Generation of an Artificial Double Promoter for Protein Expression in Bacillus subtilis through a Promoter Trap System

Bacillus subtilis is an attractive host for production of recombinant proteins. Promoters and expression plasmid backbones have direct impacts on the efficiency of gene expression. To screen and isolate strong promoters, a promoter trap vector pShuttleF was developed in this study. Using the vector, approximately 1000 colonies containing likely promoters from Bacillus licheniformis genomic DNA were obtained. Amongst them, pShuttle-09 exhibited the highest β-Gal activities in both Escherichia coli and B. subtilis. The activity of pShuttle-09 in B. subtilis was eight times of that of the P43 promoter, a commonly used strong promoter for B. subtilis. A sequence analysis showed that pShuttle-09 contained PluxS and truncated luxS in-frame fused with the reporter gene as well as another fragment upstream of PluxS containing a putative promoter. This putative promoter was a hybrid promoter and its β-Gal activity was higher than PluxS. Reconstructing the hybrid promoter from pShuttle-09 to PlapS further improved the β-Gal production by 60%. The usefulness of our promoter trap system is likely due to random shuffling and recombination of DNA fragments and adoption of a rapid and high-throughput screening. Thus, our data provide additional evidence to support the concept of using a promoter trap system to create new promoters.


Introduction
Bacillus subtilis is a non-pathogenic gram-positive bacterium and has been an attractive host for production of recombinant proteins from both prokaryotic and eukaryotic origins [1], [2], [3]. The suitability and popularity of B. subtilis as one of main hosts for recombinant protein production is due to several reasons: generally recognized as safe as a result of its lack of pathogenicity and the absence of endotoxins; capable of secreting functional proteins directly into culture media, easy for genetic manipulation and handling, and capable of large-scale fermentation [2], [4], [5], [6].
B. subtilis own proteins can be over expressed in B. subtilis in the order of several grams per liter [3]. However, secretion of heterologous proteins in B. subtilis is usually low [7]. Thus, efficient production of high-value recombinant proteins in B. subtilis remains a major challenge. To circumvent the problem, strong promoters, a variety of translation/secretion signals, and transcription terminators are continuously being explored. In particular, several strong promoters have been reported [8], [9], [10], [11], [12], [13], [14], [15]. In spite of the progress, it is still beneficial to find even more potent promoters.
Promoters are important regulatory elements in a genome and control spatial and temporal expression of genes and their expression strength. In bacteria, a promoter is recognized by RNA polymerases and associated sigma factors, which may be recruited to the promoter by regulatory proteins binding to specific sites in the promoter. There are several ways to find new promoters. The advent of DNA sequencing has significantly accelerated sequencing of many bacterial genomes and prediction of promoters. However, it is not yet possible to scan a bacterial genome sequence and readily predict the expression behavior of genes belonging to a regulon, largely because an accurate prediction of promoters is difficult [16]. Alternatively, a promoter trap vector, which is a plasmid containing a multiple cloning site at the 59 end of a promoter-less marker gene, can be used to identify promoters [17], [18], [19]. An unidentified DNA fragment is cloned into the multiple cloning site and expression of the marker gene is monitored to identify active promoter elements in the unidentified DNA fragment. Promoter trap systems have been successfully used to identify promoters for B. subtilis either from B. subtilis itself or from other bacillus species [20], [21], [22], [23]. A promoter trap system could perform shuffling and recombination of DNA fragments and screen of strong promoters simultaneously. However, this has never being reported. Fortuitously, when a promoter trap system was used in this study to screen Sau3A I cut fragments from B. licheniformis genomic DNA for potential promoters, a strong double promoter was identified, containing a normal promoter and a hybrid promoter due to shuffling and recombination. The putative double promoter was further improved and used for protein expression both in E. coli and B. subtilis. Our results show that promoter trap systems could be suitable for finding novel strong promoters.

