Light-dependent attenuation of phycoerythrin geneexpression reveals convergent evolution of greenlight sensing in cyanobacteriaRyan P. Bezya,b, Lisa Wiltbanka, and David M. Kehoea,c,1

aDepartment of Biology, Indiana University, Bloomington, IN 47405; bDepartment of Natural and Applied Sciences, Mount Mercy University, Cedar Rapids,IA 52402; and cIndiana Molecular Biology Institute, Indiana University, Bloomington, IN, 47405

Edited by Susan S. Golden, University of California at San Diego, La Jolla, CA, and approved October 4, 2011 (received for review May 10, 2011)

The colorful process of chromatic acclimation allows many cyano-bacteria to change their pigmentation in response to ambient lightcolor changes. In red light, cells produce red-absorbing phycocy-anin (PC), whereas in green light, green-absorbing phycoerythrin(PE) is made. Controlling these pigment levels increases fitness byoptimizing photosynthetic activity in different light color environ-ments. The light color sensory system controlling PC expression iswell understood, but PE regulation has not been resolved. In thefilamentous cyanobacterium Fremyella diplosiphon UTEX 481, twosystems control PE synthesis in response to light color. The first isthe Rca pathway, a two-component system controlled by a phyto-chrome-class photoreceptor, which transcriptionally repressescpeCDESTR (cpeC) expression during growth in red light. The sec-ond is the Cgi pathway, which has not been characterized. Wedetermined that the Cgi system also regulates PE synthesis byrepressing cpeC expression in red light, but acts posttranscription-ally, requiring the region upstream of the CpeC translation startcodon. cpeC RNA stability was comparable in F. diplosiphon cellsgrown in red and green light, and a short transcript that includedthe 5′ region of cpeCwas detected, suggesting that the Cgi systemoperates by transcription attenuation. The roles of four predictedstem–loop structures within the 5′ region of cpeC RNA were ana-lyzed. The putative stem–loop 31 nucleotides upstream of thetranslation start site was required for Cgi system function. Thus,the Cgi system appears to be a unique type of signal transductionpathway in which the attenuation of cpeC transcription is regu-lated by light color.

light regulation | photosynthesis | post-transcriptional control |phycobilisome | phenotypic plasticity

Photosynthetic gene expression is precisely regulated in re-sponse to environmental conditions and controlled at the

transcriptional and posttranscriptional levels in plants, algae, andcyanobacteria (1–3). In some cyanobacteria, genes encodingphotosynthetic light-harvesting proteins are regulated by lightcolor by chromatic acclimation (CA) (4–6). CA-mediated accu-mulation of these proteins, which contain covalently attachedbilin chromophores, allows the tailoring of the absorption profileof these structures to match the spectral distribution of ambientlight. These changes maximize photon capture for photosyn-thesis, providing a selective advantage in changing light colorenvironments (7). Two forms of CA, type 2 and type 3, exist inspecies that contain the two light-harvesting proteins phycocya-nin (PC), which maximally absorbs red light, and phycoerythrin(PE), which maximally absorbs green light (8). CA is widespread,as nearly 75% of the species containing PC and PE are capableof one of these two types of CA (8). CA2-capable cyanobacteriaproduce more PE in green light than in red light but do not alterPC levels in response to light color. CA3-capable species regu-late PE production in response to light color similarly to CA2species, but also make more PC in red light than in green light.Two signaling pathways control CA3 in Fremyella diplosiphon

(6). The Rca two-component system transcriptionally regulates

PC and PE production by repressing PE-encoding genes andactivating PC-encoding genes in red light. The Rca sensor is thephotoreceptor RcaE, the founding member of the cyanobacter-iochromes (9–11). It contains a histidine kinase domain thatmodulates the activity of the OmpR-class transcription factorRcaC, which binds to direct DNA repeats upstream of CA3-regulated genes called the L box, activating red light expressedgenes and repressing green light expressed genes (12–14). TheCgi system controls only the activity of genes that are highlyexpressed in green light, and may operate by controlling thecpeCDESTR (cpeC) operon (6), which encodes light-harvestingantenna proteins and a CA3 activator that regulates additionalgreen light expressed genes (15, 16).The pattern of gene regulation by the Rca and Cgi systems led

to the proposal that CA2 and CA3 capabilities have evolvedby loss of the Rca system from a CA3 species to create a CA2species or by acquisition of the Rca system by a CA2 species tocreate a CA3 species (6, 17). This proposal was explored usingthe CA2 species Nostoc punctiforme, where sensor kinase/responseregulator encoding genes were found adjacent to cpeCGR1 (18).These proteins, CcaS/R, are highly similar to RcaE and RcaCbut appear to function in an opposite manner in red and greenlight. It was proposed that the F. diplosiphon Cgi system, which isuncharacterized at the molecular level, is equivalent to the CcaS/CcaR transcriptional regulatory system. It was also proposed thatCA3 capability evolved through CcaSR duplication to create theRca system, which then diverged to transcriptionally regulate bothPC- and PE-encoding genes.In the descendants of endosymbiotic cyanobacteria, chlor-

oplasts, posttranscriptional processes play an important role inregulating gene expression (19–23), often through the mRNA 5′leader. Stem–loop structures located near the translation initia-tion site are frequently used as binding sites for multiple proteinsthat influence the translation rate (22, 24). In cyanobacteria,posttranscriptional regulation of photosynthetic gene expressionis also an important control mechanism, but is only known tooccur via changes in mRNA stability (25–28).Here, we characterize the molecular basis of the Cgi system

regulation of cpeC in F. diplosiphon, finding that this systemoperates at the posttranscriptional level and, unlike known post-transcriptional regulation in cyanobacteria, does not involvedifferential mRNA stability. The region of cpeC upstream of theCpeC translation start site and downstream of the cpeC tran-scription start site (the “5′ leader region”) contains a sequence

Author contributions: R.P.B., L.W., and D.M.K. designed research; R.P.B. and L.W. per-formed research; R.P.B., L.W., and D.M.K. analyzed data; and R.P.B., L.W., and D.M.K.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: dkehoe@indiana.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1107427108/-/DCSupplemental.

