Sir, Soon after the advent of gene cloning, plasmid expression vectors were developed in which variants of the lac promoter were used to direct high level IPTG- (or lactose-) inducible synthesis of recombinant proteins in Escherichia coli (Brosius, 1984, Gene27: 161–172; Amann and Brosius, 1985, Gene40: 183–190). Typically, the products of genes cloned for expression in such vectors (e.g. PTAC and PTRC vectors) account for 5–30% of the total cell protein after induction. However, such high-level synthesis often results in inclusion body formation, and this prevents phenotypic analysis and, in some cases, isolation of biologically functional product. Moreover, with toxic products (for example many membrane proteins), it often proves impossible to subclone the gene into these vectors because the basal level of expression is sufficiently high that the growth of the host cells is severely compromised even in the absence of inducer (Grisshammer and Tate, 1995, Q Rev Biophys28: 315–422. Recently, arabinose-inducible expression vectors were described that use the araBAD promoter (Cagnon et al. 1991, Prot Eng4: 843–847; Guzman et al. 1995, J Bacteriol177: 4121–4130). The pBAD vectors (Guzman et al. 1995, ibid.) have an inherently low basal level of expression (particularly in the presence of glucose, which causes catabolite repression) and hence they facilitate the successful subcloning of toxic genes. Induction by arabinose is rapid and results in 100- to 1000-fold increase in the rate of synthesis of the target protein, and good, but not massive, yields of it. The low basal levels and high inducible levels of gene products have led to the use of pBAD vectors both for depleting E. coli of specific gene products (in mutant hosts lacking the corresponding gene) and for overexpression studies when it is preferable to keep the protein soluble (Guzman et al. 1995, ibid.). It has also been shown that by incubating cultures with increasing submaximal (non-saturating) concentrations of arabinose, the yields of the target protein can be reproducibly increased over a greater than 100-fold range, and this finding forms the basis for advocating the use of ara promoter vectors in experiments in which fine control over the levels of a toxic, or marginally soluble, product needs to be imposed (Guzman et al. 1995, ibid.; Invitrogen catalogue, 1998). However, experiments in which the synthesis of green fluorescent protein (GFP) was induced with non-saturating concentrations of arabinose have revealed that, although the final yield of GFP within cultures increases as inducer concentration is increased, the levels of GFP in different individual cells within a culture vary considerably, from high to intermediate to no detectable GFP. Moreover, within different populations both the proportion of cells making detectable GFP and the maximum level of GFP present in individual cells increased with increasing concentrations of arabinose (Siegele and Hu, 1997, Proc Natl Acad Sci USA94: 8168–8172). In fact, this pattern of partial induction is expected to apply to any control circuit in which expression of the gene encoding the uptake system for the inducer is itself subject to induction. Indeed, over 40 years ago an analogous autocatalytic induction phenomenon was seen when the lac operon was subjected to induction with submaximal levels of its natural inducer, lactose (Novick and Weiner, 1957, Proc Natl Acad Sci USA43: 553–566). Lactose is dependent upon the presence of the lac permease in the envelope for its uptake into cells, and when the permease is present it actively transports the inducer into the cytoplasm. But the levels of permease present in the envelope of uninduced cells vary from none to a few molecules, depending on the recent history of the cell (i.e. when its lac operon last underwent a burst of transcription). Hence, when limited amounts of lactose are added to the medium only some of the cells will be capable of accumulating it, and in these cells the consequence of its uptake is further synthesis of the permease and thus more rapid uptake of any remaining lactose and, subsequently, full induction of the lac operon. It is important to appreciate that after exposure to low levels of autocatalytic inducers the intermediate levels of gene product seen within a cell population belie a wide variation in levels of product from cell to cell. But when working with toxic or marginally soluble gene products and, indeed, with various other gene products that display more subtle overproduction phenotypes it can be crucial to ensure a low-to-intermediate level of expression in all cells. For example, many membrane proteins fail to assemble efficiently into the E. coli cytoplasmic membrane when highly overproduced (Ito and Akiyama, 1991, Mol Microbiol5: 2243–2253). Autocatalytic induction with limiting inducer would result in high-level synthesis in some cells and failure of a proportion of the membrane protein molecules therein to assemble into the membrane. In contrast, low-to-moderate rates of synthesis in every cell would be required to ensure membrane assembly of all