Expression of UGA-Containing Mycoplasma Genes in Bacillus subtilis

Abstract

We used Bacillus subtilis to express UGA-containing Mycoplasma genes encoding the P30 adhesin (one UGA) of Mycoplasma pneumoniae and methionine sulfoxide reductase (two UGAs) of Mycoplasma genitalium. Due to natural UGA suppression, these Mycoplasma genes were expressed as full-length protein products, but at relatively low efficiency, in recombinant wild-type Bacillus. The B. subtilis-expressed Mycoplasma proteins appeared as single bands and not as multiple bands compared to expression in recombinant Escherichia coli. Bacillus mutants carrying mutations in the structural gene (prfB) for release factor 2 markedly enhanced the level of readthrough of UGA-containing Mycoplasma genes.

The Mollicutes represent a class of unique cell wall-less procaryotes that includes members of the genus Mycoplasma, a distinct subgroup pathogenic for humans and animals (3, 12). Gene structure and function analyses of mycoplasmas and other genera taxonomically categorized under the Mollicutes are limited by the lack of classical genetic systems and the inability to express cloned genes in native Mycoplasma hosts (7). A fundamental limitation of Mycoplasma gene expression is the unusual codon usage pattern displayed by mycoplasmas (24). Like mitochondria, Mycoplasma species utilize the UGA codon to encode tryptophan (9) rather than to serve as a stop codon. In general, Mycoplasma genes containing one or more UGA codons will not be expressed in commonly used expression systems that strictly adhere to the universal genetic code, due to premature termination. An exception is the very limited readthrough of the UGA stop codon in Escherichia coli and Salmonella enterica serovar Typhimurium (18, 19). Interestingly, although UGA also functions as a termination codon in Bacillus subtilis, the efficiency of UGA readthrough in this gram-positive bacterium is high due to the presence of a tRNA that reads the UGA termination codon as tryptophan (13). Therefore, UGA readthrough in B. subtilis is substantially leakier than that in members of the family Enterobacteriaceae (11). Furthermore, B. subtilis mutants with mutations in the structural gene (prfB) for release factor 2 (RF2) demonstrated increased readthrough levels by overcoming UGA-mediated termination (11). These observations enabled us to use B. subtilis as a host to express specific UGA-containing Mycoplasma genes.

Mycoplasma strains and growth conditions.

Wild-type virulent Mycoplasma pneumoniae strain B9 and Mycoplasma genitalium strain G37 were grown as described earlier (4, 6). B. subtilis strains and plasmids listed in Table 1 were grown in Luria-Bertani (LB) medium at 37°C. Except for B. subtilis PY22, all other Bacillus strains were derivatives of B. subtilis 168 strain BR151. E. coli strain INVαF′ (Invitrogen, Carlsbad, Calif.) was used for the transformation and expression of the TGA-corrected p30 gene. Chromosomal DNA from M. pneumoniae and M. genitalium was isolated as described earlier (4, 22).

TABLE 1.

B. subtilis strains and plasmids

Strain or plasmid	Relevant genotype or plasmid feature(s)	Construction, source, or referencea
PY22	trpC2	23
MLK1143	trpC2 metB10 lys-3 prfB2	11
MLK1144	trpC2 metB10 lys-3 prfB3	11
MLK1145	trpC2 metB10 lys-3 prfB4	11
MLK1147	trpC2 metB10 lys-3 prfB5	11
BS-28	trpC2 pDG28	pDG28→PY22
BS-P30	trpC2 pDG28-30	pDG28-30→PY22
BS-MSR	trpC2 pDG28-MSR	pDG28-MSR→PY22
BS-CP30	trpC2 pDG28-C30 (p30TGA→TGG)	pDG28-C∗30→PY22
BS2-P30	trpC2 metB10 lys-3 prfB2 pDG28-30	pDG28-30→MLK1143
BS3-P30	trpC2 metB10 lys-3 prfB3 pDG28-30	pDG28-30→MLK1144
BS4-P30	trpC2 metB10 lys-3 prfB4 pDG28-30	pDG28-30→MLK1145
BS5-P30	trpC2 metB10 lys-3 prfB5 pDG28-30	pDG28-30→MLK1147
BS2-MSR	trpC2 metB10 lys-3 prfB2 pDG28-MSR	pDG28-MSR→MLK1143
BS3-MSR	trpC2 metB10 lys-3 prfB3 pDG28-MSR	pDG28-MSR→MLK1144
BS4-MSR	trpC2 metB10 lys-3 prfB4 pDG28-MSR	pDG28-MSR→MLK1145
BS5-MSR	trpC2 metB10 lys-3 prfB5 pDG28-MSR	pDG28-MSR→MLK1147
PCR2.1-30	Apr Kmr 864-bp PCR-amplified fragment with p30 in PCR2.1	This study
PCR2.1-MSR	Apr Kmr 570-bp PCR-amplified fragment with msr in PCR2.1	This study
pDG28	Apr EmroriBS PSPAC lacI	23
pDG28-30	Apr EmroriBS PSPAC::p30 lacI	This study
pDG28-MSR	Apr EmroriBS PSPAC::msr lacI	This study
4P3	AprnirB::p30TGA→trpA	John Glassb