18542–18547 | PNAS | November 8, 2011 | vol. 108 | no. 45 www.pnas.org/cgi/doi/10.1073/pnas.1107427108

required for Cgi control with predicted secondary structure andlocation that is similar to cis elements that regulate gene ex-pression in plant and green algal chloroplasts. A short transcriptat the 5′ end of cpeC was also detected. These findings indicatethat the Cgi pathway is a transcription attenuation system that isregulated by light color and provide insights into the evolution oflight regulated signal transduction pathways in CA2- and CA3-capable cyanobacteria.

ResultsTo identify cpeC cis element(s) needed for Cgi system operationand determine whether this pathway functions independently ofthe Rca system, several translational fusions were created usingthe cpeC upstream region, the non-light–regulated cpc1 5′ leader(29, 30), and the gusA reporter gene (31) (Fig. 1A). cpeC up-stream and 5′ leader regions (pRB7) conferred a 10-fold increasein β-glucuronidase (GUS) activity in green versus red light (Fig.1B), mirroring the previously measured 10-fold change in cpeCmRNA levels (15, 32–34). Thus, all elements needed for CA3regulation of cpeC expression are contained in the region from−412 to +196 in pRB7. Replacing the L box direct repeat(pRB8) increased GUS activity levels in red light but not greenlight (Fig. 1B), confirming the Rca system’s repressing effect oncpeC expression in red light (14). The ∼2.5-fold green light in-duction remaining after L box replacement, identical to thatmeasured for cpeC mRNA in rcaC and rcaE mutants (15, 33), isdue to the Cgi system. This demonstrates that the Cgi pathway re-quires the upstream and/or 5′ leader regions of cpeC. Replacingthe cpeC 5′ leader with the cpc1 5′ leader in pRB7 to create

pRB1 also led to increased GUS activity in red light but onlya slight increase in green light (Fig. 1B). This ∼2.5-fold greenlight induction was due to Rca regulation, since cpeC light colorregulation was completely lost after the L box was replaced inpRB1 (pRB6) (Fig. 1B). These results demonstrate that althoughthe Cgi system operates independently of the Rca system, itprovides additional repression of cpeC expression in red light.They also show that the Cgi system requires the cpeC 5′ leaderbut not the region upstream of the transcription start site, sug-gesting that unlike the Rca system, the Cgi pathway acts post-transcriptionally.We measured rates of cpeC mRNA loss during growth in red

and green light to determine if differential abundance of thistranscript could be explained by light color-dependent mRNAdegradation activity. An rcaE null mutant (11) was used for thisexperiment to eliminate the influence of the Rca system. Therewas no significant difference in cpeC mRNA stability duringgrowth in red versus green light (red light half life = 14.1 min,green light half life = 16.3 min, P= 0.523) (Fig. 2). Thus, changesin cpeCmRNA levels mediated by the Cgi pathway are not due todifferent RNA degradation rates in red and green light.The Cgi system does not appear to operate at the transcrip-

tional level or the level of mRNA stability. We examined thepossibility that it works by transcriptional attenuation, whichshould produce a short species of cpeCmRNA encompassing the5′ leader region. Analysis of cpeC expression using a probe madeto the coding region of the first gene in this operon revealed twotranscripts, of ∼2.2 and 3.2 kilo-nucleotides (knt), that corre-spond to two- and three-gene transcription units (“long tran-scripts”) (Fig. 3 A and B). These RNAs are 10 times moreabundant in green light than red light in wild-type (WT) cells and2.5 times more abundant under the same light conditions in anrcaC null mutant, due primarily to greater expression in red light(Fig. 3 A and C), as previously noted (34). However, whena probe made to the cpeC 5′ leader region was used (Fig. 3B), anadditional RNA (“short transcript”) was detected that was muchsmaller than the long transcripts (Fig. 3A Center). Poly-acrylamide gel electrophoresis was used to size the short tran-script at ∼125 nt (Fig. 3A Right). Quantification of theabundance of the long transcripts detected using the cpeC 5′leader probe in WT and the rcaC mutant revealed patterns ofexpression in red and green light comparable to those obtainedwhen using a probe from within the coding sequence, with a 10-fold difference for WT and a ∼2.5-fold difference for the rcaCmutant (Fig. 3C) (34). In WT, the short transcripts were ap-proximately three times more abundant in green light than red

0

20

40

60

80

100

120

pRB8 pRB1 pRB6pRB7RL GL

L box

RL GLRL GLRL GL

A

B

Rela

�ve

Ac�v

ity (%

)

gusA-421 +286

pRB1

pRB6

pRB7

pRB8

+196

+38/+10

cpeC

cpeC

cpeC

cpeC

gusA

gusA

gusA

-421

cpc1

cpc1

Fig. 1. Role of the cpeC 5′ leader in Cgi regulation. (A) Diagrams of theconstructs used. Upstream and 5′ leader regions of cpeC with (pRB7) andwithout (pRB8) the L box (black triangle) were joined translationally to gusA.The CpeC start codon begins at +188, so these translational fusions includethe first three amino acids of CpeC. The region upstream of cpeC with andwithout the L box, joined to the cpc1 5′ leader, was also joined transla-tionally to gusA (pRB1 and pRB6). Bent arrows, transcription start site. (B)Relative mean rates of GUS activity from cell lysates of F. diplosiphontransformed with the indicated plasmids and grown in red light (RL) orgreen light (GL). The mean value (302 nmol of product per mg of proteinper min) derived from cells transformed with pRB1 and grown in green lightwas set at 100%. At least five independently transformed lines were testedfor each plasmid and light condition. Error bars show SE.