Materials and Methods
Bacterial strains, plasmids and growth conditions B. subtilis 1A747 (B. subtilis PY79) was a gift from Bacillus Genetic Stock Centre of the Ohio State University. B. pumilus BYG, B. licheniformis A061, B. amyloliquefaciens and B. megaterium W-5 were isolated by our laboratory. The bacterial strains were isolated from effluent of a paper mill in Hubei province of China and partial 16S rDNA sequencing was used for identification of species in the genus Bacillus. The plasmids used in this study are listed in Table 1. E. coli DH5a was purchased from Novagen (Darmstadt, Germany). The bacterial strains were cultured in the Luria-Bertani (LB) medium at 37uC. The following concentrations of antibiotics were used for selection: 100 mg/mL ampicillin (Amp), 5 mg/mL chloramphenicol (Cm) and 50 mg/mL spectinomycin (Spec).

PCR and DNA manipulation
Polymerase chain reaction (PCR) primers and oligonucleotides used in this study are listed in Table 2. Isolation and manipulation of recombinant DNA was performed using standard techniques [24]. All enzymes were from Promega China (Beijing, China) and used as recommended by the manufacturer. The transformation of E. coli and B. subtilis was performed by electroporation [24], [25].

Construction of a promoter trap vector
Using the primer pair bgaB-1 and bgaB-2, the bgaB coding region was PCR amplified from the plasmid pDL [6]. The amplified fragment was flanked by BamH I and Sac I at 59 and 39 ends, respectively. The promoter trap vector was assembled as follows. Firstly, the obtained bgaB digested by BamH I and Sac I was cloned into corresponding sites of pGJ103, yielding pGJ-Bga. Then, the Apa I-BamH I-treated spectinomycin resistance gene, which was PCR amplified from the pDG1728 [26] using the primer pair spec-1 and spec-2, was inserted into corresponding sites of pGJ-Bga, resulting in a promoter trap vector pShuttleF.

Cloning of promoter fragments
The genomic DNA of B. licheniformis was partially digested with Sau3A I, and then ligated with the promoter trap vector pShuttleF treated with the BamH I and alkaline phosphatase. The ligation mixture was transformed into E. coli DH5a and recombinants harbouring promoter fragments were screened on the LB solid medium supplemented with X-Gal (20 mg/mL).

Construction of expression vectors
Using primers apr-1 and laps-down, the promoter P LapS flanked by Apa I and EcoR I was PCR amplified from pShuttle-P LapS . The resultant promoter was digested with Apa I and EcoR I and cloned into corresponding sites of pGJ103, yielding the expression vector pLus-Hyb. To introduce 6-His tag, synthetic oligonucleotides Tag-1 and Tag-2 were annealed. The annealed fragment with adhesive ends of Xba I and Sac I at 59 and 39 ends respectively was inserted into the Xba I-Sac I-treated pLus-Hyb, resulting in pLus-His.

b-Gal activity assay
The b-Gal activity assay was carried out as previously described [27]. Samples were taken at different time points of culture and the b-Gal activity was measured. The activity was expressed as Miller units per mL sample (Miller U/mL). For each assay, three independent experiments were performed with two replicates. Statistical tests were carried out by the SPSS (Statistical Product and Service Solutions) software. Data are presented as means6 S.D.

