In this manuscript we present a novel E. coli strain based on the trxB gor suppressor strain SMG96. We further engineered into its chromosome a dsbC gene lacking its signal sequence, under the regulation of the strong ribosomal promoter rrnB. These strains were engineered both in E.coli K12 and B strain backgrounds. A detailed characterization of the SHuffle strains along with parameters involved in protein production at bench-scale (non-high throughput) was investigated.
To expand our understanding of the mechanism of disulfide bond formation within SHuffle strains, we investigated the redox state of cytoplasmic DsbC. We showed that the majority of cytoplasmic DsbC is in its hemi-reduced state, which is essential for its disulfide bond isomerase activity. However, oxidized DsbC species were also detected when expressed within the oxidizing cytoplasm, which could result in DsbC directly oxidizing reduced substrates. This is not surprising, as mutant DsbB which have gained the ability to oxidize DsbC are in turn capable of oxidizing proteins in the periplasm
. Oxidized DsbC species may not always be beneficial to the folding of reduced proteins which require disulfide bonds. This may explain the drop in activity observed for Gaussia luciferase when expressed in cells with cytoplasmic DsbC. Similar observations were made when expressing parathyroid hormone in trxB gor strains
. In this study, co-expression of cytoplasmic DsbC had no positive influence in vivo, but did dramatically reduce the amount of misfolded species when DsbC was co-incubated in the presence of oxidized and reduced glutathione.
E. coli B strains such as BL21 are the preferred host strain for protein expression as generally give higher yields for the large majority of proteins. Some of the reasons for this may be that, unlike its K12 cousin, it has not been subjected to extensive domestication for the purpose of DNA manipulation
, and it lacks the cytoplasmic protease lon known to play a key role in protein quality control
. Similarly, when we compared the expression of three proteins in SHuffle K12 vs. SHuffle B strains, we consistently observed higher yields in the B strain backgrounds. However, we also observed differences between the two strains at the level of redox states of proteins. Unlike in SHuffle K12, a fraction of periplasmic DsbC was observed to be in its reduced state in the SHuffle B strain. Further redox differences were observed in the ahpC mutations between the two strains. While SHuffle K12 ahpC gene has the triplet TTC codon expansion, SHuffle B ahpC gene has the triplet codon contraction, lacking one of the three TTC codons. These differences highlight the distinct biological differences between the two SHuffle strains and require detailed studies to elucidate their biological roles.
To define conditions critical for the folding and correct formation of disulfide bonds, we tested the impact of the three most commonly manipulated physical parameters; temperature, time and strength of induction. We consistently observed that growth temperatures had the most profound impact on improving protein production in SHuffle cells. This may be due to the fact that SHuffle cells are under oxidative stress, and the resulting detrimental effects may be compounded by high metabolic activity during growth at high temperatures such as 37°C. This hypothesis is supported by the observation that over-expression of poorly folding proteins such as vtPA at 37°C in SHuffle cells is toxic (data not shown).
We observed very efficient production of proteins to high yields when SHuffle cells were grown overnight in Magic Media, reaching final yields of 400 mg/l in the case of a cellulase (with a single disulfide bond). To validate the role of the media, we produced vtPA in Magic Media and observed a 6 fold increase in the final yields compared to standard expression conditions using IPTG as an inducer. This form of protein expression in SHuffle cells may indeed be optimal, even though the mechanism of expression is not clear. Although the exact composition of Magic Media is not disclosed, it is designed to be used for the auto-expression of proteins under the control of the lac promoter. The principle of autoexpression is based on diauxic regulation where glucose is the preferred carbon source which results in the repression of the lac promoter and upon its consumption, cells switch to growth on lactose which results in the induction of the lac promoter
. However, β-galactosidase activity is needed to convert lactose to allolactose, the natural inducer of the lactose operon
. In the case of the SHuffle B T7 cells, the T7 RNA polymerase gene 1 is inserted into the lacZ gene, rendering it inactive. Thus, another mechanism of expression other than autoinduction must be occurring, which is why we termed this form of expression “autoexpression” instead of autoinduction.
In this study, we focused on improving folding of target substrate proteins by manipulating the strain and the conditions of expression. However, for optimal expression of proteins, many other parameters must be manipulated. For example, all proteins which require disulfide bonds for their folding will be secreted to compartments where disulfide bond formation can occur. Thus, they will all have some sort of a signal sequence at their N-terminus. However, to express these proteins in the cytoplasm, a signal sequenceless version of the target protein must be expressed. Removal of the 5’ signal sequence will alter the composition and structure of the mRNA, which is known to play a key role in the expression level of the target protein
. One remedy to this potential problem is to fuse the target protein to the carboxyl terminal of MBP, which is known to enhance solubility and can be proteolytically removed post production
. Otherwise, using the appropriate expression vector with the optimal promoter, codon usage and ribosome binding site need to be considered for optimal expression of the target protein.
