Optimization of Bacteriocin Release Protein (BRP)-Mediated Protein Release by Escherichia coli: Random Mutagenesis of the pCloDF13-Derived BRP Gene To Uncouple Lethality and Quasi-Lysis from Protein Release

Abstract

Bacteriocin release proteins (BRPs) can be used for the release of heterologous proteins from the Escherichia coli periplasm into the culture medium. However, high-level expression of BRP causes apparent lysis of the host cells in liquid cultures (quasi-lysis) and inhibition of growth on broth agar plates (lethality). To optimize BRP-mediated protein release, the pCloDF13 BRP gene was subjected to random mutagenesis by using PCR techniques. Mutated BRPs with a strongly reduced capacity to cause growth inhibition on broth agar plates were selected, analyzed by nucleotide sequencing, and further characterized by performing growth and release experiments in liquid cultures. A subset of these BRP derivatives did not cause quasi-lysis and had only a small effect on growth but still functioned in the release of the periplasmic protein β-lactamase and the periplasmic K88 molecular chaperone FaeE and in the release of the bacteriocin cloacin DF13 into the culture medium. These BRP derivatives can be more efficiently used for extracellular production of proteins by E. coli than can the original BRP.

The pCloDF13 bacteriocin release protein (BRP) is a small lipoprotein of 28 amino acids required for the release of the bacteriocin cloacin DF13 into the extracellular medium of Escherichia coli cultures (27). As a side effect of BRP expression periplasmic proteins are released. This feature of BRP expression has been used to release heterologous proteins like the human growth hormone and guar α-galactosidase from the E. coli periplasm into the extracellular culture medium (8, 25).

In addition to release of periplasmic proteins, high-level expression of wild-type BRP causes a decline in turbidity (called quasi-lysis) when cells are cultivated in liquid medium. On broth agar plates high-level expression of BRP causes a severe decrease in the number and size of colonies (called lethality) (27). Quasi-lysis and lethality are in part due to the BRP signal peptide, which is not degraded after processing and which accumulates in the cytoplasmic membrane (28). This accumulation of the stable signal peptide has early effects on protein biosynthesis and Mg²⁺ transport (24). The deleterious effects of the mature BRP and its stable signal peptide on the host cell can be counteracted in different ways. First, quasi-lysis can be prevented by the addition of divalent cations to the culture medium (15). Second, the use of a hybrid BRP (lipoprotein-BRP [Lpp-BRP]) which is targeted by the unstable murein lipoprotein signal peptide alleviates deleterious effects caused by the accumulation of the stable BRP signal peptide in the cytoplasmic membrane (26, 28). However, high-level induction of the Lpp-BRP still results in quasi-lysis, which is caused by the accumulation of mature BRP in the cell envelope.

As stated above, the pCloDF13-derived BRP has been used for extracellular production of heterologous proteins (8, 25). However, efforts to increase the level of release of proteins from the producing cells by raising the expression level of BRP have resulted in damage to the cells and release of unwanted proteins. Therefore, we started to modify the BRP gene in order to obtain a BRP derivative that is less harmful during high-level expression but is still useful for release of industrial or pharmaceutical proteins. Such a BRP derivative would be more suitable (optimized) for use in extracellular production of E. coli proteins. To optimize BRP-mediated protein release, we used a BRP derivative with the harmless labile lipoprotein signal peptide (Lpp-BRP). This construct was then further mutated by PCR-directed random saturation mutagenesis by using the region coding for the mature part of the Lpp-BRP as a template. Mutated BRPs whose expression did not inhibit the colony-forming ability of the host were selected and tested for quasi-lysis and the release of β-lactamase and the K88 fimbrial molecular chaperone from the periplasm.

MATERIALS AND METHODS

Bacterial strains, media, and plasmids.

E. coli C600 (F⁻ thr-1 leuB6 thi-1 lacY1 supE44 rfbD1 fhuA21 mcrA1) was used for cloning and as a host in all experiments. YT medium (19) containing ampicillin (100 μg ml⁻¹) and/or chloramphenicol (30 μg ml⁻¹) was used for culturing.

