After analyzing the nucleomorph genome of the cryptophyte Guillardia theta, we identified an open reading frame (orf222) with a high degree of similarity to cyanobacterial genes (Table 1), that encode soluble proteins possessing a DUF1001 domain. Alignments of the cryptophyte sequence with these cyanobacterial sequences indicated that orf222 should encode an additional transit peptide as shown by a N-terminal extension. Further orfs with homology to orf222 and the cyanobacterial homologs are additionally present in the nuclear genome of red alga 24. In higher plants, i.e. Arabidopsis thaliana, orfs with some similarity are also present 25. Even the bacteriophage S-PM2, which infects Synechococcus strains, encodes a homolog of orf222 26 (Table 1).

The number of Orf222 homologues in cyanobacteria varies in several species and does not correlate with the number of phycobiliproteins. Nevertheless, there is a strong tendency to express more than one species of Orf222 homolog in organisms containing multiple types of phycobiliproteins in the rods (Table 1). Based on amino acid sequence alignments and phylogenetic networks, four monophyletic groups can therefore been assigned (Fig. 1). Two of them resemble CpeT-like proteins (phycoerythrin operon protein); the other two groups harbor members of Slr1649-like type. Neither the Guillardia theta sequence nor any other eukaryotic sequences can be assigned to any one of the four monophyletic groups.

Additionally, clear affiliations of the Gloeobacter violaceus PCC 7421 (glr1182) and Synechococcus sp. PCC 7002 (CpcT) sequences can not be extrapolated. With the exeption of the Prochlorococcus species, at least one member of the Slr1649-like group is present in all cyanobacteria investigated to date. CpeT-like proteins were only detected in cyanobacteria encoding phycoerythrin and/or phycoerythrocyanin. Although the proteins of both groups seem to have the same function, further investigations on the corresponding genes relevant in the genomic context revealed a noticeable difference. Unlike the genes of the cpeT-group, the slr1649-group is by far less conserved in its genomic localization (Table 1). Except for Nostoc sp. PCC 7120, Anabaena variabilis ATCC 29413 and Trichodesmium erythraeum IMS 101, the localization of the homolog gene is always downstream of cpeS. In few cases it is followed by cpeR. On the other hand, genes for the slr1649-group are rather randomly distributed in the investigated cyanobacterial genomes (Table 1).

We used Synechocystis sp. PCC 6803 as a model organism and created first a slr1649 knockout strain (Δslr1649) by inserting a kanamycin resistance cassette into the slr1649 open reading frame via homologous recombination. The generated homozygous knock-out mutant showes identical features described in Shen et al. 10. Just like the characterized cpcT mutant in Synechococcus sp. PCC 7002, our knock-out mutant contains a decreased level of phycocyanin up to 60% and a resulting pale green phenotype. The knock-out cells produce smaller phycobilisomes, which could be the cause of their different migration behaviour in sucrose density gradients in comparison to wild type phycobilisomes. Furthermore, isolated phycobilisomes showed a red-shifted absorbance maxima and a slightly smaller apparent molecular mass in the β-subunit of phycocyanin on SDS-PAGE (data not shown). After the digestion of purified phycocyanin with formic acid and a phycocyanobilin addition assay, Shen et al. concluded after digestion that the cpcT gene from Synechocccus sp. PCC 7002 encodes a bilin lyase responsible for the attachment of phycocyanobilin to Cys-153 on the β-subunit of phycocyanin 10. The same is most likely true for the Synechocystis homolog Slr1649 due to the high homology and the similar phenotype between the two knock-out mutants. Thus, Slr1649 is thought to attach a bilin group to the homolog position Cys-155 of β-phycocyanin in Synechocystis sp.PCC 6803.

By isolating and analysing phycobilisomes from our knock-out mutant, we observed one additional feature besides the one already known from the characterization of the Synechococcus sp. PCC 7002 homolog. As shown in Fig. 2A two linker polypeptides, CpcC2 and CpcD, encoded by the genes cpcC2 and cpcD respectively, are missing in the isolated phycobilisomes of the mutant but could be identified in the wild type during mass spectrometry analysis. The other linker polypeptides CpcC1 (cpcC1), CpcG1 (cpcG1), ApcC (apcC) and ApcE (apcE) are present in both strains.

