Engineering of a Bacillus subtilis Strain with Adjustable Levels of Intracellular Biotin for Secretory Production of Functional Streptavidin

Abstract

Streptavidin is a biotin-binding protein which has been widely used in many in vitro and in vivo applications. Because of the ease of protein recovery and availability of protease-deficient strains, the Bacillus subtilis expression-secretion system is an attractive system for streptavidin production. However, attempts to produce streptavidin using B. subtilis face the problem that cells overproducing large amounts of streptavidin suffer poor growth, presumably because of biotin deficiency. This problem cannot be solved by supplementing biotin to the culture medium, as this will saturate the biotin binding sites in streptavidin. We addressed this dilemma by engineering a B. subtilis strain (WB800BIO) which overproduces intracellular biotin. The strategy involves replacing the natural regulatory region of the B. subtilis chromosomal biotin biosynthetic operon (bioWAFDBIorf2) with an engineered one consisting of the B. subtilis groE promoter and gluconate operator. Biotin production in WB800BIO is induced by gluconate, and the level of biotin produced can be adjusted by varying the gluconate dosage. A level of gluconate was selected to allow enhanced intracellular production of biotin without getting it released into the culture medium. WB800BIO, when used as a host for streptavidin production, grows healthily in a biotin-limited medium and produces large amounts (35 to 50 mg/liter) of streptavidin, with over 80% of its biotin binding sites available for future applications.

Because of the unusually tight binding (K_d ≈10⁻¹³ to 10⁻¹⁶ M) of biotin to streptavidin (SAV) and avidin, homotetrameric proteins with a single biotin binding site per subunit, these molecules have been widely used as capturing molecules to detect, locate, and immobilize biotinylated molecules in many applications, both in vitro and in vivo (3, 8, 9, 31). In the recent interest to develop protein and antibody arrays for high-throughput genomics and proteomics studies, SAV and avidin are important elements in the generation of high-density biochips (2, 6). Besides these in vitro applications, SAV and avidin are also applied in vivo for tumor targeting and imaging, drug delivery, and localization of infection sites (4, 24, 28). Relative to avidin (a glycoprotein with a pI of 10), SAV (a nonglycosylated protein with a pI around 6) shows lower levels of nonspecific bindings to cells or matrices under in vitro conditions and is a preferred choice for these applications. An efficient system to produce functional SAV would be desirable. Currently, SAV can be produced as a soluble secretory protein from its natural host, Streptomyces avidinii (1, 7), or from a recombinant host, Bacillus subtilis (22, 31a). It is also commonly produced from Escherichia coli intracellularly as a soluble (13) or insoluble (29) protein. For production in the insoluble form, any bound biotin in SAV can be eliminated during the in vitro refolding process. On the other hand, in the production of soluble SAV, trapping of biotin during biosynthesis of SAV can effectively reduce the number of free biotin binding sites available (13).

Our study (31a) of secretory production of SAV from B. subtilis indicates that cells overproducing SAV are suffering reduced growth, presumably because of the depletion of biotin by the SAV produced. Although supplementation of biotin in the culture medium can improve cell growth and subsequently SAV production yield, SAV thus produced will be saturated with biotin and is no longer fully functional as a biotin binding protein The requirement of the denaturation/renaturation cycle to remove the tightly bound biotins from SAV produced under this condition will make secretory production of soluble SAV no longer an attractive means to produce SAV. To overcome this problem, we constructed an engineered B. subtilis strain (WB800BIO) which has an adjustable intracellular biotin level. In this system, a higher level of biotin can be synthesized intracellularly to sustain the physiological needs of the cells without having any significant amounts of biotin released to the medium. To develop an adjustable promoter, an engineered promoter system was constructed with the strong B. subtilis groE promoter (20) fused to the B. subtilis gluconate operator sequence (12). The replacement of the natural regulatory sequence (5) in the B. subtilis chromosomal biotin biosynthetic operon (bioWAFDBIorf2) with this engineered promoter system allows the intracellular biotin level to be adjustable with gluconate. We describe in this paper the successful secretory production of SAV up to 50 mg/liter of culture. These SAVs have mostly free biotin binding sites.

MATERIALS AND METHODS

Construction of an engineered promoter system.