0

20

40

60

80

100

0 10 20 30 40 50 60

% re

mai

ning

RNA

Fig. 2. Rates of cpeC RNA loss in F. diplosiphon during growth in red andgreen light. RNA blot analyses measured cpeC RNA levels in the rcaE mutantof F. diplosiphon during growth in red light (open circles) or green light(filled circles). Rifampicin was added at time 0. For each light condition, RNAlevels are expressed as a percentage of cpeC transcripts at the zero timepoint, which was set to 100%. Means of all values were calculated afterloading normalization using ribosomal values. A one-phase decay model wasused for the nonlinear fit in the regression analysis (Prism 5, GraphpadSoftware). Six independent assays were conducted for each light condition.Error bars show SE.

Bezy et al. PNAS | November 8, 2011 | vol. 108 | no. 45 | 18543

PLANTBIOLO

GY

light. However, the abundance ratios of long to short transcriptswere different in red versus green light. In red light, it was 0.86,whereas in green light it was 3.6. Thus, there was a higher per-centage of short transcripts in red light than in green light. The lackof a functional Rca system in the rcaC mutant dramatically af-fected the overall amounts of small transcripts in red and greenlight. The small RNA was six times more abundant in this mutantthan inWT in red light, but only slightly lower than inWT in greenlight. The ratios of long to short transcripts in red and green light inthis mutant were quite similar to those measured for WT, withvalues of 0.64 in red light and 3.0 in green light. These data suggestthat the Cgi system has a role in shifting the ratio of short to longcpeC transcripts in red light versus green light, with a greaterpercentage of short transcripts in red light and a greater percent-age of long transcripts in green light. These shifts are consistentwith aCgi regulatorymechanismof cpeC transcription attenuationin which red light causes more frequent premature terminationand less accumulation of the long forms of this RNA.

Mfold RNA secondary structure analyses (35) of the cpeC 5′leader predicted the formation of four stem–loop structures(Fig. 4A). To test their role in the Cgi system, deletion constructs

~125 nt~2.2 knt

~3.2 knt~5.2 knt

3

0.5

1

2 3

0.5

1

2

knt knt

nt

100

200

300

A

B

C

Fig. 3. Detection of a cpeC 5′ leader transcript. (A) Representative auto-radiographs of RNA blot analyses for WT and the rcaC mutant in red light(RL) or green light (GL) after hybridization with a cpeC coding sequenceprobe (Left) or a probe encompassing the first 150 bp of the cpeC 5′ leader(Center and Right). The short transcript was resized using a polyacrylamidegel (right). Ribosomal loading controls (ribo) are shown. Boxed areas demarkregions used to quantify the results. (B) Diagrams of the cpeC operon, probelocations, and transcripts detected. Sizes and classifications as long or shorttranscripts are shown, as well as the genes included in the long transcripts.The ∼5.2-knt transcript is detectable after a longer exposure (Fig. S1).Dashed lines, probe locations; bent arrow, transcription start. (C) Relativemean accumulation levels of short and long cpeC transcripts [boxed areas in(A)] in WT and rcaC mutant cells grown in red or green light. Values areexpressed as a percentage of the long transcripts value for WT cells grown ingreen light after ribosomal normalization (ribo). Numbers in parentheses arethe ratios (to the nearest tenth) of long to short transcripts for each lightcondition and strain. Data are from three independent RNA blot analyses.Error bars show SE.

pRB8 cpeC

cpeC

gusA

5’ a a caaagcguauaaa ac g ccuaaaaccuugagcaucuaggaaaucac

ag

cu

uu

ca

ug

cuuauggaagcugguaa uuuggguugcaac a a a

uc

aa

uc

ca

aa g a a

uu g a u g a

gc c u g g a u g a g

c a a uc u gaaauuuauuuaca a a u

cga

aa

ca

aa

uu

uc u u a

aa

cu

uucguuaacagg ag

aa

acaauuaaug

cc

auuug

g3'

1

2

3

4

A

B -421

pRBΔ2

pRBΔ3

+196

pRBΔ4

V

V

V

2 3 4

C

pRBΔ1

1

gusA

gusA

gusA

gusA

0

20

40

60

80

100

120

pRBΔ3pRB8 pRBΔ1 pRBΔ2RL GL RL GLRL GLRL GL

Rela

�ve

Ac�v

ity (%

)

pRBΔ4RL GL

pRB22RL GL

pRB22 gusA

+10

+30

+20

+40+50

+60

+70

+80

+90

+100

+120+130

+150

+170

+160

+180

+190

+110+140

Fig. 4. Contribution of putative stem4 to Cgi function. (A) Mfold secondarystructure prediction (35) for the cpeC 5′ leader from +1 to +196. The startcodon and SD (arrow) sequences are in bold. Regions corresponding tostem1-stem4 are numbered accordingly. (B) Diagrams of the constructs usedin (C). The upstream and 5′ leader regions of cpeC without the L box werejoined translationally to gusA (pRB8), which was used to create deletions ofstem–loops 1 (pRBΔ1), 2 (pRBΔ2), 3 (pRBΔ3), or 4 (pRBΔ4), which are in-dicated in the pRB8 sequence as black boxes. Sequence between stem4 andthe SD (blue box) was also replaced (pRB22). Bent arrows, transcription start.(C) Relative mean rates of GUS activity from lysates of F. diplosiphontransformed with the indicated plasmids and grown in red light (RL) orgreen light (GL). The mean value (240.5 nmol of product per mg of proteinper min) derived from cells transformed with pRB8 and grown in green lightwas set at 100%. At least five independently transformed lines were testedfor each plasmid and light condition. Error bars show SE.

