Combinatorial mutagenesis and selection of improved signal sequences and their application for high-level production of translocated heterologous proteins in Escherichia coli

Abstract

We previously designed the consensus signal peptide (CSP) and demonstrated that it can be used to strongly stimulate heterologous protein production in Escherichia coli. A comparative study using CSP and two bacterial signal sequences, pelB and ompA, showed that the effect of signal sequences on both expression level and translocation efficiency can be highly protein specific. We report here the generation of CSP mutant libraries by a combinatorial mutagenesis approach. Degenerated CSP oligonucleotides were cloned in frame with the 5' end of the bla gene, encoding the mature periplasmic β-lactamase released from its native signal sequence. This novel design allows for a direct selection of improved signal sequences that positively affect the expression level and/or translocation efficiency of β-lactamase, based on the ampicillin tolerance level of the E. coli host cells. By using this strategy, 61 different CSP mutants with up to 8-fold-increased ampicillin tolerance level and up to 5.5-fold-increased β-lactamase expression level were isolated and characterized genetically. A subset of the CSP mutants was then tested with the alternative reporter gene phoA, encoding periplasmic alkaline phosphatase (AP), resulting in an up to 8-fold-increased production level of active AP protein in E. coli. Moreover, it was demonstrated that the CSP mutants can improve the production of the medically important human interferon α2b under high-cell-density cultivations. Our results show that there is a clear potential for improving bacterial signal sequences by using combinatorial mutagenesis, and bioinformatics analyses indicated that the beneficial mutations could not be rationally predicted.

Fig 1

Schematic representation of the expression related features of the pCSP1bla screening vector (A) and the pCSP2ifn protein expression vector (B). In these vectors, the signal peptide and target genes are easily exchanged using the NdeI/NcoI and NcoI/BstAPI (removing the c-myc-His₆ tag) or the NcoI/NotI (preserving the c-myc-His₆ tag) restriction sites, respectively. Pm, positively regulated promoter; xylS, gene encoding the Pm activator; bla, ampicillin resistance gene encoding β-lactamase; tLPP, transcriptional terminator; rrnBT1T2, bidirectional transcriptional terminator; ifnα2b, codon optimized gene encoding the human cytokine IFN-α2b. The DNA sequence corresponding to the CSP region is displayed at the top with the unique NdeI and NcoI restriction sites in bold. The vector backbones of pCSP1bla and pCSP2ifn are generally conserved but differ with respect to the selection marker (kan in pCSP1bla and bla in pCSP2ifn), and pCSP2ifn harbors the cop271C mutant of the trfA replication gene, leading to a 3- to 4-fold-increased plasmid copy number (50).

Fig 2

DNA and amino acid sequences of the original CSP and selected mutant variants. The corresponding ampicillin tolerance levels were obtained from spotting E. coli DH5α-strains harboring pCSP1bla and its derivatives with the 5 best silent CSP mutants and the 12 best random CSP mutants onto LA containing either ampicillin and m-toluic acid (20 μM, induced) or ampicillin only (uninduced). Ampicillin tolerance of DH5α (pNSP1bla) is included as a control (no signal sequence [NoSP]). Periods (.) indicate that the base or amino acid is the same as in the unmodified CSP sequence, while “–” indicates gaps in the sequence (in pNSP1bla). “+” behind the ampicillin concentration given indicates that poor growth was observed at even higher concentrations, while a “–” shows that the growth was reduced at the concentration indicated.

Fig 3

β-Lactamase production in E. coli DH5α strains harboring the plasmids pNSP1bla (no signal peptide [noSP]), pCSP1bla (CSP), and the 12 different plasmids pCSP1S2bla-pCSP1S61bla (CSP mutants S2 to S61). Gene expression was induced by 0.1 mM m-toluic acid. (a) Specific β-lactamase activities measured from cell extracts. 1 U = the amount of enzyme that catalyzes the transformation of 1 μmol of substrate min⁻¹. The results are the averages of two biological replicas performed in triplicates, and the standard errors are indicated. (b) Western blot detection of soluble β-lactamase in the same extracts as in panel a. The same total protein amount was applied in all wells. The sizes of the mature and the uncleaved β-lactamase proteins are approximately 29 and 32 kDa, respectively. The two boxes show independent blots; vertical lines within the boxes indicate that one or more samples have been removed from the data set. The M column represents the migration of ladder proteins of sizes 25, 37, and 50 kDa, respectively.

Fig 4

Relative AP activity of E. coli DH5α strains harboring relevant expression constructs. All values are relative to the CSP result, which was arbitrarily set to 1. AP, endogenous AP signal sequence (pASP1phoA construct); CSP, original sequence (pCSP1phoA construct), while S2, S38, S48, S60, and S61 indicate the variant CSP sequences in pCSP1S2phoA, pCSP1S38phoA, pCSP1S48phoA, pCSP1S60phoA, and pCSP1S61phoA, respectively. Cultures were induced by 0.1 mM m-toluic acid. All results are the averages of two biological replicas with at least three samples, and the standard errors are indicated.

Fig 5

Western blot analysis of E. coli RV308 strains (pIFN30SpelB, pCSP2ifn, pCSP2S2ifn, or pCSP2S61ifn) producing IFN-α2b fusion proteins with different secretion signal peptides in high-cell-density cultivations. Cell samples were withdrawn from each culture 6.3 h after induction. The cells were lysed by sonication, and the total crude cell extracts were subjected to analysis. Dilution series of decreasing total protein amount were loaded for each sample and are shown relative to the highest total protein concentration loaded (set arbitrarily to 1). Signal peptide–IFN-α2b–cMyc–His₆ is 24.4 kDa, while the protein complex in which a signal peptide has been cleaved off is 22.4 kDa.