To facilitate construction, the promoter-operator cassette was first assembled in E. coli. The T₁T₂ transcription terminator complex from the E. coli rrnB operon (26) was amplified by PCR using E. coli genomic DNA as template and synthetic oligonucleotides 5′-GGGCATGCGATATCAGGCATCAAATAAAACGAAAG-3′ as the forward primer and 5′-GGGTCTAGAGTTAACTAGATATGACGACAGGAAG-3′ as the backward primer. These primers have sequences corresponding to nucleotides 6609 to 6630 and 6838 to 6856, respectively, in the E. coli rrnB operon (GenBank accession number J01695). The PCR-amplified 270-bp fragment has a SmaI half site and intact SphI and EcoRV sites at the 5′ end. It also has intact HpaI and XbaI sites at the 3′ end. This fragment was digested with XbaI and inserted into SmaI/XbaI-digested E. coli Bluescript vector pBS (Stratagene) to form pBST. A synthetic groE promoter-gluconate operator was created to control biotin expression. Using pGroESL, a plasmid which carries the B. subtilis groE promoter (20) as template, the promoter sequence was amplified by PCR with the forward primer 5′-GGGAGCTCGTTTAAACGTGAAAAAGCTAACGGAAAAG-3′ and the backward primer 5′-CTGGTACCTTAATTAAGAGTATACTTGTATACAAGTATAATAAAGAATCTCCCTTCCAATTTC-3′. Sequences highlighted in bold are derived from nucleotides 184 to 204 and 258 to 281 in the B. subtilis groE operon (20), respectively. Since the backward primer carries the nucleotide sequence of the B. subtilis gluconate operator (sequence underlined), the amplified product has the gluconate operator fused downstream to the groE promoter. This 150-bp product was digested by SstI/KpnI and inserted into SstI/KpnI-digested pBST. The resulting plasmid was designated pBSPT.

Construction of pKSWT.

An integration vector pKSWT (Fig. 1) was constructed to modulate the bioW operon expression system in B. subtilis. This vector is a derivative of pK184 (16), a low-copy-number E. coli plasmid that cannot replicate in B. subtilis. The cassette containing the groE promoter-gluconate operator-T₁T₂ complex in pBSPT was transferred as an SstI/XbalI insert into pK184 to form pKPT (Fig. 1). The first gene, bioW, in the B. subtilis biotin biosynthetic operon bioWAFDBIorf2 (5) was placed under the control of the synthetic groE promoter-gluconate operator in the following manner. First, the downstream (bioW_D) and upstream (bioW_U) sequences of bioW were amplified by PCR using B. subtilis genomic DNA as template. bioW_D was amplified with the forward primer 5′-GTGGTACCGTTCTTCAGTTATCAGTGAAAG-3′ (2089 to 2110) and the backward primer 5′-GAGATATCGGACATTAAGCCAAACCGATG-3′ (3030 to 3059). bioW_U was amplified with the forward primer 5′-GTGAATTCGATTGCTTGGTGATTTGCCTG-3′ (1044 to 1064) and the backward primer 5′-GTGAGCTCGGATCGATTTATGGCAGTTGG-3′ (1974 to 1994). The numbers shown in parentheses represent the locations of these sequences in the bioW operon sequence (GenBank accession number U51868). The 980-bp bioW_D fragment containing the bioW ribosomal binding site at the 5′ end was digested by KpnI/EcoRV and inserted into pKPT to a position between the promoter-operator region and the T₁T₂ terminator. This step was followed by insertion of the 963-bp EcoRI/SstI-digested bioW_U sequence 5′ to the promoter-operator region to form pKWT. Finally, a 1.17-kb PstI fragment carrying the spectinomycin resistance marker was obtained from pGMS57 (19), blunt ended by T4 DNA polymerase (New England BioLabs Canada), and inserted into SstI-digested, end-repaired pKWT. The resulting plasmid, designated pKSWT (Fig. 1), confers spectinomycin resistance to the E. coli DH5α host, which can be selected on plates containing 100 μg of spectinomycin/ml (Sigma Canada).

FIG. 1.

