Delineation of the Transcriptional Boundaries of the Zux Operon of Demonstrates the Presence of Two New Zux Genes*

The 5’ and 3’ ends of the lux mRNA of Vibrio harveyi, which extends over 8 kilobases, have been mapped, and two new genes, ZuxG and ZuxH, were identified at the 3’ end of the lux operon. Both Sl nuclease and primer extension mapping demonstrated that the start site for the lux mRNA was 26 bases before the initiation codon of the first gene, ZuxC. The promoter region contained a typical -10 but not a recognizable -35 consensus sequence. By using Sl nuclease mapping the mRNA was found to be induced in a cell density-and arginine-dependent manner. The DNA downstream of the five known V. harveyi lux genes, ZuxCDABE, was sequenced and found to contain coding regions for two new genes, designated 1uxG and ZuxH, followed by a classical rho-independent termination signal for RNA polymerase.

The regulation of luminescence in marine bacteria has been the target of intense investigation over the last few years. Structural genes responsible for light production have been isolated from several strains of luminescent bacteria, including Vibrio harueyi, Vibrio fischeri, and Photobacterium phosphoreum (1). There are five common lux structural genes: luxC, D, and E code for the reductase, transferase, and synthetase components, respectively, of a fatty acid reductase complex; and 1uxA and B code for the (Y and p subunits of luciferase (Z-6). The fatty acid reductase complex is responsible for producing an aldehyde substrate which, along with 0, and FMNH*, is necessary for the light-emitting reaction * This research was supported by Grant MA7672 from the Medical Research Council of Canada. The costs of uublication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 USC. Section 1734 solely to indicate this fact. catalyzed by luciferase. An additional gene, luxF, has been found in the P. phosphoreum lux operon (7). Although its specific function is unknown, the protein exhibits 30% homology with the p subunit of luciferase.
The mode of regulation of only one bacterial lux system, that of V. fischeri, has been well documented (4,8). There are two (left and right) operons involved in the V. fischeri lux system which are transcribed in opposite directions. The right operon contains the 1~x1 gene, which is responsible for producing a small molecule (autoinducer) that causes induction of the luminescence system. This regulatory gene is followed by the five structural lux genes, luxA-E, in the order lux-CDABE. The left operon contains the 1uxR gene which encodes a protein that has been proposed to function as a receptor for the autoinducer.
This complex then stimulates transcription of the right operon. A positive feedback loop is therefore established, and autoinduction of the luminescent system is achieved in a cell density-dependent manner. The autoinducers of V. fischeri and V. harveyi have been purified and identified and have similar chemical structures (9, 10). Analysis and expression of V. harveyi lux DNA have shown that the luxA-E genes are arranged in the same order, lux-CDABE, as in the V. fischeri lux system. There is no gene corresponding to the 1~x1 gene of V. fischeri immediately upstream from the V. harveyi structural genes. Instead, the first open reading frame of greater than 40 codons is located more than 630 bases upstream from 1uxC. It has the same relative position and orientation as the 1uxR gene of V. fischeri (ll), but this upstream V. hurveyi gene does not correspond in sequence to either regulatory lux gene of V. fischeri (12,13). The role, if any, of this gene in the regulation of the lux operon of V. hurveyi has yet to be determined. Although the structural genes of the luminescent systems as well as the structures of the autoinducers of both bacteria are comparable, the mechanism of regulation and/or the organization of the lux regulatory genes are different. Transposon mutagenesis of V. harveyi resulting in Lux-phenotypes has demonstrated that two unlinked regions of the genome are essential for luminescence (14). Region I contains the luxCDAB and E genes, whereas region II appears to have a regulatory function, suggesting that the regulatory genes of the luminescent operon of V. harveyi are not linked to the structural genes.
The present work defines the boundaries of the transcriptional unit of the lux operon of V. hurveyi which extends over 8 kilobases and demonstrates that the lux mRNA is induced during development of luminescence. The 5' and 3' ends of the mRNA have been mapped, and two new lux genes, 1uxG and luxH, have been located at the 3' end of the operon. Moreover, evidence has been obtained that 1uxG is found not only in V. harveyi, but also in V. fischeri and P. phosphoreum.

RESULTS
A set of polycistronic mRNAs which codes for the lux proteins has been identified in V. harueyi (16,21), and it has been shown that the mRNAs detected by Northern blot analysis extending across luxD, B, A, and E are induced during development of luminescence. However, the mRNAs starting at 1uxC are not readily detected, and it could not be determined whether they are induced. Moreover, the exact upstream and downstream termini of the mRNA have not yet been determined.
