Genomic DNA Microarray Analysis: Identification of New Genes Regulated by Light Color in the Cyanobacterium Fremyella diplosiphon

Abstract

Many cyanobacteria use complementary chromatic adaptation to efficiently utilize energy from both green and red regions of the light spectrum during photosynthesis. Although previous studies have shown that acclimation to changing light wavelengths involves many physiological responses, research to date has focused primarily on the expression and regulation of genes that encode proteins of the major photosynthetic light-harvesting antennae, the phycobilisomes. We have used two-dimensional gel electrophoresis and genomic DNA microarrays to expand our understanding of the physiology of acclimation to light color in the cyanobacterium Fremyella diplosiphon. We found that the levels of nearly 80 proteins are altered in cells growing in green versus red light and have cloned and positively identified 17 genes not previously known to be regulated by light color in any species. Among these are homologs of genes present in many bacteria that encode well-studied proteins lacking clearly defined functions, such as tspO, which encodes a tryptophan-rich sensory protein, and homologs of genes encoding proteins of clearly defined function in many species, such as nblA and chlL, encoding phycobilisome degradation and chlorophyll biosynthesis proteins, respectively. Our results suggest novel roles for several of these gene products and highly specialized, unique uses for others.

For photosynthetic bacteria that rely on light as their primary source of energy, acute responsiveness to changing light conditions can provide a selective advantage in many ecological settings. While numerous species are capable of acclimation to changes in light intensity, the ability to sense and respond to variations in light color is less common among prokaryotes. One group that is capable of complex responses to changes in both light intensity and color is that of the cyanobacteria, and in these organisms the best-studied response to changes in light color is a process called complementary chromatic adaptation (CCA) (8, 42, 63). CCA was originally used to describe a photoreversible change in color phenotype that was dependent upon the color of light in which the cells were grown (27). Later work revealed that CCA resulted from shifts in the ratio of two phycobiliproteins, chromophorylated proteins involved in photosynthetic light harvesting. These were phycocyanin (PC; absorbance maximum, ∼620 nm) and phycoerythrin (PE; absorbance maximum, ∼560 nm) (6, 7).

PC and PE have open-chain tetrapyrroles (bilins) covalently attached and are located within photosynthetic light-harvesting structures called phycobilisomes (PBS) (28, 31, 58). The common hemidiscoidal forms of PBS are composed of two structurally discrete regions called cores and peripheral rods. Cores make up the inner portion of the PBS, keep the PBS associated with the photosynthetic thylakoid membrane, and transfer light energy from the peripheral rods of PBS to the photosynthetic reaction centers. Peripheral rods are cylindrical structures that are attached to and radiate from the core, capturing light energy and transferring it to the core. Peripheral rods may contain PC, PE, or both, depending upon the species and ambient light color. In addition to these phycobiliproteins, which always consist of an α and β subunit, both peripheral rods and cores contain linker peptides, proteins that are typically not chromophorylated and have important structural and functional roles in PBS (64).

Action spectra for CCA in the well-studied filamentous species Fremyella diplosiphon have demonstrated that this process is highly sensitive to red light (RL) and green light (GL): PC accumulates in RL and PE accumulates in GL (34, 67). Detailed molecular analyses of light color responses in this organism have focused on CCA and primarily on the expression of genes encoding PBS components (32, 42, 65). Allophycocyanin, the principal phycobiliprotein found in the core, is encoded by the apcAB operon and is similarly abundant in RL and GL (19, 37). F. diplosiphon synthesizes two detectable forms of PC under nutrient-replete conditions (10). The cpcB1A1 (cpc1) operon encodes constitutive PC and is expressed at similar levels in both RL and GL (18, 19, 50). The cpcB2A2 (cpc2) genes encode inducible PC (PCi). Their expression is not detectable in cells grown in GL but occurs at a high level during growth in RL (17). PCi and its associated linker peptides (encoded by cpcH2I2D2) are cotranscribed in the cpcB2A2H2I2D2 operon (17, 18, 47). The cpeBA genes, which encode PE, and the cpeCDE genes, which encode the PE linkers, are expressed at high levels during growth in GL but are made at low levels in RL (24, 25, 49). Unlike the genes encoding PCi and its linkers, the operons encoding PE and the PE linkers are unlinked (24, 25, 49). For at least cpc2 and cpeBA, the changes in transcript abundance observed during CCA primarily result from changes in transcription (52). Recently, two additional genes that are GL upregulated have been identified in F. diplosiphon within the pebAB operon, which encodes the enzymes involved in the conversion of biliverdin to phycoerythrobilin (the chromophore attached to apo-PE) (3).

Thus far, five proteins have been identified that affect expression levels of CCA-responsive genes in cells that are light acclimated. Three of these, RcaE, RcaF, and RcaC, appear to act within a single signal transduction pathway. RcaE is a phytochrome-class CCA photoreceptor with strong similarity to histidine kinases (43, 66), while RcaF and RcaC are response regulators that, along with RcaE, appear to be part of a complex phosphorelay (13, 44). Disruption of rcaE, rcaF, or rcaC dramatically affects CCA regulation of both PC and PE synthesis (13, 43, 44). A fourth component, CpeR, is required for the GL expression of pebAB and cpeBA but not for that of cpeCDE (3, 14, 57), while a fifth component, CotB, is required for proper expression of both cpeBA and cpeCDE in GL (5). The biochemical functions of CpeR and CotB have not been established.

It has been recognized for some time that the cellular responses to changes in light color in F. diplosiphon involve more than restructuring of PBS during CCA (65). As early as 1973, Bennett and Bogorad noted morphological differences between cells grown in RL and GL. In GL, filaments were longer and cells were larger and more cylindrical; after a shift from GL to RL, cells called necridia were transiently produced (7). These morphological changes and CCA responses were proposed to be controlled by different regulatory systems (8). Also, electron transport rather than a photoreceptor appears to regulate cellular differentiation, including the RL production of hormogonia (short, gas-vacuolated filaments) and synthesis of heterocysts (specialized cells for nitrogen fixation) during GL growth in the absence of fixed nitrogen, in this organism (12).

To gain a more complete perspective on the physiology and regulation of acclimation to changing light color in F. diplosiphon, we have used two-dimensional gel electrophoresis to determine that, at a minimum, nearly 80 proteins accumulate differentially in RL and GL. In addition, we created DNA microarrays to characterize genome-wide differences in RNA levels between cells acclimated to RL and GL. In prokaryotic systems, microarrays typically are created by PCR amplification of, or synthesis of oligonucleotides corresponding to, specific open reading frames (ORFs) identified in sequenced genomes. The F. diplosiphon genome is unsequenced. Thus, we created and analyzed genomic DNA microarrays, which contained uncharacterized but indexed fragments of F. diplosiphon genomic DNA. Using this approach to analyze approximately half of the F. diplosiphon genome, we have identified 17 novel genes that are differentially expressed in RL and GL.