18544 | www.pnas.org/cgi/doi/10.1073/pnas.1107427108 Bezy et al.

were created that eliminated each putative stem–loop (stem1 tostem4), using pRB8 as the base construct (Fig. 4B), which lacksthe L box and removes Rca system influence. The putative stem–

loop sequences were eliminated rather than replaced to avoidcreating alternative secondary structures. Stem1 included regions1, 2, and 3 of the 5′ leader, whereas stem2, stem3, and stem4corresponded to regions 2, 3, and 4 in Fig. 4A. Cells carryingthese plasmids were analyzed for increased GUS activity duringgrowth in red and green light. Only the removal of stem4 led toelevated GUS activity in red light (Fig. 4C), the same phenotypeas obtained by replacing the entire cpeC 5′ leader region (Fig. 1).We tested whether this effect was specific for stem4 or due to itscloseness to the putative Shine–Dalgarno (SD) region (arrow,Fig. 4A) by substituting random sequence for the region betweenstem4 and the SD region (pRB22) and measuring GUS activityin red and green light. This construct repressed GUS activity inred light as well as pRB8 (Fig. 4C), demonstrating that the effectof stem4 removal was not due to its proximity to the SD region.Thus, the stem4 region of the cpeC 5′ leader is important for Cgisystem function.Four modified forms of the 30-bp stem4 region were tested to

determine the effect of the sequence changes in the stem whilemaintaining (pRB21) or reducing (pRB19) the free energy offolding, changing the loop sequences (pRB18), and modifyinga single-nucleotide stem mismatch (pRB20) (Fig. 5A). Each was

tested for Cgi system function in red and green light, with pRB8and the stem4 deletion construct used as controls (Fig. 5B).These all resulted in the loss of red light repression of GUSactivity. Thus, minor changes in the sequence and/or structure ofstem4 disables the Cgi system, independent of the free energyof folding or whether they are located in the putative stem orloop region. Also, the position of stem4 is critical, since movingstem4 closer to or further from translation start eliminated Cgisystem function (Fig. S2).

DiscussionLight color regulation of cpeC expression by the Cgi system inF. diplosiphon operates by posttranscriptional repression in redlight. This regulation is not due to differential cpeCRNA stabilityin red and green light, but appears to operate via transcriptionalattenuation. In addition, this system requires the cpeC 5′ leaderregion adjacent to the putative SD sequence, which is predictedto form a stem–loop structure. Thus, the Cgi system is a pre-viously unidentified type of light regulated signal transductionpathway that appears to use a stem–loop structure adjacent tothe translation start site, similar to those regulating the expres-sion of many chloroplast genes (21–24).Control of cpeC mRNA levels through a mechanism not in-

volving differential RNA stability (Fig. 2) is unique among cya-nobacterial genes that are posttranscriptionally regulated. Twomechanisms of posttranscriptional control of gene expressionare known in cyanobacteria: light-regulated changes in the sta-bility of the transcripts from psbA genes (25, 28, 36–39), whichmay operate via ribosome pausing rather than at the level oftranslation initiation (26, 40, 41), and antisense RNA regulationof isiA expression, which has been proposed to alter the degra-dation rates of transcripts from this gene (42). Posttrans-criptional control of hliA and hspA gene expression has also beenreported, but the mechanism(s) through which these processesoperate are not known (27, 43).The involvement and location of the stem4 region suggests

that similarities may exist between the Cgi system and thoseregulating chloroplast translation in plants and algae, in which 5′leaders of chloroplast mRNAs contain regulatory sequences neartranslation initiation sites (44–50). Although these sequencesoperate in different ways (22, 23), they often form stem–loopstructures similar in size and distance from the translation startsite as stem4 of the cpeC 5′ leader. Proteins, some of which arelight and redox regulated, interact with these stem–loop struc-tures and activate or repress translation rates of the transcripts.For example, the rps7 and psbA 5′ leaders have protein bindingsites within stem–loop structures upstream of their translationstart codons, and these are 70–84% A–U rich, equivalent tothe 80% A–U composition of stem4 in the cpeC 5′ leader. Al-though these structural similarities suggest the possibility of re-lated mechanisms, the final responses differ. Transcription andtranslation are coupled in bacteria, so cpeC transcriptional at-tenuation may be regulated by this stem–loop. In chloroplasts,these processes are predominantly uncoupled (21), and thesestem–loop structures are important in translational regulation.Overall, our data suggest that the Cgi system mechanism may bestructurally related to a subset of mechanisms that provide animportant form of regulation of chloroplast gene expression inplants and green algae.The process through which the Cgi system might cause more

frequent cpeC transcription attenuation in red light than greenlight is not yet clear, although it seems relatively inefficient be-cause significant amounts of short transcript are still present ingreen light (Fig. 3C). It is unlikely to involve a riboswitch, be-cause these require physical or thermal inputs (51, 52) that couldnot be provided by an elicitor such as light. It is also probablynot via antisense RNA, because a single nucleotide changewithin stem4 abolished the Cgi repression of cpeC (Fig. 5). It is

