Random Insertion of a TetR-Inducing Peptide Tag into Escherichia coli Proteins Allows Analysis of Protein Levels by Induction of Reporter Gene Expression

Construction and function of InsTipα in E. coli.

We assembled the 1,816-bp-long insertion element InsTipα (Fig. 1A) from previously published vectors and oligonucleotides, amplified it by PCR, and cloned it into the PvuII-digested vector pWH1866 (17) (see Table 1 for a list of all bacterial strains, plasmids, and phages) to obtain pWH1912. (Construction details and sequences of all plasmids generated in this study are available upon request.) InsTipα is flanked by 19-bp-long mosaic elements (ME) (26), the binding sites of the hyperactive transposase Tn5 (8). The leftward end of InsTipα (Fig. 1A) contains a sequence designed to fuse a tag, termed InsTip1, to any protein encoded at the insertion site. An open reading frame spans the ME encoding 7 amino acids, followed by a sequence encoding a 7-amino-acid linker and the 16-residue Tip element terminated by an ochre stop codon. The insertion element also contains aphAIII, which confers kanamycin resistance in gram-negative and -positive bacteria (10), flanked by two Flp recombinase target sites to allow excision of the antibiotic resistance gene (4), thereby minimizing genomic interference.

• Open in new tab
• Download powerpoint

FIG. 1.

InsTipα architecture, in vivo screen, and in vivo activity. (A) Architecture of the insertion element InsTipα. Triangles represent the ME for transposase binding, the white box indicates the linker element, and the gray box encodes Tip. The ochre stop codon is symbolized by “S.” The white and black arrows represent the kanamycin resistance gene (aphAIII) and its promoter, and a lollipop represents the bidirectional transcriptional terminator from Tn10 (30). The resistance cassette is flanked by two Flp recombinase target sites (FRT) drawn as black arrows. The amino acid sequence encoded by the mosaic element at the leftward 5′ end, the linker, and tip is shown in single-letter abbreviations above the nucleotide sequence that is displayed in white type on a black background. The stop codon is marked by an asterisk. (B and C) In vivo screening system for TetR-inducing InsTip1 fusions. (B) Screening was carried out with E. coli WH207(λtet50) containing a chromosomally encoded tetA-lacZ transcriptional fusion. TetR (gray ovals) binds tetO and represses lacZ. (C) Integration of the insertion element InsTipα into an expressed open reading frame (geneX) results in a GeneX-InsTip1 fusion protein (light gray oval with X). Induction of TetR by this fusion protein is indicated by β-Gal activity. (D) In vivo activity of the TrxA-InsTip1 fusion. The leftmost pair of bars shows the repressed (white) and tetracycline-induced (gray) β-Gal activities. The next seven bars show the activities of the TrxA-Tip fusion control without (white) and with IPTG at the indicated concentrations (black; 15, 30, 60, 125, 250, and 500 μM). The right group of bars shows the β-Gal activities obtained with the TrxA-InsTip1 fusion under the same conditions. The y axis displays the β-Gal activities in Miller units (MU) (23). (E) Western blot of 10 μg of crude cell lysates separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and analyzed with an anti-thioredoxin A monoclonal antibody (Invitrogen).

To determine if InsTipα insertion into an open reading frame can generate a protein fusion that induces TetR, InsTipα was fused to TrxA of E. coli, which had previously been used to express and characterize Tip (16). The constructed plasmid, pWH1909, expressing TrxA-InsTip1 under P_tac control, and the plasmid pWH1911, expressing TetR constitutively at a low level, were cotransformed into E. coli WH207(λtet50) (35). The addition of IPTG (isopropyl-β-d-thiogalactopyranoside) to the resulting strain leads to expression of the InsTip1 fusion protein, and subsequent induction of TetR should relieve repression of lacZ (Fig. 1C). We titrated the strain with IPTG concentrations of 15 to 500 μM to express increasing amounts of TrxA-InsTip1 (Fig. 1D). Plasmid pWH2101 expressing TrxA-Tip was cotransformed instead of pWH1909 to provide a positive control for induction (16). A plateau of β-galactosidase (β-Gal) activity was obtained for concentrations above 60 μM of IPTG with both constructs (Fig. 1D). Western blot analyses carried out simultaneously demonstrated nearly identical expression levels of the fusion proteins TrxA-Tip and TrxA-InsTip1 (Fig. 1E). Thus, neither the length nor the primary structure of the peptide segment encoded by the ME and linker affects the TetR-inducing properties of Tip.