Plasmid pJL17lpp is a pBR322 derivative which encodes a hybrid pCloDF13-derived BRP targeted by the unstable murein lipoprotein signal peptide (25, 28). Expression of this hybrid, Lpp-BRP, is controlled by the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible lpp-lac tandem promoter-operator (P_lpp/lac). Plasmid pJL171pp was used as a template for PCR-mediated mutagenesis and for subsequent subcloning of the resulting mutated BRP fragments.

Plasmid pSV88-E is a pACYC184 derivative which encodes the periplasmic molecular chaperone FaeE involved in biogenesis of K88 fimbriae. The gene encoding FaeE is located downstream from the IPTG-inducible lpp-lac tandem promoter-operator (P_lpp/lac) in this plasmid. Plasmid pSV88-E is compatible with pBR322 derivatives (20).

Plasmid pJL25 is a pACYC184 derivative which is compatible with pBR322 derivatives and which codes for cloacin DF13 and its immunity protein under control of the original pCloDF13 mitomycin-inducible SOS promoter (13). This plasmid was used to study the specific release of the bacteriocin cloacin DF13 in complementation experiments with various BRP constructs. In these experiments, the unstable Lpp signal peptide of the various mutant BRPs used (encoded by A13, C09, C16, and C25) was first removed by digesting the plasmids with SphI and HindIII and then replaced with the original stable BRP signal peptide derived from pJL22-SphI (11, 12). The newly constructed BRP derivatives were checked by nucleotide sequencing.

Recombinant DNA techniques.

Purification of plasmid DNA and transformation of cells were carried out as described elsewhere (1, 18). Small DNA fragments (<200 bp) were isolated from 2% agarose gels (29) and precipitated by using linear polyacrylamide as a carrier (4). Other basic recombinant DNA techniques were performed as described elsewhere (21). Mutated BRPs were analyzed by nucleotide sequencing (22) by using a Taq DyeDeoxy terminator cycle sequencing kit and a model 373A automated DNA sequencer (Perkin-Elmer/Applied Biosystems).

Mutagenesis of the pJL17lpp-encoded Lpp-BRP.

We designed four doped oligonucleotides (Fig. 1, primers 2 through 5) complementary to different regions of the DNA sequence coding for the mature part of the Lpp-BRP (mBRP). These partially overlapping regions were flanked at their 3′ ends by a deoxyadenosine nucleotide. This allowed the use of mutated PCR fragments as primers in successive PCR experiments without the risk that certain substitutions would prevail in the newly synthesized DNA (10). In primer 5, two deoxynucleotides were changed to restore the carboxyl-terminal structure of the BRP which is part of the epitope for monoclonal BRP antiserum (Fig. 1). The carboxyl-terminal structure was previously lost by the introduction of a restriction site (11). The appropriate mutation rate (doping level) was determined by carrying out various Monte Carlo simulations of random saturation mutagenesis by using the program RAMHA (randomized algorithm for mutagenesis and histogram analysis) (23). In these simulations the sequence of the wild-type mature BRP was used as the template DNA. Antitermination strategies for triplets relatively susceptible to the formation of premature stop codons were taken into account. At a mutation rate of 1.0% within synthesized oligonucleotides, 55.4% of the proteins generated were predicted to contain the wild-type amino acid sequence. The proportion of mutant proteins with a single amino acid substitution was predicted to be 33.1%, whereas 9.6% of the proteins were predicted to have two substitutions and 1.9% were predicted to have three or more substitutions. Higher doping levels resulted in more mutant proteins with three or more substitutions, whereas lower doping levels resulted in fewer mutant proteins. Primers 2 through 5 were synthesized by using these specifications (Table 1).

FIG. 1.

Template DNA and oligonucleotides used for random saturation mutagenesis of the mature part of the Lpp-BRP encoded by pJL17lpp. The region which was subjected to mutagenesis is translated, and the amino acids are shown between the coding (upper) and noncoding (lower) strands. The regions complementary to the sequences of the doped oligonucleotides (primers 2 through 5) and the flanking primers (primers 1 and 6) are indicated. Mutations which restored the carboxy-terminal epitope and mutations which introduced a HindIII site are indicated by asterisks. For specifications of the primers see Table 1.

