Isolating Escherichia coli strains for recombinant protein production

Abstract

Escherichia coli has been widely used for the production of recombinant proteins. To improve protein production yields in E. coli, directed engineering approaches have been commonly used. However, there are only few reported examples of the isolation of E. coli protein production strains using evolutionary approaches. Here, we first give an introduction to bacterial evolution and mutagenesis to set the stage for discussing how so far selection- and screening-based approaches have been used to isolate E. coli protein production strains. Finally, we discuss how evolutionary approaches may be used in the future to isolate E. coli strains with improved protein production characteristics.

Keywords: Escherichia coli; Evolution; Mutagenesis; Protein production; Recombinant protein; Strain isolation.

Fig. 1

The Luria and Delbrück experiment. In 1943, Luria and Delbrück devised an experiment to address if mutations occur prior to selection or in response to it (‘mutation’ versus ‘acquired hereditary immunity’) [3]. Several aliquots from single E. coli cultures and from multiple, independent E. coli cultures were spread on plates containing bacteriophage T1 (‘virus α’). On these plates, only bacteria resistant (immune) to bacteriophage T1 survive and form colonies. This allowed estimating the number of bacteriophage T1 resistant bacteria in the cultures. In aliquots from the same culture, variation observed in the number of bacteriophage T1 resistant mutants was minor and could be attributed to experimental error. In contrast, the number of resistant mutants in aliquots of the multiple independent cultures varied greatly. Luria and Delbrück concluded that, in this setup, ‘resistance to virus is due to a heritable change of the bacterial cell which occurs independently of the action of the virus’ (cit. [3])

Fig. 2

Types of mutations. Mutations can cause a large variety of changes in a genome. According to the nature of the change relative to the ancestral sequence, alterations may be grouped into base substitutions (i.e., transitions and transversions), insertions, deletions, inversions and translocations. a Examples of the possible effects of a single nucleotide alteration, including a nucleotide insertion and deletion, in a coding region. In this figure, the bases constitute codons and the encoded amino acids are indicated below the DNA sequence to illustrate possible effects. b Examples of larger-scale alterations. Genes are depicted as arrows, non-coding regions as bars. c Schematic representation of a transposition using the Tn5 transposon as an example. Tn5 is a composite transposon with two flanking IS50 elements and contains multiple resistance genes [125]. A transposase (encoded by IS50R) mediates excision of Tn5 from the donor locus and integration into a new location. In the target sequence, Tn5 insertion leads to duplication of a few base pairs (indicated by asterisk). Note that transposition mechanisms differ depending on the transposable element The outline of figure 2c was taken from [156] with permission

Fig. 3

Repair of single-strand DNA lesions. Schematic representations of the modus operandi of base excision repair (BER) (a), methyl-directed mismatch repair (MMR) (b) and nucleotide excision repair (NER) (c). The lesions in the figure serve merely as examples as the aforementioned repair pathways are capable of repairing a variety of different lesions. In all examples, bases are shown as blocks using the one-letter code, the deoxyribose-phosphate moiety is depicted as a grey line. Incisions are indicated by black triangles penetrating the sugar–phosphate backbone. a Example of BER acting on a chemically altered base (denoted by the yellow star). The affected nucleotide is removed by the subsequent action of a glycosylase and an AP-endonuclease. DNAP I re-synthesizes the missing part of the DNA strand and DNA ligase closes the nick. b Example of MMR acting on a wrongly incorporated adenine (in yellow). MutS binds to the site of the distortion and subsequently recruits MutL and MutH. MutH incises the newly synthesized, non-methylated strand at the sequence GATC. Subsequently, a DNA helicase and exonuclease unwind and degrade part of the newly synthesized strand, including the non-matching nucleotide(s). DNAP III and DNA ligase fill in the missing sequence. c Example of NER acting on a pyrimidine dimer (in yellow). The UvrAB-complex binds to the site of the lesion and promotes incisions 3′ and 5′ from the lesion by UvrC. Subsequently, the UvrD-helicase promotes dissociation of the contained stretch of DNA. Also in NER, DNAP I re-synthesizes the missing part of the DNA strand, and DNA ligase closes the nick

Fig. 4

Recombination-dependent repair of double-strand breaks. The RecBCD complex has both helicase and nuclease activity. It unwinds the DNA starting from the site of the break and degrades both strands during this process. Movement of RecBCD along the DNA is indicated with an arrow. At specific sites (indicated by ‘x’), the activity of the complex is altered such that only the strand with the free 5′ end continues to be degraded. That way, a 3′ overhang is created. RecA forms a nucleoprotofilament at the 3′ overhang and promotes strand invasion at a homologous double strand. Templated by the homologous DNA, replication re-starts and the missing sequences are filled in, followed by resolution of the resulting Holliday junctions. For the sake of clarity, proteins are only depicted on one site of the double-strand break

Fig. 5

Factors affecting mutation rates and patterns. Schematic representation of how extrinsic and intrinsic factors may contribute to the observed mutation rates and patterns. Screening for or selection of a certain phenotype is based on the acquired mutations

Fig. 6

Isolation of C41(DE3) from BL21(DE3). To isolate C41(DE3), BL21(DE3) was first transformed with a T7-based expression vector harbouring the gene encoding the mitochondrial oxoglutarate malate carrier protein (OGCP) and expression of ogcp was induced with IPTG in liquid culture. Notably, the ogcp expression vector has an ampicillin resistance marker. Surviving cells were selected for on IPTG-containing agar plates and subsequently probed for efficient OGCP production. In a second step, selected clones were cured from the ogcp expression vector by culturing them for a prolonged period of time in a closed setup (modified after [143])

Fig. 7

P_lacWT, P_lacUV5 and P_lacWeak. Expression of the lac operon (lacZYA) is governed by the P_lacWT region. A variant of this well-known promoter region, termed P_lacUV5, controls the expression of the gene encoding T7 RNAP in BL21(DE3) [142]. This variant differs from P_lacWT in four positions (asterisk). For better orientation, we highlighted the relevant sites: the binding site for CRP–cAMP, the −35/−10 binding sites for E. coli RNAP, and the first bases of the O1-operator site. Note that the term region was chosen to account for all four mutations. In different BL21(DE3)-derived protein production strains including C41(DE3) [101, 142, 143, 146], P_lacUV5 has reverted to a weaker variant, designated P_lacWeak [143]. This variant still harbours the altered CRP–cAMP binding site of the P_lacUV5 region, but reverted to P_lacWT in the −10 and the O1-operator site Picture was taken from [143] with permission

Fig. 8

Combining evolutionary approaches and engineering to create E. coli strains enabling the efficient production of disulfide-containing proteins in the cytoplasm. A screening approach was used to isolate E. coli strains that allow the formation of disulfide bonds in the cytoplasm [150]. In the screen, PhoA, a periplasmic protein, which requires disulfide bonds for its activity, was produced without a signal sequence in a strain lacking chromosomal phoA. The activity of the signal-sequence-less PhoA served as an indicator for cytoplasmic disulfide bond formation. Subsequently, mutants with PhoA activity were screened for, which resulted in the isolation of trxB-deficient strains. Using an engineering approach, it was found that disulfide bond formation in the cytoplasm is even more efficient in trxB null mutants that are unable to either synthesize or reduce gluthathione (gshA ⁻ or gor ⁻) [151]. However, these double mutants grow very poorly and require an exogenous reductant to achieve a reasonable growth rate. Finally, suppressor strains were isolated that grow well and still allow stable disulfide bond formation in the cytoplasm [152]