MATERIALS AND METHODS

Strain and growth conditions used.

F. diplosiphon UTEX 481 (also called Calothrix sp. strain PCC 7601 and Tolypothrix sp. strain PCC 7601) shortened-filament mutant strain Fd33 derived from SF33 (15) was used as our wild-type strain. These cells have normal light color responses but form discrete colonies on plates. Cells were completely acclimated as previously described (57) by growth from low densities in 15 μmol m⁻² s⁻¹ of continuous RL or GL for 5 days at 30°C. For each experiment, RL- and GL-grown cell cultures were started from two colonies taken from the same plate of wild-type cells.

Two-dimensional polyacrylamide gel electrophoresis analysis of F. diplosiphon soluble proteins.

Wild-type cells were grown in 50 ml of BG-11 as described above. Cultures were diluted in the late afternoon to an absorbance at 750 nm (A₇₅₀) of 0.4 and then grown overnight to an A₇₅₀ of ∼0.6. Cells were pelleted by centrifugation at 5,200 × g for 10 min at room temperature (RT), resuspended twice in 40 ml of extraction buffer (0.65 M sodium phosphate [pH 7.5], 1 mM phenylmethylsulfonyl fluoride, 1 mM 2-sulfanyl ethanol, 0.2 mM benzamidine HCl, 50 μM of 6-aminohexanoic acid), and repelleted as described above after each resuspension. Cells were resuspended in extraction buffer to a 3-ml total volume and passed through a small French pressure cell at 18,000 lb/in² three times at RT. To obtain the soluble fraction and remove the majority of the PBS, the lysate was centrifuged at 84,000 × g for 1 h at RT. The clear upper layer (∼2 ml) was transferred to a new tube, and 300 μl of 50% (vol/vol) trichloroacetic acid (TCA) was added to precipitate the proteins. This solution was placed at 4°C for 2 h and then centrifuged at 16,000 × g for 10 min at 4°C. The pellet was resuspended in 1 ml of ice-cold 10% TCA and then recentrifuged at 16,000 × g for 10 min at 4°C. This wash was repeated once with 1 ml of 5% TCA and twice with 1 ml of acetone. Finally, the pelleted proteins were resuspended in 500 μl of an 8 M urea solution containing 2% (wt/vol) 3[(3-cholamidopropyl)-dimethyammonio]-1-propanesulfonate (CHAPS).

To separate proteins, a sample volume containing ∼100 μg of protein (10 to 40 μl) was added to 2.75 μl of 5 M dithiothreitol-1.25 μl of IPG buffer (Amersham Pharmacia Biotech AB, Piscataway, N.J.) (pH 3 to 10) and double-distilled water to achieve a total volume of 250 μl. First-dimension separation was carried out at RT by applying the sample to an IPG strip holder per the manufacturer's instructions. Immobiline DryStrip (Amersham Pharmacia Biotech AB) (linear pH 3 to 10) was placed on top of the sample. The IPG strip was covered with mineral oil and allowed to rehydrate for 10 h. Charge separation was applied (IPGphor Isoelectric Focusing unit; Pharmacia Biotech) at 500 V for 1 h, then at 1,000 V for 1 h, and finally at 8,000 V until a total of 59,000 V hours had been reached. The strip was immersed in 10 ml of equilibration buffer (50 mM Tris [pH 8.8], 6 M urea, 30% [vol/vol] glycerol, 2% [wt/vol] sodium dodecyl sulfate [SDS]) containing 65 mM dithiothreitol for 10 min and then for 10 min more in 10 ml of equilibration buffer containing 140 mM iodoacetamide. After a rinse in double-distilled water, the strip was applied to an SDS-12% polyacrylamide gel (1.5 mm in thickness) and electrophoresed at 35 mA for 4 h at 10°C. Proteins were visualized using ammoniacal silver (33). Proteins were selected as differentially abundant on the basis of visual assessment.

Genomic-DNA microarray library construction and clone amplification.

An F. diplosiphon genomic DNA library was prepared essentially as described previously (45) except that the vector used was pGEM-7Zf(+) (Promega, Madison, Wis.). More than 90% of the clones in the library contained a genomic DNA insert; the average length of these inserts was 4 kb, and the sizes ranged between approximately 1.5 and 9.0 kb (data not shown). The library was plated onto Luria-Bertani (LB) plates containing 100 μg of ampicillin (amp₁₀₀)/ml that were each overlaid with 40 μl of 20 mg of X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside)/ml and 4 μl of 200 mg of IPTG (isopropyl-β-d-thiogalactopyranoside)/ml and grown overnight at 37°C. Colonies were robotically picked using a custom-built colony-picking robot capable of >98% blue-white color discrimination, and each colony was placed into 200 ml of LB broth containing amp₁₀₀ in a separate well of a 96-well plate. A total of 22 separate plates were used. The estimate of the number of clones necessary for half-genome representation on the microarrays was made by a previously established method (56) with an estimate of a 10-Mbp genome on the basis of the approximately 9.2-Mbp genome (http://genome.jgi-psf.org/draft_microbes/nospu/nospu.info.html) of the closely related filamentous cyanobacterium N. punctiforme and an average genomic DNA insert size in the library of 4 kb. The 96-well plates were placed at 37°C overnight with vigorous shaking. Reinoculations were made by transferring 5 μl of each overnight culture into 200 μl of fresh LB broth-amp₁₀₀ in separate wells of 96-well plates. The newly inoculated cultures were grown at 37°C for 2 h with vigorous shaking and then used directly in PCR amplification as described below.