Results
Screening and isolation of the strong promoter fragments from B. licheniformis chromosomal DNA To screen and isolate strong promoter elements, a promoter trap vector pShuttleF ( Figure 1A) was constructed on an E. coli-B. subtilis shuttle vector pGJ103. The pShuttleF contained bgaB coding for a heat stable b-galactosidase (b-Gal) as a reporter gene. In addition, an alternative resistance selection marker, spectinomycin, was introduced into the pShuttleF beside the chloramphenicol resistance marker. Both spectinomycin and chloramphenicol resistance genes could confer the resistance selection either in E. coli or B. subtilis. Finally, an engineered BamH I restriction site was designed next to the start codon of bgaB in order to clone DNA fragments digested by the Sau3A I. Validation of the pShuttleF was achieved by inserting a commonly used strong promoter P43 (2, 6) in B. subtilis upstream of bgaB. The resultant pShuttle-P43 was transformed into E. coli DH5a and B. subtilis 1A747 and blue colonies on solid LB plate supplemented with X-gal were isolated. The bgaB directed by the P43 in pShuttleF was duly expressed in E. coli and B. subtilis ( Figures 1B  and 1C).
In order to identify new promoters, Sau3A I-digested B. licheniformis DNA fragments were inserted into the BamH I site of the promoter trap vector pShuttleF and resultant plasmids were electro-transformed into E. coli DH5a. Two hundred colonies were picked from about 1000 colonies according to visible coloration on the X-gal screen plate in the presence of both antibiotics. The b-Gal activities by these recombinants after 24 h of culture ranged from 500 Miller U/mL to 14016 Miller U/mL. Amongst them, the recombinant pShuttle-09 demonstrated the highest b-Gal production. The selected 200 colonies were also electro-transformed into the B. subtilis1A747. Ten blue colonies were selected to measure the b-Gal activities. As shown in Figure 1C, pShuttle-09 again showed the highest b-Gal activity with the activity of 164176300 Miller U/mL at 24 h, which was eight times of that from the P43 promoter system in B. subtilis. To further verify the reporter gene product driven by pShuttle-09, SDS-PAGE analysis of the b-Gal from the pShuttle-09 both in E. coli and B. subtilis was carried out. Coomassie blue staining revealed a distinct band with a molecular mass of approximately 70 kDa, which corresponded to the molecular weight of b-Gal (data not shown). Taken together, these results suggested that pShuttle-09 contained a putative strong promoter from B. licheniformis chromosomal DNA and drove expression of the reporter gene both in E. coli and B. subtilis.

Sequence analysis, predication and characterization of the cloned promoter fragment
To characterize the putative promoter DNA element in pShuttle-09, the inserted fragment was enzyme cut and sequenced. The sequencing result showed that the inserted fragment was 305 bp long. Sequence alignment showed that the inserted sequence in pShuttle-09 belonged to B. licheniformis genome sequence (NC_CP000002.3) with 96% of homology. According to the annotation of B. licheniformis genomic DNA sequences, the inserted sequence contained two fragments (Figure 2A) corresponding to a partial sequence of the coding region of ylyB gene (about 109 bp) and a 59end partial sequences of luxS (about 200 bp).
Sequence analyses ( Figure 2B) also showed truncated luxS gene was fused with bgaB and a typical SD was present 7 bp upstream of the start codon of the luxS gene. It contained typical 235 and 210 elements recognized by a Sigma Factor s A . Therefore, the promoter P luxS might be the control element of bgaB in pShuttle-09. However, further in silico prediction of promoters ( Figure 2B) indicated two conserved prokaryotic promoter regions with high probability scores in the cloned fragment. The first putative promoter (P luxS ) was located in the 59 non-translational region of luxS gene. Interestingly, the second putative promoter was located on two sides of the second Sau3A I-cutting site in pShuttle-09. The 235 element was located within the coding region of ylyB, while  the 210 element was located upstream of the core promoter region of luxS.
In order to determine the effect of the first putative promoter (P luxS ) on the reporter gene expression, a P luxS -i vector was constructed by sub-cloning a 200 bp P luxS fragment into pShuttleF. For facilitating a rigorous comparison, promoters of luxS were also cloned from genomic DNA of B. subtilis, B. pumilus and B. megaterium. The b-Gal production driven by P luxS from B. subtilis, B. pumilus and B. megaterium were about 78%, 63% and 70% of that from B. licheniformis (Figure 3A), respectively. However, the b-Gal production driven by P luxS from B. licheniformis was only about 27% of that from pShuttle-09 in B. subtilis ( Figure 3A). These data strongly suggested that P luxS is not a major promoter for b-Gal expression in pShuttle-09.
Analysis of the second putative promoter showed the 210 region was located at 6 bp in the second Sau3A I-cutting region and 235 region was located at 86 bp in the first Sau3A I-cutting region in pShuttle-09. A partial sequence of the coding sequence of the ylyB gene by itself could not have the promoter activity, since it did not contain complete promoter elements. It appears that the hybrid promoter was unintentionally created in pShuttle-09. Subsequently, this hybrid promoter was sub-cloned to determine whether it can direct the expression of b-Gal or not. As shown in Figure 3B, the b-Gal production driven by the hybrid promoter was about 67% of that from pShuttle-09 in B. subtilis. It seems that both the hybrid promoter and P luxS contributed additively to expression of b-Gal in pShuttle-09.