0

20

40

60

80

100

120

140

g a a a u u u a u u ua c aaaucg

aaacaaauuuc

A

stem 4

c u a a a u a c u a ua c aaaucg

auaauaauuag

g c a g u c g a c u ga c aaaucg

cagccgacugc

g a a a u u u a u u ua g t tagtc

aaacaaauuuc

g a a a u u u a u u ua c aaaucg

aaauaaauuuc

* *

*

*

*

*

*

*

**

*

** *

*

*

*

*

*

*

**

*

*

*

* * *

***

Δ stem 4

pRB20

pRB18

pRB19

pRB21

pRBΔ4

pRB8

Rela

�ve

Ac�v

ity (%

)

pRB8

RL GL

pRB20

RL GL

pRBΔ4

RL GL

pRB21

RL GL

pRB18

RL GL

pRB19

RL GL

B

ΔG

-3.4

-3.4

-6.3

-3.4

-17.1

Muta�ons to stem 4

N/A

pRB8 cpeC gusA-421 +196

Fig. 5. Effect of altering stem4 features on the Cgi response. (A) Diagramsof the changes in stem4 sequence tested in B. Changes were made in pRB8,with the upstream and 5′ leader regions of cpeC lacking the L box andjoined translationally to gusA. The deletions of and substitutions in stem4,with their corresponding free energies of folding in kcal/mol, are shown.Asterisks denote mutated bases. Bent arrow, transcription start. (B) Relativemean rates of GUS activity from lysates of F. diplosiphon transformed withthe indicated plasmids and grown in red light (RL) or green light (GL). Themean value (251.5 nmol of product per mg of protein per min) derived fromcells transformed with pRB8 and grown in green light was set at 100%. Atleast four independently transformed lines were tested for each plasmidand light condition, except for pRB8, which was tested three times. Errorbars show SE.

Bezy et al. PNAS | November 8, 2011 | vol. 108 | no. 45 | 18545

PLANTBIOLO

GY

also unlikely that this system represses cpeC expression bytranslating an ORF within the cpeC 5′ leader, because in-troducing a stop codon within each of the three small ORFswithin this leader failed to cause a major increase in reportergene expression in red light (Fig. S3). The strong effect of anystem4 modification suggests that it may interact with one or moreproteins, and its proximity to the SD and translation start codonsuggest that either stem4 or its associated components mayinteract with the ribosome. Because transcription attenuationsystems are controlled by varying the extent of coupling betweentranscription and translation, for the Cgi regulation of cpeC ex-pression, attenuation should occur within the first coding regionin the cpeC operon. Because the cpeC 5′ leader is 196 nt long, theattenuated transcript should be longer than this length. How-ever, the species detected from this region on the RNA blot wasonly ∼125 nt long (Fig. 5), apparently ending at the 3′ end ofstem1 (Fig. 4). This size discrepancy might result from exo-nuclease cleavage of the 3′ end of the attenuated transcript up towhere the stem1 secondary structure would stop any furthernuclease activity. This possibility is supported by 3′ RACE resultsusing RNA from a rcaC null mutant (34), which produced onepredominant band that, when sequenced, proved to be cpeCsequence extending 413 nt from transcription start, or 226 ntfrom translation start (Fig. S4). The 3′ end sequences showedheterogeneity, suggesting that cpeC RNAs with different 3′ endswere present in the sample. It is also possible that the proposedRNA stem–loop instead forms in the DNA; we do not currentlyhave evidence to support either possibility. However, it is unlikelythat a DNA structure is controlling transcription attenuation be-cause if it did form in the DNA, the short cpeC transcript wouldlikely end before the structure, perhaps close to +30 (Fig. 4A).Recently, it was proposed that orthologs of the N. punctiforme

CcaSR proteins, which control CA2 regulation of PE expression,might make up the Cgi system in CA3-capable species (18). Thisis unlikely because the CcaSR system apparently acts transcrip-tionally, whereas the results presented here demonstrate that theCgi system operates posttranscriptionally. It is conceivable thatcomponents such as CcaS and CcaR operate in the initial stepsof the Cgi pathway, although this is doubtful because genesencoding these components are not present in the genome ofanother CA3-capable species, Synechococcus sp. PCC 7335 (18).In addition, the differences we have found between the CcaSRand Cgi systems do not support the hypothesis that CA2 andCA3 regulatory systems are related by either the straightforwardaddition or subtraction of the Rca system (6, 17). They alsodemonstrate that CA regulation of PE expression must haveevolved more than one time and in more than one way. Thisfinding contrasts with our current understanding of the evolutionof CA regulation of PC synthesis, which uses highly conservedRca system components and operates through the L box regu-latory element in all species examined thus far (30, 53). Whetherthe differences between the evolution of PE and PC regulationare the result of more recent evolution of CA control of PCexpression, or of greater selection pressure acting on the Rcasystem, remains to be determined.The original description of the CA2 versus CA3 phenotype

noted that no CA2 species was able to completely halt PE pro-duction in green light, unlike a number of CA3 species (8). Amolecular explanation for this difference is now possible. PEproduction is very strongly suppressed by the combined effects ofthe Rca system, which operates via transcriptional repression, andthe Cgi system, which also represses production, but does soposttranscriptionally. Conversely, in the CA2 species N. puncti-forme, only a single system regulates PE expression, which is nevercompletely repressed in red light (8, 18). This initial study alsonoted significant differences in the amount of PE and PC producedin different species, and within both the CA2 and CA3 groups (8).This complexity in CA2 and CA3 responses is likely due to pro-

moter strength differences as well as the types and numbers of CAregulatory pathways used by each cyanobacterial species.

Materials and MethodsGrowth Conditions. SF33 (54) of Fremyella diplosiphon UTEX 481 (also calledTolypothrix sp. PCC 7601) was wild type. Cultures were grown as described(15) with or without 10 μg/mL kanamycin in 15 μmol photons m−2 s−1 usingred and green fluorescent lights (Industrial F40T12-Red and -Green, LightBulbs Unlimited).

