Consideration on Efficient Recombinant Protein Production: Focus on Substrate Protein-Specific Compatibility Patterns of Molecular Chaperones

Abstract

Expression of recombinant proteins requires at times the aid of molecular chaperones for efficient post-translational folding into functional structure. However, predicting the compatibility of a protein substrate with the right type of chaperone to produce functional proteins is a daunting issue. To study the difference in effects of chaperones on His-tagged recombinant proteins with different characteristics, we performed in vitro proteins expression using Escherichia coli overexpressed with several chaperone ‘teams’: Trigger Factor (TF), GroEL/GroES and DnaK/DnaJ/GrpE, alone or in combinations, with the aim to determine whether protein secondary structure can serve as predictor for chaperone success. Protein A, which has a helix dominant structure, showed the most efficient folding with GroES/EL or TF chaperones alone, whereas Protein B, which has less helix in the structure, showed a remarkable effect on the DnaK/J/GrpE system alone. This tendency was also seen with other recombinant proteins with particular properties. With the chaperons’ assistance, both proteins were synthesized more efficiently in the culture at 22.5°C for 20 hours than at 37°C for 3 hours. These findings suggest a novel avenue to study compatibility of chaperones with substrate proteins and optimal culture conditions for producing functional proteins with a potential for predictive analysis of the success of chaperones based on the properties of the substrate protein.

Introduction

Large scale expression of recombinant proteins is essential for structural and functional analysis of proteins; proteins purified from large scale bacterial cultures serve experimental and therapeutic purposes. Escherichia coli is widely used as a host for protein expression as a simple system suitable for a wide variety of proteins. However, expression of proteins in E. coli often results in various problems, such as the formation of inclusion bodies and protease degradation of the protein. These issues often are a result of improper folding of the expressed proteins (1).

In the E. coli cytosol, a fraction of a newly synthesized protein often requires the activity of molecular chaperones to assure folding to a native state; this provided basis for artificial chaperone co-expression systems to help support normal folding and isolation of biologically active proteins in E. coli systems (2). The major chaperones implicated in this folding process are the ribosome-associated Trigger Factor (TF), and the DnaK and GroEL chaperones with their respective co-chaperones, DnaJ/GrpE and GroES. TF, DnaK/J/GrpE and GroEL/ES are considered to play distinct but cooperative roles in protein folding in vivo (2–5). TF is an ATP-independent chaperone and has peptidyl-prolyl-cis-trans isomerase (PPIase) activity in vitro (6). It is composed of three domains. The N-terminal domain is involved in association with large ribosomal subunits, and the central substrate-binding and PPIase catalytic site shows homology with the FK506 binding protein. The function of the C-terminal domain is unknown (7–9). TF exists near the nascent protein exit of the ribosome and has an affinity for short nascent chains of 57 amino acid residues or less (10), so it is considered to be the chaperone that first binds to the nascent polypeptide chains. DnaK is an ATP-dependent chaperone and acts in cooperation with DnaJ and GrpE to refold misfolded proteins through repeated cycles of substrate binding and release (11). At 30°C, DnaK associates with approximately 9–18% of newly synthesized proteins including nascent polypeptides (12). GroEL is an ATP-dependent oligomeric chaperone system which encapsulates nonnative substrate proteins in a central cavity capped by a co-chaperone, GroES, composing a safe folding cage. GroEL was shown to associate post-translationally with at least 10–15% of newly synthesized polypeptides (13). The collaborative actions of these chaperones are complicated and, at present, the functional relationship between these chaperone systems is only partially understood and merits further study.

A molecular interplay between nascent proteins and molecular chaperones in the crowded environment of E. coli cytosol is another unsolved mechanism. How nascent proteins are accurately selected by chaperones and what dictates a good compatibility to them is an emerging question central to accurate protein biogenesis (14). This complicated co-translational and post-translational protein targeting pathway consists of multiple stages including initial binding, ribosome delivery to the membrane, and enforcement of a timer for protein targeting (15). However, with lack of accurate method for structural analysis of chaperone-substrate protein complex, the details of the mechanisms have not been fully elucidated and a proper choice of chaperone for a particular protein is still a trial-and-error matter.