Construction of the integration vector pKSWT for modulating biotin synthesis in B. subtilis. First, the B. subtilis groE promoter (P_groE) and gluconate operator (O_gnt) were inserted in pK184 containing the E. coli rrnB transcription terminator (T₁T₂) to form pKPT. Then, the downstream (bioW_D) and upstream (bioW_u) sequences of B. subtilis bioW were inserted sequentially to form pKWT. Finally, the spectinomycin resistance marker (Sp^r) was inserted as shown to generate the integration vector pKSWT.

Construction of B. subtilis WB800BIO.

To replace the regulatory region of the wild-type bioW operon by bioW with the engineered groE promoter-gluconate operator, pKSWT was linearized by EcoRI and transformed to B. subtilis WB800, an eight-protease-deficient strain (32). Transformants were selected on tryptose blood agar base plates (TBAB; Difco) containing, per milliliter, 170 μg of spectinomycin. The successful replacement of the regulatory sequence of the chromosomal bioW was confirmed by colony PCR (30) using Taq DNA polymerase (Amersham Pharmacia Biotech), and the resulting engineered B. subtilis strain was designated WB800BIO. To use WB800BIO as a host to produce SAV, pSSAV-Tcry (Wu et al., submitted), the plasmid which carries a synthetic sav gene, was transformed to WB800BIO to generate WB800BIO(pSSAV-Tcry).

Construction of an integration vector carrying lacZ under the control of the groE promoter-gluconate operator.

The sequence containing the groE promoter-gluconate operator in pBSPT was transferred as an EcoRI/SmaI insert to EcoRI/SnaBI-digested pDH32 M (17). pDH32 M contains a promoterless lacZ that can be integrated to the amyE locus. The resulting plasmid, pDH32MEP, was linearized by PstI and transformed to B. subtilis 168. Transformants were selected on TBAB plates containing, per milliliter, 5 μg of chloramphenicol. Cm^r transformants were further confirmed by their inability to digest starch on TBAB plates containing 1% (wt/vol) soluble starch (Sigma Canada) (33). The resulting clone was designated WB1200. B. subtilis 168 was also transformed with PstI-linearized pDH32 M to give WB1201, which served as the negative control.

Cell growth and production of SAV.

B. subtilis harboring pSSAV-Tcry was cultivated in CA medium, a semidefined medium which was based on a previously reported Casamino Acids medium (22) but with minor modifications. This medium contained 1× Castenholz medium (22), 25 mM potassium phosphate, 2% vitamin-free Casamino Acids (BDH), 1% glycerol, 0.01% yeast extract, kanamycin (10 μg per ml) and, per 100 ml, 0.1 gm of asparagine and 5 mg each of lysine, methionine, histidine, and tryptophan. The final pH of the medium was adjusted to 7.5. For medium containing gluconate, sterilized concentrated sodium gluconate (Sigma Canada) was added to the medium after autoclaving. This medium is considered to be a semidefined medium because it contains low levels of yeast extract and Casamino Acids. Cells were cultivated in a shake flask at 32°C. Cell growth was monitored using a Klett-Summerson photoelectric colorimeter (Klett Manufacturing Co.).

Biotin assays.

The intracellular amount of biotin in B. subtilis was estimated using a competition enzyme-linked immunosorbent assay (ELISA) method described below. B. subtilis cells cultivated in CA medium were harvested by centrifugation at 10,000 × g for 5 min, washed once with phosphate-buffered saline (PBS; 0.1 M sodium phosphate, 0.15 M NaCl, pH 7.2), and lysed with a French press. The cell lysate was centrifuged at 15,000 × g for 20 min and the soluble fraction was collected as the supernatant. Competition ELISAs followed a standard ELISA protocol. The wells of a Reacti-bind maleic anhydride-activated polystyrene strip plate (Pierce) were coated with 300 ng of recombinant SAV (Roche Molecular Biochemicals) in PBS according to the manufacturer's instructions. Nonspecific sites were blocked by adding 200 μl of 0.5% bovine serum albumin in PBST (PBS containing 0.05% Tween 20) to each well. After 1 h at room temperature, wells were washed with PBST and incubated for another hour with a mixture containing an appropriate volume of the cell lysate and 800 ng of biotinylated horseradish peroxidase (HRP; Pierce) per well. After washing with PBST, the amount of bound biotinylated HRP was determined with the use of a substrate designated “1-step slow TMB (3,3′,5,5′-tetramethylbenzidine)” from Pierce as the color development reagent. End-point readings were taken at 450 nm using an ELISA plate reader (CERES 900; Bio-Tek Instruments, Inc.). d-Biotin (0 to 1,000 ng in an appropriate volume of PBS) was used in place of the cell lysate to generate a biotin calibration plot.