Each F. diplosiphon genomic DNA insert contained within the 2,071 clones from the library was PCR amplified using primers that hybridized to the T7 and SP6 priming sites within pGEM-7Zf(+). Expand Long Template PCR mix was used (Roche Applied Science, Indianapolis, Ind.). The final concentrations for each reaction mixture were 50 mM Tris-HCl (pH 9.2), 16 mM (NH₄)₂SO₄, 1.75 mM MgCl₂, 0.35 mM deoxynucleoside triphosphates, 0.3 pmol of each primer ml⁻¹, and 1 unit of Long Template Taq polymerase. For each reaction, 2 μl of a fresh cell culture was added just prior to initiation of PCR amplification. Amplification parameters were as follows: (i) 4 min at 92°C, (ii) 40 sec at 92°C, (iii) 40 sec at 58°C, (iv) 10 sec at 60°C, and (v) 6 min 30 sec at 68°C (vi) nine times to step ii; (vii) 40 sec at 92°C, (viii) 40 sec at 58°C, (ix) 10 sec at 60°C, and (x) 4 min 30 sec plus 20 sec per cycle at 68°C (xi) 29 times to step 7; (xii) end. All PCR products were checked for insert size and amplification efficiency on agarose gels. The reaction conditions used allowed sufficient amplification of the inserts contained within ∼90% of the clones for use on the microarrays. PCR amplification mixtures were cleaned using a BioRobot 3000 (QIAGEN, Valencia, Calif.) and 96-well plate format QIAGEN PCR clean-up preps (QIAGEN). The cleaned PCR products were transferred to 96-well V-bottom plates (Greiner Bio-One, Inc., Frickenhausen, Germany) and dried at 37°C for 3 to 4 h. The dried PCR products were resuspended in 10 μl of 50% dimethyl sulfoxide, and 5 μl of each mixture was transferred to a new plate, which was used for printing the microarrays. The minimum concentration of PCR product used in printing the microarrays was approximately 0.25 mg/ml. The steps described above are summarized in Fig. 1.

FIG. 1.

Outline of the experimental approach used to identify 17 novel genes in F. diplosiphon that are responsive to changes in light color. Steps involved in the construction and analysis of genomic DNA microarrays, subarrays, Northern blots, and slot blots are shown.

Printing of half-genome microarrays.

Half-genome microarrays were printed using an Omnigridder (GeneMachines, San Carlos, Calif.) (Fig. 1). Each DNA sample was printed twice per slide. Control spots included PCR products representing cpc1, cpc2, apcAB, cpeBA, cpeCDE, pGEM-7Zf(+) DNA, and ribosomal DNA (57) and a sample that contained no DNA. The control samples were printed at a concentration of 0.25 μg/μl. In addition, two dilution series of cpc1 from 0.03 to 1 μg/μl were printed positioned in opposite corners of each microarray. The microarrays were printed on GAP slides (Corning Inc., Corning, N.Y.). DNA was fixed to the slides by baking at 80°C for 3 h.

RNA isolation and analysis.

TriReagent (Molecular Research Center, Inc., Cincinnati, Ohio) was used following the manufacturer's protocol for the isolation of RNA from completely light-acclimated cells. RNA quality was assessed via gel electrophoresis, and concentration was determined via spectrophotometric analysis. RNA was fixed to a Nytran SuperCharge Nylon membrane (Schleicher & Schuell, Keene, N.H.) either by slot or Northern blotting. From 5 to 10 μg of RNA per slot was used for slot blots, and 10 to 20 μg of RNA per lane was used for Northern blots. RNA analyses were conducted as described previously, including the use of ribosomal probe controls for normalization of all values (57). The statistical significance (at α = 0.05) of the light-responsive expression of each ORF tested by Northern and slot blot analyses was determined using Student's one-tailed t test.

Hybridization of genomic microarrays.

Labeled cDNAs were generated via reverse transcription of 10 μg of total RNA in the presence of amino-allyl modified dUTP (Sigma, St. Louis, Mo.) following the protocol for amino-allyl dye coupling found at www.microarrays.org/protocols.html except that random hexamers were used for priming. Cy3 or Cy5 monofunctional reactive dyes (Amersham Biosciences) were then conjugated to the cDNAs via the amino-allyl adduct. Labeled cDNAs from both light conditions were combined and resuspended in 40 μl of hybridization solution (3× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 0.75 μg of tRNA/μl, 30 mM HEPES [pH 7.0], 0.2% SDS) and incubated at 100°C for 2 min. The slide was placed into a hybridization chamber (GeneMachines). The microarray was covered with a LifterSlip (Erie Scientific Company, Portsmouth, N.H.), and the hybridization solution was pipetted under the LifterSlip. Slides were hybridized for 18 h at 65°C. Slides were washed at RT for 5 min each in 2× SSC-0.1% SDS followed by 2× SSC and then briefly rinsed in water. Slides were dried via centrifugation for 5 min at 40 × g at RT. The steps described above are summarized in Fig. 1.

Microarray data analysis and identification of light-responsive clones.

Slides were scanned using a GenePix4000 (Axon Instruments, Inc., Foster City, Calif.) scanner, and data were collected using Axon GenePix Pro 3.0 (Fig. 1). Slides were typically scanned two to three times (with a maximum of five times), and photomultiplier tube voltages were adjusted to produce a ratio of total Cy5/Cy3 intensity of 1 ± 0.1; only the data from scans with this ratio were analyzed. Clone selection criteria are described in Results.

Sequencing of clones and identification of ORFs.

Sequences of all clones were obtained using an ABI 3700 apparatus (Perkin Elmer, Foster City, Calif.). Initial sequencing of clone inserts used T7 and SP6 priming sites in pGEM-7Zf(+). Additional, clone-specific primers were then generated near the 3′ ends of the sequences obtained from the initial sequencing runs. This process was continued until both strands of the insert DNA were sequenced (Fig. 1). The sequences of all primers and insert DNAs from the 32 clones that were sequenced in this study are available at http://camd.bio.indiana.edu.

ORFs were identified by ORFinder from the National Center for Biotechnology Information (NCBI) with the use of the bacterial codon usage set (Fig. 1). Only those ORFs greater than 300 bp were annotated except for clone 3092E12, which did not contain any ORFs greater than 250 bp, and the group E-G gene, which has been previously identified as nblA (48; GenBank accession no. X04592). The putative function of the translated protein from each ORF was explored using BLASTX 2.2.4 software from NCBI by comparisons to the nonredundant GenBank database (2). Proteins from ORFs that were similar to a previously identified protein and had an expectation value of less than 0.05 were assigned that putative protein function or were designated similar to a protein of unknown function. Conceptual proteins for which similarity searches produced no matches with an expectation value of less than 0.05 were designated hypothetical proteins of unknown function.

Generation of ORF-specific microarrays (subarrays).

After ORF identification, primers specific to the 5′ and 3′ regions of each ORF were generated to permit PCR amplification (Fig. 1). The sequences of these primers are available at http://camd.bio.indiana.edu/files/supplementary. PCR amplification conditions for individual ORFs were as follows: (i) 4 min at 94°C, (ii) 40 s at 94°C, (iii) 40 s at variable temperature, and (iv) 1 min at 68°C (v) 20 times to step ii; (vi) 40 s at 94°C, (vii) 40 s at variable temperature, and (viii) 1 min plus 5 s per cycle at 68°C (ix) 20 times to step vi; (x) end. The annealing temperatures in this program were adjusted slightly for each primer pair to optimize product yields. PCR amplification products were checked using gel electrophoresis, cleaned using QIAGEN PCR cleanup kits (QIAGEN), randomly distributed into wells of 96-well plates, dried down, and then resuspended in a 10-μl solution of 50% dimethyl sulfoxide prior to printing.