^aAn arrow indicates construction of the strain by transformation, and an asterisk indicates TGA→TGG site-directed mutagenized p30. 

^bEli Lilly, Indianapolis, Ind. 

Expression of Mycoplasma genes in B. subtilis and E. coli.

The entire structural gene (p30) of M. pneumoniae adhesin P30 is comprised of an open reading frame of 825 nucleotides, encoding 275 amino acids with a predicted molecular mass of 29,743 Da (4). The single UGA codon that encodes tryptophan is located at amino acid position 16 (4). The p30 gene was amplified using primers listed in Table 2, and the resulting 864-bp DNA fragment was cloned into the PCR2.1 vector (Invitrogen) and transformed into E. coli INVαF′ cells (20). Plasmid PCR2.1-30 was digested with XbaI and SalI. The excised 851-bp fragment was cloned downstream of P_SPAC into the E. coli and B. subtilis shuttle vector pDG28 to generate the recombinant plasmid pDG28-30. Transformation in B. subtilis and E. coli was accomplished as reported earlier (20, 25), and the transformants were selected on LB agar plates containing ampicillin at 100 μg/ml for E. coli and erythromycin at 1 μg/ml for B. subtilis.

TABLE 2.

Primers designed to amplify Mycoplasma genes

Gene	Primer	Sequence (5′-3′)a	Restriction enzyme
p30	MP30F	TCCCTTTCTAGAGTGCATGAAGTTACCACCTC	XbaI
	MP30R	TATCTGTCGACTTGAAGCACTAATTGAAGTTTT	SalI
msr	MGMSRF	ATTAATCTAGATAATATGAAAGAAATTTA	XbaI
	MGMSRR	TAAACATAGTCGACGGATTAATGATTATTGCTA	SalI

^aRestriction sites are indicated by underlining. 

The recombinant B. subtilis wild-type strain PY22 containing the p30 gene under the control of P_SPAC (BS-P30) was grown in the presence or absence of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). Bacterial cells were pelleted and sonicated (40 W for 15 s three times with 45-s intervals), and samples were analyzed by sodium dodecyl sulfate (SDS)–12% polyacrylamide gel electrophoresis for visualization using Coomassie blue staining and Western immunoblotting. No visible differences were observed between the total protein profiles of the induced and uninduced cultures in the Coomassie blue-stained SDS gels (data not shown). The P30 protein, however, was detected by immunoblotting with P30 monoclonal antibody in the induced B. subtilis strain BS-P30 but not in the uninduced strain (Fig. 1A) (5). This monoclonal antibody recognizes an epitope in the middle region of the P30 protein. As expected, P30 was absent in control cells (BS-28) carrying plasmid pDG28 without p30. These data indicated that P30 synthesis was under the control of the inducible P_SPAC promoter and that at least limited readthrough of the p30 gene is possible in B. subtilis. In contrast, no immunologically detectable P30 was induced in recombinant E. coli (data not shown). Apparently, the efficiency of UGA readthrough in members of the Enterobacteriaceae is very low.

FIG. 1.