Localization of the 5' End of the lux mRNA-To elucidate the 5' terminus of the lux mRNA, total mRNA was isolated from V. harveyi, hybridized to a DNA probe encompassing the 5' region of the luxC gene, and treated with Sl nuclease. The probe used was a SacI-BamHI restriction fragment (see Fig. l), 5' '*P labeled at the BamHI site. As shown in Fig. 2, only mRNA isolated from V. harveyi cells after induction of luminescence partially protected this fragment, demonstrating that the 5' terminus of the induced lux mRNA occurs between the Sac1 and BamHI sites. Since arginine is known to augment luminescence in minimal medium (22), the Sl nuclease patterns of mRNA obtained from V. hnrveyi grown in the presence and absence of arginine in minimal medium were also investigated. Only mRNA from cells grown in minimal medium containing arginine protected the DNA probe, and the same sized fragment was obtained as during induction in complex medium. The partially protected DNA fragment was sized on a sequencing gel to determine the 5' end of the mRNA accurately. The Sl nuclease reactions were carried out with 20, 50, 100, and 300 units of enzyme (Fig.  3a). The sharpest band can be seen when 50 units of Sl nuclease are used (lane 2, Fig. 3a), placing the start of the message 16 nucleotides before the initiation codon of the ZuxC gene. However, start sites ranging from 11 to 26 nucleotides in front of luxC! can be measured depending on the amounts of Sl nuclease used. With 20 units of Sl nuclease, a minor band (arrow, Fig. 3a) can be seen corresponding to a start site 26 nucleotides in front of luxC (position +I, Fig. 3~). Because of the somewhat ambiguous nature of the Sl nuclease mapping results, the 5' end of the mRNA was also determined by primer extension mapping (Fig. 3b). A loo-base ssDNA fragment (ClaI-SspI) 32P labeled at the 5' end was hybridized to RNA and was extended with reverse transcriptase. The lOObase primer migrated as predicted with respect to the M13mp18 sequence ladder (data not shown), whereas the with sizes of 184 and 185 bases. From the size of the larger fragment (Fig.  3b), the 5' end of the message could be assigned to nucleotide +l (Fig. 3c) in agreement with the results obtained by Sl nuclease mapping using the lowest amount of enzyme. It is interesting to note that an increase in Sl nuclease concentration results in the removal of the AT-rich 5' end by as much as 13 nucleotides.
Upon examining the DNA sequence, a putative -10 promoter sequence can be recognized just upstream from the message start site (Fig. 3~). There does not, however, appear to be any recognizable -35 promoter consensus sequence.
Nucleotide Sequence of Two New Genes Found within the lux Operon-Since the mRNAs extend 2-3 kilobases downstream from the luxE gene, the last known gene of the lux operon, the downstream region was sequenced (using the strategy outlined in Fig. 4) in order to determine the specific 3' terminator site and whether or not other lux genes are encoded in this area. The DNA sequence downstream from 1uxE was found to contain two previously unrecognized genes, luxG and lunH, transcribed in the same direction as 1uxE (Fig.  5). The first gene, luxG, starts just one nucleotide after the stop codon of the luxE gene, and consequently the Shine-Dalgarno (23) sequence of luxG resides in the 3' end of the coding region of the luxE gene. This gene codes for a protein of 233 amino acids with a molecular weight of 26,108. The second gene, lunH, starts 22 nucleotides after luxG and codes for a protein of 230 amino acids with a molecular weight of 25,326. The first 150 codons of luxG were compared with an open reading frame found just downstream from the luxE gene of a different genera of luminescent bacteria, P. phosphoreum' (Fig. 6) nucleotides downstream from the stop codon of luxH (Fig.  7~). It cont,ains a classical GC-rich hairpin loop followed by a string of Ts. In order to determine if this signal is active in. uiuo, the 3' end of the lux mRNA was mapped using Sl nuclease. A DNA probe '*P labeled at the 3' end with Klenow was hybridized to total RNA isolated from V. harueyi cells.
The Sl nuclease reactions and subsequent treatment of the protected fragments were performed as before for the 5' end of the mRNA. It is of interest to note that in order to determine accurately the size of the Sl nuclease reaction products, it was necessary to run the sequencing gel at a very high temperature. The DNA ran anomalously at lower temperatures possibly due to the secondary structure of the hairpin loop in the probe. The same size product was obtained in Sl nuclease reactions containing 20, 50, and 100 units of enzyme (Fig. 7a), indicating that Sl nuclease was not removing residues from the duplex at the 3' end as was observed for the 5' end. According to Sl nuclease mapping, the mRNA terminates after the hairpin structure, at one of two nucleo- tides within the run of Ts (Fig. 7~). When the probe was hybridized to mRNA isolated from V. harveyi cells before and after induction of luminescence (Fig. 76), more lux mRNA could be seen in the cultures after induction of luminescence, consistent with the 5' Sl nuclease mapping results. Just downstream from the termination signal for the lux operon is found an analogous, rho-independent termination signal in the complementary strand (Fig. 7~). The 3' termination of an open reading frame that extends at least 400 bp occurs 19 nucleotides upstream from this signal. When a probe (BumHI-SacI, 3' labeled at the Sac1 site) was hybridized to RNA isolated from uninduced and induced cultures of V.