The same controls were used on the subarrays as on the half-genome microarrays except that dilution series of cpc1, apcAB, cpc2, and cpeBA DNA were spotted onto the subarrays. In addition, PCR products corresponding to the genomic insert of 25 non-light-responsive clones identified on the half-genome microarrays that were expressed at least twofold above the background level were included on the subarrays. In each experiment using subarrays, the ratio of the mean Cy3 and Cy5 intensities of all the spots corresponding to these 25 clones was adjusted to 1 prior to data collection. Four other PCR amplification products, pebAB, rcaC, cotB, and cpeSTR, were spotted onto the subarrays. The light responsiveness of the latter three has not been previously established. Subarrays were printed on Corning GAP II slides, processed, and hybridized in the same manner as the half-genome microarrays.

Nucleotide sequence accession numbers.

All clone and group DNA sequences obtained in this study have been submitted to GenBank at the NCBI under accession no. AY548433 through AY548463 and AY552548.

RESULTS

Protein profiles differ in RL- and GL-grown cells.

The cellular response of F. diplosiphon to growth in RL versus GL was initially tested by examining the patterns of accumulation of soluble proteins under these two conditions. Visual inspection of gels after two-dimensional gel electrophoresis revealed that nearly 80 soluble proteins were altered in abundance in cells grown under the two light conditions (Fig. 2). A total of 37 were more abundant in RL-grown cells than in GL-grown cells, while 42 proteins were more abundant in GL-grown cells. The proteins that differentially accumulated under the two light conditions spanned a wide range of isoelectric points and molecular weights.

FIG. 2.

Two-dimensional gel electrophoresis of proteins extracted from cells acclimated to GL or RL. (A) Protein harvested from F. diplosiphon cells fully acclimated to GL. (B) Protein harvested from F. diplosiphon cells fully acclimated to RL. In both cases the proteins were separated in two dimensions, by isoelectric point in the horizontal direction and by mass in the vertical direction, and then visualized by silver staining. Proteins were identified as differentially abundant by visual assessment of the gels. Proteins identified as more abundant in RL are marked with a red arrow, while proteins identified as more abundant in GL are marked by green arrows. Gels shown are representative of at least three replicate analyses using independently isolated protein samples.

Genomic DNA microarray analysis identifies 42 regions of the F. diplosiphon genome that appear to be light responsive.

Each 4,200-spot microarray contained two replicates of each PCR amplification product from 2,071 independent library clones, providing approximately 50% genome coverage (“half-genome microarrays”). Total RNA from cells completely acclimated to GL or RL was labeled using Cy3 for the GL sample and Cy5 for the RL sample. Experiments using half-genome microarrays were conducted with three sets of independently isolated RL and GL RNA samples, and duplicate microarrays were hybridized for each set of samples. Since every clone was spotted twice per microarray, 12 spots were analyzed per clone. Clones that were selected for further study had (i) expression levels at least twice the background level in all three biological replicates; (ii) at least eight usable data points; (iii) an average light induction of 1.5-fold or greater in at least six experiments; and (iv) a standard error of the mean of less than 25% of the average (fold) induction. A cutoff level of 1.5-fold induction was used, because the average genomic DNA insert length for the clones was 4 kb. Thus, each clone insert was expected to contain multiple genes, and genes that were expressed but not light regulated were expected to dampen the light responsiveness measured for that clone. Induction ratios were expressed as spot intensity in RL divided by spot intensity in GL for RL-induced clones and the inverse for GL-induced clones.

A total of 42 clones met the above-described criteria in the experiments using half-genome microarrays. Of these, the smallest clone insert was 1.5 kb, the largest was 7.8 kb, and the mean insert size was 4.3 kb. A total of 16 clones contained genes that were RL induced, and 26 contained genes that were GL induced. Expression values for each of the 42 clones and control spots are provided in Table 1. Overall, the values for the control spots corresponded well with previous quantitative Northern blot analysis measurements (3, 5, 9, 57). All primary intensity and ratio data and statistical analyses for every clone on the half-genome microarray can be accessed at http://camd.bio.indiana.edu.

TABLE 1.

Potential light-color-responsive clones

Group and clonea	Array result		RL RNA intensity	GL RNA intensity	Length (bp)	No. of ORFs	Slot blot result
	FIb	SE					FIb	SE
GL
3091F11A	1.74	0.06	1,719	2,786	4,134	6	3.33	0.12
3097E11A	1.73	0.06	368	626			3.00	0.1
3102C5A	1.74	0.06	1,413	2,116			1.89	0.24
3107A11A	1.64	0.05	1,557	2,433			2.29	0.08
3087C3B	1.63	0.1	476	804	5,146	6	1.20	0.43
3091E9B	1.70	0.06	637	1,234			1.04	0.29
3095H1B	1.52	0.09	549	981			1.67	0.20
3087B6C	1.57	0.06	5,275	7,421	4,166	4	1.13	0.04
3087B9C	1.86	0.06	2,411	5,799			1.13	0.06
3087E2E	3.34	0.03	2,638	10,702	6,485	7	7.69	0.02
3090A5E	7.59	0.02	3,379	21,956			8.130	0.04
3101C7F	1.60	0.11	414	937	3,861	5	2.99	NAc
3101F11F	1.59	0.09	587	1,009			1.48	0.30
3102F8F	1.53	0.1	485	868			2.02	0.23
3085B1	1.52	0.09	748	988	3,717	3	2.04	NA
3088E12	2.13	0.09	338	1,097	3,135	4	1.67	0.16
3091A7	1.59	0.05	1,494	2,757	3,210	2	1.01	0.23
3091B3	1.60	0.09	698	1,807	3,825	1	3.61	0.09
3092A8	1.80	0.07	1,018	1,980	4,096	2	1.35	0.36
3092A9	1.50	0.06	1,745	2,920	7,835	6	2.88	0.15
3093E3	1.90	0.03	2,089	4,336	3,501	3	2.16	0.27
3094E9	1.71	0.04	1,520	2,552	7,501	7	2.50	0.06
3098E8	1.52	0.06	2156	3,335	3,478	5	2.16	0.27
3104G9	1.60	0.04	1034	1,882	3,839	3	1.76	0.30
3106E11	1.64	0.07	1076	2,315	5,596	4	1.72	0.54
3109A9	1.72	0.05	742	1,282	5,834	6	1.53	0.14
cpeBA	3.41	0.05	19,004	64,208	1,100	2	—d
cpeCDE	14.76	0.01	1,681	32,971	3,500	3	—
RL
3104H2D	2.48	0.28	14,332	5,093	4,623	5	1.11	0.14
3104H3D	1.98	0.14	18,783	8,561			0.41	0.09
3084A10	1.98	0.22	10,139	5,660	1,504	2	NDe	NA
3092E12	5.98	1.42	11,029	1,289	1,745	5	2.74	0.44
3092A2	1.53	0.22	7,270	4,727	6,688	7	1.10	0.31
3092F10	1.95	0.24	4,996	3,235	4,143	4	1.29	0.36
3095F6	1.92	0.15	2,077	1,107	6,075	3	0.84	0.13
3098H5	1.73	0.39	1,159	518	4,714	4	1.22	0.66
3101B8	1.69	0.09	973	563	6,912	5	1.54	0.43
3101D4	1.57	0.11	3,379	1,941	2,714	2	0.72	0.19
3104F2	1.85	0.12	2,798	1,566	3,620	3	2.07	0.95
3105B10	2.22	0.13	4,780	2,112	3,222	3	2.00	0.49
3106A6	2.73	0.50	3,373	1,232	3,843	3	1.24	0.32
3106F10	1.57	0.14	2,618	1,657	6,465	2	1.50	0.58
3107H12	1.87	0.12	2,776	1,528	2,805	3	1.38	0.01
3109D10	1.89	0.22	1,094	545	4,544	5	1.07	0.12
cpc2	24.07	2.73	38,714	2,139	1,400	2	—
cpc1	1.17	0.18	12,058	11,661	3,500	4	—
apcAB	1.38	0.02	9,979	7,331	600	1	—

