In a previous study, Brz was identified as a second regulator of bop transcription in Hbt. salinarum 10. This changed a simple model of bop transcription control from one using only one activator, Bat, to a more complex system with two indispensible activators that probably interact with each other. The same study also showed that the regulatory effects of Brz on bop are not mediated via Bat 10. In the current study, the activation effect of Bat on bop via direct regulation of Brz by Bat was excluded by the phenotype of Δbat mutants, in which transcription of brz was found not to be down-regulated (data not shown). These results strongly suggest that Bat and Brz cooperate at the protein level. In addition, a new candidate for cooperation with Brz and Bat was found to be encoded by a small ORF on the brz mRNA. The existence of this ORF was recognized after 3'-end determination of the brz transcript. For this, the mRNA was circularized and the joined ends reverse-transcribed and PCR amplified according to the method described by Brenneis; 2007 17. The 3'-ends of the brz mRNA were located 173-228 bp downstream of the brz stop codon (Figure 1), in some cases downstream of the TATA box of the adjacent bop gene. Although mRNA degradation cannot be excluded as a cause of the multiple 3' ends detected by RACE, previous studies have shown that ragged transcription termination is common in Archaea 17. These results correlate well with the position of the 3'-end of the brz mRNA reported previously by Koide; 2009 18, who used a different approach that averaged the signals from the mRNA population. They located the 3'-end of the message at 226.95 bp from brz stop codon. In both cases, there are sufficient nucleotides downstream of the brz stop codon to encode a small protein of 5.7 kDa (55 amino acids long) with an alkaline pI of 11.4 (Figure 1). The gene was named brb (bacteriorhodopsin-regulating basic protein) as, later in the current study, it was found to be involved in regulation of bop transcription. According to RT qPCR results, brz-brb transcript level is as low as the level of bat transcripts (data not shown).

To test the translation activity of the start codon of the brb gene, a mutant producing an N-terminal BgaH fusion protein was constructed by insertion of a suicide plasmid into the brz-brb locus. The construct (pVYT1 plasmid) contained the 5'-end of the brb gene fused to the bgaH reporter gene (lacking its own translational start codon), thus translation could only start from the start codon of the brb gene (Figure 2A). Enzymatic activity of the Brb-BgaH fusion protein was detected on agar plates containing X-gal. Colonies of the mutant became blue (Figure 2B). In a control experiment, a mutation eliminating the brb start codon (GAG instead of GTG; pVYT2 plasmid) resulted in red colonies (BgaH^-) on the plates with X-gal (Figure 2C), proving no other potential start codons were active upstream or downstream of the tested brb start codon.

Attempts to detect the gene product, Brb, in various WT cell fractions using mass spectroscopy were not successful, possibly due to low levels of this protein, consistent with the low level of brz-brb transcripts by RT qPCR. The small size and high pI value may also be contributing factors limiting the number of endoproteolytic peptides available for the identification. We also tried to produce samples for analysis that were enriched in Brb. For this purpose, a mutant was constructed which carries an insertion of the DNA fragment encoding the