the protein molecules synthesized. Unlike lactose, the gratuitous inducer, IPTG, is not dependent upon the lac permease for its uptake into E. coli cells. Provided a permease-deficient, lacY, host is used, IPTG will enter the cytoplasm in a concentration-dependent manner. Hence, partial induction of the lac promoter, using IPTG in lacY hosts, can be used to fine-tune the levels of toxic gene products within individual cells, and not just within the population. Although early lac promoter-based expression vectors were inappropriate for the controlled expression of toxic genes (because they directed relatively high basal levels of expression) this problem can be overcome by positioning the lac promoter at a site in the vector where there is minimal readthrough transcription and by overexpressing the lacI gene (encoding the lac repressor) in the host cells. Hence, we propose that appropriately modified lac promoter vectors should be used (in conjunction with IPTG induction in lacY hosts) when it is important to synthesize low or intermediate levels of a particular protein within every cell. Ara promoter vectors will remain inappropriate until either a host strain is made that synthesizes the arabinose transporters constitutively or a gratuitous inducer is discovered that does not use/require them. We have developed expression vectors in which the lacUV5 promoter directs a low basal level of expression and a high (but not massive) level of IPTG-inducible expression (see Fig. 1). As we favour the direct expression approach, we have incorporated a ribosome binding site and modified lacZ′ allele (with a unique NcoI site spanning its initiation codon, and further unique restriction sites within lacZ′), downstream of the lacUV5 promoter, in both the kanamycin-resistant plasmid, pEH1, and its chloramphenicol-resistant analogue, pEH3. Therefore, coding regions from any organism can be subcloned for IPTG-inducible expression in E. coli (and transformants containing the recombinant expression plasmids can be readily identified using α-complementation). pEH1 and pEH3 also code for constitutive synthesis of the lac repressor, ensuring a low basal level of synthesis of the cloned gene product in any E. coli host strain, not just in lacI q strains. In addition, they contain a T7lac promoter upstream of the lacUV5 promoter. By supplying T7 RNA polymerase in trans the yield obtained from full induction with IPTG can be boosted, and (in the presence of rifampicin) the cloned gene product can be metabolically labelled (Studier et al., 1990, Methods Enzymol185: 60–89). Using a pair of vectors, with different selectable markers, facilitates rapid subcloning, expression and engineering of target genes. Moreover, as antibodies are commercially available against the kanamycin resistance gene product (neomycin phosphotransferase) and the chloramphenicol resistance gene product (chloramphenicol acetyl transferase), these proteins can serve as internal standards in expression studies or as markers for cytoplasmic localization in analyses of membrane protein topology and protein export. . Salient features of the lacUV5 promoter, direct expression vector, pEH1. Further information on the kanamycin-resistant expression vector, pEH1, and its chloramphenicol-resistant analogue, pEH3, is provided in the text. pEH1 contains the following unique restriction sites in lacZ′: NcoI, BamHI, HindIII, XbaI, KpnI, SacI, SmaI and EcoRI. Abbreviations: lac p, lacUV5 promoter; T7 p, T7lac promoter; lacZ′, LacZ α-peptide coding region; f1 ori, phage f1 origin for single-stranded DNA replication; ori, pMB1 replication origin; npt, kanamycin resistance gene encoding neomycin phosphotransferase; lacI, lac-repressor gene. We have subcloned the regions encoding various toxic bacterial and eukaryotic membrane proteins (for example integral membrane proteins such as the mannitol permease of E. coli, and the human β2-adrenergic receptor) into pEH1 and pEH3 for direct, IPTG-inducible, bacterial expression. LacY host cells carrying such plasmids produce low but detectable levels of these membrane proteins, and grow and form colonies normally. However, colony formation is completely prevented by the inclusion of submaximal concentrations of IPTG in L agar plates (in the case of the mannitol permease, the size of colonies is progressively reduced by the inclusion of 50 and 100 μM IPTG, respectively, and their formation is completely prevented by the inclusion of 250 μM IPTG). Finally, as with pBAD vectors, globular proteins and domains thereof (for example the DNA binding domain of the human Pax-5 transcription factor; personal communication, Dr B. Adams) can be produced in good yields and in a soluble and active form after full IPTG induction of the T7lac and/or lacUV5 promoters of these pEH vectors. The DNA sequences of pEH1 and pEH3 have been submitted to the EMBL database under the accession numbers AJ007659 and AJ007660, respectively. pEH1 and pEH3 are available from the authors on request.
No takes yet. Share an insight, caveat, or question.
Hashemzadeh‐Bonehi et al. (1998) studied this question.