Determination of number of biotin binding sites on SAV produced by WB800BIO(pSSAV-Tcry).

SAV was purified from the culture supernatant of WB800BIO(pSSAV-Tcry) grown in CA medium, using a cation-exchange column (Macro-Prep High S; Bio-Rad) as described elsewhere (Wu et al., submitted). The number of biotin binding sites on the purified tetrameric SAV was determined using a competition ELISA method similar to the one established for the biotin assay but with modifications. Wells were coated with 200 ng of SAV (Roche Molecular Biochemicals). After blocking with bovine serum albumin, wells were incubated with a mixture containing 25 ng of biotinylated HRP (Pierce) and different amounts of the purified SAV or commercial SAV from Roche Molecular Biochemicals.

Other methods.

All PCRs (except colony PCR) were carried out using Vent DNA polymerase (New England BioLabs Canada). The amplified products were confirmed by oligonucleotide sequencing, based on the dideoxy method, to be free of PCR errors. B. subtilis colony PCR was performed as described previously (30) with Taq polymerase. Protein gel electrophoresis followed a standard protocol based on the Laemmli system. Purified SAV was quantified spectrophotometrically at 280 nm using a molar extinction coefficient of 41,820 M⁻¹ cm⁻¹ (14). The activity of β-galactosidase in B. subtilis WB1200 and WB1201 was determined by using o-nitrophenyl-β-d-galactopyranoside (ONPG) (Sigma, Canada) as the substrate, as described previously (23, 33). All the β-galactosidase assays (see Fig. 2 and 3) were carried out in at least three independent runs. The standard deviation in activity was consistently less than 10%.

FIG. 2.

Action of the groE promoter-gluconate operator in B. subtilis. (A) Action of the promoter. WB1200 harboring the engineered groE promoter-gluconate operator (•) and WB1201 (negative control) (▴) were cultured in superrich medium containing 0.2% sodium gluconate. (B) Action of the operator. WB1200 was cultured in superrich medium with 0.2% (•) or without (▴) sodium gluconate. At different hours of culture at 37°C, the cells were harvested and disrupted by French press. β-Galactosidase activity in the cell lysates was determined using ONPG as substrate. One Miller unit = (A₄₂₀ × 1,000)/(reaction time [minutes] × A₅₉₅) (23).

FIG. 3.

Effect of gluconate dosage on the expression of the engineered groE promoter-gluconate operator system in WB1200. WB1200 was cultured at 37°C in superrich medium containing different amounts of sodium gluconate. At 4 h of culture, cells were harvested and disrupted by French press. β-Galactosidase activity in the cell lysates was determined using ONPG as substrate. Miller units were as defined in the legend for Fig. 2.

RESULTS

Design of an engineered promoter-operator system with adjustable expression levels.

A strong promoter with its expression adjustable would be an ideal system to control the biosynthesis of intracellular biotin in B. subtilis. Since the B. subtilis groE promoter is a relatively strong promoter and gluconate is an affordable inducer, a combination of the groE promoter with the gluconate operator should be an attractive candidate for our study. In this system, a sequence corresponding to −59 to −9 of the groE promoter (20) was fused immediately upstream of an 18-bp gluconate operator sequence (12). Since the two nucleotides corresponding to −8 and −7 in the groE promoter are actually identical to the first two nucleotides of the gluconate operator, the resulting promoter contains exactly the same hexameric sequence for the −10 region as the original groE promoter. Insertion of the wild-type B. subtilis groE promoter in high-copy-number E. coli plasmids was observed to frequently induce plasmid rearrangement. This instability was likely to be caused by the high promoter strength of the groE promoter, since insertion of the E. coli rrnB T₁T₂ terminator complex downstream of the promoter could greatly stabilize the plasmid. Therefore, terminator-containing vectors were used for the cloning of this engineered regulatory sequence (Fig. 1).