c

b

d

Two series of mutants in the brz-brb region were produced to demonstrate the involvement of Brb in regulating bop transcription. One series used a wild-type brz background, and the second used a brz inactive mutant (brzS9) as the parental background. Although BrbCBD fusion protein was unable to be detected, the brbCBD mutant produced less bop mRNA and BR than wild-type cells, indicating a role for Brb in bop expression (Figure 3; Table 1; Table 2). Given the negative effect of the BrbCBD fusion on bop gene expression, the role of Brb was assessed next by translational knockouts. Two knockout mutants having stop codons in the brb gene were constructed: brbstop1 and brbstop2 (Figure 4). The brbstop mutants were constructed instead of the brb deletion mutant to avoid any possible polar effects of the deletion on transcription of the neighboring genes. The brbstop1 mutant had a stop codon in the middle of the brb gene and so should produce only the first half of the Brb protein. Mutant brbstop2 contained two stop codons at the beginning of the brb gene. The bop mRNA (Figure 4) and BR levels (Table 1, Table 2), were found to be the same as in wild-type cells in both mutants. The lack of a phenotype could be explained either by the residual activity of a proposed protein complex that regulates bop transcription in the absence of Brb protein, or by the ability of the putative complex Bat/Brz to recruit other proteins able to substitute Brb functionally. In the first case, appropriate mutations in the Brb sequence could have more deleterious effects than a knock-out of brb. To test this, mutants were constructed based on old observations of changes in brz and brb that were previously published in a study of Hbt. salinarum strain S9, long before these ORF's were recognized to be involved in bop regulation 316. The S9 strain is a constitutive BR-overproducer, and mutations in Bat were suggested to be responsible for this phenotype 7. However, if there is a cooperation of Bat with Brz and Brb, mutations in brz and brb could also contribute to the phenotype in a bat wild-type background. Therefore, the naturally selected mutations in brz and brb of S9 were introduced separately or together into R1 wild-type cells, using the pop-in/pop-out gene replacement method. The brz mutations S9 converted Val to Ala, and Ala to Thr located in a region of high homology with Brz from the archaeon Halorhabdus utahensis (additional file 2). The brb gene mutations S9 resulted in a frame shift and translation of a protein with the N-terminal half of the native Brb followed by a new C-terminal sequence (additional file 2).

In the initial round, two single mutants, brzS9 and brbS9 as well as a double mutant brzS9brbS9 were produced (Figure 5). Northern blot analysis showed the mutated brb gene (brbS9) had no effect on bop transcription if the brz gene was from wild-type (Figure 5; Table 2; Table 3). In contrast, cells with the wild-type brb gene but mutations in the brz gene (brzS9) showed down-regulated levels of bop mRNA and BR expression (Figure 5; Table 2; Table 3). The effect was more prominent at the stationary (OD₆₀₀ = 1.1-1.2) stage of growth and did not depend on the presence of light. Surprisingly, the double mutant carrying both changes, brzS9brbS9, showed bop mRNA level elevated up to that of wild-type and BR expression up to half of the wild-type level at the stationary stage of growth (Figure 5; Table 2; Table 3). Apparently, the Brb protein can be omitted genetically at the level of the wild-type Brz either completely or with sequences like that of the brbS9 mutation. However, at the level of the brzS9 mutant knocking out BR regulation, the BrbS9 protein can rescue the S9 mutation of brz to restore bop mRNA levels up to that of WT. Since the naturally selected C-terminal sequence of Brb in the brbS9 strain shows this compensatory effect, a further two control mutants were produced, brbM1 and brzS9brbM2, with frame shifts changing only the sequence of the C-terminal part of the Brb protein beyond amino acid 23 in an arbitrary manner (additional file 3). The single mutant, brbM1, possessed the wild-type Brz, and the double mutant, brzS9brbM2, carried the S9 mutation in Brz (Figure 6). The phenotypes of both mutants confirmed the dominant role of Brz and specificity of S9 mutations in Brb, as the M2 C-terminal sequence in contrast to the S9 C-terminal sequence in S9 restored the WT phenotype only to a small extent (Figure 6; Table 1; Table 2).

Taken together, these results suggest that Brb participates in the regulation of bop transcription and, further, that Brz and Brb proteins depend on each other for the activation of bop transcription.

While Bat and Brz are crucial regulators of bop transcription, the results described above indicate that Brb is also at least marginally involved in this process. To further test a possible cooperation between the Brz, Brb and Bat proteins, a GFP reporter assay was established. Reporter plasmids contained the gfp gene and different sets of the brz, brb, bat genes in a single operon under the bop promoter control (Figure 7A). At the 3'-end a terminator sequence from the A flagellin operon was added. The reporter plasmids included replication origins of Hbt. salinarum and E. coli, and resistance markers for novobiocin and ampicillin. A control plasmid was identical except for using the OE3095R gene (coding for a conserved hypothetical protein in Hbt. salinarum, http://www.halolex.mpg.de) instead of brz, brb or bat, since its deletion did not effect bop transcription (data not shown). Expression of GFP in these constructs is controlled by the bop promoter, allowing the effects of Brz, Brb and Bat on bop promoter activity to be directly assessed.

