Primary production in the oligotrophic regions of the oceans is dominated by the cyanobacterial components of the picophytoplankton organisms Synechococcus and Prochlorococcus1. The amounts of visible and ultraviolet solar radiation in oceanic ecosystems may be sufficient, particularly in surface waters, to cause photo-inhibition2. The primary cause of photo-inhibition in chloroplasts and cyanobacteria is damage to PSII, a large protein–pigment complex that catalyses the light-dependent oxidation of water to molecular oxygen.

At the core of PSII is a dimer of two related proteins, D1 and D2, which binds the pigments and co-factors necessary for the complex's primary photochemistry. During photosynthesis, D1 and, to a lesser extent, D2 turn over rapidly as a result of light-induced damage and are replaced by newly synthesized polypeptides in a repair cycle. When the rate of photo-inactivation and damage of D1 exceeds the capacity for repair, photo-inhibition occurs, resulting in a reduction in the maximum efficiency of PSII photochemistry3.

Viruses in general, and bacteriophages (viruses that infect bacteria) in particular, are abundant in marine ecosystems and are thought to exert major biogeochemical and ecological effects on the marine environment4. We analysed the genome sequence of S-PM2 (Fig. 1), a bacteriophage that infects marine Synechococcus strains5 and is about 194 kilobases long6, and found that it encodes the D1 and D2 proteins of PSII.

Figure 1: The bacteriophage S-PM2 (here artificially coloured blue), which infects marine cyanobacteria.
figure 1

REF. 6

Scale bar, 100 nm.

The genome contains a region of about 3.8 kilobases (GenBank accession no. AY329638) that extends from just over 100 base pairs upstream of the D1 gene (psbA), through two genes that are unrelated to photosynthesis, to a point some 100 base pairs downstream of the carboxy terminus of the D2 gene (psbD). The two intervening genes encode a homologue of the bacteriophage T4 gp49 (also known as recombination endonuclease VII) and a protein that has some similarity to a putative membrane protein present in the bacterium Escherichia coli.

The psbA gene appears to be interrupted, as the translated product of the gene aligns with other cyanobacterial and plant D1 proteins up to residue 276 (see supplementary information). There is then a region of 212 base pairs that encodes an amino-acid sequence without any similarity to known D1 proteins; this is followed by a region that encodes the remaining 25 amino acids of D1. This suggests that the psbA gene in S-PM2 contains a self-splicing intron. Introns occur in the psbA genes of the protozoan Chlamydamonas, and there is evidence for light/redox-regulated splicing of psbA precursor-messenger RNAs7. We detected copies of psbA genes after amplification by the polymerase chain reaction in five out of eight other Synechococcus viruses, although these genes seem to lack the putative intron.

The complete D1 protein of S-PM2 is similar to the D1 proteins of the marine Synechococcus sp. WH8102 (see supplementary information). There is homology in the DNA sequences, indicating that S-PM2 might have acquired the gene horizontally from its Synechococcus host. Presumably, psbD was acquired independently, given the presence of two unrelated intervening genes.

The expression of virus-encoded D1 and D2 proteins in infected cells would allow a repair cycle to operate in PSII after the host's protein synthesis had been shut down, thereby maintaining the cells' photosynthetic activity and the concomitant evolution of oxygen, and ensuring the provision of energy for the continued replication of the virus. This survival strategy resembles one used by a virus that infects the green alga Chlorella, which enhances the mechanism used by the host cell to rid itself of surplus light energy to avoid photo-inhibition8.