TABLE 1.

Oligonucleotides used for random saturation mutagenesis, PCR, and nucleotide sequencing

Primer	Sequence (5′→3′)a
1	GT TCT ACT CTG CTG GCA GCA T
2	GC VAGGCAAACTAXATCCGGGAT G
3	GCCACCGTTCYACCCTZAAC A
4	GGCACCATYZTCCTCTCTG
5	A CTZAACCGCGATCCCCGTCAGTTY A
6	CTC AAG AAG ATC CTT TAA GCT TTT CT
a	CAA TCT AGC TAG AGA GGC
b	TTA GCA CGA GCT GCG TCA
c	GTG AAT ACA AGG AAG GTA T
d	GTT CAC GTC GTT GCT CAG

^aDoped deoxynucleotides are underlined. The level of doping was 1.0%. For example, the deoxynucleotide mixture used for G contained 99% G, 0.33% C, 0.33% A, and 0.33% T. Positions for which antiterminating mixtures were used are in boldface type. V contained 99% C, 0.5% A, and 0.5% G (C^AG). Y and Z are C^TG and G^CT, respectively, and were polluted at the same rate as V. X contained 99% T and 1% C. 

To allow the generation of PCR products containing the mBRP flanked by unique restriction sites, we designed two additional oligonucleotides (primers 1 and 6) complementary to the sequences flanking the region coding for mBRP (Fig. 1).

The pJL17lpp-encoded mBRP was mutated with primers 1 through 6 in seven successive PCR experiments (Fig. 2). A typical PCR setup consisted of 30 cycles, with each cycle consisting of 1 s at 95°C and 30 s at 60°C to allow annealing of primers and small fragments or 1 min at 60°C to allow annealing of PCR fragments larger than 100 bp, followed by 10 s at 72°C. To allow annealing of PCR fragments in PCR experiments E and F, the annealing step of this scheme was modified. The temperature was programmed to decrease from 70 to 50°C at a ramping rate of 1°C per 9 s. The annealing step was prolonged 50 s instead of 30 s at 50°C. After 10 of these cycles, 20 normal cycles were programmed. The product of PCR experiment G was purified and digested with SphI and HindIII. The resulting 93-bp fragment was used to replace the wild-type 176-bp SphI-HindIII fragment of cloning vector pJL17lpp. We designed additional oligonucleotides (Table 1) complementary to P_lpp/lac and to a region 50 bp downstream from the HindIII site (primers a and b, respectively). These primers were used to generate PCR fragments for DNA sequencing and to distinguish between cloning vector pJL17lpp and derivatives containing the mutated SphI-HindIII fragment on the basis of the sizes of the fragments generated (627 and 544 bp, respectively). Primers c and d, complementary to a region directly upstream from the signal peptide coding region and to a region directly downstream from the HindIII site, respectively, were used to sequence the mutated BRP coding region from either side.

FIG. 2.

Schematic representation of the seven successive PCR experiments which were carried out to mutate the region coding for mBRP. Plasmid pJL17lpp was used as the template DNA for the subsequent PCRs performed with four doped oligonucleotides (primers 2 through 5) and two flanking oligonucleotides (primers 1 and 6). The product from PCR experiment G contains the region coding for the mature part of the mutated Lpp-BRP, flanked by SphI and HindIII sites.

The efficiency of the mutagenesis procedure was tested by using potentially mutated plasmid DNA from colonies as the template DNA for PCR experiments with primers a and b. The resulting PCR fragments of approximately 540 bp indicated that there was a mutated plasmid with the 93-bp SphI-HindIII fragment originating from the product of PCR experiment G. A sequence analysis of these PCR fragments was carried out to further identify the mutations. The results of these tests showed that the efficiency of the procedure was in good agreement with the 45% mutant proteins predicted by RAMHA.

Detection of β-lactamase.