Here we studied E. coli expression of two distinct, artificially designed, fusion enzyme proteins (Protein A and Protein B) using co-transformation of chaperone expression plasmids and examined the differential interaction of the chaperones and proteins. We hypothesized that protein properties such as size, structure and isoelectric point may inform the choice of a successful chaperone system. We then tested our predictions on five additional recombinant proteins with varied characteristics.

Materials and Methods

All chemicals and reagents were obtained from Millipore-Sigma (Burlington, MA, USA) unless stated otherwise.

Protein characterization.

Both Protein A (molecular weight; 60937.35) and Protein B (molecular weight; 93081.33) are synthetic proteins consisting of a common short sequence containing a His-tag followed by an enzymatic part which functions as a DNA demethylase. Secondary structures of Protein A and Protein B were predicted by PSIPRED 4.0 (University College London, Bioinformatics Group: http://bioinf.cs.ucl.ac.uk/psipred/&uuid). Molecular weight and isoelectric point (IEP) of the proteins were calculated using ExPASy (SIB Swiss Institute of Bioinformatics: https://web.expasy.org/compute_pi/). In addition, five non-fusion natural proteins with various properties (Proteins C-G) were also subjected to the experiment to reconfirm the results from Protein A and B. To create these expression constructs, cDNAs of all the proteins were directly cloned into the pET30a vector. The amino acid sequences of the proteins are provided as Supplement 1.

Bacterial strain and plasmids.

E. coli BL21 (DE3) strain [F– ompT gal dcm lon hsdSB(rB–mB–) λ(DE3 [lacI lacUV5-T7p07 ind1 sam7 nin5]) [malB+]K-12(λS)] (Themo Fisher Scientific, Waltham, MA, USA) was used in the recombinant protein expressions. Chaperone plasmids constructed with the bacterial expression vector pAR3 were purchased from Takara Bio USA (Mountain View, CA, USA); they are summarized in Figure 1 (6, 16)

Fig.1.

A, Gel staining for SDS-PAGE of lysates from E. coli BL21 (DE3) transformed with chaperone vectors. To clarify the degree of separation, large chaperones (DnaK, DnaJ, GroEL and Tf) and small chaperones (GrpE and GroES) were observed separately on acrylamide gels of 7.5% and 15%, respectively. Naïve BL21 (DE31) without chaperon was served as a control (CTR). Gel staining was performed with GelCode Blue Stain Reagent. B, Summary of the chaperone plasmids constructed with the bacterial expression vector pAR3.

Transformation.

Competent BL21 E. coli was first transformed with the chaperone vectors, followed by second transformations with the recombinant proteins after recovering the competence of the chaperone transformed BL21 by standard methods (17).

Culture conditions.

Co-transformed E. coli was sub-cultured in 25 mL volume in LB with kanamycin (40 mg/mL) and chloramphenicol (50 μg/mL) at 37°C overnight and then inoculated to 1L of LB broth in Erlenmeyer flasks with antibiotics and chaperone inducers (0.5 mg/mL L-arabinose and/or 5 ng/mL of tetracycline). The bacteria were incubated at 37°C with a 2-inch stir bar spun at approximately 1400 rpm. The growth of the bacteria was monitored by sampling a small amount of the solution and taking an optical density reading at 600 nm (OD₆₀₀) every thirty minutes. When the OD₆₀₀ reached 0.5–0.6, 1 mM of Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to the medium for protein induction and the culture was continued at 37°C for 3 hours or 22.5°C for 20 hours. To determine the appropriate incubation time, the time course of protein induction efficiency at 37 °C and 42 °C were performed: see Supplement 2A, B. Also, chaperone’s effects on Protein A expression at 22.5°C were examined and are shown in Supplement 2C. We confirmed that the chaperones’ effects on the Protein A induction at 22.5 °C were similar to those at 37 °C. After the induction, the cells were harvested by centrifugation at 7000 rpm for 15 minutes.

