be produced from Met-113 and not by prote-olysis. Furthermore, comparison of PBS purified from FS1 tothose from the WT clearly shows that the FNR was no longerassociated with PBS in the absence of the rod-linker domain(Fig. 4B, PBS), in agreement with previous experiments in which75 aa were deleted from FNRL (14).

We tested the effect of nitrogen starvation on the FNR inthese mutants. Fig. 5 clearly shows that FNRS did not accumulatein MI6 under nitrogen starvation, although clear PBS trimmingoccurred. FS1 contains only FNRS and its levels seem to increase

during nitrogen starvation. FS1 cultures appeared to bleachfaster than the WT and MI6.

Compared with the WT, no difference in growth rate wasobserved when MI6 was grown under photoautotrophic conditions

Fig. 4. FNRS is translated from Met-113 in Synechocystis. (A) (Upper) Schematic representation of the FNRL polypeptide with the linker, hinge (H), and enzymatic(FNR) domains. Start Met-1 and Met-113 are shown. Met-113 was changed to isoleucine (ATC) in MI6. Frameshifts in FS1 and FS2 result in translation termination(*). (Lower) Translated polypeptides are indicated. (B) Gel electrophoresis and immunoblot of total protein extracts and purified PBS. In the WT FNRL and FNRS

(unusually abundant in this sample) are detected, whereas only FNRL is present in MI6, the week secondary band is the product of either proteolysis or functionallyinsignificant translation initiation at codon 102; GTG). Only FNRS is present in FS1 and FS2 (the week secondary bands are probably aggregation products of FNRS;they are absent in the FS1 extracts presented in Fig. 5). Comparison of PBS purified from WT to those from FS1 confirms the absence of FNRL in FS1.

Fig. 5. Impact of nitrogen starvation (for 0, 6, and 34 h) on the FNR in totalextracts from the WT and mutants producing only FNRL (MI6) or FNRS (FS1). (A)Coomassie-stained PAGE of total extracts. (B) Western blot analysis of the FNRin the same extracts separated on an identical gel loaded with half as muchsample per well.

18370 � www.pnas.org�cgi�doi�10.1073�pnas.0607718103 Thomas et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

whereas FS1 growth was significantly slowed down (25% longerdoubling time). Under photoheterotrophic conditions MI6 growthwas slowed down whereas FS1 growth was similar to that of the WT.

FNRS Is Absent in Cyanobacteria Lacking a Second Methionine. FNRcontains a PBS-linker domain in all PBS-containing cyanobacterialstrains from which petH has been sequenced, except for G. violaceus(8) and Synechococcus spp. strains OS-A and OS-B� (9). Wetherefore wondered whether a second translation initiation site wasassociated with the presence of a linker domain in the FNR.Alignment of FNR sequences from different strains indicates thata putative initiating Met is present in Anabaena sp. strain PCC7120and Synechococcus sp. strain PCC7002, whereas it is absent in theobligate photoautotrophes: Synechococcus elongatus and Thermo-synechococccus elongatus (Fig. 6). The absence of a putative initi-ating site is also observed in all marine cyanobacteria of the genusSynechococcus sequenced to date (F. Partensky, personal commu-nication), whereas it is present in petH of the marine diazotrophTrichodesmium erythraeum (Joint Genome Institute gene ID 4667).We probed cell extracts from strains lacking a putative second

initiating site (T. elongatus and S. elongatus) and from G. violaceusas a control. Fig. 7 shows that no trace of a second FNR signal isdetected in any of these strains even under nitrogen-starvationconditions. As expected G. violaceus contained only FNRS. Cellextracts from Anabaena PCC7120 and Synechococcus PCC7002exhibit, as in Synechocystis, two FNR isoforms (data not shown).