The wild-type cells R1 were transformed by the reporter plasmids and transformants were selected on plates containing novobiocin. Prior to measuring GFP activity, the transformants, grown in medium with novobiocin, were washed and then resuspended in novobiocin-free medium. This was to avoid the negative effects of novobiocin on bop promoter activity 1920. For each transformant, cells were grown under light or dark conditions, and the fluorescence measured and compared. Plasmid copy number should not depend on the presence and absence of light, so the results for the same transformant should be directly comparable. Since light is an inducer of bop transcription, GFP fluorescence would only be expected to be increased in light versus dark conditions by the correct combination of Brz, Brb, Bat proteins expressed from the reporter plasmid. As shown in Figure 7, the enhancement of GFP fluorescence by growth of cells under light was only observed when all three genes: brz, brb and bat were present in the reporter plasmid. This indicates their cooperation for light mediated activation of the bop promoter. While the absolute values of GFP fluorescence varied between transformants carrying the same plasmid construct, as well as between transformants carrying plasmids with different constructs of brz, brb and bat, these variations do not effect the interpretation of light versus dark behavior of individual transformants, and were probably due to differences in plasmid copy number. This data clearly identifies a requirement for all three genes, brz, brb and bat, in bop induction.

ORFs OE2447F and OE3131F encode hypothetical proteins that carry zinc finger motifs like Brz, and might also be involved in the bop gene regulation. For this reason, searches near these ORFs were conducted to check if they also possessed brb-like ORFs encoding proteins potentially able to substitute for the Brb protein. The 5' and 3' UTRs of many genes in Hbt. salinarum have been determined by Koide; 2009 18 and this data indicated that transcripts of both OE2447F and OE3131F were unusually long. The transcription start site (TSS) of OE2447F (OE2447F, 62 aa, pI = 4.2) lies 134.51 bp upstream of its start codon, while the transcription termination site (TTS) is located 741.21 bp downstream from the stop codon. The 134.51 bp region, upstream of OE2447F contains a small ORF in the same orientation as OE2447F (7 bp overlap), and can code for a small, basic protein (45 aa, pI = 9.4) (Figure 8; additional file 4). We designated this protein as

b

p

4

b

p

2

To test whether Bp2 and Bp4 are functional analogues of Brb, and also influence bop transcription, mutants with stop codons aborting translation of the bp2 and bp4 genes, bp2stop and bp4stop, were constructed from wild-type cells. The introduced stop codons did not effect BR expression (Table 3). To test a possible substitution of Brb by these proteins, the brbstop2 mutant was used as the parental strain to construct a double mutant having stop codons aborting translation of the bp2 or bp4 genes. Repeated attempts to make such double mutants were not successful even though the same pop-in/pop-out method was used that had readily been able to generate the previous bp2 or bp4 stop codon mutations. The results indicate that bp2 and bp4 are not essential by themselves, but may become significant in a brb knockout background.

In Haloquadratum walsbyi, ORFs encoding a small zinc finger (HQ1083B) and a small basic protein (Bp) were also found upstream of the bat gene. The gene encoding the basic protein overlaps with the bat gene, so they are likely to be co-transcribed. The HQ1083B gene may be transcribed separately or together with bp and bat genes (Figure 8; additional file 6). The only close homologue of Hbt. salinarum Brz, is that carried by Hrb. utahensis. In this archaeon, the brz gene appears to be transcribed as a monocistronic mRNA, but an ORF encoding a small basic protein can be annotated between the bat and crtY genes (Figure 8; additional file 7). This ORF could either be part of an operon starting from the brp and ending just before crtY, or transcribed together with the crtY gene.