Purified hyperactive transposase and PvuII-restricted, gel-purified InsTipα DNA were incubated to form transposomes (2) and electroporated (8) into E. coli WH207(λtet50)/pWH1911, resulting in 6,400 kanamycin-resistant candidates on MacConkey agar plates. After 36 h of incubation at 37°C, 5 colonies displayed a yellow color indicating induction of TetR. Chromosomal DNA from the five strains was prepared and sequenced to determine their InsTipα integration sites (2). The diverse sequences obtained at the InsTipα integration loci agree in indicating its random insertion (12). Table 2 lists the strains obtained and the genes with their respective InsTipα insertion sites (see also Fig. 2), together with information about the function and subcellular localization of the affected proteins.

• Open in new tab
• Download powerpoint

FIG. 2.

InsTipα insertion sites and expression of the resulting InsTip1 fusion proteins. (A to E) Results for the five strains with InsTipα insertions in (A) atpD, (B) ppiD, (C) yraI, (D) yicO, and (E) yfdX. Top panels, schematic depiction of the insertion site and orientation of InsTipα into each target gene. The respective gene is represented by a black arrow and its designation, and its total length is indicated in bp. InsTipα is drawn schematically above the open reading frame, and its insertion site is indicated by a dotted line with the exact position of the nucleotide preceding the inserted sequence. Bottom panels, growth curves (filled squares) and β-Gal activities (filled bars) of WH207(λtet50)/pWH1911 and the respective candidate strains (open squares and open bars) grown at 37°C in MacConkey medium without indicator. Growth is expressed by the optical density at 600 nm (OD₆₀₀) on the left y axis. The right y axis displays the β-Gal activities in Miller units (MU) (23). Each value was determined in triplicate, and the standard deviations are shown (error bars).

Determining fusion protein expression by InsTip1-induced β-Gal expression.

We first examined the growth curves of the five candidates and their parent strain WH207(λtet50)/pWH1911 at 37°C in MacConkey medium lacking the indicator bromocresol purple (Fig. 2A to E). The growth rates of all strains were identical (Fig. 2B to E), except that for WH851, which was clearly growth defective (Fig. 2A). We assumed that the InsTipα insertion in atpD inactivates ATP synthase (29). We therefore analyzed the growth of WH851 and the atpD⁺ parent strain on minimal medium with a nonfermentable carbon source. In contrast to the atpD⁺ strain, WH851 displayed no growth on agar plates containing minimal medium with succinate as the sole carbon source (data not shown). Since growth on this medium depends on a functional oxidative phosphorylation system, this phenotype is consistent with a nonfunctional ATP synthase (14, 18). This assumption was further supported by the growth phenotype in liquid minimal medium supplemented with limiting amounts of glucose as the sole carbon source (18, 34); in this medium, WH851 reached a maximal optical density of 0.3, in contrast to an optical density of 1.0 for the atpD⁺ reference strain (data not shown), again indicating the absence of oxidative phosphorylation.

At regular intervals during the growth periods, cells were collected from all candidate strains and the parent strain to determine the β-Gal activities during exponential and stationary growth (Fig. 2A to E). The parent strain showed no change in β-Gal activity throughout the entire experiment, although the β-Gal activity could be induced 40-fold by the addition of Tc (Fig. 1D) (35). Only background levels of β-Gal activity were detected for the candidate strains during exponential growth (Fig. 2A to E). However, for PpiD-InsTip1, YraI-InsTip1, YicO-InsTip1, and YfdX-InsTip1, a rapid increase in β-Gal activity was observed at the onset of the stationary growth phase, reaching a plateau by the 20-h time point of the experiment (Fig. 2B to E). With a value of approximately 1,100 units (23), WH852 has demonstrated the highest level of β-Gal activity. In contrast, we detected an increase in β-Gal activity within the stationary growth phase for WH851, but it was delayed and not as pronounced as the increases for the other strains. In addition, it did not reach a plateau during the 40-h time frame of the experiment (Fig. 2A).