DiscussionBy the use of site-directed mutants in Synechocystis, we demonstratethat two FNR isoforms are produced by differential initiation oftranslation. That mutant MI6 produced only FNRL whereas mu-tants FS1 and FS2 produced only FNRS argues against the widelyheld view that FNRS is a product of proteolytic cleavage of FNRLand demonstrates that the small isoform is the result of a secondtranslation initiation, at 337-nt distance from the first. To ourknowledge this demonstration of IRES is unique in photosyntheticprokaryotes. Another bacterial example is found in E. coli wherecheA (third locus in the mocha operon) encodes two proteins(CheAL and CheAS) translated from two in-frame start sites, thefunctional significance of these isoforms and the translation-initiation mechanism involved are not clear (29).

In Anabaena spp. petH produces two mRNAs; one is consti-tutive (starts at nucleotide �63 from Met-1) and the other oneis used in the absence of combined nitrogen (starts at nucleotide�188 from Met-1) (30). This larger transcript is expected toproduce FNRS by a yet-unknown posttranscriptional mechanismthat would prevent initiation at Met-1. The organization re-ported for petH in Synechocystis is markedly different becausevan Thor et al. (25) mapped only one transcription-start point atnucleotide �523 from Met-1. Although the nucleotide se-quences incorporating the initiation codon are known to deter-mine translation initiation efficiency in prokaryotes (31), initi-ation cannot be evaluated from knowledge of the sequencealone. An SD sequence is involved in translation initiation butmight not be required when the initiation codon is located withinan AT-rich sequence that forms no stable secondary structure(32), which is the case for the sequence incorporating Met-113in the Synechocystis petH. It is feasible for proteins or othersRNA to bind to an mRNA and alter its translation efficiency.Alternatively, a low level of translation can also allow RNAsecondary structures to form, causing premature termination,frameshifting, and initiation of transcription at codons that arenot used when translational level is high (33). The invertedrepeats present at the 5� end of petH mRNA in Synechocystis

Fig. 7. Cyanobacteria lacking the second putative initiating methionine, T.elongatus (T) and S. elongatus (S), contain only FNRL even after 38 h ofnitrogen starvation (�). G. violaceus contains only FNRS (G).

Fig. 6. Sequence alignments of six cyanobacterial FNR (N-terminal part) exhibits the three domains (rod-linker, hinge, and catalytic). Strains: ANA, AnabaenaPCC7120 (P58558); SP6, Synechococcus elongatus (Q5N4L3); TEL, Thermosynechococccus elongatus BP-1 (Q93RE3); S70, Synechococcus PCC7002 (P31973); SY3,Synechocystis PCC6803 (Q55318); and GVI, G. violaceus (Q7NI88). Amino acids are shown in boldface at positions where they are identical in �50% of thesequences compared (by Clustal W). The first initiating Met is italicized and the second one is underlined when present. A PEST-like sequence (highlighted) isfound only in Synechocystis.

Thomas et al. PNAS � November 28, 2006 � vol. 103 � no. 48 � 18371

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

23,

202

1

suggest that secondary structures are likely to form; mutationswithin these repeats could clarify the issue.

Proteolysis occurs in the linker domain of FNRL in the absenceof PC, just as PBS linkers are known to be highly sensitive toproteolysis when not associated to PBPs. This phenomenonoccurs in Synechocystis mutants lacking PC and during heterol-ogous expression of FNRL in E. coli (22); we found similarproteolytic fragments in extracts from PC-deficient mutants ofSynechocystis and E. coli expressing petH (Fig. 9, which ispublished as supporting information on the PNAS web site),which indicates the action of similar proteases in both bacteria.The sizes of the FNRL proteolytic fragments also suggest thatproteolysis occurs within the linker domain of the FNR and notonly at a putative PEST site as has been proposed (6, 10).Furthermore, heterologous expression of the mutant alleles MI6and FS failed to produce FNRS and FNRL, respectively, indi-cating that also in E. coli translation initiates at both sites (Fig.9). We also constructed a PC-deficient mutant of Synechocystiscarrying the MI6 mutation; in this strain FNRS is absent andFNRL exhibits the proteolytic profile characteristic of the ab-sence of PC (Fig. 10, which is published as supporting informa-tion on the PNAS web site).