^aClones were selected as described in the text. Roman capitalized superscript characters indicate group affiliations. All members of a group are clustered within the table. Control genes cpeBA, cpeCDE, cpc1, cpc2, and apcAB were included for comparison.

^bFI (fold induction) values were determined by dividing RL intensity by GL intensity for RL-induced genes and GL intensity by RL intensity for GL-induced genes. For control genes apcAB and cpc1, FI values are given as RL intensity divided by GL intensity.

^cNA, not available.

^d—, control genes were not analyzed by slot blotting in this series of experiments. See reference 57 for details of their expression characteristics.

^eND, not done.

We further tested the expression of these 42 clones by analysis with Northern slot blots, with the PCR product of the genomic clone used as a probe (Table 1). The expression ratio for each clone was the mean obtained from three separate experiments, with each using RNA samples that were independently isolated and different from the RNA samples used in the microarray experiments. The expression of only one clone, 3084A10, was measurable in the microarray experiments but not detectable via slot blot analysis. Overall, 24 of the 42 clones identified as light responsive in microarray experiments were also found to be light responsive by slot blot analysis. It is not clear why only just over half of the 42 clones were confirmed as light responsive in slot blot studies. However, the slot blot probes were produced by labeling of each clone with random primers. Thus, it is possible that the radioactively labeled fragments corresponding to light-regulated ORFs were not produced in sufficient quantities to be detected over the background level generated by fragments corresponding to ORFs on the same clone that were not light responsive.

Sequencing of the 42 regions of the F. diplosiphon genome that appear to be light responsive leads to identification of 132 ORFs and multiple redundant clones.

We tested for redundancy within the 42 clones by sequencing the ends of each with vector priming sites flanking the insertion site. Inserts within 16 of the 42 clones had sequences in common with insert DNA from at least one other clone and were therefore condensed into six groups named group A through group F. The number of clones per group ranged from two to four. Groups C, D, and E contained two clones, groups B and F contained three clones, and group A contained four clones. The assignment of clones to groups is provided in Table 1. The redundancy present in the light-responsive clone population we identified supported the validity of the use of genomic DNA microarrays. Overall, the half-genome microarray studies led to the isolation of 26 individual clones and six additional groups of clones for a total of 32 unique regions of the F. diplosiphon genome that appeared to contain one or more genes that were regulated by light color.

All 32 of these regions of the F. diplosiphon genome were entirely sequenced. These data are available at our website (http://camd.bio.indiana.edu) and GenBank at NCBI. Both strands of approximately 148 kb of genomic DNA were sequenced, and all ORFs 300 bp or larger were identified. This analysis detected 124 ORFs within 31 of the regions but did not identify any ORFs within clone 3092E12. The minimum ORF size was therefore adjusted to 100 bp for this clone, which resulted in the identification of five ORFs.

Three additional ORFs of less than 300 bp were identified by comparison of clone or group sequences to DNA sequences in GenBank. The group E insert contained DNA whose sequence had been submitted to GenBank previously (40, 48, 49, 61; GenBank accession no. X04592). The annotation for that sequence included identification of a 168-bp ORF which, due to its size, we did not identify in our initial analysis. We designated this ORF group E-G. Two more ORFs of less than 300 bp were identified after comparison of the genomic DNA within clone 3088E12 to the sequence from Anabaena sp. strain PCC 7120. In total, 132 ORFs were identified within the 32 regions. The mean number of ORFs per region was 4.2 and ranged from 8 ORFs in clone 3094E9 to 1 ORF in clone 3091B3. Each ORF was named with a clone identification number or group letter followed by a letter, starting with “A” for the first ORF and ending with the letter corresponding to the last ORF in that clone or group.

BLAST analysis of the conceptual translation product from each ORF allowed the ORFs to be placed into putative functional classes (supplemental data Table S1 [http://camd.bio.indiana.edu/files/supplementary]) (41). BLAST analysis results for each of the 132 ORFs are available at http://camd.bio.indiana.edu/files/supplementary.

A total of 17 ORFs are newly identified as light responsive by subarray and Northern blot analyses.

Light-regulated changes in RNA levels were initially measured for these ORFs through the creation of subarrays that contained 129 of the 132 ORFs identified above (ORFs of less than 300 nucleotides on clones in group E and clone 3088E12 were not printed). The products resulting from PCR amplification of each ORF ranged from 132 bp to 3,600 bp and averaged 723 bp. The subarrays were used to analyze four sets of independently isolated RNA samples from RL- and GL-grown cells. Each sample set was used for hybridization to two or more subarrays. In total, 11 hybridizations were conducted to generate the data from the subarrays. Because each PCR product was spotted four times on each microarray, a total of 44 spots were analyzed per ORF. The fluorescence intensity values for all ORFs and controls from each experiment are provided at http://camd.bio.indiana.edu/files/supplementary.