SDS-PAGE and western blotting.

We extracted the bacterial protein with NP40 lysis buffer (50 mM Tris, pH 8.5, 150 mM NaCl, 1% NP40) with protease inhibitor cocktail VII (Research Product International, Mt. Prospect, IL, USA) and performed SDS-PAGE with 50 μg/lane of proteins in 7.5 or 12% acrylamide gel. Then, some gels were stained directly with GelCodeBlue Stain Reagent (Thermo Fisher) and others were transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA, USA) using Mini Trans-Blot Cell and Criterion Blotter (Bio-Rad). Efficiency of the transfer was confirmed by a brief staining of the membranes with Ponceau S solution. Expression levels of the proteins in bacteria were evaluated by western blotting with specific antibodies. Primary antibodies used in the immunoblotting are summarized in Table 1. Secondary antibody (HRP-conjugated anti mouse IgG) was obtained from Cell Signaling Technologies (#7076, Danvers, MA, USA) and used in 2000 times dilution. Western blots with chemiluminescence (SuperSignal West Pico PLUS; Thermo Fisher) were developed by ChemiDoc Imager (Bio-Rad) and then analyzed using Image Lab software (Bio Rad).

Table 1.

List of primary antibodies

	Providers	Ref #	Working dilutions
Anti-Protein A	Santa Cruz Biotechnologies	sc-376652	1:100
Anti-Protein B	Santa Cruz Biotechnologies	sc-293186	1:100
Anti-His Tag (H-3)	Santa Cruz Biotechnologies	sc-8036	1:100

Ni-NTA column purification.

6x-His tagged proteins were purified using HisPur Ni-NTA Spin Columns (Thermo Fisher) following the manufacturer’s instruction. Briefly, 10 mL of cell lysate from 1 L of culture was added to the column and incubated for 30 min with continuous rotation at RT. After the incubation, the column was washed three times with 6 mL of wash buffer (PBS with 25 or 45 mM imidazole; pH 7.4) followed by three times elution with 3 mL elution buffer (PBS with 250 mM imidazole; pH 7.4). Yields of the proteins of interest were examined by BCA protein assay (Thermo Fisher) and SDS-PAGE.

Statistics.

Data in Figure 4 and 5 were expressed as an average ± standard error of the mean (SEM) of at least three independent experiments. The data passed the Shapiro-Wilk normality test (P=0.14) and D’Agostino & Pearson omnibus normality test (P=0.47) but failed Kolmogorov-Smirnov test, therefore the values were normalized to control to additionally make sure parametric tests apply. In pairwise comparisons in Fig. 6 F-test showed that the variances were not significantly different between groups (P=0.22) also allowing for the use of parametric criteria. For Figure 4 the data were analyzed using one-way analysis of variance (ANOVA) followed by LSD, Bonferroni and Holm multiple comparison tests using Js-STAR (KISNET; http://www.kisnet.or.jp/nappa/software/star/). An unpaired, two-tailed Student’s t-test was applied for Figure 6 data using Excel in Office 365 (Microsoft). Statistical significance was accepted when p < 0.05.

Fig.4.

Western blotting of the supernatants of protein lysates from chaperone vectors co-transformed E. coli BL21(DE3) with induction at 37°C for 3 hrs, using proteins (A: Protein A and B: Protein B) and His-tag specific first antibodies. A lysate from naïve (non-transformed) BL(DE31) served as control. Gel staining of SDS-PAGE using the same sample at the same time (lower panels) shows the consistency of sample loading. Arrows indicate expected molecular weights of the recombinant proteins. C, Bar graphs represent ratio of band intensity to control in Protein A (left) and Protein B (right) western blotting (n = 4). *P < 0.05. **P < 0.01. Bar graphs are Mean±SEM.

Fig.5.

