Encoded by the RNH1 gene, ribonuclease inhibitor is a protein expressed in the cytosol; it inhibits pancreatic ribonucleases, merging with them into complexes and protecting cellular RNA from premature degradation [1, 2]. The inhibiting effect stems from non-covalent binding of protein ribonuclease molecules, which leads to blocking of both RNA-binding and catalytic ribonuclease domains [3]. Ribonuclease inhibitor is a fairly conservative protein, with homology over 70% in mammals - humans, mice, horses, and pigs. It consists of 15–16 leucinerich repeats (the exact number of repeats depends on the type of mammal), and each repeat includes 28–29 amino acids, has one alpha helix and one beta sheet. The repeats make up a U-shaped structure weighing approximately 50 kDa [4, 5]. Physico-chemical properties of ribonuclease inhibitor put it among acidic proteins, with isoelectric point in the region of 4.5. Thirty-two cysteine residues, which enable the enzyme, are another important characteristic of ribonuclease inhibitor. In native protein, cysteines are in a reduced state; they participate in the formation of a hydrophobic protein core. Oxidation of cysteine residues is accompanied by appearance of disulfide bridges between them, and conformational changes weaken the activity of ribonuclease inhibitor [6, 7].

Protein preparations of the ribonuclease inhibitor are of biotechnological value for the purposes of both research (in vitro RNA synthesis, in vitro translation, cDNA production) and mass production of medicines based on artificial RNA molecules. Therefore, since its discovery, there have been attempts to synthesize this enzyme in large quantities. Historically, the first of the said attempts sough to make the ribonuclease inhibitor from mammalian tissues [8]. As the recombinant DNA technologies matured, the focus of attention shifted to the most widely used E. coli bacterial system. However, recombinant ribonuclease inhibitor tends to aggregate in E. coli, like other proteins containing a large number of hydrophobic amino acids, and with cysteine residues in particular. Several other factors, including the high rate of translation associated with transcription in bacteria, bring accumulation of such recombinant proteins as inclusion bodies formed from aggregates of improperly folded protein molecules [9]. Therefore, early experiments designed to produce a functionally active soluble ribonuclease inhibitor in E. coli often ended with the entire protein accumulating as inactive aggregates, and when these were refolded in vitro, the resulting protein most often lost its potency, and the yield thereof was very low [10].

Various approaches have been developed to solve this problem. One of them suggests employing a vector with a weak promoter (for example, Ptrp) and a minimal medium for expression of the ribonuclease inhibitor gene. Such conditions weakened the expression of the target protein gene significantly, which allowed yielding considerable amounts of the recombinant inhibitor in a soluble form, but the overall level of the final product turned out to be significantly lower compared to expression systems where the target gene is controlled by the T7 promoter [2]. Another approach involves enriching the culture medium with DTT (dithiothreitol), thus keeping cysteines in a reduced state, which has a positive effect on the folding of many proteins prone to aggregation, including ribonuclease inhibitor [11]. This technique was further developed with cultivation under anaerobic conditions at a reduced temperature and with co-expression of the GroELS cellular chaperone, which brings the production of ribonuclease inhibitor to a fairly high level [12]. The downsides of this method are its complexity and the large number of additional components required, both complicating the technological process when the goal is to synthesize the protein at scale. Finally, another approach aims at producing the target product as a chimeric molecule with an fusion polypeptide that improves solubility of the target protein. In an earlier study, MBP fusion polypeptide was shown to positively affect production of the ribonuclease inhibitor in soluble form in E. coli [4].

The purpose of this work was to identify the conditions of production of the recombinant human ribonuclease inhibitor in E. coli that would simplify the process in the laboratories and facilities synthesizing the product at semi-industrial scale.

MATERIALS AND METHODS

Plasmid design

The genetic constructs used in this work were produced by the restriction ligase DNA cloning method. The DNA sequence of the RNH1 gene encoding human ribonuclease inhibitor was synthesized de novo with codon optimization for E. coli (IDT, https://www.idtdna.com /). Amplification of all DNA sequences followed the manufacturer's protocol (NEB) and employed Q5 high-fidelity polymerase. Cloning of the ribonuclease inhibitor gene into the set of vectors with fusion polypeptides based on pET28a (Novagen; USA) and pSol (Lucigen; USA) was done by restriction sites NdeI and NotI, in an enzymatic reaction, with two corresponding restriction endonucleases (Sibenzyme; Russia); the procedure lasted for 1 hour at the temperature of 37 °C. Next, we did electrophoresis in agarose gel and purified DNA fragments therefrom with the help of a GeneJET Gel Extraction Kit (Thermo; USA). These fragments, corresponding to the vector part and the ribonuclease inhibitor gene, were then used in the ligation reaction with T4-DNA ligase (NEB), which lasted for 30 minutes at room temperature. Competent E. coli 10G cells (Lucigen; USA) were transformed with the help of electroporation (Bio-Rad; USA) as prescribed by the manufacturer. Plasmid DNA was isolated from clones containing correct inserts, with confirmed correctness by the Sanger sequencing reaction.