FNRL might have been created in an ancestral cyanobacterium(like G. violaceus) by fusion of genes encoding PBS linker andFNRS. We show here that the second initiating methionine is notrequired for photosynthesis, which might explain its loss incyanobacteria that are obligate phototrophs. Its presence infacultative heterotrophs suggests a separate function for thesmall isoform in these organisms. We also show that differentphysiological conditions alter the relative amounts of the twoFNR isoforms, which has implications for their possible func-tions. FNRL seems clearly related to photosynthetic electrontransfer, while FNRS accumulates under heterotrophic andstarvation conditions where catabolism is stimulated and anab-olism slowed down.

Plastidic FNR seem to have adapted their catalytic efficiencyto the host tissue; leaf FNR satisfy the requirement of electronflow to sustain CO2 fixation, while root FNR acts as a shuttlebetween the NADPH generated by heterotrophic metabolismand electron-accepting enzymes, i.e., nitrite reductase (34).FNRL attachment to the PBS, in cyanobacteria, might provide away for the FNR to optimize electron flow for CO2 fixation(NADPH production); the slow photoautotrophic growth ofmutants lacking FNRL (FS) supports this hypothesis. FNRSaccumulation, in the WT during heterotrophic growth, and theimpairment of heterotrophic growth in MI6, suggest that theattachment to PBS is a hindrance for the FNR activity requiredunder these conditions (NADPH consumption). It is possiblethat in mutants such as CK, AB, and PAL, containing a higherphotosystem II/photosystem I ratio caused by PBS deficiencies(20, 21), FNRS accumulation (visible in Fig. 1C�) is caused by anexcess of NADPH. This hypothesis is strengthened by theaccumulation of FNRS in the WT grown under high light (Fig.11, which is published as supporting information on the PNASweb site). These results strongly suggest that FNRL is an NADP�

reductase, whereas FNRS is an NADPH oxidase.The fact that the obligately phototrophic cyanobacteria tested

here express only FNRL and that FNRS accumulates in Synecho-cystis (and most probably other facultatively heterotrophic cya-nobacteria) under conditions where heterotrophic metabolism isneeded raises the old issue of FNR being the dehydrogenase partof the NDH-I complex in cyanobacteria and plant plastids (3, 5, 18).

MethodsStrains and Growth Conditions. WT and mutants of Synechocystiswere grown as in ref. 19. For nitrogen starvation, cells wereharvested by centrifugation and resuspended in a medium whereNaCl replaced NaNO3.

Preparation of Cell Extracts. Cells were centrifuged, washed with 50mM EDTA, and resuspended in 20 mM Tricine, pH 8 containingComplete protease inhibitor (Roche, Meylan, France), thenbroken by vortexing 6 min with glass beads. Unbroken cells andglass beads were removed by centrifugation at 7,500 � g for 2min. The supernatant was used as total cell extract. PBSs werepurified as in ref. 35.

Gel Electrophoresis and Western Blotting. Proteins were separatedby using Tris-Tricine lithium dodecyl sulfate-12% PAGE. Pro-teins were trichloroacetic acid-precipitated, and the pellet wasresuspended in the loading buffer. Chlorophyll concentrationwas used to ensure equivalent loading of cell extracts, 2 or 4 �gchl was loaded per 6-mm well for blotting or Coomassie staining,respectively. For immunoblots, proteins were transferred toPVDF membranes by using semidry transfer. Blots were blockedwith Tris-buffered saline supplemented with 0.1% Tween and0.5% dry skimmed milk and incubated with the primary antibody(1:5,000 dilution). After washing, blots were incubated 2 h atroom temperature with a 1:15,000 dilution of peroxidase-conjugated anti-rabbit IgG (Promega, Madison, WI). The signalwas visualized by using ECL chemiluminescent substrate (Am-ersham, Piscataway, NJ) and autoradiography films. Images weregenerated by using a CCD camera and GraphicConvertersoftware.