An inhibitory effect of novobiocin on bop transcription was reported for the NRC-1 strain of Hbt. salinarum 1920, but its influence in R1 strain is not known. Although experimental care was taken not to allow novobiocin to interfere with transcriptional experiments reported here, we were interested to see the genome wide influences of novobiocin on transcription. Therefore, a DNA microarray analysis of the transcripts from R1 (and R1M1, a bacterioruberin deficient strain derived from the R1) was performed, comparing cells grown in the presence or absence of novobiocin (additional file 8). Surprisingly, bop transcription in the R1 (and R1M1) strain was not inhibited, but was 5-fold increased by the presence of 0.1 μg/ml novobiocin. This represents a clear difference to the behavior of the NRC-1 strain, even though R1 and NRC-1 are extremely similar in sequence 21. It most likely reflects a strain-specific mutation in the B subunit of gyrase in R1 cells (Lys-544-Asn; 21), as novobiocin binds to and directly affects the gyrase. Attempts to generate an Asn-544-Lys mutant in R1 cells failed, possibly because of the important role of the B subunit of gyrase in recombination events during the pop-in/pop-out procedure.

Besides the novobiocin-induced changes in bop transcription, the expression of many other genes were affected, including stimulation of gyrase and type I topoisomerases. These results are similar to the effects described previously in E. coli, where topoisomerases are known to be regulated by DNA supercoiling in a negative feedback loop 22. It has been reported also, that the expression of several heat-shock genes depends on the level of DNA supercoiling 23, and we detected as well an up-regulation of heat-shock gene hsp20. Finally, novobiocin exerted an up-regulating effect on genes of the basal transcription apparatus, while some genes encoding ribosomal proteins and many metabolic enzymes were down-regulated (additional file 8).

Hbt. salinarum R1, R1M1 and mutant strains derived from R1 were grown in complex medium as described previously 25, either in the dark or under light. For all experiments, cells were cultivated in 35 ml of medium contained in 100 ml volume flasks, and the flasks closed by aluminium foil. Under these conditions, oxygen in the cultures is naturally depleted during cell growth. The E. coli strains DH5α and DH10β were used for transformation according to the supplier's instruction (Invitrogen).

brbCBD, brbstop1, brbstop2, brzS9, brbS9, brzS9brbS9, brbM1, brzS9brbM2, bp2stop and bp4stop mutants were produced using the pop-in/pop-out method 26. For this, the plasmids were constructed by inserting of corresponding DNA fragments into the pVT plasmid 10 using HindII, BamHI, XbaI, PstI restriction sites. The fragments were produced by PCR using corresponding primers (additional file 9) and verified by sequencing. To create the brzS9brbM2 mutant, the corresponding plasmid was constructed by cloning the brzS9brbM2 fragment selected from PCR fragments, amplified by using a primer having a randomly brbM2 mutation instead of the brbS9 mutation. Transformations were carried out by the PEG method 2728. Transformants were selected using blue/red screening 26 by plating the cells onto agar growth medium containing 0.1-0.2 μg ml^-1 of novobiocin and 40 μg ml^-1 of X-gal 29. Cells from single blue colonies (pop-in) were propagated in culture medium without novobiocin and plated on agar plates containing 40 μg ml^-1 of X-gal without novobiocin. Red colonies (pop-out) were checked for the presence of the respective mutations by sequencing of PCR fragments. For the amplification and sequencing of these fragments the corresponding primers were used (additional file 9). The brbM1 mutant was produced as a random mutant during selection of the brbS9 mutant.

The brb-bgaH mutants were produced in result of the pVYT1 and pVYT2 plasmids insertion in brz-brb genes region (Figure 2; additional file 9). The plasmids were constructed by cloning bgaH and brb gene fragments into the pAN plasmid 28. The bgaH and brb fragments were PCR amplified using corresponding primers (additional file 9). Transformations were carried out by the PEG method 2728. Transformants were selected by plating the cells onto agar growth medium containing 0.1-0.2 μg ml^-1 of novobiocin and 40 μg ml^-1 of X-gal 29.