hurueyi, no message could be detected (data not shown). It appears then that the mRNA corresponding to this gene is in low abundance and is not coordinately induced with light production, indicating that this open reading frame is not likely to code for a lux gene involved in the luminescent system. The presence of the two termination signals in opposite strands with converging coding regions at the end of luxH along with in viva verification of the 3' end confirms that the end of luxH is the 3' terminus of the lux operon. New Genes Encoded by lur mRNA 3517 DISCUSSION In this paper, the transcriptional end points of the mRNA from the lux operon of V. harueyi have been defined, and two new genes encoded by the lux mRNA have been identified. Sl nuclease and primer extension mapping were the two techniques used to map the 5' end of the mRNA and gave identical results providing that the amount of Sl nuclease was carefully controlled. Just upstream from the startpoint a -10 but no corresponding -35 recognition sequence for RNA polymerase could be found. This may suggest that a regulatory protein is required for proper transcription initiation by the RNA polymerase. The promoter region for the right operon of the V. fischeri luminescent system also lacks a -35 consensus sequence and is believed to require a positive regulator for transcription (12).
Previous studies using Northern blots have shown that polycistronic messages of varying lengths exist for the V. hnroeyi lun operon (16,21). Although it is clear that the mRNAs extending across the luxD, A, B, and E genes and downstream DNA were induced, those starting at the luxC gene were not readily detected. Nor could it be determined whether or not they were induced during the development of luminescence. By application of Sl nuclease in these experiments, it has been possible to show that the mRNA originating at the 1unC gene is indeed induced, consistent with the synthesis of all the proteins within this operon being coregulated during induction of light emission. Similarly, Sl nuclease mapping has shown that arginine causes an increase in the lux mRNA level in V. hnrueyi grown in minimal medium, indicating that arginine acts to stimulate luminescence at the transcriptional level.
The DNA located downstream of the 1urE gene was sequenced in an effort to understand why the mRNA extends beyond the last known gene of the lux operon. Two new genes were found, designated as luxG and 1uxH. Downstream from IuxH, two classical rho-independent termination signals for RNA polymerase on opposite strands and separated by less than 30 bp could easily be identified. The termination signal for the lux mRNA, which has an energy of -17.6 kcal, was confirmed in uiuo using 3' Sl nuclease mapping. The other termination signal, with an energy of -11 kcal, is located just after the end of a convergent open reading frame coding for a protein of unknown function. The next best candidate that could exhibit a hairpin loop structure in the downstream region after 1uxE has an energy of only -6.9 kcal and lies within the 1uxH gene. The presence of the termination signal immediately after 1urH along with the induction of the corresponding mRNA with light production provide strong evidence that the 1uxG and 1uxH genes are part of the lux operon.
Elucidation of the functions of the 1uxG and 1uxH genes may provide a key to understanding the role of luminescence in bacteria. Homologies between the proteins coded by 1uxG and 1uxH and sequences of proteins of known functions have not yet been detected. The presence of these genes in free living bacteria is not essential for light production since clones containing only the luxC, D, A, B, and E genes are able to emit light (16, 25), and transposon mutagenesis has failed to produce any Lux-phenotypes with insertions in these genes (14). It is possible that 1uxG and 1uxH are regulatory proteins, but this is difficult to test because Escherichia coli is unable to support regulated light generated by V. harveyi DNA (16). Preliminary sequence data have shown that V. fischeri contains a homologous gene to 1uxG at the same relative position in the operon. This result suggests that the 1uxG gene product is not essential for regulation by autoinduction since clones containing only the 1uxCDABE and regulatory genes of V. fischeri are able to produce regulated light in E. coli (5). Transposon mutagenesis of V. harueyi (14) has provided further evidence that 1uxG and 1uxH are not required for regulation since disruption of the transcription of downstream genes does not affect induction of the reporter gene, @-galactosidase. Comparison of the first 150 amino acids of luxG with an amino acid sequence found just downstream from luxE in P. phosphoreum demonstrated 39% identity. 1uxG is therefore common to and located in the same relative position in the lux operons of V. harueyi, V. fischeri, and P. phosphoreum. It is possible that lunG and lurH may produce proteins that fine tune the expression or properties of the light-emitting reaction without affecting induction of light production. Alternatively, it is possible that the downstream genes are required for an essential function relating the Zux system to the survival and/ or symbiosis of luminescent bacteria in the marine environment.