Production of recombinant proteins in __E. coli__

We used the following strains intended for production of recombinant proteins in E. coli: BL21 (DE3) (Invitrogen; USA); KRX (Promega; USA); Rosetta (DE3) pLysS (Novagen; USA); BL21 Star (DE3) (Invitrogen; USA), Origami 2 (DE3) (Novagen; USA). Using plasmids containing genes of fusion polypeptides and the ribonuclease inhibitor, we electroporated each strain and seeded the LB liquid medium in test tubes (35 ml, including selective antibiotics) with a transformation mixture (100 µl) . The incubation lasted for 12–14 hours at 37 °C, with the stirring speed of 180 rpm. Next, we transplanted the grown culture in a ratio of 1:200 into test tubes or flasks containing the required amount of the LB culture medium (10 - 500 ml) with selective antibiotics. The cell culture was built up to the optical density (OD) of = 0.5–0.6 in a shaker at 180 rpm at 37 °C. Once the needed OD was reached, we cooled the culture on ice for 10 minutes, added the 1 mM IPTG inductor for pET28a-based vectors or 0.2% L-rhamnose for pSol-based vectors, and incubated at different temperatures while stirring (120 rpm) for 10 hours (16 °C) or 2 hours (37 °C). Co-expressed cellular chaperone genes were induced by the introduced L-arabinose (0.5%) or tetracycline (5 ng/ml), with pre-incubation before adding the target gene expression inducers. Then, bacterial cells were deposited by centrifugation for 20 minutes at 6000 g and 4 °C.

To measure solubility of the resulting protein, we transferred the deposited cells into buffer A (50 mM Tris-HCl pH 7.5; 0.3 M NaCl; 0.005 M imidazole) and destroyed them using a Qsonica Q700 ultrasound disintegrator until the cell suspension on ice was clarified (ultrasonic pulses for 10 s, cooling — 60 s). The destroyed cells were deposited by centrifugation (15000 g), which lasted 30 minutes at 4 °C. Next, we collected samples of the supernatant containing the target protein in the soluble form, and the precipitate with aggregates was dissolved in urea (2 M) for further electrophoretic analysis. PAAG electrophoresis (10%) was used to separate protein under denaturing conditions, as prescribed by the standard methods. After staining with Coomassie blue dye and washing it off, we photographed the gels and quantified the soluble and insoluble forms of protein. Image Lab application (Bio-Rad; USA) enabled densitometric analysis of the gels.

Protein purification using affinity chromatography

After ultrasonic destruction of the biomass and centrifugation of cellular debris, we filtered the remaining supernatant through a membrane with 0.22 µm pores. The resulting preparation for chromatographic purification contained the soluble fraction of proteins in buffer A (50 mM Tris-HCl pH 7.5; 0.3 M NaCl; 0.005 M imidazole); it was applied on Nuvia IMAC Resin (BioRad), a chromatographic sorbent. The proteins were eluated by chromatographic buffer B (50 mM Tris-HCl pH 7.5; 0.5 M imidazole). In a Spin-X centrifuge concentrator (pore size 10 kDa), purified preparation was concentrated from 5 to 1 ml. If necessary, we further treated it with TEV protease (following manufacturer's recommendations) overnight at 4 °C, and then repeated chromatographic purification, but after that only one fraction, that which did not bind to the sorbent and contained purified ribonuclease inhibitor, was collected. The target fraction was dialized against the storage buffer (40 mM HEPES-KOH pH 7.6, 100 mM KCl) at 4 °C. Glycerin (50%) and DTT (8 mM) were added for long-term storage at –20 °C.