Upon further investigations, the absence of the linker proteins was proven not to be a transcriptional effect. In exemplarily reverse transcription experiments for cpcC2, the presence of identical cpcC2 transcripts was confirmed in both the mutant and the wild type strain by sequencing the obtained RT-PCR products (data not shown).

In order to investigate if the nucleomorph-specific reading frame orf222 from the cryptophyte Guillardia theta is able complement the effects of the slr1649 loss-of-function, we integrated this potential gene without its putative transit peptide into the cyanobacterial genome of Synechocystis sp. PCC 6803. This simultaneously affected the reading frame of slr1649 and its cis-acting upstream signals (Fig. 3A). In the complemented strain, slr1649 is separated from its natural upstream region, generating a promoter-less truncated gene, in which the translational initiator codon and the next two codons are no longer present in the reading frame. The loss of the slr1649 gene product and the complete segregation of the mutation were shown by immunoblot experiments using polyclonal antibodies generated against Slr1649 (Fig. 3B). Here, cross-reactions of the antibody were shown to be present in the wild type but not in the mutant strain extract. Additional analysis of the complemented strain by RT-PCR and sequence analysis showed that the integrated cryptophytic orf222 is transcribed (data not shown).

Interestingly the phenotype of the complemented strain is similar to that of the wild type strain, as indicated by the greenish color of the culture (data not shown). During sucrose density-gradient separation of isolated phycobilisomes from the complemented strain, we noticed no difference in the migration behaviour of the prominent band in respect to the wild type strain in contrast to the migration behaviour of the knock-out mutant (Fig. 4). This indicates that the size of the phycobilisomes is identical in both strains and could indeed be confirmed by a proteome and mass spectrometry analysis. Here we showed that the missing linker proteins of the slr1649 knock-out strain were present in the complemented strain (Fig. 2A). Additionally, there was no molecular mass shift in the β-subunit of phycocyanin of the complemented strain on SDS-PAGE visible. Further analyses revealed that the chromophore group, missing at positions Cys-β153 and Cys-β155 in the knock-out mutants of Synechococcus sp. PCC 7002 and Synechocystis sp. PCC 6803 respectively, most probably reappeared in the complemented strain, because Zn²⁺ stainings of phycobilisomes separated by SDS-PAGE showed an equal signal intensity of the phycobiliproteins of the wild type and complemented strain (Fig. 2B).

To clarify this we digested isolated phycobiliproteins with formic acid. In doing so, CpcB is cleaved at a single site while all other phycobiliproteins remain unaffected. The expected sizes for fluorescent fragments are 15.36 kDa with a chromophore group at position Cys-β84 and 2.78 kDa for the fragment with the chromophore group at position Cys-β155. As shown in Figure 5, these expected fragments were obtained. In addition to a signal at 15.36 kDa, a signal at 2.78 kDa was detected in the lane containing wild type strain protein and protein from the complemented strain but not in the one containing Δslr1649 protein. These data confirmed that the chromophore group, missing in the knock-out mutant, is present in the complementation. This together with the identification of the linker protein spectrum in the complemented strain indicated a wild type phycobilisome structure.

Synechocystis sp. PCC 6803 strains, wild type, Δslr1649 and the complemented strain, were grown at 30°C in Erlenmeyer flasks containing BG-11 media 34 with gentle swirling under standard light conditions (70 μE) and atmospheric CO₂ levels. For growth on plates, BG-11 medium was supplemented with 1% Agar. Plates were incubated under the same conditions as liquid cultures.

Two pairs of primers were used to amplify the flanking regions of the knock-out construct: 1649a_f (5'-GGT TAC TGC TCG AGG CGC ATC A-3') and 1649a_r (5'-GGA CGG CAA GGG ATC CTA TCT GG-3') generate fragment slr1649a, 1649b_f (5'-GGA CGG CAA GGG ATC CTA TCT GG-3') and 1649b_r (5'-CAG AAA TTG CCG CGG CCA ATC TC-3') fragment 1649b. Both were ligated into the pGEM-T vector (Promega, Mannheim) and after verification of the sequence, transferred into the pBluescript II SK (Stratagene, Amsterdam) vector. Escherichia coli strain MRF' XL-1 blue was used as plasmid host for cloning steps.

Using the BamHI restriction site (inserted by the primer 1649a-r and 1649b_f), a kanamycin resistance gene was cloned between the two fragments resulting in plasmid pΔ1649. After transforming into wild type Synechocystis sp. PCC 6803 cells with this plasmid, transformants were selected on BG-11 agar plates supplemented with kanamycin (5 μg/ml starting concentration). Kanamycin resistant clones were transferred to BG-11 liquid media. A homozygous culture was achieved by increasing kanamycin concentrations (50 μg/ml final concentration). Complete knock-out was confirmed via Southern blot analysis.

