Discussion

The LH-Ⅰ and RC complexes from R.sphaeroides are highly organized within the photosynthetic membrane for efficient transfer of light energy［1，2，20］.The expression of the genes coding for the polypeptides of these complexes is also highly regulated with respect to growth conditions such as light and oxygen as demonstrated in this study.Light was shown to regulate the puf operon in R.sphaeroides，with high intensity light partly repressing the transcription of the puf operon approximately 3-fold.In addition，overall expression of the puf operon was repressed an additional 3-fold by oxygen（see Table 1，Figure 2）.

A striking aspect of puf operon expression is that the LH-Ⅰ（βand α），RC-L，and RCM polypeptides are synthesized in vivo at widely different levels.Several possibilities may explain this observation；multiple promoters，attenuation control，or different translation efficiencies have been observed to regulate differential gene expression within individual operons of other procaryotic systems［21-24］.

Previous［10］and present studies have shown that there are two mRNA species homologous to the structural genes for the LH-Ⅰ complex polypeptides of R.sphaeroides.They were reported in the previous study as 2.6 and 0.8 kb in length based on their mobility on agarose gels.According to more accurate measurements on polyacrylamide gels performed in this work and DNA sequence determinations for this region［1］the small transcript is 0.5 kb in length.The larger and less abundant mRNA（2.6 kb）is a polycistronic mRNA homologous to all four structural genes（pu f B，-A，-L and-M）whereas the smaller more abundant mRNA（0.5 kb）is also polycistronic but specific only for the pu f B and A polypeptides，with the ratio of the smaller to the larger mRNA ranging from 7—18 to 1（［10］this study）with an average value of approximately 12∶1.The ratio of the two transcripts approximates the ratio of LH-I to RC complexes found within the photosynthetic membrane.Similar results were recently reported in R.capsulatus［11］，and Belasco et al.［11］concluded that the ratio of the small to large transcript encoding the LH-Ⅰand RC genes in the analogous rxcA（puf）locus of R.capsulatus results predominantly from the selective degradation of a specific segment of the initial large transcript at the 3'-end of the molecule resulting in a stabilized small transcript.

Our previous［10］and present studies also show different stabilities for the two transcripts of the puf operon in R.sphaeroides.The half-lives of the small and large transcripts were 20 and 9 min，respectively，after inhibition of transcription initiation with rifampicin.These results are very similar to those reported by Belasco et al.［11］and Dierstein［25］for R.capsulatus.It is interesting that exposure of photosynthetic cells to oxygen significantly decreased the half-lives of both of these mRNA species.When photosynthetic cells were exposed to aerobic growth conditions，the half-life of the transcript specific for pufL and pufM decreased to 5 and 6.5 min，respectively，while the value for the t1/2 of the mRNA for LH-Ⅰ structural genes declined to 10 min.The rapid loss of these mRNA species under these conditions is not surprising based on the abrupt cessation in the insertion of photosynthetic membrane components when photosynthetic cells are subjected to aerobic conditions［26］.An accelerated degradation of the mRNA specifying photosynthetic membrane proteins would be an efficient regulatory mechanism for photosynthetic cells when they are exposed to oxygen.At present we have no information as to how the transcripts are degraded and whether oxygen，as a signal for aerobic growth，directly or indirectly affects ribonuclease activities responsible for turnover of these transcripts.However，we can propose a degradation pathway for the large transcript based on the existence of intermediate size mRNA species observed in the Northern hybridization studies and the differential effect of 02 on mRNA degradation.It would appear that processing of the large mRNA（2.6 kb）occurs from the 5'-region.In addition，there appear to be regions within the 2.6-kb transcript which are more resistant to exonuclease degradation.Although the experiments reported here were not designed to specifically follow the degradation pathway of the mRNA，the observed results are consistent with the above interpretation.

The differential degradation of the two transcripts of the puf operon undoubtedly results in different steady-state concentrations of the mRNA specific for LH-I and RC-L and M structural genes.However，the different ratios of these two mRNA species（10—20-fold）cannot be explained solely by differences in mRNA stability（2-fold）.Our ability to demonstrate two different 5'-ends，with the small transcript being 29 bases longer at the 5'-end than the large transcript，would preclude the large transcript（as defined here）being a precursor of the small transcript（again as defined here）.Furthermore，it appears that breakdown of the large transcript proceeds in a 5'to 3'direction，which would further mitigate against it being a precursor of the small transcript as suggested for R.capsulatus［11］.The simplest interpretation of these data is that two different metabolically stable transcription initiation sites exist，the frequency of use of each being dependent upon upstream promoter affinities for RNA polymerase.Further，the selection of one initiation site relative to the other would result in transcription of the complete operon while selection of the other would result in termination of transcription between pu f B，A and pu f L，M.The secondary structure of the puf operon mRNA between puf A and puf L would be in accord with such a hypothesis［1］.Additionally，we have been unable to detect any mRNA sequences within 500 bp upstream to the Pstl site of the large and small transcripts reported here（see lane 5，Figure 8）.