RNA Analysis. For RNA half-life measurements, rcaE cells were grownto midlogarithmic phase in red or green light. Rifampicin was dissolvedin 100% methanol and added to 150 μg/mL final. Cultures were kept inthe same light after rifampicin addition. Samples (50 mL) were takenimmediately after rifampicin addition (time zero) and after 1, 2, 5, 10, 20,40, and 60 min. RNA was isolated and analyzed as described (15) exceptthat 0.9 mL of Tri-Reagent was used during isolation and RNA was pre-cipitated by adding a 0.5× vol isopropanol and loading onto a QiagenRNeasy Mini kit column, then eluted per the manufacturer’s instructions.The same method, minus rifampicin, was used to isolate RNA from wild-type, rcaE, and rcaC cultures grown in red and green light. RNA wasseparated either by electrophoresing 15 μg of each sample for 2 h at 100 Von 1% agarose-formaldehyde gels (55) then transferring overnight toImmobilon Nytran-NY+ membrane (Fisher Scientific) using 10× SSC or byresuspending 5 μg of each sample in 10 μl of Sample Buffer (20 mM Mops,pH 7.0/1 mM di-sodium EDTA/5 mM sodium acetate/50% formamide/0.7% formaldehyde/40 μg/mL ethidium bromide) plus 2 μl of Sample Dye(50% glycerol/10 mM di-sodium EDTA/2.5 mg/mL bromophenol blue/2.5mg/mL xylene cyanol), heating to 65 °C for 5 min and electrophoresing for1.5 h at 100 V on a 6%, 8 M urea-Tris-borate-EDTA polyacrylamide gel.RNA was then transferred to Immobilon Nytran-NY+ at 15 V for 1 h (55).After UV cross-linking, membranes were probed as previously described.PCR amplification of F. diplosiphon genomic DNA using primers cpeC-Land cpeC-R (primers shown in Table S1) generated the cpeC coding se-quence probe and primers cpeC5′L and cpeC5′R were used to amplifya cpeC 5′ leader probe (15). A Molecular Dynamics SP PhosphoImager wasused to quantify probe hybridization and ribosomal values were used tonormalize mRNA values.

Plasmid Construction. All numbering is relative to the transcription start site ofthe gene involved. pRB7wasmade by PCR amplification using primers 400cpeCand cpeCUTR, cutting with SphI and BamHI, and insertion into similarly cutpRB1 (also called p400cpeCGUS) (14, 29, 56). pRB8 was created using primers400cpeC and CDEpcboxmut1 in one PCR amplification and cpeCUTR andCDEpcboxmut2 in another. The two resulting PCR products were annealedand PCR amplified using 400cpeC and cpeCUTR, and the product cut with SphIand BamHI and inserted into similarly cut pRB1. Two-step PCR amplificationwas also used to create pRB13 (primer pairs 400cpeC/cpeCstem1-1 and cpe-CUTR /cpeCstem1-2), pRB14 (primer pairs 400cpeC/cpeCstem2-1 and cpeCUTR/cpeCstem2-2), and pRB15 (primer pairs 400cpeC/cpeCstem3-1 and cpeCUTR/cpeCstem3-2). pRB12 was synthesized using primer pairs 400cpeC/cpeCstem4-1in a PCR amplification. The product was used as a template for a secondamplification, using primer pairs 400cpeC/cpeCstem4-2. The final PCR ampli-fication products for pRB12-pRB15 were cut and ligated into pRB1 as de-scribed for pRB7 and pRB8. Stem–loop 4 mutations were made by PCRamplification using pRB8 and primer 400cpeC paired with one of the fol-lowing four primers: Stem4onlymut, StrStem4, 4LoopMut, and Allpair4.Products were used in a second PCR amplification with primer pair 400cpeC/cpeCstem4-2. The resulting fragments were cut with SphI and BamHI andcloned into the same sites in pRB1. Sequences 3′ of stem–loop 4 were changedusing pRB8 and PCR amplifying with primer pair 400cpeC/Down4mut1.The first amplification product was reamplified with primer pair 400cpeC/Down4mut2. The product was cut with SphI and BamHI and cloned intothe same sites in pRB8. Plasmid transformations into F. diplosiphon were byconjugation (16, 54). All PCR-amplified DNA and ligation junctions weresequenced.

GUS Assays. GUS assays were modified from previous protocols (29, 31).Transformants were grown in BG-11 with kanamycin to an A750 of ∼0.7 ineither red or green light. A total of 350 μl was centrifuged for 4 min at5,000 × g at room temperature. Pellets were resuspended in 1 mL of GUSassay buffer (50 mM NaPO4, pH 7.0/1 mM EDTA) containing 12 μg/mLchloramphenicol and centrifuged as before. Pellets were resuspended in1 mL of GUS assay buffer, 20 μl of 0.1% SDS, and 40 μl of chloroform, then

18546 | www.pnas.org/cgi/doi/10.1073/pnas.1107427108 Bezy et al.

vortexed for 10 s. At least three technical replicates were conducted foreach transformant, and each construct was assayed using at least fourindependently transformed lines. For each replicate, 20 μl of cell lysate wasmixed with 180 μl of GUS assay buffer containing 1.25 mM α-p-nitrophenylβ-D-glucoronide (PNPG, Sigma) and incubated at room temperature. Ab-sorption was measured at 405 nm and recorded every 2 min for 30 min witha Molecular Devices SpectraMax 190 (Molecular Dynamics). Protein con-centrations determined using a Pierce BCA protein assay reagent kit per themanufacturer’s instructions. Activity was quantified as nmol of product permg of protein per min.

3′ RACE Analysis. Methods for 3′ RACE analysis are described in SI Materialsand Methods.

cpeC 5′ Leader ORF Disruption. Methods for cpeC 5′ leader ORF disruption aredescribed in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank the D.M.K. laboratory members forthoughtful discussions and comments on the manuscript. This workwas supported by National Science Foundation Grant MCB-1029414(to D.M.K.).