Gluconate-mediated induction of the engineered promoter system.

Addition of gluconate is expected to increase expression from the engineered promoter system. To examine the induction effect, the engineered regulatory sequence was inserted into an integration vector, pDH32 M (17), which carries a promoterless lacZ as the reporter. The resulting plasmid, pDH32MEP, was integrated into the chromosomal amyE locus of B. subtilis 168 to generate WB1200. Figure 2A compares the β-galactosidase activity in the cell lysate from WB1200 with that of the negative control, WB1201. Gluconate (0.2%) was supplemented in the culture medium in both cases. Near zero activity of β-galactosidase was detected with WB1201 throughout the culture period. In contrast, the activity in WB1200 increased with time of culture, reached a peak around 4 to 5 h, and declined thereafter. When measured at 4 h, the activity of β-galactosidase in WB1200 was 36-fold that of the negative control. In a parallel study, action of the gluconate operator was examined, using the cell lysate of WB1200 grown in the absence or presence of 0.2% gluconate in the culture medium. Figure 2B shows that a basal, albeit low level of promoter activity existed in the absence of gluconate. Addition of gluconate induced a higher activity throughout the culture period. At 4 h, the activity was sixfold that of the basal level. These results indicate that the engineered promoter-operator system functions properly in B. subtilis.

Adjustable expression from the engineered groE promoter.

It is vital that the expression level of the engineered promoter system should be adjustable by varying the amounts of gluconate. To examine the effect of gluconate dosage on the expression level of lacZ, WB1200 was cultivated for 4 h with 0 to 0.5% gluconate in the culture medium. Figure 3 shows that the promoter activity did vary with the level of gluconate present. The activity increased with increasing amount of gluconate and reached an optimum at 0.3% gluconate. This information allows one to fine-tune the level of expression of the target gene by varying the amount of gluconate in the culture medium.

WB800BIO produces higher levels of intracellular biotin in the presence of 0.2% gluconate.

In B. subtilis, important genes involved in biotin biosynthesis are organized in a seven-gene operon, bioWAFDBIorf2 (5). Since the first gene of this operon is bioW, this operon is designated the bioW operon. The engineered promoter system which we have shown to be properly functioning in B. subtilis was applied to replace the natural regulatory region of the B. subtilis bioW operon. An engineered strain, WB800BIO, was constructed in the genetic background of WB800 by using pKSWT. WB800BIO is expected to produce intracellular biotin at higher levels in the presence of gluconate. To validate this assumption, the intracellular level of biotin was estimated by a competition ELISA study in which free biotin competes with biotinylated HRP for the limiting amount of SAV on the wells of a microtiter plate. As more biotin is present, less biotinylated HRP will be retained by SAV. Thus, an assay of the bound HRP activity allows one to estimate the amount of biotin present in the sample. Using this method and a calibration curve established with known amounts of free biotin as the competitor for biotinylated HRP, we estimated the amounts of intracellular biotin produced by WB800 and WB800BIO. The lowest detection limit was 3 μg/liter. We could not observe any detectable levels of intracellular biotin in the cell lysate of WB800, whether gluconate was present in the culture medium or not. As well, no intracellular biotin was detected with WB800BIO grown in the absence of gluconate. However, when given 0.2% gluconate in the growth medium, WB800BIO accumulated 3.5 μg/liter of biotin intracellularly at 5 h and 5.5 μg/liter at 9 h of culture.

WB800BIO(pSSAV-Tcry) grows healthily in biotin-depleted medium in the presence of gluconate.