Since bottlenecks in the folding pathway of any given protein are specific to that protein, we explored whether we could increase protein yield by co-expressing various helper proteins. We chose a subset of helper proteins based on either prior experimentation which has shown their utility, or in assumptions based on the helper proteins properties. Redox-active helper proteins had the biggest effect. Co-expression of mutant thioredoxins and PDI homologs were the most successful class of helper proteins. Surprisingly, co-expression of the catalase katG improved the activity of vtPA 10-fold. This observation supports the notion that the SHuffle cells are under oxidative stress and boosting the cell’s defenses against oxidative damage can increase the capacity of the cells to produce correctly folded disulfide bonded proteins. However, the decrease in vtPA activity when additional DsbC was expressed from the helper plasmid accentuates the fact that, for each individual protein, there can be an optimum level of a redox helper, with a decrease in activity at amounts higher or lower than that optimum. A similar decrease in activity was observed in the case of periplasmic expression of vtPA
. Overexpression of periplasmic DsbC resulted in loss in vtPA activity and eventually resulted loss of viability. The authors attributed the loss in viability to a dramatic reduction in the oxygen uptake rate when DsbC was over-expressed
. It is plausible that a similar interaction is occurring in the cytoplasm. This drop in activity was not observed when the putative disulfide bond isomerase from Aquifex aeolicus (cAaDsbC) was co-expressed. This difference highlights the protein specificities that govern the interaction between the oxido-reductase and its substrate protein.
Expression of proteins in the cytoplasm instead of in the periplasm is of great advantage. Not only does one avoid the complication of having to secrete the target substrate, the periplasm is devoid of ATP, has only a few ATP-independent chaperones, and is only ~20% of the volume of cytoplasm
. The advantage of cytoplasmic expression was observed in the case of vtPA, which had two fold increase in activity when expressed in the cytoplasm
. Similarly, we observed a 7 fold increase in the activity of an α1,3 Galactosidase from Xanthomonas manihotis having a single disulfide bond, when expressed in the cytoplasm instead of the periplasm (data not shown).
Although cytoplasmic expression may improve the activity of certain proteins, cytoplasmic disulfide bond formation may sometimes be detrimental to certain biological processes. For example, cytoplasmic assembly of the E. coli phage M13 appears to be problematic, as SHuffle strains were incapable of forming infective phage (data not shown). In addition, SHuffle cells grown in minimal media under high dissolved oxygen rates showed poor growth when glycerol was the sole carbon source (data not shown). This may be due to altered redox state of SHuffle cells’ metabalome. For example, the cydAB operon, which is under the regulation of the ArcAB two component system
, shows a delayed response in transcriptional activity when shifting from aerobiosis to anerobiosis in SHuffle cells (data not shown). This is most likely due to the silencing of ArcB kinase activity by the oxidation of its cytoplasmic redox-active cysteine residues
. These observations highlight our current lack of understanding of the redox biology of SHuffle cells, with many important questions remaining unanswered. How do SHuffle cells cope with oxidizing and reducing conditions within cytoplasm? Which reductases are involved in the oxidation of substrate proteins? What is the role of cytoplasmic oxidized DsbC in disulfide bond formation? How do SHuffle cells perform in high density fermentations? Proteomic and mass spectrometric approaches to address these questions are now in progress.
The SHuffle strains and the expression conditions presented in this report represent the first detailed analysis of the conditions required for efficient cytoplasmic expression and folding of disulfide bonded proteins. The results should allow the expression of previously inaccessible production of proteins in E. coli. These SHuffle strains greatly expand the cell biologists toolkit by enabling the use of bacterial production in place of more cumbersome eukaryotic expression systems.
We have demonstrated the value in engineering an E. coli trxB gor suppressor strain which expresses active cytoplasmic DsbC. We found that temperature is of paramount importance and should be optimized for the optimal expression of a substrate protein. Autoexpression of proteins using Magic Media was also very helpful in improving yield. We found several intriguing redox related differences between the E. coli B and K12 versions of this strain and showed that the E. coli B version of SHuffle strains were superior to its K12 counterpart. Further improvements were made by co-expressing various helper proteins. These SHuffle strains along with the knowledge gained regarding their use will be of great use to the protein expression community.