RESULTS

Our first step in identification of the optimal conditions for production of soluble recombinant ribonuclease inhibitor was extension of the list of screened fusion polypeptides that increase solubility of this protein, since this approach has already been tested earlier [4]. For this purpose, we have constructed a set of twenty vectors based on two expression plasmids, pET28a (Novagen; USA) and pSol (Lucigen; USA). Two vectors contained the human ribonuclease inhibitor (RNH1) gene codon optimized for expression in E. coli, with its sequence encoding polyhistidine tag at the protein's N-terminus, which ultimately encoded the target protein molecule with a histidine tail for affine purification. The remaining plasmids encoded chimeric proteins consisting of fusion polypeptides (MBP, TIG, TSF, FH8, PpiB, YrhB, SUMO, TRX, GST) and a recombinant ribonuclease inhibitor. The induction of expression from T7 (pET28a) or rhamnose (in the case of pSol) promoter in these plasmids enables production of chimeric proteins comprised of combinations of a ribonuclease inhibitor with different fusion polypeptides carrying the polyhistidine tag on the N-terminus for affinity purification (fig. 1). It should be noted that a sequence encoding the proteolysis site for TEV protease was inserted between the ribonuclease inhibitor gene and genes of the fusion polypeptides. Thus, after primary affinity purification, it is possible to split off the fusion polypeptide and use affinity chromatography to produce a pure ribonuclease inhibitor preparation.

All the resulting genetic constructs were tested under the typical conditions of production of recombinant proteins in E. coli; for the purpose, they transformed cells of BL21 (DE3) strain, and enabled the said production at 37 °C. As expected, incubation of bacterial cultures at 37 °C yielded no soluble form of the ribonuclease inhibitor but only insoluble aggregates, regardless of the fusion peptide and from what promoter the target gene's expression was initiated (data not provided). When the incubation temperature was lowered to 16 °C, variants without fusion peptides also formed inclusion bodies, but in the end, there were several chimeric proteins in the soluble form (fig. 2). Therefore, all our further attempts at production of the recombinant ribonuclease inhibitor were conducted at lower temperature (16 °C) . In particular, we registered good results when using the following fusion polypeptides: MBP (70–75% of the target product in soluble form), TIG (about 90% of the target product in soluble form) and, to a lesser extent, PpiB (about 60% of the target product in soluble form).

Stronger promoter T7 ensured considerably bigger yield of the protein in pET28a vector systems compared to the pSol vector, which relies on a weaker rhamnose promoter (fig. 2). Nevertheless, part of the target protein still transitioned into a functionally inactive state in the inclusion bodies, a problem that could be remedied by further optimization of the operating conditions.

The next step was to compare the yield of soluble ribonuclease inhibitor in different E. coli strains designed specifically for the production of recombinant proteins. As a rule, previous attempts on the task incorporated the experience of working with recombinant producers based on the E. coli BL21 (DE3) strain, but, in most cases, it is not the optimal choice in case of proteins prone to aggregation: a significant proportion of the product are insoluble aggregates. Therefore, seeking to refine the conditions of production of ribonuclease inhibitor, we compared the amounts of protein yielded by four more E. coli purpose-designed strains. Thus, in addition to the BL21 (DE3) strain, the following commercially available strains were experimented with (all with various design specifics conditioned by their intended use in the context of production of recombinant proteins). Origami strain 2 (DE3), with mutations in the genes of thioredoxin reductase (trxB) and glutathione reductase (gor), which promote formation of disulfide bonds and correct folding of some recombinant proteins in the cytoplasm of bacterial cells. BL21 Star (DE3) strain, with a mutation in the gene encoding ribonuclease E (rne131), which improves stability of the target gene's mRNA and may boost production of recombinant proteins. It was also interesting to try the Rosetta (DE3) pLysS strain, which can produce proteins without codon optimization to E. coli and also allows reducing background expression from the T7 promoter by additional expression of the T7 lysozyme gene, a natural inhibitor of DNA-dependent RNA polymerase of the T7 phage, which is of particular importance in the context of production of proteins potentially toxic to bacteria. Finally, the E. coli KRX strain, with a chromosomal copy of the T7 DNA-dependent RNA polymerase gene controlled by the rhamnose promoter, which governs basal expression better, and is also well titrated by rhamnose while producing high amounts of recombinant proteins, as ensured by the T7 promoter [13].

