The necessary blood feeding behavior of the Ae. aegypti mosquito facilitates the transmission of potentially deadly and harmful viruses to uninfected human hosts. Zika, Dengue, Yellow Fever, and Chikungunya are mosquito-borne viral diseases that have become or are becoming a global health concern [4]. At the moment, there are no treatments, limited vaccines or therapeutics available to combat these mosquito-borne viral infections, which have led to high endemics and epidemics observed over the past few years [32]. Therefore, the only effective strategy still remains to be vector control, and with Ae. aegypti resistance to chemical compounds and effects to other insect species, new and more effective strategies are needed. A potential strategy may be to focus on blood meal digestion and the proteases involved in this process [1, 13,14,15,16]. With knockdown studies on three midgut proteases (AaSPVI, AaSPVII and AaLT) leading to a decrease in fecundity [1], may provide a potentially new vector control strategy. However, for this to be realized, in vitro biochemical studies focusing on midgut proteases must first be conducted. Even then, producing soluble recombinant mosquito proteases must first be achieved before these studies can be initiated.
Initial work to recombinantly express the zymogen (inactive) and mature (active) Ae. aegypti midgut proteases led to insoluble expression [15]. Furthermore, due to unknown activation of the midgut proteases, a strategy using a pseudo propeptide region with an unnatural enterokinase sequence was developed in order to facilitate activation of the proteases in vitro. To rescue the enzymes from inclusion bodies, a denaturation/renaturation scheme was developed [15]. Although this approach was successful in isolating bona fide active midgut proteases for initial enzyme kinetic analysis, the process is tedious and time consuming, and may not lead to yields comparable to proteins that can be solubly expressed [33]. In addition, the mode of activation of each zymogen is still unknown because the native propeptide region was removed [15]. Therefore, the purpose of this work is to describe the approach taken to recombinantly and solubly express the zymogen AaET, AaSPVI, AaSPVII, and AaLT midgut proteases with the native propeptide region using E. coli as the host. This will provide a much faster and facile starting point to researchers who have difficulty in producing solubly recombinant mosquito proteases in E. coli.
The field of Biochemistry has been revolutionized by the success of producing recombinant proteins using bacteria [20]. Without the molecular techniques, vast commercially available expression vectors, engineered bacterial strains, and rich media cultivation methods, large amounts of blood fed Ae. aegypti mosquitoes would be required to isolate midgut proteases that are only present once a mosquito has imbibed a blood meal [1, 15]. Because of the ease of manipulation, growth, and for institutions with limited funding, the cost effectiveness of recombinantly expressing proteins using E. coli, has made this organism the most widely preferred [24, 34]. Of course, there are limitations to recombinantly expressing proteins in E. coli (such as low expression, protein aggregation, plasmid instability, and protein degradation), but for each case there are available published troubleshooting strategies that address each potential problem (see [20, 24, 34]). As such, every protein to be recombinantly expressed will have its own problems and must be individually optimized to ensure the production of soluble and active protein. For the production of the most abundant zymogen midgut proteases, we have taken the troubleshooting ideas described in several review articles and highlight the most important parameters required to successfully producing soluble proteases using E. coli, which as we found were the type of bacterial cells and bacterial growth and induction temperature.
A major issue when attempting to recombinantly express eukaryotic proteins in bacteria is aggregation, especially proteases dependent on disulfide bridge formation for structure, stability, and function. Amino acid sequence analysis on AaET, AaSPVI, AaSPVII, and AaLT revealed the presence of six cysteine residues, which are predicted to form three disulfide bridges in each protease. It was not surprising that expression of these proteases in BL21(DE3) and Rosetta(DE3) bacterial strains led to insoluble protein (inclusion bodies) [15]. The cytoplasm of these E. coli strains are highly reducing and the reducing environment is caused by the thioredoxin and the glutathione/glutaredoxin reductase pathways, reducing disulfide bonds in proteases leading to misfolding [20, 33]. Furthermore, the initial attempt at expressing the zymogen midgut proteases included the protein leader (signal) sequence [15], a portion of amino acids on the N-terminus of the protein that is recognized and targeted to the endoplasmic reticulum for secretion into the cytoplasm. The problem, however, is that this polypeptide is usually hydrophobic in nature and has been shown to cause protein aggregation, thermodynamically destabilizing the recombinant expression of the protein of interest in E. coli [26]. To circumvent this issue, the signal sequence of each midgut protease was removed using PCR, keeping only the natural propeptide region (Fig. 1). This was a similar approach taken in the 2011 study where the signal sequence was removed when the unnatural EK site was introduced into the recombinant midgut proteases [15].