1. Kendrick RE, Kronenberg GHM (1994) Photomorphogenesis in Plants (Kluwer Aca-demic Publishers, Dordrecht), 2nd Ed.

2. Bryant D (1994) The Molecular Biology of Cyanobacteria (Kluwer Academic Publish-ers, The Netherlands).

3. Stern D, Witman G (2009) The Chlamydomonas Sourcebook (Academic Press, Am-sterdam, Boston), 2nd Ed.

4. Tandeau de Marsac N (2003) Phycobiliproteins and phycobilisomes: The early ob-servations. Photosynth Res 76:193–205.

5. Grossman AR (2003) A molecular understanding of complementary chromatic adap-tation. Photosynth Res 76:207–215.

6. Kehoe DM, Gutu A (2006) Responding to color: The regulation of complementarychromatic adaptation. Annu Rev Plant Biol 57:127–150.

7. Stomp M, et al. (2008) The timescale of phenotypic plasticity and its impact oncompetition in fluctuating environments. Am Nat 172:169–185.

8. Tandeau de Marsac N (1977) Occurrence and nature of chromatic adaptation in cy-anobacteria. J Bacteriol 130:82–91.

9. Kehoe DM, Grossman AR (1996) Similarity of a chromatic adaptation sensor to phy-tochrome and ethylene receptors. Science 273:1409–1412.

10. Rockwell NC, Lagarias JC (2010) A brief history of phytochromes. ChemPhysChem 11:1172–1180.

11. Terauchi K, Montgomery BL, Grossman AR, Lagarias JC, Kehoe DM (2004) RcaE isa complementary chromatic adaptation photoreceptor required for green and redlight responsiveness. Mol Microbiol 51:567–577.

12. Chiang GG, Schaefer MR, Grossman AR (1992) Complementation of a red-light-in-different cyanobacterial mutant. Proc Natl Acad Sci USA 89:9415–9419.

13. Kehoe DM, Grossman AR (1997) New classes of mutants in complementary chromaticadaptation provide evidence for a novel four-step phosphorelay system. J Bacteriol179:3914–3921.

14. Li L, Alvey RM, Bezy RP, Kehoe DM (2008) Inverse transcriptional activities duringcomplementary chromatic adaptation are controlled by the response regulator RcaCbinding to red and green light-responsive promoters. Mol Microbiol 68:286–297.

15. Seib LO, Kehoe DM (2002) A turquoise mutant genetically separates expression ofgenes encoding phycoerythrin and its associated linker peptides. J Bacteriol 184:962–970.

16. Cobley JG, et al. (2002) CpeR is an activator required for expression of the phycoerythrinoperon (cpeBA) in the cyanobacterium Fremyella diplosiphon and is encoded in thephycoerythrin linker-polypeptide operon (cpeCDESTR). Mol Microbiol 44:1517–1531.

17. Kehoe DM (2010) Chromatic adaptation and the evolution of light color sensing incyanobacteria. Proc Natl Acad Sci USA 107:9029–9030.

18. Hirose Y, Narikawa R, Katayama M, Ikeuchi M (2010) Cyanobacteriochrome CcaSregulates phycoerythrin accumulation in Nostoc punctiforme, a group II chromaticadapter. Proc Natl Acad Sci USA 107:8854–8859.

19. Deng XW, GruissemW (1987) Control of plastid gene expression during development:The limited role of transcriptional regulation. Cell 49:379–387.

20. Leon P, Arroyo A, Mackenzie S (1998) Nuclear control of plastid and mitochondrialdevelopment in higher plants. Annu Rev Plant Physiol Plant Mol Biol 49:453–480.

21. Choquet Y, Wollman FA (2002) Translational regulations as specific traits of chloro-plast gene expression. FEBS Lett 529:39–42.

22. Marín-Navarro J, Manuell AL, Wu J, P Mayfield S (2007) Chloroplast translation reg-ulation. Photosynth Res 94:359–374.

23. Rochaix J-D (1996) Post-transcriptional regulation of chloroplast gene expression inChlamydomonas reinhardtii. Plant Mol Biol 32:327–341.

24. Fedoroff NV (2002) RNA-binding proteins in plants: The tip of an iceberg? Curr OpinPlant Biol 5:452–459.

25. Kulkarni RD, Schaefer MR, Golden SS (1992) Transcriptional and posttranscriptionalcomponents of psbA response to high light intensity in Synechococcus sp. strain PCC7942. J Bacteriol 174:3775–3781.

26. Kulkarni RD, Golden SS (1997) mRNA stability is regulated by a coding-region elementand the unique 5′ untranslated leader sequences of the three Synechococcus psbAtranscripts. Mol Microbiol 24:1131–1142.

27. Salem K, van Waasbergen LG (2004) Light control of hliA transcription and transcriptstability in the cyanobacterium Synechococcus elongatus strain PCC 7942. J Bacteriol186:1729–1736.

28. Horie Y, et al. (2007) Dark-induced mRNA instability involves RNase E/G-typeendoribonuclease cleavage at the AU-box and SD sequences in cyanobacteria. MolGenet Genomics 278:331–346.

29. Casey ES, Grossman A (1994) In vivo and in vitro characterization of the light-regu-lated cpcB2A2 promoter of Fremyella diplosiphon. J Bacteriol 176:6362–6374.

30. Alvey RM, Bezy RP, Frankenberg-Dinkel N, Kehoe DM (2007) A light regulated OmpR-class promoter element co-ordinates light-harvesting protein and chromophore bio-synthetic enzyme gene expression. Mol Microbiol 64:319–332.