Two procedures were used to investigate whether β-lactamase was released from the periplasm into the extracellular medium. The first procedure consisted of a iodometric plate assay (2). Colonies were transferred with a toothpick to plates which contained ampicillin to maintain the plasmid, IPTG (1 mM) to induce the BRP, and starch as an indicator in the iodometric assay. After 16 h of growth, the plates were flooded with an indicator solution (containing I₂, KI, and penicillin G). The diameters of cleared zones were measured after 4 min. In the second procedure, the presence of β-lactamase in culture supernatants was determined. Five-milliliter cultures of cells were induced with 0.1 mM IPTG for the expression of BRP. After 3 h of growth, the cells were collected by centrifugation, and the proteins present in the spent medium were precipitated by adding trichloroacetic acid to a final concentration of 10% and using bovine serum albumin as the carrier protein. The precipitated proteins were analyzed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) with 11% polyacrylamide gels, followed by immunoblotting. β-Lactamase was detected on the immunoblots by using specific antiserum and enhanced chemiluminescence (Amersham).

Detection of FaeE.

Cells containing pSV88-E (encoding FaeE) and one of the plasmids encoding a BRP derivative (A13, C09, or C25) were cultured in the presence of ampicillin and chloramphenicol and induced with 0.1 mM IPTG in the early exponential phase of growth. At various times after induction samples were collected, and the presence of the K88 chaperone FaeE in cells and in cell-free culture supernatant fractions was determined by immunoblotting by using a specific antibody against FaeE (26).

Detection of BRP.

The presence of the BRP in cells was analyzed by tricine SDS-PAGE (28) and immunoblotting by using a specific monoclonal antibody (11, 12). The localization of mature BRP in cytoplasmic (inner) membranes and in outer membranes was analyzed by disrupting induced cells expressing BRP or a derivative and separating inner and outer membranes by isopycnic sucrose density gradient centrifugation, followed by tricine SDS-PAGE of collected membrane fractions and immunoblotting, essentially as described previously (11, 12, 28).

RESULTS AND DISCUSSION

Random saturation mutagenesis.

To create mutant BRPs which still induce the release of proteins from the periplasm but which are less deleterious to the host than wild-type pCloDF13 BRP, the mature part of the BRP gene was subjected to random mutagenesis. The DNA sequence coding for mBRP was used as a template in seven successive PCR experiments performed with four doped oligonucleotides and two flanking primers (Fig. 1 and 2). The efficiency of the mutagenesis procedure was tested by analyzing plasmid DNA from 30 colonies. Seventeen potential mutants containing PCR-generated DNA were found on the basis of a restriction fragment analysis (see Materials and Methods). Eleven of these potential mutants were sequenced to get some idea of the mutation rate. Five of these plasmids (A series; represented by A13) encoded the wild-type BRP (Table 2), whereas with the other six plasmids (A01, A07, A12, A16, A17, and A20) the BRP amino acid sequence encoded was different (Table 2). Hence, the mutation rate was in good agreement with the calculated and predicted rates (see Materials and Methods).

TABLE 2.

Primary structure and functioning of mutant Lpp-BRPs

Plasmid(s)	Amino acid sequence of mature BRPa	Growth inhibition and quasi-lysisb	β-Lactamase releasec
pJL17lpp	CQANYIRDVQGGTVAPSSSSELTGIAVN	++	+
A13	CQANYIRDVQGGTVAPSSSSELTGIAVQ	++	+
A01	CQANYLRDVQGGTVAPSPPLN	−	−
A07	CQANYIRDVQGGTVAPSPPLN	−	−
A12	CQANYIGMFRVERWHHPPPLN	−	−
A16	CQANYSRDVQGGTVAPSSSSELTGIAVQ	±	±
A17	CQANYIRDVQGGTVAPSSSTELTGIAVQ	++	+
A20	CQANYIRDVQGGTAAPSSSSELTGIAVQ	++	+
D02	CQANY	−	−
D03	CQANNIRDVQGGTAAPSSSSELTGIAVQ	−	−
D04	CQANYIRDVQGGTAAPSPPLN	−	−
D05	CRANYIRDVQGGTVAPSSSSELTGIAVQ	−	−
B04	CPANYIRDVQGGTVAPSSSSELTGIAVQ	−	−
B39	CLANYIRDVQGGTVAPSSSSELTGIAVQ	−	−
B41	CKANYIRDVQGRTVAPSSSSELTGIAVQ	−	−
B50	CKPNYIRDAQGGTVAPSSSSELTGIAVQ	−	−
B54	CQANYIRDVQGGTGHHRPPLN	−	−
B92	CQPNYIRDVQGGRWHHRPPLN	−	−
B96	CQANCIRDAQGGTVAPSSSSELTGIAVQ	−	−
C03	CQANYIRGVQGGTVAPSSSSELTGISVQ	+	+
C09, C17, C18	CQANYIRDVQGGTEAPSSSSELTGIAVQ	±	+
C16	CQANYIRDVQGGTVASSSSSELTGIAVQ	±	+
C20	CQANYIRDVQGGTVASSSSSELTGIAVR	±	+
C25	CQANYIRDVQGRTVAPSTSSELTGIAVQ	−	+