Western blotting of the supernatants of protein lysates from E. coli co-transformed with chaperones and additional five proteins. A. Lysates from five proteins (Protein C-G) and chaperone vectors co-transformed BL21(DE3) E coli were run on SDS-PAGE, and then blotted with anti-His-tag first antibody. B. For each chaperone vector, the relationship between the helix content of the produced protein and the production efficiency is summarized in a graph. In the line graphs, the ratio of the western blot band intensity to control (non-transformed BL21) for each chaperone is plotted in the Y-axis and the helix contents of protein A-G is the X-axis. (n=3); Mean±SEM. C. A summary plot integrating the efficiency of chaperones and helix content of all substrates. Purification of proteins in Fig. 5A and 4 was subjectively scored blindly by a separate investigator on a scale of 1 to 5, 5 being the highest gel band intensity in the group and 1 being the lowest; ‘0’ was used when no band was visible. These scores were assembled into a data matrix and plotted as line charts with color coding for each chaperon and with area under the curve coloring.

Fig.6.

Comparison of protein expression efficiencies between two conditions (3 hrs 37°C and 20 hrs 22.5°C) for supernatants from chaperone vector co-transformed E.coli BL21(DE3). A, pTf16 / Protein A co-transformed. B, pKJE7 / Protein B co-transformed. Arrows indicate expected molecular weights of the recombinant proteins. Bar graphs represent ratio of band intensity to 37°C 3hrs culture from densitometric analysis (n = 3). **P < 0.01. In the bar graphs, means are shown; error bars represent SEM.

Results

Specifications of Protein A and Protein B.

Specifications of Protein A and Protein B are summarized in Table 2. As shown, the Protein A was smaller (MW; 60937.35) and had higher IEP (9.43) in comparison to the Protein B (MW; 93081.33, IEP; 8.97). A more noteworthy difference in features was that Protein A had a helix-based secondary structure, while Protein B had a higher proportion of strands (Figure 2).

Table 2.

Specifications of Protein A-B

				Secondary	Structure*
	# of aa	M.W.	IEP**	Helix	Strand
Protein A	544	60937.35	9.43	149 (27.4%)	39 (7.2%)
Protein B	841	93081.33	8.97	139 (16.5%)	100 (11.9%)

^*Number of the amino acids to form indicated conformations

^**iso electric point

Fig.2.

Expected secondary structures of the Protein A (544 amino acids) and Protein B (841 amino acids)

Protein expression without chaperones.

First, protein expression was attempted without the help of chaperones. Figure 3 depicts chaperone-less expression of the Protein A. As shown, much more target protein was present in the pellet lysate than in the cell lysate, which indicated that the protein was contained as the inclusion body of the bacterial cytosol. This suggested that this protein was insoluble because it was not normally folded. Protein B showed the same tendency (data not shown).

Fig.3.

SDS-PAGE analysis of Protein A constructed pET30a expression in E.coli BL21(DE3). Lane 1: Supernatant of cell lysate without induction, Lane 2: Supernatant of cell lysate with induction for 16 hrs at 15°C, Lane 3: Supernatant of cell lysate with induction for 3 hrs at 37°C, Lane 4: Pellet of cell lysate without induction, Lane 5: Pellet of cell lysate with induction for 16 hrs at 15°C, Lane 6: Pellet of cell lysate with induction for 3 hrs at 37°C. An arrow indicates an expected molecular weight of the recombinant protein

Chaperone expression in BL21 (DE3) E. coli.

After transformation with the chaperone vectors, we added the inducers (L-arabinose and/or tetracycline) to the bacterial suspension and incubated for 4 hours at 37°C with continuous shaking. After that, cell lysate was extracted, and the expression of chaperones was confirmed by SDS-PAGE. Expected size of the chaperones was observed by GelCode Blue gel staining (Figure 1A).

Protein expression with chaperones.