Expression of M. pneumoniae gene p30 in B. subtilis and E. coli. B. subtilis strains BS-28 (control) and BS-P30 (p30) were cultivated in LB medium and induced with 1 mM IPTG. E. coli strain INVαF′ carrying the 4P3 plasmid (UGA-corrected p30) was grown in LB medium under anaerobic conditions for the expression of P30 from the nirB promoter. Sonicated protein extracts (120 μg) were fractionated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting with P30 antibodies. (A) Immunoblot analysis of UGA readthrough of p30 gene in wild-type B. subtilis PY22. Lanes: 1, BS-28 with no IPTG; 2, BS-28 with 1 mM IPTG; 3, BS-P30 with no IPTG; 4, BS-P30 with 1 mM IPTG. (B) Immunoblot analysis of multiple P30-related proteins from the expression of the UGA-corrected p30 gene in E. coli. Lanes: 1, M. pneumoniae; 2, E. coli (4P3, UGA corrected); 3, E. coli (pDG28-30, not UGA corrected).

It is noteworthy that B. subtilis-expressed recombinant P30 protein appeared as a single band. Mycoplasma genes previously expressed in E. coli generated multiple protein bands (16). For example, the Spiroplasma gene encoding spiralin contains no UGA codons but generates multiple recombinant proteins in E. coli (15). Expression of multiple proteins exceeding the coding capacity of a cloned genomic fragment from Mycoplasma capricolum has also been described for E. coli (1). Differences in UGA codon usage and initiation sites for translation in E. coli may account for these variations in protein size (16). In order to determine whether p30 underwent a similar mechanism of recombinant expression in E. coli, we used the TGA-corrected p30-carrying plasmid construct designated 4P3. Expression of the TGA-corrected p30 gene under the nirB promoter in E. coli (17) resulted in the appearance of multiple P30-related proteins. One major protein band had the mobility of authentic P30 (30 kDa), while a second major band had the mobility of a larger protein exceeding the p30 coding capacity (45 kDa) (Fig. 1B).

In order to determine the efficiency of UGA readthrough of p30 in B. subtilis (Fig. 1A), we compared the level of expression of the wild-type p30 with that of a p30 gene in which the TGA was changed to TGG by site-directed mutagenesis. A recombinant plasmid carrying the TGA-corrected p30 gene (strain BS-CP30 [Table 1]) was constructed and transformed into B. subtilis. When the UGA-corrected p30 was expressed in Bacillus, the protein was overexpressed, similar to the level in E. coli, and without multiple proteins. Quantitation of translational products by immunoblot analysis was performed on a Macintosh computer using the public-domain NIH Image program (developed at the U.S. National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image/). To calculate the level at which the internal UGA codon of p30 was read as tryptophan, we plotted a standard graph of P30 protein concentrations (the values generated by the image analyzer on the immunoblot) versus total cell protein (the amount of IPTG-induced BS-CP30 protein loaded on SDS-polyacrylamide gels). Using 100 μg of IPTG-induced BS-CP30 protein concentration as 100%, we calculated the level of P30 expression in test samples by immunoblotting. By this method, the level of UGA suppression for the wild-type strain (BS-P30) that carried the UGA-containing p30 gene was calculated as 6%. This is similar to the readthrough levels observed with cat genes containing UGA codons in B. subtilis (13).

To further determine the capacity of B. subtilis to translate through UGA codons, we selected the methionine sulfoxide reductase (msr) structural gene of M. genitalium with two UGA codons as a test gene. The Mycoplasma gene msr has an open reading frame of 471 nucleotides, which encodes a protein of 157 amino acids, with a calculated molecular mass of 18,414 Da. The two UGA codons occur at amino acid positions 12 and 99 (8). Recently, we generated antibodies reactive against the MSR (methionine sulfoxide reductase) protein of M. pneumoniae (S. Dhandayuthapani and J. B. Baseman, unpublished results) that cross-react with the M. genitalium MSR protein. The M. genitalium DNA fragment encoding msr was amplified using primers outlined in Table 2 and cloned into the pDG28 expression vector to create pDG28-MSR. Recombinant B. subtilis cells carrying the pDG28-MSR (BS-MSR) were grown to mid-log phase and induced with 1 mM IPTG. Immunoblot analysis revealed the synthesis of a full-length msr gene product (Fig. 2B).

FIG. 2.