^aThe amino acid sequences of mutated Lpp-BRPs are shown. The amino acid sequences were deduced from the nucleotide sequence data. For comparison, the primary structure of the pJL17lpp-encoded Lpp-BRP is given. The mutations are indicated by boldface type. 

^bThe ability of the various BRP derivatives to induce growth inhibition and a decline in culture turbidity (quasi-lysis) was tested in liquid medium by measuring the culture turbidity at regular time intervals after moderate induction. Moderate induction means induction with intermediate concentrations of the inducer (100 μM). This concentration of inducer is supposed to be high enough to have a strong effect on growth in cells expressing the wild-type BRP or a BRP with a wild-type phenotype, but low enough to allow less growth inhibition in cultures expressing a mutant BRP with a less deleterious phenotype. ++, severe growth inhibition shortly after addition of IPTG (inducing agent), followed by a decline in culture turbidity; +, strong growth inhibition after induction and no apparent decline in culture turbidity; ±, growth inhibition several hours after induction and no decline in culture turbidity; −, some or minor growth inhibition upon induction and no decline in culture turbidity. 

^cThe release of β-lactamase as a periplasmic marker protein was studied in liquid medium after moderate induction with 100 μM IPTG (see above) of the BRP derivatives. β-Lactamase was detected by immunoblotting. +, significant amount of β-lactamase in the culture supernatant; ±, intermediate amount of β-lactamase; −, no β-lactamase detectable. 

Selection of mutated Lpp-BRPs.

To select mutated Lpp-BRPs which were less deleterious to the host, cells transformed with mutated plasmids were plated onto broth agar containing 1 mM IPTG in order to strongly induce BRP gene expression. Cells expressing wild-type BRP or BRP derivatives with a wild-type phenotype could not grow on these plates due to the lytic effect of the BRP. Mutant BRPs that are less deleterious than the original BRP should allow growth of colonies. Large IPTG-resistant colonies were isolated and tested for their ability to induce the release of the periplasmic marker protein β-lactamase. Using the iodometric plate assay, we selected several mutants that caused relatively large clearing zones (B series mutants B04, B39, B41, B50, B54, B92, and B96), whereas growth and release experiments performed with liquid cultures resulted in several other mutants (C and D series mutants) (Table 2).

Analysis of mutated Lpp-BRPs.

Upon induction of the various Lpp-BRP derivatives (Table 2), cells harboring mutated plasmids C03, C09, C16, C17, C18, C20, C25, A17, and A20 released amounts of β-lactamase into the culture medium comparable to the amount released by cells expressing the Lpp-BRP encoded by plasmid A13. A number of representative examples of this release are shown in Fig. 3. None of the other strains, including the B series mutants, released significant amounts of β-lactamase into the culture medium (data not shown). Noninduced control cells harboring plasmid A13 did not release significant amounts of β-lactamase either. These results are summarized in Table 2.

FIG. 3.

