As I stated in the introductory post on this topic, it was thought for a long time that prokaryotes were incapable of generating circadian rhythms. When it was discovered, in 1994 , that one group of prokaryotes, the cyanobacteria, possess a circadian clock, the news was greeted with great excitement. This was the first definitive demonstration of a circadian clock in a bacterium (I intend to revisit the E.coli saga in a later post).
All three hypotheses for the origin of the circadian clock suppose that it first evolved in an aquatic, unicellular organism. While protists fit the bill quite nicely, having a bacterium with a circadian clock pushed the origin of the clock further back into the past. This made the researchers happy as it supported the notion that the clock was a universal property of life, as well as that it evolved only once in the history of Life on Earth. This also suggested that clocks in all organisms use same or similar intercellular mechanisms for generation of circadian rhythms.
At the time that clocks were discovered in cyanobacteria, only two circadian genes were characterized: period in fruitflies and frequency in Neurospora crassa. The second fly gene, timeless, was discovered the following year, and the first mammalian gene Clock and the first plant gene Toc were discovered some years later. Thus, at the time, it was still plausible that all of life used the same mechanism for the circadian clock, just as all of life uses ATP for energy storage and DNA for information storage.
However, studying genetics in bacteria is a much quicker and easier task than in the large multicellular eukaryotes. Very soon, the cyanobacterial clock genes were discovered and it turned out that they had no resemblance to fly or mold genes. KaiA, KaiB and KaiC (as they were discovered in Japan, they were named "kaiten", which implies a cycle of events reminiscent of the turning of the heavens) have no homologies with any of the clock genes found in any other group of organisms and the internal logic of the bacterial clock is different from that in plants, fungi and animals, i.e., it is not a typical transcription-translation feedback loop.
The clock in cyanobacteria is better thought of as a relay switch. It turns about 2/3 of the genome on in the morning (and off in the evening) and turns on the remaining 1/3 of the genome at dusk (and off at dawn). Recent findings about bacterial, plant, protist, fungal and animal clocks suggests as many as five separate events of the origin of a circadian clock on Earth - one for each major group of organisms.
Mutations and deletions [1,2 5,6] of either one of the three Kai genes affect the circadian phenotype, either by altering the inherent period of the freerunning rhythm, or by abolishing rhythmicity altogether. Interestingly, Synechococcus cells appear to have a "memory" of the circadian phase in which they find themselves and this memory gets transmitted from parental to daughter cells during cell division. Actually, under certain conditions, cell division is a much more rapid process than a circadian cycle. In other words, Synechococcus may undergo several cell divisions over a period of a single day, yet the colony as a whole keeps its circadian rhythms running all along [2,3].
Next time, I will focus on the contributions of cyanobacteria to the understanding of the origin, evolution and adaptive function of circadian clocks.
References and further reading:
 Circadian clock mutants of cyanobacteria by Kondo T, Tsinoremas NF, Golden SS, Johnson CH, Kutsuna S, Ishiura M., Science.266(5188):1233-6 (1994, Nov 18)
 Circadian clocks in prokaryotes by Carl Hirschie Johnson, Susan S. Golden, Masahiro Ishiura & Takao Kondo, Molecular Microbiology, Volume 21 Page 5 (July 1996).
 Circadian Rhythms in Rapidly Dividing Cyanobacteria by Takao Kondo, Tetsuya Mori, Nadya V. Lebedeva, Setsuyuki Aoki, Masahiro Ishiura and Susan S. Golden, Science, Vol. 275. no. 5297, pp. 224 - 227 (10 January 1997)
 Independence of Circadian Timing from Cell Division in Cyanobacteria by Tetsuya Mori and Carl Hirschie Johnson, Journal of Bacteriology, p. 2439-2444, Vol. 183, No. 8 (April 2001)
 CYANOBACTERIAL CIRCADIAN CLOCKS -- TIMING IS EVERYTHING by Susan S. Golden & Shannon R. Canales, Nature Reviews Microbiology 1, 191-199 (2003)
 Circadian rhythms: as time glows by in bacteria by Johnson CH, Nature 430, 23-24 (2004)
If bacteria have cercadian rhythms, then do viruses also have them?
It might be interesting if viruses did manage to have some sort of kinase-based clock, a la the Kaiten proteins. Seems kind of far-fetched, though. I can't imagine that viruses have too much ATP in stock, and much in the way of metabolism for processing ADP and ATP.
But viruses might be able to hijack the bacterium's clock when it hijacks the bacterium itself...