The proteins of interest were co-transformed with each chaperone, and induction was performed at 37 °C for 3 hrs. Levels of protein in the cell lysates (i.e., the expression levels of soluble protein) were then evaluated by western blotting. Protein A had the highest soluble expression when co-transformed with pGro7 (groEL-groES) and pTf16 (TF) vectors, while Protein B had the highest soluble expression observed with pKJE7 (dnaK-dnaJ-grpE) vector (Figure 4A–C). As shown in Figure1B, the chaperone vectors used in this study expresses three groups of chaperones, GroE, Dna, and Tf, alone or in combination. Of note, the co-expression of groES/EL and TF (pG-Tf2) with Protein A erased the single effects of each chaperone (pGro7 and pTf16). Alternatively, the co-expression of groES/EL and dnaK/J/grpE (pG-KJE8) abolished the effect of groES/EL alone (pGro7). Also, for Protein B the co-expression of groES/EL and DnaK/J/GrpE (pG-KJE8) reversed the single effect of DnaK/J/GrpE (pKJE7) and rather inhibited soluble expression of the protein in comparison to the control (Figure 4A–C). That is, it seems that the effects of individual chaperones may be offset depending on the combination.

These results suggest that proteins with different properties, specifically, differences in the ratio of helix and strand, IEP and size may have different affinities for chaperones. In order to further investigate the relationship between protein properties and affinity to chaperones, five additional protein expression experiments (Protein C-G) were performed. Specifications of Protein C-G are summarized in Table 3. As shown in Figure 5A and B, proteins with >25% helix in their secondary structure (Protein C and D) showed better compatibility to chaperones pGro7 and pTf16. On the other hand, proteins with a helix ratio of 10–20% (protein E and F) showed a particularly strong ‘affinity’ for pKJE7. These results were consistent with the findings from the experiments using Protein A (27.4% helix) and Protein B (16.5% helix). As an additional finding, Protein G, which has an exceptionally low helix content of 7.9%, showed different chaperone selectivity than either Protein A or B. Thus, there was some regularity between the helix content of the protein and the chaperone selectivity, whereas the relevance was much more ambiguous with the strand content, protein size or the IEP (Supplement 3A–C). These results summarized in Figure 5 C indicate that the helix content of the protein may be a more critical factor in predicting chaperone success.

Table 3.

Specifications of Protein C-G

	Name	Description	RefSeq
Protein C	mitogen-activated protein kinase 7	serine/threonine-specific protein kinase	NM_139032.3
Protein D	protein kinase C, zeta	serine/threonine-specific protein kinase	NM_001033581.1
Protein E	platelet-derived growth factor receptor, beta	receptor tyrosine kinase	NM_002609.4
Protein F	forkhead box O1	transcription factor	NM_002015.4
Protein G	epidermal growth factor receptor	receptor tyrosine kinase	NM_201282.2

Culture condition and soluble expression.

In order to find more efficient culture conditions for soluble expression, low temperature (22.5 °C) expression induction for 20 hrs was tried in addition to the usual expression conditions of 37 °C and 3 hours. Both Protein A and Protein B showed higher soluble expression under the condition of 22.5 °C for 20 hrs induction, but the effect was more remarkable for Protein B (Figure 6AB).

Discussion

There are various molecular chaperones in the cell, each of which exerts different properties and activities, and regulates processes such as folding, transport, and degradation by interacting with nascent proteins (18, 19). However, the detailed mechanism is unknown. This knowledge gap can present a substantial impediment to recombinant protein expression, since the choice of a chaperone system must be determined by trial and error for every protein. Chaperones act on denatured state proteins that do not have a higher-order structure. Although they function by direct interaction with proteins, the relatively weak interaction between chaperones and proteins and the high flexibility of proteins makes high-resolution conformational analysis difficult. Therefore, the knowledge regarding the binding property of chaperones to their substrate proteins is limited, and it has been virtually impossible to predict the affinity to the specific protein from its structure.