31. Jefferson RA, Burgess SM, Hirsh D (1986) β-Glucuronidase from Escherichia coli asa gene-fusion marker. Proc Natl Acad Sci USA 83:8447–8451.

32. Federspiel NA, Grossman AR (1990) Characterization of the light-regulated operonencoding the phycoerythrin-associated linker proteins from the cyanobacteriumFremyella diplosiphon. J Bacteriol 172:4072–4081.

33. Alvey RM, Karty JA, Roos E, Reilly JP, Kehoe DM (2003) Lesions in phycoerythrin chro-mophore biosynthesis in Fremyella diplosiphon reveal coordinated light regulation ofapoprotein and pigment biosynthetic enzyme gene expression. Plant Cell 15:2448–2463.

34. Li L, Kehoe DM (2005) In vivo analysis of the roles of conserved aspartate and histidineresidues within a complex response regulator. Mol Microbiol 55:1538–1552.

35. Zuker M (2003) Mfold web server for nucleic acid folding and hybridization pre-diction. Nucleic Acids Res 31:3406–3415.

36. Li R, Golden SS (1993) Enhancer activity of light-responsive regulatory elements in theuntranslated leader regions of cyanobacterial psbA genes. Proc Natl Acad Sci USA 90:11678–11682.

37. Agrawal GK, Kato H, Asayama M, Shirai M (2001) An AU-box motif upstream of theSD sequence of light-dependent psbA transcripts confers mRNA instability in darknessin cyanobacteria. Nucleic Acids Res 29:1835–1843.

38. Mulo P, Sicora C, Aro EM (2009) Cyanobacterial psbA gene family: optimization ofoxygenic photosynthesis. Cell Mol Life Sci 66:3697–3710.

39. Asayama M (2006) Regulatory system for light-responsive gene expression in pho-tosynthesizing bacteria: cis-elements and trans-acting factors in transcription andpost-transcription. Biosci Biotechnol Biochem 70:565–573.

40. Tyystjärvi T, Herranen M, Aro EM (2001) Regulation of translation elongation in cy-anobacteria: membrane targeting of the ribosome nascent-chain complexes controlsthe synthesis of D1 protein. Mol Microbiol 40:476–484.

41. Tyystjärvi T, Sirpiö S, Aro EM (2004) Post-transcriptional regulation of the psbA genefamily in the cyanobacterium Synechococcus sp. PCC 7942. FEBS Lett 576:211–215.

42. Dühring U, Axmann IM, Hess WR, Wilde A (2006) An internal antisense RNA regulatesexpression of the photosynthesis gene isiA. Proc Natl Acad Sci USA 103:7054–7058.

43. Kojima K, Nakamoto H (2005) Post-transcriptional control of the cyanobacterial hspAheat-shock induction. Biochem Biophys Res Commun 331:583–588.

44. Mayfield SP, Cohen A, Danon A, Yohn CB (1994) Translation of the psbA mRNA ofChlamydomonas reinhardtii requires a structured RNA element contained within the5′ untranslated region. J Cell Biol 127:1537–1545.

45. Klinkert B, Elles I, Nickelsen J (2006) Translation of chloroplast psbD mRNA in Chla-mydomonas is controlled by a secondary RNA structure blocking the AUG start codon.Nucleic Acids Res 34:386–394.

46. Zou Z, Eibl C, Koop HU (2003) The stem-loop region of the tobacco psbA 5’ UTR is animportant determinant of mRNA stability and translation efficiency. Mol Gen Genom269:340–349.

47. Zerges W, Auchincloss AH, Rochaix JD (2003) Multiple translational control sequencesin the 5′ leader of the chloroplast psbC mRNA interact with nuclear gene products inChlamydomonas reinhardtii. Genetics 163:895–904.

48. Fargo DC, Boynton JE, Gillham NW (1999) Mutations altering the predicted secondarystructure of a chloroplast 5′ untranslated region affect its physical and biochemicalproperties as well as its ability to promote translation of reporter mRNAs both in theChlamydomonas reinhardtii chloroplast and in Escherichia coli. Mol Cell Biol 19:6980–6990.

49. Nickelsen J, Fleischmann M, Boudreau E, Rahire M, Rochaix JD (1999) Identification ofcis-acting RNA leader elements required for chloroplast psbD gene expression inChlamydomonas. Plant Cell 11:957–970.

50. Hirose T, Sugiura M (1996) Cis-acting elements and trans-acting factors for accuratetranslation of chloroplast psbAmRNAs: Development of an in vitro translation systemfrom tobacco chloroplasts. EMBO J 15:1687–1695.

51. Winkler WC, Breaker RR (2005) Regulation of bacterial gene expression by ri-boswitches. Annu Rev Microbiol 59:487–517.

52. Smith AM, Fuchs RT, Grundy FJ, Henkin TM (2010) Riboswitch RNAs: Regulation ofgene expression by direct monitoring of a physiological signal. RNA Biol 7:104–110.

53. Bezy RP, Kehoe DM (2010) Functional characterization of a cyanobacterial OmpR/PhoB class transcription factor binding site controlling light color responses. J Bac-teriol 192:5923–5933.

54. Cobley JG, et al. (1993) Construction of shuttle plasmids which can be efficientlymobilized from Escherichia coli into the chromatically adapting cyanobacterium,Fremyella diplosiphon. Plasmid 30:90–105.

55. Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A Laboratory Manual(Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY), 2nd Ed.

56. Jefferson RA, Kavanagh TA, Bevan MW (1987) GUS fusions: β-glucuronidase asa sensitive and versatile gene fusion marker in higher plants. EMBO J 6:3901–3907.

Bezy et al. PNAS | November 8, 2011 | vol. 108 | no. 45 | 18547

PLANTBIOLO

GY