Overall, the induction levels measured using subarrays in the experiments were dampened relative to those measured using the half-genome microarrays (Table 2; supplemental data Table S2 [http://camd.bio.indiana.edu/files/supplementary]) and from Northern blot analyses (see below). The cpeSTR operon was upregulated in GL, as was pebAB, as previously reported (data not shown) (3). There were no detectable differences in cotB or rcaC expression levels in RL and GL (http://camd.bio.indiana.edu/files/supplementary). It is not clear why such weak responses were measured in the experiments using the subarrays. This dampening was not due to the DNA concentrations of the spotted ORF samples, since these were a minimum of 0.25 mg/ml, above the concentration needed to be in excess of that of the probe even for highly expressed genes (supplemental data Fig. S1 [http://camd.bio.indiana.edu/files/supplementary]).

TABLE 2.

Seventeen novel F. diplosiphon genes regulated by light color

Light response category and gene or groupa	Subarray result		Product size (bp)	Avg RL intensity	Avg GL intensity	Northern blotting result		tc	nd
	FIb	SE				FIb	SE
GL
3088E12-C	1.56	0.09	357	1,843	3,128	1.76	0.12	3.60	6
3091A7-A	1.24	0.05	1,641	9,036	11,525	1.63	0.07	5.58	3
3104G9-B	1.79	0.04	338	3,009	5,688	1.39	0.11	2.59	5
3109A9-C	1.32	0.03	806	4,630	5,995	2.24	0.07	7.59	6
Group A-C	1.36	0.05	309	16,584	22,219	2.00	0.08	5.83	5
Group A-D	1.33	0.04	267	9,789	13,046	1.92	0.09	9.55	4
Group C-E	1.71	0.04	849	11,324	17,906	2.18	0.11	5.10	5
Group E-A	2.66	0.04	553	3,884	12,440	6.54	0.04	23.87	6
Group E-B	2.05	0.06	271	1,495	2,896	5.81	0.05	16.79	6
Group E-G	#e					7.52	0.05	19.39	5
cpeBA	11.4	0.02	1,100	6,103	56,877	—f
cpeCDE	5.127	0.05	3,500	10,662	54,176	—
cpeSTR	2.77	0.05	3,000	8,591	16,735	—
RL
3091A7-B	1.33	0.10	605	1,816	1,534	2.09	0.41	2.67	4
3092A9-C	1.27	0.07	1,591	2,908	2,524	2.27	0.58	2.19	4
3098H5-C	1.30	0.12	497	1,626	1,426	1.74	0.36	2.07	6
3105B10-A	1.55	0.18	1,595	1,621	1,237	2.05	0.09	11.72	4
3106A6-A	1.27	0.05	2,020	6,715	5,971	1.91	0.33	2.77	4
3106F10-A	1.37	0.10	1,932	2,445	2,221	1.98	0.31	3.14	4
3106F10-B	1.33	0.23	2,794	1,793	1,688	1.54	0.23	2.34	5
cpc2	11.99	1.95	1,400	59,661	6,511	—
Nonresponsive
apcAB	0.89	0.02	600	4,932	6,132	—
cpc1	0.98	0.10	3,500	4,949	5,280	—

^aControl genes cpeBA, cpeCDE, cpc1, cpc2, and apcAB were included for comparison.

^bFI (fold induction) values were determined by dividing RL intensity by GL intensity for RL-induced genes and GL intensity by RL intensity for GL-induced genes. For control genes apcAB and cpc1, FI values are given as RL intensity divided by GL intensity.

^ct, absolute value from a Student's one-tailed t test.

^dn, number of biological replicates for the Northern blot analysis.

^e#, ORF group E-G is a homologue of Nb1A and was not printed on the subarrays.

^f—, control genes were not analyzed by Northern blotting in this series of experiments. See reference 57 for details of their expression characteristics.

Our subarray results did not allow us to accurately determine which of the 132 ORFs were light responsive. Therefore, we conducted Northern blot and slot blot analyses to further characterize patterns of light-responsive expression of ORFs that were at least 1.25-fold induced in three or more independent experiments using subarrays. cpeBA and cpeYZ were excluded from further analysis. A total of 30 ORFs met this criterion. Of these ORFs, the experiments using subarrays indicated that 20 were RL induced and 10 were GL induced.

RNA accumulation was measured for each of the 30 ORFs in at least three sets of independently isolated RNA samples from RL- and GL-grown cells. These RNA samples were different from those used in the experiments that utilized half-genome microarrays and subarrays. A summary of the results from Northern blot analyses is provided in Table 2. For most ORFs, greater induction levels were measured by Northern analysis than by subarray analysis (Table 2; http://camd.bio.indiana.edu). ORFs defined as light responsive were at least 1.4-fold induced and had a P value of 0.05 or lower in a Student t test of the null hypothesis that there were no significant differences in RNA levels under the different light conditions. A total of 17 ORFs were identified as light responsive (10 were GL induced and 7 were RL induced). These included an nblA homologue that was not included in the subarray analyses. The expression patterns of two additional ORFs, 3092E12-D and 3107H12-A, showed consistent RL responsiveness, but the degree of induction was not statistically significant for either. Overall, 82% of the 17 ORFs were 1.4- to 2.5-fold induced whereas the remaining 18% were induced more than 2.5-fold. The induction levels were from 1.4- to 7.5-fold for the GL-induced ORFs and from 1.5- to 3.1-fold for RL-induced ORFs.

The 17 ORFs have not been previously identified as RL or GL responsive and encode a diverse group of proteins.

Assignments of putative function to the proteins encoded by the 17 ORFs found to be light responsive on the basis of the results of BLAST analyses are provided in Table 3. This table also contains the accession number of the GenBank entry that is most closely related to that ORF and the light induction level for each ORF. None of these ORFs had been previously identified as being light responsive in F. diplosiphon.

TABLE 3.

Putative functions of 17 new genes that are regulated by light color

Light response category and ORF	Putative function	FIa	Locus	E value	Accession no.
GL
3088E12-C	50S ribosomal subunit L22	1.76	rpl22	1 e−72	NP_488250
3091A7-A	Unknown	1.63	al15036	3 e−38	NP_489076
3104G9-B	Unknown	1.39	alr4922	7 e−67	NP_488962
3109A9-C	DPOR subunit L	2.24	chlL	1 e−146	NP_489118
Group A-C	Unknown	2.00
Group A-D	Unknown	1.92
Group C-E	ATPase β subunit	2.18	atpB	1 e−153	NP_489079
Group E-A	Tryptophan-rich sensory protein	6.54	tspO-L	3 e−96	CAD28152
Group E-B	Unknown	5.81
Group E-G	NblA1	7.52	nblA1	1 e−23	CAD28153
RL
3091A7-B	Biopoymer transport protein	2.09	exbB	1 e−75	NP_489087
3092A9-C	NADH dehydrogenase subunit 4	2.27	ndhD2	0.0	NP_489090
3098H5-C	Carbonic anhydrase	1.74	cynT	3 e−75	NP_665783
3105B10-A	d-Aminoacylase	2.05	ndaD	1 e−28	NP_125834
3106A6-A	Unknown	1.91	all0432	0.0	NP_484476
3106F10-A	Two-component sensor kinase	1.98	all0886	0.0	NP_484929
3106F10-B	Two-component sensor kinase	1.54	alr0710	0.0	NP_484753

^aFI (fold induction) values were determined by dividing RL intensity by GL intensity for RL-induced genes and GL intensity by RL intensity for GL-induced genes.