Improvement of the cloned double promoter through reconstruction of the hybrid promoter element
In order to further improve efficacy of the double promoter, the second Sau3A I-cutting region containing P luxS and another 210 region was assembled with 235 regions of P Apr from B. subtilis, B. pumilus, B. licheniformis and B. amyloliquefaciens, respectively. The corresponding promoters, P LapS , P LapP , P LapL and P LapA , were used to drive the expression of b-Gal. The results ( Figure 3C) indicated that b-Gal activity driven by P LapS was 1.6 fold higher than that from pShuttle-09 after 24 h culture. On the other hand, the production of b-Gal from P LapA , P LapP and P LapL was only about 81%, 32% and 33% of that from pShuttle-09, respectively. Subsequently, P LapS was used for construction of expression vectors.

Construction of expression vector using the improved promoter P LapS
To further exploit its application in B. subtilis, the strong promoter P LapS was sub-cloned as a truncated fragment flanked with engineered Apa I and EcoR I restriction sites, with the EcoR I site close to the SD. The resultant promoter was assembled to E. coli-B. subtilis shuttle vector pGJ103, generating an expression vector pLus-Hyb ( Figure 4A). In order to easily detect and purify expressed protein, another vector pLus-His ( Figure 4B) was constructed through introducing 6-his tag at the end of MCS in pLus-Hyb.
To determine efficiencies of pLus-Hyb and pLus-His, bgaF-1 and bgaF-2 were cloned into pLus-Hyb and pLus-His, respectively, to facilitate b-Gal expression. The production of b-Gal from the pLu-bga reached 2603761037 Miller U/mL (Figure 4C), indicating that the b-Gal was highly expressed. SDS-PAGE analyses of b-Gal from crude extract of B. subtilis harbouring pLu-bga ( Figure 4D) or purified b-Gal from B. subtilis harbouring pLuHis-bga ( Figure 4E) further verified that b-Gal was successfully expressed. Similarly, bioI gene involved in biotin biosynthesis pathway of B. subtilis was successfully expressed by using two similar expression vectors (pLu-biol and pLuHis-biol) in B. subtilis ( Figure 5). Therefore, these results further demonstrated the effectiveness of the promoter systems for protein expression.