Immunoblot analysis of the release of β-lactamase (Bla) by E. coli C600 cells harboring a mutated BRP-encoding plasmid (A13, C09, C16, or C25). Cells were induced for production of the Lpp-BRP with 0.1 mM IPTG for 3 h, and then cells and the supernatant fraction (medium) were separated by centrifugation and equivalent amounts (0.2 optical density at 660 nm unit) were analyzed. Lane 1, A13 (wild type); lane 2, C09 (C17 and C18 gave the same results as C09); lane 3, C16; lane 4, C25. In some samples the β-lactamase appeared as two bands; these bands represent two different conformations of β-lactamase, a phenomenon which was caused by heating in SDS sample buffer.

Induction of cells expressing wild-type BRP with intermediate concentrations of the inducer IPTG (50 to 100 μM) (moderate induction) results in severe growth inhibition but should allow better growth of a bacterial culture when the BRP is less deleterious due to mutations (27). We examined the growth effects after induction of the selected mutant plasmids. The results are summarized in Table 2. Furthermore, a few selected examples of this growth inhibition and/or quasi-lysis after induction are shown in Fig. 4. Cells harboring C09, C16, C17, C18, C20, or C25 are particular interesting because they are defective in quasi-lysis and show little growth inhibition but still release significant amounts of β-lactamase.

FIG. 4.

Growth, growth inhibition, and quasi-lysis of E. coli C600 cells harboring a plasmid encoding the wild-type Lpp-BRP (plasmid A13) or a plasmid encoding a mutated BRP (C25, C09, C16, C20, or C03). Cells were cultured in medium lacking Mg²⁺ and induced for production of one of the BRP derivatives with 0.1 mM IPTG. Symbols: ★, A13 without IPTG; ▪, C25; □, C09; ▴, C16; •, C20; ▵, C03; ○, A13 induced with IPTG. A plus sign indicates that the preparation was induced with IPTG.

Sequence analysis of mutant BRPs.

All of the mutated Lpp-BRPs selected (see above) were further analyzed by nucleotide sequencing (Table 2). Most of the mutated Lpp-BRP derivatives which did not cause quasi-lysis and which showed little or no growth inhibition after moderate induction of the BRP but still provoked the release of β-lactamase from the periplasm into the culture medium appeared to be mutated in the central section of the mature BRP.

In general, mutations in the first six amino-terminal residues and all types of truncations seemed to eliminate growth inhibition and quasi-lysis. This is consistent with the observation that truncated pCloDF13 BRPs consisting of 20, 16, 9, or 4 amino acid residues (11) and truncated pCloA BRPs consisting of 18 and 16 amino acid residues (7) cause growth inhibition and quasi-lysis, but to a lesser extent than the respective wild-type BRPs (3, 11). Since these truncated BRPs are targeted by stable signal peptides, their expression causes lethality, and the stable signal peptides are at least partly responsible for the observed decline in culture turbidity.

Replacement of the Val residue at position 14 of the mature BRP by a negatively charged Glu residue affected functioning of the BRP. In contrast, replacement of the Val residue at this position of the pColA BRP by a Gln, Leu, or positively charged Arg residue did not affect functioning of the pColA BRP (7). In addition, the wild-type ColE1 BRP does not contain a Val residue at this position but contains an Ile residue. The effects of these substitutions on the conformation of the BRPs remain to be elucidated.

Although the seven B series mutants caused large clearing zones in the iodometric plate assay, they did not induce the release of β-lactamase into the extracellular environment when they were tested in liquid medium. The reason for this phenomenon is not clear. It is noteworthy that the iodometric selection procedure resulted essentially in two types of mutants. The first type (B54 and B92) are frameshift mutants. Apparently, these mutants lack information located at the C-terminal end of BRP important for causing lethality, quasi-lysis, and the release of β-lactamase. In the second type of B series mutants, the hydrophilic Gln residue at position 2 is replaced by a Pro, Leu, or Lys residue. The Gln residue at position 2 of the mature BRP is conserved in all other known BRPs (27). This residue is important, since mutations in the amino terminus of lipoproteins affect lipid modification, processing, and localization in the cell envelope (5, 6, 9, 17, 30, 31). Whether the B series mutants are affected in expression, lipid modification, processing, and localization remains to be investigated.

Expression of mutant BRPs.