Production and Purification of FNR Antibody. A 75-aa N-terminaltruncated form of the Synechocystis FNR (37 kDa) was expressedin E. coli BL21 (DE3) from a plasmid provided by J. J. van Thor(University of Amesterdam, Amsterdam, The Netherlands) (36).An isopropyl �-D-thiogalactoside-induced E. coli culture wasused to purify the FNR to homogeneity as follows. A 50–70%ammonium sulfate precipitation of the crude soluble extract wasfollowed by column chromatography purifications: DE52 (What-man, Middlesex, UK), Hitrap Q Sepharose, HiLoad PhenylSepharose (Amersham Pharmacia), and a final blue Sepharoseaffinity. The purified FNR was then used to raise antibodies inrabbits. A pure Ig fraction was obtained from the crude sera ofrabbits as in ref. 37.

Construction of the Mutagenic Plasmid. A 318-bp fragment (213 bpupstream to 105 bp downstream of the petH first ATG codon)was amplified from genomic DNA by using primers SB and BS.This fragment was cloned between XbaI and BamHI sites ofpBC, creating pSB7. A 2,561-bp BamHI–EcoRV fragment con-taining the Synechocystis petH gene (710 bp upstream of the ATGto 612 bp downstream of the stop) was amplified from genomicDNA by using primers BE and EB. This fragment was clonedbetween BamHI and EcoRV sites of pSB7, yielding pSBH. Theomega cassette was inserted in the unique BamHI site of pSBH,yielding the mutagenic plasmid pSBH� (Fig. 8). Sequences ofprimers used in this study are listed in Table 1, which is publishedas supporting information on the PNAS web site.

Site-directed mutagenesis was performed by PCR on a plasmidcontaining a 570-bp XbaI–MscI fragment. The base transversion inMI6 was created with the mutagenic primer pair IFN and WRN,their overlapping 5� ends containing a silent mutation eliminatingan AlwNI site; the mutation was carried by IFN. Frameshiftmutations in FS1 and FS2 were created by cytidine insertion anddeletion, respectively, at the AlwNI site. Primers SHF and SHRwere used for FS1, and SIF and SIR were used for FS2. Plasmidswere sequenced to ensure that no error occurred during the PCRand that the desired mutations were created.

Transformation of Synechocystis PCC 6803. The WT Synechocystiswas transformed by mutagenic plasmids carrying each of thethree mutations MI6, FS1, and FS2. Transformants were se-lected on plates containing 50 �g�ml�1 spectinomycin and 20 mM

18372 � www.pnas.org�cgi�doi�10.1073�pnas.0607718103 Thomas et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

1

glucose. Complete segregation was confirmed by PCR andrestriction analysis.

We thank T. Ogawa (National Laboratory of Plant Molecular Genetics,Shanghai, China) for the generous gift of mutant M55; C. Richaud(Ecole Normale Superieure, Paris, France) for providing the N1 mutant;J. J. van Thor for sharing plasmids; A. Boussac for help in growing T.elongatus; D. Picot and C. Reiss for fruitful discussions; and the Kazusa

DNA Research Institute and the Department of Energy Joint GenomeInstitute for the use of genomic sequences. G.A. is indebted to B. Robertfor support and trust and A. A. Pascal for encouragement and criticalreading of the manuscript. This work was supported by the CentreNational de la Recherche Scientifique Unite de Recherche Associee2096 and Formation de Recherche en Evolution 2910, the Commissariata l’Energie Atomique, Saclay, Joliot-Curie Biology Department, and theAgence Nationale de Recherches program PhycoSyn.

1. Bowsher CG, Hucklesby DP, Emes MJ (1993) Plant J 3:463–467.2. Neuhaus HE, Emes MJ (2000) Annu Rev Plant Physiol Plant Mol Biol

51:111–140.3. Gonzalez de la Vara L, Gomez-Lojero C (1986) Photosynth Res 8:65–78.4. Scherer S, Alpes I, Sadowski H, Boger P (1988) Arch Biochem Biophys

267:228–235.5. Guedeney G, Corneille S, Cuine S, Peltier G (1996) FEBS Lett 378:277–280.6. van Thor JJ, Jeanjean R, Havaux M, Sjollema KA, Joset F, Hellingwerf KJ,

Matthijs HC (2000) Biochim Biophys Acta 1457:129–144.7. Karplus PA, Faber HR (2004) Photosynth Res 81:303–315.8. Nakamura Y, Kaneko T, Sato S, Mimuro M, Miyashita H, Tsuchiya T,

Sasamoto S, Watanabe A, Kawashima K, Kishida Y, et al. (2003) DNA Res10:137–145.