To see how selection of the strain affects production of the soluble form of the inhibitor, we continued working with the variants that showed best results at the previous stage, i.e., pET28a-based vectors encoding chimeric proteins, with fusion polypeptides MBP and TIG. As mentioned above, all comparative experiments with these strains implied bacterial culture incubation at 16 °C. Ultimately, it was the KRX strain that delivered the best yield (fig. 3). The possible reason behind this result is a more suitable mode of expression by the T7 DNA-dependent RNA polymerase gene controlled by the rhamnose promoter, which refines the dose of T7 DNAdependent RNA polymerase accumulated in cells; accordingly, the recombinant ribonuclease inhibitor gene expression is not "overloaded," as in the case of direct induction of the T7 promoter with IPTG of lacUV5 promoter, when it is virtually impossible to weaken expression by titration. The BL21 (DE3) and BL21 Star (DE3) strains are also good options for the recombinant ribonuclease inhibitor production when combined with the selected fusion polypeptides and cultivated at low temperatures, but their yield of the soluble product is 10–15% (depending on the polypeptide) lower than that of the KRX strain. The remaining two strains performed considerably worse, with Origami 2 (DE3) being the outsider, probably due to its specific mutations designed to create intracellular conditions for the formation of disulfide bridges. Another possible reason for the poor results is the lower amount of biomass produced by these strains compared to KRX, BL21 (DE3) and BL21 Star (DE3), which may apparently be attributed to the introduced mutations affecting viability and metabolic processes in the cells. Thus, the second stage of the work revealed that in the context of production of the soluble form of ribonuclease inhibitor, the best "strain — fusion polypeptide" combinations are E. coli KRX-MBP and E. coli KRX-TIG.

There is one more tool used against formation of insoluble aggregates in production of recombinant proteins in E. coli: cellular chaperone genes co-expression, which helps proteins to adopt a native conformation. We tested the efficacy of this tool on chaperones dnaK, dnaJ, drpE, groES and groEL. Earlier, researchers reported a positive effect of some chaperones on the production of functional ribonuclease inhibitor [14, 15]. We decided to test them in the best combinations of the KRX E. coli strain and fusion polypeptide MBP, TIG, PpiB. This experiment revealed that the maximum effect — about 10% uptick of the target protein yield — was achieved against coexpression of genes of dnaK, dnaJ, grpE (fig. 4). An earlier study mentioned that GroELS chaperone positively influenced production of the soluble form of ribonuclease inhibitor [12]; in this regard, the likely reason behind the differences with our findings is the use of different strains.

Thus, the highest yielding ribonuclease inhibitor production combination identified in this work includes the KRX E. coli strain, TIG and MBP fusion polypeptides, co-expressing dnaK, dnaJ, grpE chaperone genes, and cultivation temperature of 16 °C. To accurately assess the yield of the target protein given by the tested combinations, we used affinity chromatography on a metal chelate carrier and subsequent proteolysis reaction with TEV protease. The total amount of soluble ribonuclease inhibitor was 60 mg/l in case of E. coli KRX-TIG-dnaK, dnaJ, grpE, and 45 mg/l for E. coli KRX-MBP-dnaK, dnaJ, grpE.

DISCUSSION

The rapid development of therapeutic applications of synthetic mRNAs underpins the need for preparative amounts of proteins required to produce large amounts of RNA in vitro [16, 17]. Apart from the enzymes that synthesize and modify RNA in vitro, another important component of the reaction mixture is ribonuclease inhibitor, a protein that prevents degradation of the RNA molecules caused by ribonucleases, which is crucial for medicines. Ribonuclease inhibitor is also widely used in research tasks related to the synthesis of RNA, cDNA, and in vitro translation. The production of this protein in recombinant E. coli strains is complicated by the specificity of its structure with abundant cysteine residues, and the tendency to aggregate in bacterial cells. Previous attempts at production of the soluble form of ribonuclease inhibitor in E. coli require special conditions that significantly complicate the technological process and make it costly to scale [11, 12]. This work aimed to find the ribonuclease inhibitor production conditions and method that would deliver large yields and be cost-effective from the technological point of view.

As the first stage, we screened a number of fusion peptides that could increase the proportion of soluble ribonuclease inhibitor produced in E. coli. Previous studies have shown that polypeptides can significantly increase the yield of soluble form of recombinant proteins, including ribonuclease inhibitors [4, 18]. The ability to improve solubility of various recombinant proteins stems from the formation of chimeric molecules, the fusion part of which (polypeptide) can later be removed by proteolysis [9, 19]. We tested nine different fusion polypeptides: MBP, TIG, TSF, FH8, PpiB, YrhB, SUMO, TRX, GST; they differ significantly in their physico-chemical properties. Additionally, we tested them in two expression systems, one based on the commercially available plasmid pET28a (target gene expression controlled by a strong T7 promoter), and another on the pSOL vector (expression controlled by a rhamnose promoter). The best results were ensured by MBP, TIG, and PpiB, with the yield highest in the systems utilizing MBP and TIG, and that with PpiB performing considerably worse (fig. 2). Overall, all the experiments of this work have shown that larger polypeptides, such as MBP (42 kDa) and TIG (49 kDa), provide more of the final target product than fusions with smaller molecular weight, like PpiB (19 kDa). Comparing the yield of target products at the level of vector lines (pET28a and pSol, in this study), the unambiguous system of choice for production of ribonuclease inhibitor is that where a strong T7 promoter controls target gene expression (fig. 2).