Nucleotide sequence of orf222 was amplified from Guillardia theta DNA without its putative transit peptide by using the primers 222komp2_f (5'-CAT ATG AAT TAA AAC CAA TCC TTA ATT G -3') and 222komp_r (5'-GTT AAA ATT AAA TGA ATT CTA ATA A-3'). Two pairs of primers were used to amplify the flanking regions: 1649a_f (5'-GGT TAC TGC TCG AGG CGC ATC A-3') and 1649kompa1_r (5'-CAA TAA CTA CAT ATG TCC CAT TCC-3') generated the fragment Compa, which includes the upstream region for slr1649, 1649kompa2_f (5'-TTT ATG TCG AAT TCC ACT GAT C-3') and 1649b_r (5'- GAG ATT GGC CGC GGC AAT TTC TG-3') generated fragment Compb. All three fragments were ligated into the pGEM-T vector (Promega, Mannheim) and after verification of the sequence, transferred into the pBluescript II SK (Stratagene, Amsterdam) vector using different restrictions sides inserted by the primers leading to a precursor construct. By using EcoRI restriction sites, a spectinomycin cassette was cloned between the two fragments Compa/orf222 and Compb resulting in plasmid pComp222 Δslr1649. After transforming the Synechocystis sp. PCC 6803 wild type strain with this plasmid, transformants were selected on BG-11 agar plates supplemented with spectinomycin (5 μg/ml starting concentration). Spectinomycin resistant clones were transferred to BG-11 liquid media. A homozygous culture was achieved by increasing spectinomycin concentrations (30 μg/ml final concentration).

Synechocystis sp. PCC 6803 cells were collected by centrifugation of 5 ml cell culture at 3200 × g. For DNA isolation, the pellet was resuspended in 400 μl TE buffer pH 7.0. After addition of breaking buffer (10% sodium dodecyl sulfat (w/v), 5% sodium lauryl sulfat (w/v)), 200 μl glass beads (0.2 mm diameter) and 400 μl phenol, cells were lysed by vortexing the suspension three times for 10 s. The suspension was then centrifuged at 12 000 × g and the resulting upper phase transferred to a new cup. This sample was treated twice with phenol-chloroform-isoamylalcohol (25:24:1) and centrifuged as before. By adding 1/10 Vol. NaAc pH 4.8 and two Vol. 96% ethanol, the DNA was precipitated for 1 h at -20°C. Afterwards, an additional washing step with 70% ethanol was performed. The pellet was dried and resuspended in H₂O.

RNA was isolated from Synechocystis cells with Trizol^© (Invitrogen, Karlsruhe) according to the manufacturer protocol. Northern Blot and Southern Blot analysis were performed according to standard protocols (Sambrook). Probes were constructed using the PCR DIG Probe Synthesis Kit (Roche, Mannheim).

To generate an antibody against Slr1649 we used the primers ex1649_f (5'-GGA TCC TTA TGT CCC ATT CCA CTG-3') and ex1649_r (5'-CTC GAG GCT GGC TAA AAA CTA ACT-3') to amplify the slr1649 gene, which was finally cloned in the pGEX-5X-3 vector (GE Healthcare Biosciences). After overexpression and purification of the Slr1649 GST fusion protein, immunization steps were executed by the Eurogentec company (Seraing).

The IgG fraction was purified from serum by protein A sepharose beads (GE Healthcare Biosciences).

Phycobilisome isolation was performed according to Gray et al. 35. Cells were collected by centrifugation at 5000 rpm for 10 min at room temperature. After an additional washing step with BG-11 media, cells were resuspended in 0.75 M potassium-phosphate buffer pH 7.0 (PPB), containing a protease inhibitors cocktail (PIC, 2 mg/ml Antipain, 5 mg/ml Chymostatin, 2 mg/ml Aprotinin, 5 mg/ml Trypsin-Inhibitor, 2 mg/ml Pepstatin, 5 mg/ml Leupeptin, 1 mg/ml Elastatinal and 2 mg/ml Na₂EDTA in HEPES/KOH. Final concentration 200 μg/ml Inhibitor) and afterwards broken by two passes through a French press (Aminco) at 124 MPa. The lysates were incubated with Triton X-100 (2%) for 15 min at room temperature and subsequently centrifuged at 20 000 rpm for 1 h to pellet unbroken cells and membrane debris. The supernatant was immediately loaded on a 10%–40% linear sucrose gradient, solved in PPB and centrifuged at 18 000 rpm for 16 h at 15°C.