In order to avoid solubility expression issues resulting from improper disulfide bond formation, we turned to SHuffle® T7 Competent E. coli cells (NEB) [21]. These cells are BL21(DE3) derivatives that carry mutations in the reductase pathways (thioredoxin and glutathione/glutaredoxin) leading to a more oxidizing cytoplasm, which should allow formation of disulfide bridges [21]. In addition, the cells are engineered to constitutively express a disulfide bond isomerase (DsbC), a protein that aids in correcting of mis-oxidized disulfide bonds. Although the cytoplasmic conditions are ideal for promoting disulfide bridge formation, expression of the midgut proteases still led to insoluble expression when grown at the recommended 30 °C temperature. This was observed for all protease as seen in Fig. 2. This 30 °C temperature is 7 °C lower compared to wild type E. coli, BL21(DE3), and other closely related strains. Regardless, whether using BL21(DE3) or its derivatives, including SHuffle® T7 cells, when attempting to express eukaryotic proteins in a prokaryotic system there is a chance that at the optimal temperature, the eukaryotic proteins may be expressed insolubly. Because transcription and translation happen simultaneously in the bacterial cytoplasm, the rate of protein synthesis is approximately ten times faster than that of a eukaryotic cell [33]. And since a eukaryotic protein is being synthesized in a foreign prokaryotic environment, the rate of folding of the recombinantly expressed protease is not ideal. Prokaryotic proteins tend to fold at a much faster rate than their eukaryotic counterparts at the optimal growth conditions, and with the combination of a speedy rate of synthesis and slow folding in recombinant bacterial expression, the eukaryotic protein could aggregate and be insolubly expressed [35]. This provides a plausible explanation for why the midgut proteases were expressed insolubly at 30 °C in cells with a more oxidizing cytoplasm. To overcome this problem, we tested lower temperatures starting at 23 °C and going as low as 10 °C. Successful soluble expression was observed for AaET-NL and AaSPVII-NL at 23 °C (Fig. 3), but no soluble expression was observed for AaSPVI-NL and AaLT-NL. Interestingly, a second band was observed for AaET, which we hypothesized to be the active mature form. We attempted BApNA activity assays of the crude lysates, but since expression was only successfully visualized using WB, not enough solubly expressed AaET was present and could not reach the lower level of detection of p-nitroanilide formation (> 0.0125 abs units) [15].