9. Steunou AS, Bhaya D, Bateson MM, Melendrez MC, Ward DM, Brecht E,Peters JW, Kuhl M, Grossman AR (2006) Proc Natl Acad Sci USA 103:2398–2403.

10. Schluchter WM, Bryant DA (1992) Biochemistry 31:3092–3102.11. Sauer J, Schreiber U, Schmid R, Volker U, Forchhammer K (2001) Plant

Physiol 126:233–243.12. Allen MM, Smith AJ (1969) Arch Mikrobiol 69:114–120.13. Yamanaka G, Glazer AN, Williams RC (1978) J Biol Chem 253:8303–8310.14. van Thor JJ, Gruters OW, Matthijs HC, Hellingwerf KJ (1999) EMBO J

18:4128–4136.15. Gomez-Lojero C, Perez-Gomez B, Shen G, Schluchter WM, Bryant DA (2003)

Biochemistry 42:13800–13811.16. Yao Y, Tamura T, Wada K, Matsubara H, Kodo K (1984) J Biochem (Tokyo)

95:1513–1516.17. Rowell P, Diez J, Apte SK, Stewart WD (1981) Biochim Biophys Acta

657:507–516.18. Matsuo M, Endo T, Asada K (1998) Plant Cell Physiol 39:263–267.

19. Ughy B, Ajlani G (2004) Microbiology 150:4147–4156.20. Ajlani G, Vernotte C, DiMagno L, Haselkorn R (1995) Biochim Biophys Acta

1231:189–196.21. Ajlani G, Vernotte C (1998) Plant Mol Biol 37:577–580.22. Martinez-Julvez M, Hurley JK, Tollin G, Gomez-Moreno C, Fillat MF (1996)

Biochim Biophys Acta 1297:200–206.23. Richaud C, Zabulon G, Joder A, Thomas JC (2001) J Bacteriol 183:2989–2994.24. Ogawa T (1991) Proc Natl Acad Sci USA 88:4275–4279.25. van Thor JJ, Hellingwerf KJ, Matthijs HCP (1998) Plant Mol Biol 36:353–363.26. Komar AA, Lesnik T, Cullin C, Merrick WC, Trachsel H, Altmann M (2003)

EMBO J 22:1199–1209.27. Maier D, Nagel AC, Preiss A (2002) Proc Natl Acad Sci USA 99:15480–15485.28. Smith RA, Parkinson JS (1980) Proc Natl Acad Sci USA 77:5370–5374.29. Sanatinia H, Kofoid EC, Morrison TB, Parkinson JS (1995) J Bacteriol

177:2713–2720.30. Valladares A, Muro-Pastor AM, Fillat MF, Herrero A, Flores E (1999) FEBS

Lett 449:159–164.31. Draper DE (1996) in Escherichia coli and Salmonella, ed Neidhardt FC (Am

Soc Microbiol, Washington, DC), Vol 1, pp 902–908.32. Kozak M (2005) Gene 361:13–37.33. Engelberg-Kulka H, Schoulaker-Schwarz R (1996) in Escherichia coli and

Salmonella, ed Neidhardt FC (Am Soc Microbiol, Washington, DC), Vol 1, pp909–921.

34. Ceccarelli EA, Arakaki AK, Cortez N, Carrillo N (2004) Biochim Biophys Acta1698:155–165.

35. Elmorjani K, Thomas J-C, Sebban P (1986) Arch Microbiol 146:186–191.36. van Thor JJ, Geerlings TH, Matthijs HC, Hellingwerf KJ (1999) Biochemistry

38:12735–12746.37. McKinney MM, Parkinson A (1987) J Immunol Methods 96:271–278.

Thomas et al. PNAS � November 28, 2006 � vol. 103 � no. 48 � 18373

PLA

NT

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

July

23,

202

1