Comparison of UGA readthrough of Mycoplasma genes p30 and msr among the different prfB mutant strains of B. subtilis. The B. subtilis strains were cultivated in LB medium with 0.5% glucose and induced with IPTG (1 mM). (A) Immunoblot analysis of the P30 protein induction. Lanes: 1, BS5-P30 (prfB5); 2, BS3-P30 (prfB3); 3, BS2-P30 (prfB2); 4, BS4-P30 (prfB4). (B) Immunoblot analysis of MSR protein induction. Lanes: 1, BS5-MSR (prfB5) uninduced; 2, BS5-MSR (prfB5); 3, BS2-MSR (prfB2); 4, BS4-MSR (pfrB4); 5, BS3-MSR (prfB3); 6, BS-MSR (wild type); 7, BS-MSR (wild type) uninduced.

Expression of Mycoplasma genes in prfB mutants of B. subtilis.

Mutations in RF2 genes (prfB) fail to terminate translation efficiently and allow misreading of UGA (2). The combination of tRNA UGA suppression and RF2 mutation has moderately increased UGA readthrough in E. coli. However, the expression of Mycoplasma genes under these conditions was limited due to the synthesis of multiple-sized, related Mycoplasma proteins for unknown reasons (21). Since B. subtilis permits readthrough of UGA, we used B. subtilis RF2 mutants (prfB) in an attempt to increase the level of UGA readthrough of Mycoplasma genes. A recent study describing prfB mutants of B. subtilis demonstrated that such mutants could suppress a catA86-TGA mutation between 19 and 54%, compared to the 6% level detected in wild-type Bacillus strains (11).

Specific B. subtilis prfB mutant strains with impaired RF2 function and enhanced UGA suppression were selected (11). Plasmids pDG28-30 and pDG28-MSR were transformed individually into strains with novel prfB alleles (mutations prfB2 and prfB4 affect ribosome-binding site, mutation prfB5 affects frameshifting, and mutation prfB3 has an unknown activity due to an alteration from Glu to Lys at amino acid residue 2). The recombinant RF2 strains were then induced with IPTG. The level of expression of p30 was increased two- to ninefold in these mutants compared to wild-type B. subtilis (Fig. 2A and Table 3). The maximum level of UGA suppression was observed with the BS5-P30 (prfB5) strain. Bacillus strains BS2-P30 (prfB2), BS4-P30 (prfB4), and BS3-P30 (prfB3) permitted readthrough of p30 in decreasing order (Table 3). A similar pattern of Mycoplasma protein expression was observed with msr readthrough (Fig. 2B). In this case, the level of msr expression was increased from two- to sevenfold in the RF2 mutants compared to wild-type B. subtilis (data not shown). Interestingly, no truncated MSR-related protein of 98 amino acids was observed, suggesting that the truncated peptide was not stable. A possible explanation for the synthesis of full-length MSR by B. subtilis, with no truncated peptides, could be the presence of adenine next to the UGA (at amino acid 99) of msr. It has been observed that the nucleotide 3′ to the UGA triplet increases UGA suppression (increased readthrough) in E. coli in the order A > G > C > U (14). In p30, the nucleotide 3′ to the UGA triplet (i.e., at amino acid 16) is adenine; in msr, the nucleotide next to amino acid 12 is guanine (i.e., UGAG) and at amino acid 99 is UGAA. Thus, the synthesis of intact P30 and MSR proteins is favored.

TABLE 3.

Relative levels of expression of Mycoplasma p30 gene (one TGA) in prfB mutant strains of Bacillus

Bacillus strain	Level of UGA suppression (%)a
BS2-P30	27
BS3-P30	12
BS4-P30	21
BS5-P30	54

^aThe level of UGA suppression in wild-type B. subtilis for p30 was calculated as 6%. 

Again, the expected level of msr expression in prfB was consistent with p30 expression for all Bacillus mutants, except for BS5-(prfB5), in which expression of p30 was significantly higher. The difference in readthrough of p30 and msr could be due to the influence of factors like codon bias, the position of UGA in the coding sequence, and the size of the open reading frame. It may be possible to increase UGA readthrough more than the level observed (54%), although this may be detrimental to cell growth or even result in lethality (10). However, further enhancement of Mycoplasma gene expression might be accomplished by cloning selected UGA-containing Mycoplasma genes under a strong Bacillus promoter to heighten mRNA levels. Based on our study, we predict that Mycoplasma open reading frames containing up to three TGA codons can be readily expressed in RF2 mutants of B. subtilis, suggesting that the obstacle of readthrough of Mycoplasma genes containing UGAs can be overcome using this expression host.