Having established that the most productive vector is that based on pET28a, encoding chimeric ribonuclease inhibitor variants with MBP and TIG fusion polypeptides, we moved to the next stage of optimization: selection of the suitable strain of E. coli from several commercially available options. For this purpose, we compared the ribonuclease inhibitor production capacity of BL21 (DE3), Origami 2 (DE3), BL21 Star (DE3), Rosetta (DE3) pLysS, and KRX E. coli strains, all of which are designed for production of recombinant proteins, but have certain specifics, which enables the comparison and allows selecting that which maximizes production of the soluble form ribonuclease inhibitor [20]. Under the conditions of this study, the highest ribonuclease inhibitor yield was delivered by the KRX E. coli strain (fig. 3), likely because of the rhamnose promoter control over the T7 gene of DNA-dependent RNA polymerase, since this promoter has a low basal activity level and is better titrated by L- rhamnose. Apparently, this design feature allows finding the optimal balance of conditions for bacterial production of toxic or aggregation-prone recombinant proteins, while also ensuring a high yield of the target protein by reducing the amount of protein loss in inclusion bodies, which is common for production in strains with the chromosomal copy of T7 DNA-dependent RNA polymerase controlled by the lacUV5 promoter and induced by IPTG [21].

Finally, having established that the best "strain — fusion polypeptide" combinations are E. coli KRX-MBP and E. coli KRX-TIG, we decided to investigate the effect of the additional cellular chaperone genes co-expression on the yield of the target protein. For this purpose, we used a commercial kit (Takara; USA) that includes vectors carrying the genes of chaperones dnaK, dnaJ, grpE, groES and groEL, which, when produced additionally in the E. coli cells, often help "problematic" recombinant proteins to adopt the correct conformation, thus reducing the formation of insoluble aggregates [14, 15]. For recombinant ribonuclease inhibitor, the positive effect of groES and groEL on the protein production under anaerobic conditions has already been shown [12], but this approach is not always convenient when scaling technological processes to produce larger amounts of recombinant proteins. Interestingly, in our experimental conditions, aerobic cultivation of E. coli at a reduced temperature (16 °C), the best ribonuclease inhibitor yields were given by two combinations: a) MBP fusion polypeptide, dnaK, dnaJ, grpE, chaperone genes coexpression, KRX E. coli strain, and b) TIG fusion polypeptide, dnaK, dnaJ, grpE chperone genes co-expression, KRX E. coli strain (fig. 4). To accurately quantify recombinant ribonuclease inhibitor in such combinations, we performed affinity purification and then compared the the yield of the target product. The first combination produced 45 mg/l. However, the maximum yield of the final product, that is, the soluble form of ribonuclease inhibitor, was achieved in aerobic cultivation of the KRX E. coli strain at a reduced temperature (16 °C), transformed by the expression vector based on pET28a with a codon optimized ribonuclease inhibitor gene, in combination with the TIG fusion polypeptide gene, against the background of dnaK, dnaJ, grpE chaperone genes co-expression. This combination gives a yield of 60 mg/l when culturing bacteria in flasks on a standard LB medium, which is significantly higher than the results achieved earlier, and also easier from the bacteria cultivation perspective [4, 11, 12, 22].

The results of our study confirm the need for an integrated approach to the search for optimal conditions of production of aggregation-prone recombinant proteins in E. coli, especially when the goal is to scale up the process of synthesis of enzymes, therapeutic proteins, cytokine preparations or antigens for vaccines [23].

CONCLUSIONS

This work shows that a systemic approach to identification of conditions and components for production of enzymes and therapeutic recombinant proteins in E. coli, given that they are prone to aggregation and form inclusion bodies under normal bacterial cultivation conditions, allows finding combinations delivering significantly higher yields of the target protein in soluble form. Moreover, it is possible to find the balance of conditions and components needed to establish production at semi-industrial and industrial scales, and to minimize protein loss to aggregation. It should be noted that the final ribonuclease inhibitor production protocol compiled in this study can be optimized further by incorporating the experience of earlier works and opting for a nutrient media other than the standard LB, as well as introduction of cultivation in a fermenter, which will further increase the final product yield. However, even in its present form, the protocol can support semi-industrial production of this protein in preparative quantities for further use in enzymatic RNA synthesis in vitro.