Of the 10 ORFs that were GL induced, 3 encoded proteins similar to hypothetical proteins in F. diplosiphon and 2 encoded proteins related to unknown or hypothetical proteins in both F. diplosiphon and the filamentous cyanobacterium Anabaena sp. strain PCC 7120. The protein sequences encoded by three ORFs were significantly related to previously identified proteins in Anabaena sp. strain PCC 7120. ORF 3088E12-C encoded a protein with similarity to an L22 ribosomal protein, ORF 3109A9-C encoded a protein homologous to the L subunit of a light-independent protochlorophyllide reductase (DPOR), and group C-E encoded a protein most similar to the β subunit of an ATP synthase. The final two ORFs were both within group E. Group E-G encoded a homologue of NblA from Synechococcus sp. strain PCC 7942, and group E-A coded for a protein with similarity to the tryptophan-rich sensory protein TspO from Rhodobacter capsulatus.

Five of the ORFs that were RL induced encoded proteins that had their closest counterparts in Anabaena sp. strain PCC 7120. ORFs 3106F10-A and 3106F10-B encoded serine/threonine kinase class proteins containing two-component sensor domains, ORF 3091A7-B encoded a protein similar to a biopolymer transport protein, ORF 3092A9-C coded for a homologue of subunit 4 of NADH dehydrogenase, and ORF 3106A6-A encoded a protein similar to an Anabaena hypothetical protein. ORF 3098H5-C encoded a protein with similarity to carbonic anhydrase encoded by cynT from Synechococcus sp. strain PCC 7942. ORF 3105B10-A encoded a protein most closely related to a putative d-aminoacylase from Pyrococcus abyssi.

The level of expression of each GL-induced ORF was classified as either high or low. The highly expressed ORFs (20 to 50% of the levels measured for cpeBA/cpeCDE), ORF group C-E (encoding an ATPase-β-related protein), ORF group E-A (encoding a TspO-related protein), and cpeSTR, encoded three unknown or hypothetical proteins. The remaining GL-induced ORFs were weakly expressed (5 to 10% of the levels measured for cpeBA/cpeCDE). The GL-induced ORFs also differed in their extent of GL induction. Three ORFs from groups E, A, B, and G were 5.8- to 7.5-fold induced, while the remaining seven were induced only 1.4- to 2.2-fold. The cpeSTR operon was also upregulated by GL only 2.8-fold. The RL-induced ORFs were all expressed in RL at 3 to 10% of the expression level of cpc2. The RL induction values for all of these ORFs were also low and ranged from 1.5- to 2.3-fold, much lower than the 11.9-fold measured for cpc2.

DISCUSSION

In this study we have used a unique functional genomics approach, the construction and analysis of genomic DNA microarrays, to uncover 17 new genes within the F. diplosiphon genome that have not been identified previously as light color responsive. Our data suggest that the majority of changes in RNA levels occurring during light color acclimation in F. diplosiphon are relatively subtle (Table 2). These results contrast markedly with the induction levels for cpeBA, cpeCDE, and cpc2, the CCA-regulated genes encoding PBS components. For these operons, a shift to the inducing light condition leads to a 10- to 20-fold or greater increase in RNA abundance (3, 5, 9, 52, 57), not surprising in light of the fact that they encode proteins that make up major portions of the PBS, which can constitute over 60% of the total soluble protein in these cells (7).

Although none of the 17 genes have been previously identified as light color responsive, homologs are responsive to environmental cues in other cyanobacterial species. In Synechocystis sp. strain PCC 6803, rpl22 (corresponding to ORF 3088E12-C) is downregulated in high-intensity UV-B light in microarray experiments (38). Similar experiments demonstrated that expression of atpB (ORF group C-E) is altered by light to dark transitions, exposure to UV-B or high-intensity white light (WL), iron deprivation, and changes in the redox status of intermediates in the photosynthetic electron transport chain (30, 36, 38, 60). exbB (ORF 3091A7-B), whose product apparently forms a complex with TonB and ExbD and is required for transport of ferric siderophores and vitamin B-12 across the outer membrane of Escherichia coli, is cold regulated in Synechocystis sp. strain PCC 6803 (62). The ndhD2 gene (ORF 3092A9-C), which encodes an NADH dehydrogenase that is involved in transferring electrons from NADPH to plastoquinone in Synechocystis sp. strain PCC 6803 (53), is upregulated during inorganic carbon limitation but not during acclimation to high-intensity light (51). Microarray experiments demonstrated that this gene is regulated by cold treatment, osmotic stress, UV-B and high-intensity WL treatment, and changes in the redox state of components in the photosynthetic electron transport chain and during acclimation to high intensities of light (1, 35, 36, 38, 39, 62).

The regulation of the cyanobacterial genes chlL (ORF 3109A9-C), chlN, and chlB, which encode DPOR, is not well understood, although they are apparently expressed in darkness and repressed under nitrogen-fixing conditions (26). Recent microarray studies have also shown that the expression of combinations of chlB, chlL, and chlN is altered by high-intensity WL and UV light, iron deficiency, and changes in the redox state of the photosynthetic electron transport chain in Synechocystis sp. strain PCC 6803 (35, 36, 38, 60). Our data suggest that as in many other cyanobacterial species, chlL and chlN are cotranscribed whereas chlB is located elsewhere in the genome (data not shown; http://camd.bio.indiana.edu) (26). Our finding that chlL and chlN homologs are GL upregulated in F. diplosiphon suggests that rapid growth in GL may require greater amounts of chlorophyll a than can be synthesized by the light-dependent form of protochlorophyllide reductase alone, which requires RL for its activity (26). This raises the interesting possibility that F. diplosiphon produces DPOR, which does not require light for its activity, under high-intensity-GL and low-intensity-RL conditions to produce sufficient chlorophyll a for rapid growth in this light environment.