B. subtilis cells overproducing SAV tend to grow very slowly and lyse occasionally, even in superrich medium. Supplementation of biotin in the medium could often lead to much healthier growth, suggesting that the depletion of biotin in the culture medium by SAV secreted by the cells is limiting the normal cell growth. As WB800BIO can increase the intracellular biotin content if given gluconate, we wonder if WB800BIO(pSSAV-Tcry) could grow better in a semidefined medium containing gluconate, thus allowing us to obtain high-quality SAV which has its biotin binding sites fully available for future applications. Figure 4 compares the growth profiles of WB800(pSSAV-Tcry) and WB800BIO(pSSAV-Tcry) in CA medium containing vitamin-free Casamino Acids. WB800(pSSAV-Tcry) grew very slowly in this medium. Typically, the growth had a very long lag phase and the cells were still in the early stage of growth at 20 h of culture. Addition of biotin (40 μM) to the culture medium enhanced the growth dramatically, suggesting that biotin deficiency is the limiting factor behind the poor growth. WB800BIO(pSSAV-Tcry) grew very poorly in the absence of gluconate; its growth was even worse than WB800(pSSAV-Tcry) with prolonged culture (>25 h). However, in the presence of gluconate, it could grow even faster than WB800(pSSAV-Tcry) cultured in medium containing 40 μM biotin, and it could maintain a high-cell-density, long stationary phase.

FIG. 4.

Growth of WB800 and WB800BIO in a semidefined medium. WB800 and WB800BIO were cultured in CA medium containing vitamin-free Casamino Acids (see Materials and Methods for composition of CA medium). □, WB800 with no biotin added to the medium; □, WB800 with 40 μM biotin included in the medium; ▪, WB800BIO with no biotin and no sodium gluconate included in the medium; •, WB800BIO with no biotin but 0.2% sodium gluconate included in the medium. Three independent runs of these cell cultures were carried out and the standard deviation for each growth curve was within the 10% range.

Yields of SAV secreted by WB800, WB800BIO, and WB700SPO.

Yields of SAV secreted by WB800(pSSAV-Tcry), WB800BIO(pSSAV-Tcry), and WB700SPO(pSSAV-Tcry) cultured in CA medium containing vitamin-free Casamino Acids and gluconate were compared (Fig. 5). WB700SPO(pSSAV-Tcry) is a seven-protease-deficient sporulation mutant that produces more SAV than WB700 (34). As the cultures grew at different rates, the SAV yields were studied at the early stationary phase, which took place at different times for the different cultures. The Coomassie-stained sodium dodecyl sulfate (SDS)-polyacrylamide gel showed that the yields of SAV are similar for WB800(pSSAV-Tcry) and WB700SPO(pSSAV-Tcry). WB800BIO(pSSAV-Tcry), on the other hand, has superior performance to the other two. As we have not been able to detect any biotin content in the regular Casamino Acids but cell growth in medium containing regular Casamino Acids is much faster than that using the vitamin-free Casamino Acids, we also examined the production yields of SAV from CA medium containing the regular Casamino Acids. The amounts of SAV secreted were estimated by quantifying the SAV band intensities on a Coomassie-stained gel using the Fuji bio-imaging analyzer system in reference to a calibration curve generated with known amounts of SAV. At 15 to 20 h of culture, the yield of SAV for the various hosts was estimated to be 20 to 25 mg/liter for both WB700SPO and WB800 and 35 to 50 mg/liter for WB800BIO. Hence, WB800BIO(pSSAV-Tcry) remains the superior performer of the three.

FIG. 5.

SAV produced by various B. subtilis hosts. Cells were cultured in CA medium containing vitamin-free Casamino Acids and harvested at the early stationary phase. SAV in the culture supernatant was analyzed on a 12% polyacrylamide gel containing SDS and stained with Coomassie blue. M, molecular weight marker; lane 1, WB800BIO(pSSAV-Tcry); lane 2, WB800(pSSAV-Tcry); lane 3, WB700SPO(pSSAV-Tcry); lane 4, WB800(pWB980) (negative control). The SAV band is marked by an arrow.

WB800BIO(pSSAV-Tcry) leaked little biotin to the culture medium.