It is also possible that transcription of the puf operon initiates at only one site，and post-transcriptional modification（s）results in degradation at the 5'-end of some fraction of the primary transcripts，resulting in the two 5'-ends observed here.These posttranscriptional modifications would，in some unknown manner，be involved in the formation of the two transcripts（2.6 and 0.5 kb）in very precise stoichiometries.Clearly，if this is the route by which these two transcripts are derived and the observed stoichiometry achieved，the mechanism（s）of post-transcriptional processing of the primary puf transcript of R.sphaeroides is far more complex than that envisioned for R.capsulatus［11］.(https://www.daowen.com)

If differential transcription initiation gives rise to the large and small transcripts，then the question of the factors regulating termination of the two puf operon transcripts at the molecular level need to be investigated.The data presented in this manuscript show that the small puf operon transcript，if it were to initiate 104 nucleotides upstream of the start of the pu f B structural gene，terminates prior to the start of the puf L structural gene.DNA sequence analysis of the intercistronic region between pu f A and-L reveals the existence of two possible stem-loop structures which，by analogy to other well studied procaryotic systems，would be classified as factor-dependent transcription terminators［27，28］.Whether either or both of these structures function in termination of the small puf operon transcript remains to be determined.Conversely，the results obtained would indicate that transcription of the large puf operon mRNA，which has a 5'terminus 75 base pairs upstream of the puf B structural gene，continues past any possible regulatory signals involved in terminating transcription of the small puf operon mRNA.No information is currently available on the existence of any transcription termination structures 3'of the pu f M structural gene.In order to more fully understand the factors regulating puf operon expression in R.sphaeroides，experiments probing the molecular mechanisms which control transcription initiation and/or termination in this bacterium need to be undertaken.In addition，the existence of protein factors which may interact either directly or indirectly with either or both of the puf operon transcripts to aid in transcription termination［29，30］or antitermination［29，31］remains to be determined.Finally，we may conclude that the promoter structure for the R.sphaeroides puf operon could be complex.The possible existence of two precise and differentially regulated sites for transcription initiation of this operon would undoubtedly be determined by selective recognition or binding of RNA polymerase to different domains within the puf operon promoter.

Figure 8　Selective Northern hybridization of the small and large puf operon transcripts with specific deoxyoligonucleotide probes

The top of the figure illustrates the structure of the R.sphaerides puf operon，the sequence of the 5'-ends of the two transcripts encoded by this operon，and the specific deoxyoligonucleotide probes LHI-4 and LHI-5 are also shown.Bulk RNA（30 μg）from R.sphaeroides 2.4.1 cells grown photosynthetically in the presence of low light resolved on a 1.2%agarose gel was used for all samples.Shown in lane 1 is the ethidium bromide fluorescent material from a typical bulk RNA sample along with the migration of bacteriophage λ DNA molecular weight standards.Lanes 2 and 3 show the results obtained after a low（room temperature）and high（65℃）stringency wash using the LH-Ⅰ-5 probe，respectively.Lane 4 is the same as lane 3 except for a longer exposure to demonstrate the absence of the 2.6-kb transcript.Lane 5 shows the results obtained using a PstI-XorII restriction fragment probe for the LH-Ⅰ genes.The PstI site is approximately 600 bp upstream of the start of puf B，and the XorII site is within puf B.Lanes 6 and 7 show the two puf operon transcripts visualized after a low（room temperature）and high（50℃）stringency wash using the pufB-specific LHI-4 probe，respectively.The high stringency wash temperature for each deoxyoligonucleotide probe was based on the theoretical melting temperature（52℃for LH-Ⅰ-4，66℃for LH-Ⅰ-5）of these molecules.Theoretical Tm were calculated assuming 2C of melting temperature for each A-U base pair and 4℃for each G-C base pair［15］.

Acknowledgment—We wish to thank Dr.B.Orozco for his advice on the S-1 nuclease and primer extension analysis.