The nblA gene (ORF group E-G), which encodes a small peptide that was first identified as being involved in PBS breakdown during periods of nutrient limitation in Synechococcus sp. strain PCC 7942 (16), has subsequently been found to have two homologs in Synechocystis sp. strain PCC 6803 whose expression levels are upregulated during nitrogen, sulfur, and iron starvation (4, 46, 55, 59) as well as by UV-B light but not by high-intensity WL (38). In addition, nblA transcripts in this species were reduced after dark-to-light transitions (30).

Recently, an nblA homologue has also been identified in F. diplosiphon 5′ of the cpeBA operon and 3′ of a tspO-like gene (48; GenBank accession no. X04592). This genomic DNA region is the same as the region we have identified as group E (Table 1; Fig. 3). The expression patterns of these genes, which were designated tspO-like, orf114, and nblA1 and which correspond to our group E-A, group E-B, and group E-G genes, were analyzed (48). In that study, the tspO-like and orf114 regions appeared to be cotranscribed whereas two differently sized transcripts were produced for nblA1. The expression of all of these genes was altered during growth in medium that contained nitrate or was nitrogen deficient versus medium in which ammonium was present. In addition, in contrast to our finding that the nblA1 gene is highly induced by GL (Table 2 and Table 3), Luque et al. found that the transcript levels of the nblA1 gene were not responsive to light color whereas the RNA levels of the tspO-like and orf114 genes were not examined for light responsiveness. The reason for the differences measured in the GL responsiveness of nblA1 is not presently clear. However, the high level of GL expression and induction of nblA1 measured here suggests a novel cellular role for this family member, perhaps an involvement in the synthesis or assembly of PE-containing PBS. This proposal is supported by the fact that the cells used in our experiments were completely acclimated to the ambient light conditions and, thus, that GL-grown cells completely lacked PCi. This eliminates the possibility that NblA1 was transiently expressed immediately following a shift to GL to degrade PCi-containing PBS.

FIG. 3.

Location of the ORFs identified within GL-responsive group E. The seven genes and their direction of transcription are identified by the arrows. Group E (this study) and gene (40, 48, 49) designations are provided above the arrows. The thick line represents DNA found within group E; nucleotide numbering is provided below the line. The clone numbers that make up group E are shown to the left, with the corresponding DNA indicated with thin lines and corresponding nucleotide numberings shown below each line. A scale bar is provided. Maps of all ORFs on all clones are available at http://camd.bio.indiana.edu/files/supplementary.

The expression of tspO in various bacterial species is regulated by a range of environmental cues. The first protein identified as TspO was in Rhodobacter capsulatas, where it was highly expressed in anaerobic cultures grown in light (68). A gene which encodes a related protein in Sinorhizobium meliloti and was first identified on the basis of its expression during nodule development (54) responds to nutrient limitation (23). Finally, as previously noted, the expression of the tspO-like gene from F. diplosiphon is also nutrient regulated (48). The fact that the TspO-like protein is highly upregulated in cells completely acclimated to GL suggests that, like NblA1, it may be directly or indirectly involved in the synthesis of PE-containing PBS.

The possibility that the NblA1 and TspO-like proteins have novel roles in F. diplosiphon is supported by the location of the genes encoding these proteins in the genome (Fig. 3). These genes are clustered with the cpeBA and cpeYZ operons (encoding putative attachment enzymes of the PE chromophore) (40, 48, 49; GenBank accession no. X04592), suggesting that they may play an important role in the production of PE or in the incorporation of PE into PBS in GL. NblA1 has been reported to interact with components of the PBS (48).

Since the 17 new genes that were light responsive were identified using half-genome microarrays (Tables 1 and 2), whole-genome microarrays might be expected to uncover approximately 34 such genes, which is less than half of the 79 soluble proteins whose abundance was regulated by RL and GL (Fig. 2). This difference may be due to our underestimating the size of the F. diplosiphon genome. It may also have resulted from the presence of clones on the half-genome microarrays in which the elevated expression of a gene(s) that was not light responsive masked the light responsiveness of an adjacent gene(s). This is supported by the fact that RL induction that was statistically significant was observed for two ORFs, 3091A7-B and 3092A9-C, even though they were isolated from GL-induced clones. For 3091A7-B, the other ORF on the clone was found to be induced by GL whereas none of the other ORFs contained within clone 3092A9-C were significantly regulated by light. We also cannot exclude the possibility that RNAs that do not encode proteins may be produced from sequences within some of the clones we have isolated. Previous work has identified the production of such RNAs in the gvpABC operon of this organism (20). It is also possible that some of the 132 ORFs that we initially identified were actually light responsive but were not among the 30 selected for further analysis. Other possibilities include posttranscriptional regulation and the presence of light-responsive ORFs that encoded proteins of less than 100 amino acids other than nblA1. However, we identified only six ORFs within all of our clones that encoded proteins between 50 and 100 amino acids that were significantly related to other proteins in the NCBI database. The majority of these were hypothetical proteins in other cyanobacterial genomes (data not shown).

This study addresses the differences in protein profiles and gene expression patterns in F. diplosiphon cells that are completely acclimated to RL or GL. We refer to this fully acclimated physiological state as the steady-state condition. Previous studies have demonstrated that the physiology of F. diplosiphon cells in the steady-state condition differs markedly from that of cells that are in the process of adjusting to new conditions of light color, which we refer to as the acclimation state. Acclimation-state responses that have been examined at the molecular level thus far include changes in profiles of protein accumulation (29), the formation of hormogonia, gas vesicles, pili, and necridia, and adjustments of photosynthetic electron flow between photosystems II and I (12, 21, 65). In addition, morphological differences between RL- and GL-grown cells are likely to require transient synthesis of enzymes involved in determining cell shape and size, although none have yet been described. F. diplosiphon genes known to respond predominantly or exclusively during the acclimation state; for example, the gpv genes, which encode proteins that are involved in the formation of gas vesicles (20, 21, 22), should not have been (and were not) identified in our study.

At least one response in the acclimation state, hormogonium formation, is controlled by the redox state of intermediates in the photosynthetic electron transport chain, while one response in steady-state conditions, CCA, is controlled by a photoreceptor(s) (8, 12, 43). This is a logical regulatory scheme, because large changes in the redox state of components of the electron transport chain driven by shifts in light color would be transient, mostly dissipating as the light acclimation process occurred (11, 12) and thereby attenuating signals generated by any redox sensing system. Conversely, steady-state responses require the continued signaling provided by photoreceptors long after CCA has largely rebalanced the redox state of the components of the photosynthetic electron transport chain to that existing prior to the light shift (11). In F. diplosiphon, genes may be responsive to light color either exclusively under acclimation-state or steady-state conditions or may be responsive under both conditions. It will be interesting to determine the mechanism(s) that regulate the expression of the 17 genes that have been identified as light responsive in this study.