The most interesting BRP derivatives obtained based on potential application are those that provoke release of periplasmic proteins like β-lactamase but are less effective than the original BRP in causing quasi-lysis and growth inhibition (derivatives encoded by A16, C09, C16, C20, and C25). To investigate whether the amino acid changes in a number of these mutant BRPs affected their expression, stability, and/or subcellular localization, an immunoblot analysis of whole cells and of isolated inner and outer membranes was carried out (Fig. 5). Cells expressing C09-, C16-, or C25-encoded BRP contained an amount of BRP comparable to the amount in cells expressing the wild-type BRP (A13) (Fig. 5A). The C09-encoded mutant BRP showed a somewhat lower mobility than wild-type BRP upon tricine SDS-PAGE; this might be explained by the change in primary structure (V14E). An analysis of inner and outer membrane fractions of cells expressing A13 BRP, as well as C09 and C25 BRPs, revealed that the amounts of BRP in the two membrane fractions and also the distributions between the membranes are comparable (Fig. 5B). These findings indicated that the expression and subcellular localization of these mutant BRPs are similar to those of wild-type A13 BRP.

FIG. 5.

Immunoblot analysis of BRP in induced cells and in isolated inner and outer membrane fractions. (A) Whole cells. The plasmids encoding BRP (A13) or a mutant derivative (C09, C16, or C25) are indicated above the lanes. C17 and C18 are identical to C09. Equivalent amounts of cells were electrophoresed in the lanes of the tricine gel 3 h after induction. (B) Inner membrane (im) and outer membrane (om) fractions. Membranes were separated, and fractions containing inner membranes, as well as fractions containing outer membranes, were pooled and analyzed. Similar amounts were loaded onto the gel.

Release of periplasmic chaperone and cytoplasmic bacteriocin.

A previous study showed that the pCloDF13-encoded BRP can be used to release large amounts of the periplasmic K88 fimbrial molecular chaperone FaeE into the extracellular culture medium (26). To study the ability of a number of interesting mutant BRPs to support the release of FaeE, E. coli cells containing either a control plasmid without the BRP gene or plasmid A13, C09, or C25 were complemented with a plasmid containing the faeE gene. Cultures of these cells were induced for the expression of both (mutant) BRP and FaeE, and at various times samples of cells and cell-free supernatant fractions were analyzed for the presence of the chaperone FaeE (Fig. 6). Under the conditions used no severe growth inhibition or quasi-lysis was observed except with cells induced for the expression of wild-type A13 BRP. Control cells without BRP did not release FaeE into the culture supernatant fraction, as expected. Like cells containing the wild-type A13 BRP, cells expressing C09 BRP or C25 BRP released significant amounts of FaeE into the culture medium. This release was detectable 2 h after induction. Noninduced cells did not release any FaeE (data not shown). In control cultures (no BRP) and in C09- and C25-containing cultures cytoplasmic marker protein P48 (Ffh) (14) was detected only in the cells, not in the culture supernatant fraction, indicating that no quasi-lysis had occurred. In the culture supernatant fraction of induced A13-containing cells significant amounts (10 to 20%) of cytoplasmic marker protein P48 were detected as a result of quasi-lysis (data not shown). These cells appeared to release more FaeE protein than the other cells. However, this could have been the result of quasi-lysis, and the culture supernatant fraction could have been contaminated with other unidentified proteins (Fig. 7).

FIG. 6.

Immunoblot analysis of the release of periplasmic chaperone FaeE by cells expressing wild-type BRP (encoded by A13) or a mutant derivative (encoded by C09 or C25). Control cells contained no BRP-encoding plasmid. Samples of cells (lanes C) and supernatant fractions (lanes S) were analyzed 2, 4, and 6 h after induction.

FIG. 7.