The proteins A and B used in this study are experimental recombinant proteins in which an enzyme is connected to a common short leading sequence and have some degree of difference in properties as shown in Figure 2 and Table 2. Here we report that each protein showed unique affinities for chaperones. Specifically, Protein A showed a strong interaction with GroEL/ES or TF, while Protein B had the best compatibility with the combination of DnaK/J/GrpE. We then tested the chaperone affinity with five more proteins with various properties and found that proteins with similar helix content in their secondary structure tend to be successfully aided by similar chaperones. Interestingly, our plotting has identified a clear dependence: proteins with more helix in their secondary structures showed better compatibility to chaperones pGro7 and pTf16, those with minimal helix percentage seemed more affine to chaperones KJE8 and pG-Tf2, while the ‘mid-range’ was taken by pKJE7 (Fig. 5C). On the other hand, no clear regularity to chaperone selectivity was found with respect to strand content, size or IEP of the proteins (Supplement 3). The cause of this difference in chaperone selectivity has remained unknown, but it was impressive that the specificity of the chaperone-substrate protein compatibilities were so clearly distinct.

We noted that co-expression of several chaperones at once not only did not improve the protein purification but in fact made things worse (Fig. 4). Possible explanations for this inhibitory effect may include, we speculate, cytoplasm overcrowding, competition for co-factors like co-chaperones and for ATP as energy source. In eukaryotes, active transport (including ATP-dependent diffusive-like motion) involves protein motors and cytoskeletal filaments. In the absence of cytoskeletal motor proteins, bacteria are thought to primarily rely on diffusion for molecular transport and cytoplasmic mixing. Therefore, diffusion determines the mobility of cytoplasmic constituents and hence sets the limits at which molecular interactions can occur. However, the bacterial cytoplasm is an aqueous environment that is extremely crowded (20–22), and therefore not the best place for protein-protein interaction to take place. Overexpression of extraneous proteins may further increase the congestion within the bacterial cytosol, which may be the reason for the negative effect of co-expression of multiple chaperones. Alternatively, there may be competition between the different mechanisms employed by different chaperones on the nascent substrate polypeptide; competition for energy substrates is another possibility. Elucidation of how the properties of substrate proteins are related to these hypotheses in chaperone-assisted protein expression will be an important goal for future research.

To pursue more effective protein expression conditions, we tried low temperature (22.5 °C) and long hours (20 hrs) cultures, and obtained positive results with both Protein A and Protein B. It has been experimentally known that inclusion body formation and protein disintegration can be avoided by lowering the culture temperature after expression induction from the normal 37°C to 20 to 30°C. In this low temperature condition, the transcription rate from the promoter and the translation rate in the ribosome are slow, therefore the time required to carry out the proper folding for the original structure can be secured (23). It is also speculated that various factors such as a decrease in the proteinase activity also involved in protein stability to conform the overall structure. In addition, by culturing at low temperature, the expression of contaminating proteins in the host cell is suppressed and the efficiency of subsequent protein purification is improved. Although the lower limit of growth temperature of E. coli is about 7.5 °C (24), in general, lowering the cultivation temperature under 15 °C has not been applied to E. coli expression systems. However, there are a few examples of successful expression of recombinant proteins with temperature of lower than 15 °C (21, 25–27). Therefore, it seems that there is still room for consideration regarding the setting of culture temperature.

Although co-expression of molecular chaperones is an effective means of obtaining functional soluble proteins, determination of optimal conditions by trial and error is time and labor intensive. The ability to predict the combination of chaperones for a target protein in the context of optimal expression conditions (temperature, time, etc.) is therefore greatly desirable. In that sense, it is strongly anticipated that a high-resolution three-dimensional structure analysis of chaperone-protein complexes will become possible with advances in NMR methods and other technical innovations. In addition, over the last decade, several methods to assess the effects of chaperones in vitro have been reported. For example, Niwa et al. utilized a reconstituted cell-free translation system for this purpose (18), while Hristozova et al. developed semi-high throughput method, based on measuring enzyme activity in a 96-well plate format, where the protective activity of a putative chaperone is defined by the retention of enzyme activity (28).

Here we provide an insight into predicting the potential effect of chaperones based on the aspects of the substrate protein structure; while this isn’t a comprehensive analysis and further refinement will be needed, we find our study will be helpful in bacterial protein expression technology.