A concern one has regarding the use of the biotin-overproducing strain WB800BIO is that it may produce so much biotin that biotin can get released to the medium and bind to the secreted SAV. To address this concern, we took two approaches. First, we quantified the biotin content in the culture medium before and after the cultivation of WB800BIO(pSSAV-Tcry) in the presence of 0.2% gluconate. No biotin could be detected in the medium for all these samples, including samples from WB800BIO(pSSAV-Tcry) cultivated for 0, 15, and 30 h. Second, we checked the number of biotin binding sites for SAV produced by WB800BIO(pSSAV-Tcry). A competition ELISA in which reference SAV (from Roche Molecular Biochemicals) competed with SAV purified from WB800BIO(pSSAV-Tcry) for a limiting amount of biotinylated HRP was used for this purpose (Fig. 6). The reference SAV has been reported to contain approximately four biotin binding sites per SAV tetramer (10). SAV from WB800BIO(pSSAV-Tcry) cultured in CA medium containing either vitamin-free or regular Casamino Acids was estimated to have around 3.3 biotin binding sites per tetramer (Fig. 6). In comparison, the number of biotin binding sites per tetramer of SAV produced by WB800(pSSAV-Tcry) is 3.2 (Wu et al., submitted). That most of the SAV secreted by WB800BIO(pSSAV-Tcry) is not bound by biotin is further reflected indirectly by a Western blotting study (Fig. 7). SAV produced by WB800(pSSAV-Tcry) (lane 1) and WB800BIO(pSSAV-Tcry) (lane 2) cultured in CA medium existed as mostly monomeric when boiled. In contrast, some SAV molecules produced by WB800(pSSAV-Tcry), grown in the presence of added biotin, remained as the tetrameric form even when the sample was boiled (lane 3). Thus, SAV produced by WB800BIO(pSSAV-Tcry) is not significantly bound to any biotin present extracellularly.

FIG. 6.

Determination of biotin binding sites of SAV purified from WB800BIO(pSSAV-Tcry) by competitive ELISA. SAV was purified from WB800BIO(pSSAV-Tcry) cultured in CA medium containing vitamin-free Casamino Acids. Bound HRP activity was determined using a one-step slow-TMB ELISA (Pierce) as the substrate. The number of biotin binding sites on SAV was estimated at 50% of bound-HRP activity using SAV from Roche Molecular Biochemicals (for which the number of biotin binding sites is 4) as the reference. SAV purified from WB800BIO(pSSAV-Tcry) cultured in CA medium containing regular Casamino Acids showed an almost identical competition profile as that observed for SAV purified from CA medium containing vitamin-free Casamino Acids. For clarity, data for SAV purified from CA medium containing regular Casamino Acids are not presented. •, SAV purified from WB800BIO(pSSAV-Tcry); ▴, reference SAV from Roche Molecular Biochemicals.

FIG. 7.

Western blot analysis showing SAVs produced by hosts WB800 and WB800BIO. Cells were cultured in CA medium containing vitamin-free Casamino Acids. SAVs in the culture supernatants were analyzed on a 12% polyacrylamide gel containing SDS, and the blot was probed against SAV-specific antiserum. All samples were boiled prior to loading onto the gel. M, molecular weight marker; lane1, WB800BIO(pSSAV-Tcry); lane 2, WB800(pSSAV-Tcry). No biotin was included in the medium for lanes 1 and 2. Lane 3, WB800(pSSAV-Tcry) cultured with biotin added to the medium.

DISCUSSION

Supplementation of biotin in the culture medium significantly promotes the growth of the SAV-producing WB800(pSSAV-Tcry). This reflects that overproduction of SAV can deplete biotin to such an extent that biotinylated enzymes are not present in sufficient quantities to carry out metabolic reactions. Two biotinylated proteins are found in B. subtilis (21). One is the biotin carboxyl carrier protein subunit of the acetyl coenzyme A carboxylase, which catalyzes the first committed step for fatty acid biosynthesis. The second one is suggested to be pyruvate carboxylase. As both enzymes are important for many metabolic pathways in B. subtilis, biotin deficiency can be disastrous to cell survival. However, to produce SAV with free biotin binding sites, the use of defined media containing minimal levels of biotin is essential. To overcome this dilemma, we hypothesize that an increase in the intracellular level of biotin alone should improve the cell growth, and a higher production yield of SAV should follow. As the SAV is secretory, most of it should be biotin-free, as no external biotin is administered. Indeed, the results of this study show that our hypothesis is correct. Using CA medium containing regular Casamino Acids, WB800BIO(pSSAV-Tcry) consistently produced SAV at higher levels (35 to 50 mg/liter) than WB800(pSSAV-Tcry) (20 to 25 mg/liter). Meanwhile, SAV from these preparations showed comparable (≈3.3) biotin binding sites per tetramer.