Release of cloacin DF13 by cells expressing wild-type BRP or a mutant derivative. Cells containing pJL25 (encoding cloacin DF13) and plasmids A13, C09, C16, and C25, encoding BRP targeted by the stable BRP signal sequence, were cultured and induced with 100 ng of mitomycin per ml of medium and 0.1 mM IPTG. Samples were taken 5 h after induction, cells and medium were separated by centrifugation, and the presence of cloacin DF13 in the supernatant fractions was determined by SDS-PAGE, followed by protein staining. Lane 1, cells (A13) (all other cell fractions produced a similar protein pattern); lane 2, molecular mass markers from New England Biolabs (molecular masses, from top to bottom, 212, 158, 116, 97.2, 66.4, 55.6, 42.7, 36.5, 26.6, and 20.0 kDa); lanes 3 through 6, supernatant fractions of cultures of cells expressing A13, C09, C16, and C25 BRP, respectively. Equivalent amounts of the supernatant fractions were electrophoresed on the gel. The position of cloacin DF13 (Clo) (about 66 kDa) is indicated.

E. coli cells expressing the wild-type BRP targeted by its own stable signal peptide release significant amounts of the bacteriocin cloacin DF13 into the culture medium (12). To investigate whether BRP derivatives are able to induce the release of the bacteriocin cloacin DF13 from the cytoplasm of E. coli cells, the unstable Lpp signal peptide of the A13, C09, C16, and C25 BRPs was replaced by the original stable BRP signal peptide. Cells expressing one of the new constructs were transformed with pJL25, encoding cloacin DF13, and the release of the bacteriocin into the culture medium was studied after induction of both the bacteriocin and the BRP derivative (Fig. 7). The results showed that the mutant BRPs were effective in releasing cloacin DF13 into the culture medium, but there was lower background release of other proteins than with the wild-type BRP (encoded by A13).

Concluding remarks.

The pCloDF13-encoded BRP is widely used in industry, scientific institutions, and universities for extracellular production of homologous as well as heterologous proteins by E. coli. The main objective of the use of the BRP is to obtain the protein of interest in the culture supernatant fraction with a minimal amount of other contaminating proteins. This results in an easier purification procedure and may also prevent intracellular inclusion body formation. Plasmids encoding the wild-type BRP under control of an inducible promoter are commercially available. However, these plasmids encode a wild-type BRP with an unfavorable stable signal peptide. This stable signal peptide and the mature wild-type BRP cause severe growth inhibition of strongly induced cells and even cell death. These harmful side effects of course hamper the use of the BRP. The main objective of this study was to create a new type of BRP that can be more useful for extracellular production of interesting proteins.

As described above, we selected several mutated Lpp-BRPs that were defective in causing lethality and quasi-lysis and in causing significant growth inhibition but still functioned in the release of the periplasmic protein β-lactamase, the periplasmic molecular chaperone FaeE, and the cytoplasmic bacteriocin cloacin DF13. Apparently, pCloDF13 BRP-mediated quasi-lysis, lethality, and leakage of periplasmic proteins are not as strictly coupled as previously assumed. Lethality upon BRP expression is caused both by the stable BRP signal peptide (27) and by the mature portion of the BRP (12). The stable pCloDF13 BRP signal peptide accumulates exclusively in the cytoplasmic membrane (27, 28), whereas the mature BRP is located in the outer membrane, as well as in the cytoplasmic membrane (16). Accumulation of the stable signal peptide affects protein biosynthesis and Mg²⁺ transport, similar to the effect of expression of the wild-type pCloDF13 BRP, which suggests that lethality is caused in part by effects on the cytoplasmic membrane (24). Since all mutated BRPs constructed in this study are targeted by the unstable Lpp signal peptide, the cleaved signal peptides are not deleterious to the host cells. Possibly, the mutated Lpp-BRPs defective in causing lethality and quasi-lysis are less deleterious to the host because their signal peptide does not accumulate in the cytoplasmic membrane and most of the mature mutant BRP is localized in the outer membrane. This would allow activation of the detergent-resistant outer membrane phospholipase A and thus permeabilization of the outer membrane (15) without significantly affecting the integrity of the cytoplasmic membrane. As a result, cells would be able to release periplasmic proteins into the culture medium, and less growth inhibition, quasi-lysis, and lethality would occur.

The mutated IPTG-resistant Lpp-BRPs which are hampered in causing quasi-lysis but still function in the release of the periplasmic proteins may prove to be useful for construction of a BRP secretion vector to achieve efficient release of heterologous proteins from the E. coli periplasm into the culture medium without concomitant growth inhibition and lysis of the host cells.