Discussion
The major finding from this study is that an artificial double promoter was obtained from B. licheniformis through a promoter trap system. The double promoter contained P luxS and a hybrid promoter and can be used for protein expression in B. subtilis. These results support the notion that it is possible to shuffle, generate and select a strong prokaryotic promoter simultaneously, by applying a promoter trap system to randomly cut DNA fragments from a genome.
The interest in promoters stems from myriad opportunities for controlling gene expression. Our promoter trap system was designed to increase the stringency of screening by employing the coding region of bgaB rather than the bgaB gene with a SD as a reporter. So far, the promoter-less bgaB gene has been used as a reporter in several promoter trap systems [11], [22]. However, others usually used bgaB gene with a SD. In our system, promoter fragments inserted into the upstream of bgaB could not drive expression of b-Gal unless the cloned promoter fragments contain SD. This design might limit the number of recombinant clones which can be screened. Nevertheless, approximately 1000 blue recombinant colonies were observed in this study. The strength of cloned promoters was conveniently determined via coloration caused by production of b-Gal when a recombinant harbouring a promoter fragment was cultured on solid LB supplemented with X-gal. Further quantitative determination of putative promoter strength was easily achieved by measuring the activity of b-Gal.
B. licheniformis genomic DNA was used as a potential promoter source in this study. Recombinant protein expression in B. subtilis can be driven by promoters from B. subtilis or from other bacillus species. To date, several strong exogenous promoters for the B. subtilis system have been isolated and used to construct expression vectors [3], [12], [28]. For example, exogenous promoters P xyl and P spac have been widely used as expression control elements in B. subtilis [12], [28]. Our results support the notion that other bacillus species could be good sources of strong promoters for B. subtilis.
Our results exemplified the potential of a promoter trap system to create a strong hybrid promoter. A hybrid promoter is defined in this study as a promoter consisting of different core elements from two different genes. One promoter identified in this study is a hybrid promoter assembled artificially from a 235 region from the coding sequence of ylyB gene and a 210 element from non-coding sequence of luxS gene. Our hybrid promoter is artificially created and quite different from reported hybrid promoters, which contain elements from two different promoters. It is well known that a hybrid promoter could be far more efficient than either one of the parental promoters [29]. Recently, Kim [30] constructed a hybrid promoter, BJ27UP, from a strong promoter BJ27D88 and a fragment of the tac promoter in order to express B. licheniformis aminopeptidase in B. subtilis and found that the activity of the hybrid promoter was increased approximately threefold in comparison with the tac promoter. Nevertheless, the study on hybrid promoters is limited, presumably due to lack of suitable ways to create hybrid promoters.
The results from this study also demonstrated the capability of a promoter trap system to create a double promoter. Interestingly, both the hybrid promoter and P luxS promoter in our study contributed to transcription of the reporter gene in an additive way. A double promoter in this study is defined as the combination of two promoters which control transcription of same gene. This definition is different from most overlapping promoters often seen in bacteria. Analyzing the database of the E. coli genome, Bendtsen [31] found 14% of the identified 'forward' promoters overlap with a promoter oriented in the opposite direction. These overlapping promoters are arranged in three different ways: (i) each promoter transcribes a gene and one or more regulatory proteins are identified which affect transcription of at least one of the promoters directly; (ii) each promoter transcribes a gene but no regulatory proteins are known to bind the promoter region; and (iii) only one of the promoters transcribe an annotated gene, while the other promoter could transcribe a regulatory antisense RNA or act as a regulatory RNAP binding site which interferes with the promoter transcribing the annotated gene [31]. However, some overlapping promoters do control same gene. For example, Wang and Doi [32] detected two overlapping promoters for the chloramphenicol acetyltransferase gene and these two promoters were recognized by two different RNA polymerase holoenzymes. The authors further proposed that genes expressed during both growth and sporulation may be regulated by tandem overlapping promoters whose transcription initiation points are either identical or very close. The possible benefit of double promoters to increase transcription has been explored in biotechnological application. Most studies have showed that two or more tandem promoters could significantly improve the expression level of heterogeneous genes. Widner [33] found that gene expression was distinctly increased under the control of two or three tandem promoters in contrast to one alone in B. subtilis. Kang [34] also reported that two expression systems constructed by sequential alignment of a constitutive promoter for either a-amylase from B. subtilis NA64 or maltogenic amylase from B. licheniformis downstream of the Hpa II promoter elevated the TSaGT productivity by 11-and 12-fold, respectively, in comparison with the single Hpa II promoter system. Li [35] tried to construct multiple core-tac-promoters (MCPtacs) in tandem and found that integration of the phaCAB genes with the 5 copies of core-tac-promoters resulted in an engineered E. coli that can accumulate 23.7% polyhydroxybutyrate of the cell dry weight in batch cultivation. On the other hand, Wu [36] did not observe improvement of expression, when two tandem-linked promoters were used. Our results support the idea to use double promoters to increase production efficiency of recombinant proteins. Whether the double promoter in this study is targeted by same RNA polymerase or different RNA polymerases is not known at this time and worthy of further investigation.
In conclusion, our data support the concept of using a promoter trap system to discover new promoters. The usefulness of a promoter trap system is likely due to random shuffling and recombination of DNA fragments and adoption of a rapid and high-throughput screening. It is tempting to speculate that more novel promoters could be found if reaction conditions for cutting and ligation could be adjusted to increase chances for shuffling DNA fragments.