We calculated an approximately sixfold increase in β-Gal activity of WH853 with the YraI-InsTip1 fusion for the transition from mid-exponential growth (4 h) to full stationary phase (20 to 40 h) (Fig. 2C). This observation is in excellent agreement with published data from transcriptome profiling in which a 4.1-fold increase in the mRNA level was observed due to the growth phase shift (11).

In WH854, we established expression of the YicO-InsTip1 fusion in stationary phase. yicO promoter activity was previously reported to be induced in the early stationary phase by overexpression of the two-component regulatory system BaeSR using a transcriptional reporter gene fusion (1). In contrast, a whole-genome microarray analysis of BaeSR-dependent gene expression from mid-exponential-phase cultures showed no increase of yicO expression (24). Our results for YicO-InsTip1 expression are in full agreement with the results of both of these earlier reports and explain their apparently diverse results.

In WH855, InsTipα had integrated into the yfdX gene. Since yfdX is an essential gene (6), both the insertion and the identical growth curves of WH855 and the yfdX⁺ reference strain at 37°C indicate that residues in a C-terminal position from Gly158 are not needed for growth under these conditions (Fig. 2E). A transcriptome comparison of exponentially grown cells with stationary-phase cells of E. coli showed that yfdX is induced significantly in the stationary phase (see supplemental data from reference 31 as cited in reference 22). Our data from the YfdX-InsTip1 fusion corroborate this result (Fig. 2E).

Besides being able to detect the expression of InsTip1 fusion proteins on MacConkey and minimal media, we can also use InsTip1 to determine quantitative differences in protein expression. A threefold-higher expression level of a ppiD-lacZ transcriptional fusion has been reported for exponentially growing E. coli in minimal medium with pyruvate as the sole carbon source than with glycerol as the sole carbon source (5). Therefore, we analyzed β-Gal activity of WH852 cells under these conditions and found a twofold-higher β-Gal activity for growth with pyruvate than with glycerol within the exponential growth phase. Thus, the result obtained with this translational fusion is in excellent agreement with the data from a transcriptional fusion (Fig. 3).

• Open in new tab
• Download powerpoint

FIG. 3.

Carbon source-dependent differential expression of PpiD-InsTip1. β-Gal activities of WH207(λtet50)/pWH1911 (closed bars) and WH852 (open bars) grown at 37°C to an optical density at 600 nm of 0.2 in M9 minimal medium supplemented with 50 mM glycerol or pyruvate. The y axis displays the β-Gal activities in Miller units (MU) (23). Each value was determined in triplicate and is shown with standard deviations (error bars). The dotted lines indicate the background β-Gal activities determined for the untagged parent strain, which were subtracted from the β-Gal activities of the tagged strain to determine the induction factor.

Conclusions.

We have successfully used the peptide Tip (16) as a protein expression tag. We constructed an in-frame fusion which indicates the expression of the tagged target gene by readout of a separated lacZ gene. In contrast to direct lacZ fusions, this introduces only a 30-amino-acid-long extension after the insertion site, turning the protein fusion into an inducer for TetR. The induction of a separated reporter gene should lead to amplification of the signal. We constructed an InsTipα insertion element, formed the complex with hyperactive tranposase (8), randomly mutated (12) the model organism E. coli, and obtained five candidates out of 6,400 insertion mutants. The tagged proteins were detected during the stationary growth phase, extending and confirming published expression data from transcriptional fusions or microarrays for atpD, ppiD, yraI, yicO, and yfdX, thereby validating the Tip tag approach. We could also measure expression of ppiD during the exponential growth phase. Together with the ability to detect proteins of essential and nonessential genes, as well as proteins localized in the cytoplasm, the inner membrane, or the periplasm with the tag, our results clearly demonstrate the applicability of InsTipα as a global tag. These results encourage us to further improve the insertion efficiency and to expand the detection of active InsTip1 fusion proteins by avoiding polar effects on downstream genes through inclusion of an outward promoter in the element (15). We also aim to improve detection of protein expression within the exponential phase via enhanced screening, for example, by reduction of the relatively high basal level of β-Gal activity under repressed conditions to enlarge the detection window (20) or by the use of selection systems.