Northern blot hybridizations were done as described 28. DIG-labeled bop gene probe generated by PCR were used for the chemiluminescence detection performed with the DIG luminescence detection kit (Roche) according to the supplier's instructions. Densitometry was performed using program ImageJ64. For the bop gene PCR amplification primers 1 and 2 were used (additional file 9). Total RNA was prepared from cells, taken in every time series experiment, using the peqGold RNAPure kit (Peqlab Biotechnology) according to the supplier's instruction. For RNA gel electrophoresis 4-5 μg of total RNA were used per lane.

To determine the 5' and 3'ends of transcripts 5'-3'RACE (rapid amplification of cDNA ends) was performed by the use of an RNA circularization mediated method according to Brenneis; 2007 17. For cDNA synthesis and PCR amplification from cDNA primers 3 and 4, 5 were used, respectively (additional file 9). The obtained PCR-amplified fragments were cloned (TOPO TA cloning Kit, Invitrogen) and sequenced using M13 forward and M13 revers primers.

The GFP reporter plasmid, pVTR, was constructed by cloning of following fragments into the pAN plasmid 28: multiple cloning site, pGRB origin of replication, bop promoter, GFP gene (smRSGFP), terminator of A flagellin (flgA) operon. The fragments were PCR amplified by using the corresponding primers listed in an additional file 9. The flgA terminator has previously been shown to be effective in plasmids described by Furtwängler; 2010 30. Different sets of the brz, bat, brb, OE3095R genes were cloned into the pVTR plasmid between the gfp gene and the flgA terminator. Transformations of wild type R1 cells by the reporter plasmids were carried out by the PEG method 2728. Transformants were selected by plating the cells onto agar growth medium containing 0.1-0.2 μg ml^-1 of novobiocin. Integrity of the reporter plasmids in the transformants was checked by a restriction analysis of the plasmids isolated from these transformants.

For GFP fluorescence measurement, cells were centrifuged and lysed in 3 ml of buffer (10 mM Tris-HCl, pH = 7.5). For each measurement, equal amounts of cells were used. The absolute number of cells corresponded to 3 ml of culture at OD₆₀₀ = 0.4. The wavelengths of excitation and emission were 588 nm and 610 nm, respectively.

For BR measurement, cell samples taken in every time series experiment were centrifuged and lysed in 1 ml of buffer (10 mM Tris-HCl, pH = 7.5). For each measurement, equal amounts of cells were used. The absolute number of cells corresponded to 10 ml of culture at OD₆₀₀ = 0.7. The lysates were treated with DNase I for 20 min at 37°C. Absorbance spectra were recorded from 200 to 800 nm and BR concentration was quantified from spectral decomposition performed to estimate the compositions of bacteriorhodopsin-bacterioruberin cell mixtures.

Total RNA was prepared using peqGold RNAPure solution (Peqlab Biotechnology) from cells grown to an OD₆₀₀ of 0.5. RNA (5 μg) was reverse transcribed into Cy3/Cy5-labeled cDNA using CyScribe First-Strand cDNA Synthesis Kit with enclosed random nonamer primers and Cy3-/Cy5-dUTP (both Amersham Biosciences, Freiburg). Labelled cDNA was hybridized to in-house fabricated whole genome DNA-microarrays 31 at 64°C overnight. To determine the fluorescence ratios the slides were scanned (GenePix 4000 B, Axon Instruments) and the data were extracted using the GenePix Pro 6 software. After background subtraction, pinwise normalization and data evaluation by a Student's T-test, those transcripts displaying a p-value equal or lower than 5 × 10^-5 and a ratio of +/- 1.6 were selected as significantly regulated. A detailed description of the microarray design, experimental procedure and data-evaluation is described in 3132. The data obtained from the microarray experiment for R1 strain were deposited at http://www.ebi.ac.uk/miamexpress under the accession number (E-MEXP-3219).