Standard SDS-PAGE was performed with an Hoefer SE 250 apparatus (83 mm × 101 mm, 0,75 mm thick) or a custom made system (250 mm × 150 mm and 1,0 mm thick) using the Laemmli buffer system 36. The polyacrylamide content in the separating gel was a gradient of 10% to 15%. The stacking gel contained 6% polyacrylamide. To achieve a better resolution of polypeptides with masses less than 15 kDa, the SDS-Tricine gel system was used 37. Staining of gels was generally carried out with Coomassie brilliant blue G-250 dissolved in solution A (2% phosphoric acid v/v, 10% (NH₄)₂SO₄ w/v) and methanol (40:9:1). To visualize the bilin carrying proteins gels were incubated in a 0.2 M ZnSO₄ solution 3839 and highlighted with UV in a transilluminator (Bio-Rad).

Phycobilisomes were precipitated with Methanol/Chloroform 40 and resuspended in cleavage buffer. Cleavage was done according to Piszkiewicz et al. 41. 30 μg of phycobilisomes were incubated for 16 h at 37°C with 70% formic acid in before adding SDS sample buffer and analysis by Tricine SDS-PAGE on a 17% polyacrylamide gel.

Synechocystis sp. PCC 6803 cells were grown and harvested as described above. The cell pellet was resuspended in TEN100 buffer 42, containing PIC. Cell lysis was performed as described above.

The protein spots were subjected to in-gel trypsin digestion before mass spectrometry analysis as described previously 43. The peptide mixtures from the tryptic digests were desalted and concentrated using ZipTips™ columns made from the reverse chromatography resins Poros and Oligo R3 (Applied Biosystems). The bound peptides were washed with a solution of 0.5% formic acid and eluted from the column in 1 μl of 33% (v/v) acetonitrile/0.1% trifluoroacetic acid solution saturated with α-cyano-4-hydroxycinnamic (Bruker Daltonics) directly onto a MALDI target plate and air dried before analysis in the mass spectrometer. Mass spectrometry measurement was performed on an Ultraflex-TOF TOF tandem mass spectrometer (Bruker Daltonics). Peptide mass fingerprint spectra were acquired in the reflectron positive mode with a pulsed extraction using approximately 100 laser shots. The spectra were acquired after an external calibration using reference peptides (Peptide mixture II Bruker Daltonics). The acquired spectra were further internally calibrated using trypsin autolysis peaks as internal standards (842.5100, 2211.1046 Da). Monoisotopic masses were assigned and processed using Biotools ™ and FlexAnalysis ™ software (Bruker Daltonics) before submitting them to the Mascot program 44 for searches against the non-redundant NCBI database. The parameters used in the Mascot peptide mass fingerprint searches were as follows: Taxonomy, Synechocystis; search all molecular masses and all isoelectric points; allow up to one missed proteolytic cleavage site and a peptide mass tolerance of 50 ppm. Methionine oxidation was considered as an optional modification and cysteine carbamidomethylation as a fixed modification in all the searches. Matches to Synechocystis proteins were considered unambiguous when the probability score was significant using the Mascot score with a p value < 0.05 and when there was a minimum of five peptides and with a sequence coverage greater than 20%. For each protein the identity was further validated by tandem MS-MS analysis of selected peptides.

Blast search analyses were done by NCBI protein-protein blast 45 (see Additional file 1). This program was also used for conserved domain predictions. Transmembrane domains were predicted by using TMHMM server v. 2.0 46 and the SOSUI protein prediction server 47. Genome data from Synechocystis sp. PCC 6803 were obtained from CyanoBase 48. In silico cleavage was performed by PeptideMass 49.

The 41 sequences were aligned using MUSCLE 50 with 16 iterations. The output format was set to the standard ClustalW format 51. The alignment contained 307 sites including 191 gapped sites that were excluded from the analysis. From the remaining 116 sites logdet distances were estimated using LDDist 1.3.2 52. These distances were used to construct a Neighbor-net splits graph 53 that was visualized with SplitsTree4 54.