WB800(pSSAV-Tcry) shows some degree of instability with prolonged culture in the absence of biotin supplements. The cells tend to lyse. In contrast, WB800BIO(pSSAV-Tcry) grows healthily in the absence of added biotin for even a long period of time, consistently and reproducibly. SAV can stably accumulate in the culture for at least 30 h. To achieve these results, the use of the engineered promoter system is crucial. This system has several important designs tailored for controlled biotin production. First, this promoter system should allow higher levels of biotin production when induced. Intracellular biotin of WB800BIO induced with 0.2% gluconate reached 5.5 μg/liter in a 9-h culture, while no biotin could be detected from WB800BIO grown in the absence of gluconate or from WB800 cultured under the same condition. Second, since biotin is essential for cell survival, the engineered promoter system should have a basal level expression even in the absence of the inducer to produce certain levels of biotin to maintain cell viability. Although no biotin could be detected in the absence of gluconate induction within our detection limit, study of promoter activity shows that there is a basal level expression from this promoter system (Fig. 2B), which allows WB800BIO to survive in a defined medium without induction or biotin supplement (Fig. 4). Third, the level of expression from this engineered promoter system should be adjustable, since overproduction of biotin can result in the release of intracellular biotin to the culture medium. In E. coli, release of intracellular biotin to the external environment has been demonstrated to be an energy-independent event, although the biotin uptake process is energy dependent (27). In B. subtilis, overproduction of biotin can result in an accumulation of biotin in the culture medium as high as 250 μg/liter in certain engineered strains (25). Since WB800BIO(pSSAV-Tcry) cultivated in 0.2% gluconate grew even better than WB800(pSSAV-Tcry) given 40 μM (9.76 mg/liter) biotin (Fig. 4) while no biotin could be detected in the culture supernatant, 0.2% gluconate was chosen as an appropriate induction condition. The amount (40 μM) of external biotin supplemented for the growth of WB800(pSSAV-Tcry) is already in excess. A similar degree of growth enhancement was observed with 12 μM biotin added to the culture medium.

The E. coli lac operator has been applied successfully to develop inducible expression systems in B. subtilis (15, 18). In this study, the gluconate operator was selected instead as the negative regulatory component for this engineered promoter system because it offers two advantages. (i) Gluconate is a cheaper inducer relative to isopropyl-β-d-thiogalactopyranoside (IPTG). In comparison to the cost for a gram of IPTG, the cost for a gram of gluconate is 600 times less. (ii) The gluconate repressor gene is already in the B. subtilis chromosome (11). In contrast, the use of the lac operator would require the transfer of lacI to B. subtilis. One concern with using gluconate as the inducer is that gluconate could get metabolized by B. subtilis so that its level in the culture medium could decrease with time. Indeed, we observed a decline in promoter activity in WB1200 cultivated in superrich medium 6 h after induction (Fig. 2). However, addition of extra gluconate 6 to 15 h after induction improved neither the cell growth nor the production yield of SAV in WB800BIO(pSSAV-Tcry) (data not shown).

In this study, two slightly different semidefined media were used. One contains vitamin-free Casamino Acids, while the other uses regular Casamino Acids. The use of regular Casamino Acids increases the SAV production yield by twofold. Therefore, CA medium containing regular Casamino Acids is the preferred choice for SAV production. Determination of the biotin content from either supply of Casamino Acids did not show any detectable biotin. This suggests that components (possibly other vitamins) other than biotin present in the regular Casamino Acids but absent from the vitamin-free Casamino Acids could account for the improved cell growth and SAV production. This implication is further strengthened by our observation that supplementation of biotin to medium containing vitamin-free Casamino Acids did not show any improvement of SAV production from that of WB800BIO(pSSAV-Tcry) (data not shown). The CA medium contains a low level of biotin, which comes from the yeast extract. Quantification of the biotin content in the concentrated yeast extract stock indicates that CA medium should contain biotin at a level of 2 μg/liter. With our biotin assay, we could reliably detect biotin content at a level of 3 μg/liter or higher. This explains why we could not observe detectable amounts of biotin in the CA medium. Overall, this engineered expression system is suitable to produce high levels of high quality biotin-free SAV using a biotin-depleted medium.