The lower temperatures attempted (23 °C, 15 °C, and 10 °C) resulted in the soluble expression of all proteases (Figs. 3, 4 and 5). However, the preferred temperature differed for each. For example, AaLT-NL was only solubly expressed at 10 °C, while all the other proteases were solubly expressed at 15 °C. Importantly, at this temperature we were able to observe the possible auto-activation of AaET-NL, AaSPVI-NL, and AaSPVII-NL (Fig. 4). In each case, the presence of the active species (based on BApNA activity assays) seems to be dependent on protease concentration, which has been true for bovine and porcine trypsinogen [36]. In general, after induction with IPTG, protein expression concentration increases linearly with time at early time-points, but may reach a point where the expression is constant. This is the case for the midgut proteases. Expression of the zymogen form is initially observed, but over time, as expression concentration increases, leads to activation of the protease as observed in Fig. 4. At the moment, it is unknown if any proteases or enzymes in the bacterial crude lysates may be activating the recombinant midgut proteases, but the no induction growth experiment samples have no detectable BApNA activity (Additional files 1, 2 and 3: Figures S1, S2 and S3). Work is currently underway to determine if the midgut proteases are autocatalytic or if enzymes in the bacterial lysate are aiding in the process. Nonetheless, the reduced and colder temperature at which the proteases were expressed helped with promoting proper folding. By dropping the temperature at induction, the rate of protein synthesis, as well as the temperature-dependent hydrophobic interactions involved in protein folding are reduced, increasing the chances of proper folding when utilizing E. coli [37]. This temperature reduction approach led to successfully solubly expressing the four-zymogen (no leader) midgut proteases. It is important to note that caution should be taken when dropping the temperature lower than necessary because traditional promoter systems, bacterial transcription and translational machinery, and chaperones may not be as efficient compared to the optimal E. coli growth temperatures (37 °C or in the case of the SHuffle® cells, 30 °C) [20, 37]. Nonetheless, lowering the temperature at which recombinant proteases are expressed should be strongly considered before manipulation of any other variable.
Due to the length of time needed to solubly express the proteases of interest, we utilized Terrific Broth for all of our experiments. TB media contains yeast extract and tryptone at higher concentrations compared to LB, glycerol (an extra carbon source), and phosphate salts to help with culture acidification, making this much superior than LB [20]. In addition, the cell densities of the bacterial growths in TB are typically much higher compared to LB, which is important because the lower temperatures lead to a reduction in bacterial metabolism [25, 27, 28]. For each growth, the bacterial cells were grown at 30 °C to reach the proper induction at OD600nm ~ 0.5–1.0 (this allows the density of the cells to increase at a much faster rate than if growing at the lower induction temperature), and then the temperature of the growths dropped to the determined value. At the colder temperatures, the length of time at which the cells are grown has to be extended, and as such the available nutrients may be depleted, not producing enough soluble protease. In general, E. coli growth in LB media stops at relatively low cell densities because of the limited nutrients and carbon sources [34]. Therefore, a richer media is preferred when growing at lower temperatures and for a longer extended period of time, which helped improve the amount of expression for all midgut proteases.
As a proof of principle, we proceeded to purify two zymogen proteases one expressed at 23 °C (AaSPVII-NL) and the other at 10 °C (AaLT-NL). With the observed possible auto-activation of the proteases, we wanted to ensure that halting and harvesting bacteria at an earlier time-point before the presence of the active species, would lead to successful purification and isolation of the inactive zymogen. This would be especially problematic for AaET, AaSPVI, and AaSPVII since auto-activation is observed (Fig. 4), but not problematic for AaLT since no auto-activation observed. We therefore, Nickel purified AaSPVII-NL (cells harvested at 5 h post-induction) and AaLT-NL (cells harvested at 48 h post-induction) to near homogeneity utilizing a modified three-step gradient approach to ensure separation of non-specific binding of proteins and our proteases of interest. The his6-tag was utilized in order to easily purify the proteases in one step, which as seen in Fig. 6, was achieved. Since the purification was done in the presence of DTT and cold (4 °C) buffer conditions, no auto-activation of the proteases was observed, even though the pH of the buffer was 7.2. Normally, this would be problematic since eukaryotic trypsins have been shown to auto-activate between pH 7 and 9 [30, 31]. Furthermore, to avoid any further auto-activation of the purified AaSPVII-NL and AaLT-NL zymogen proteases, buffer exchange dialysis into Sodium Acetate buffer pH 5.2 and protein concentration under these conditions did not lead to precipitation or loss of purified protease. More importantly, these conditions prevented auto-activation of the AaSPVII-NL zymogen protease. Work is currently underway to purify the other midgut zymogen proteases. Once we have isolated and purified all proteases, we will be able to determine the mode of activation and compare the kinetic parameters between solubly expressed recombinant proteases and the isolated refolded proteases from [15].