Enhanced performance of polymer-polymer aqueous two-phase systems using ionic liquids as adjuvants towards the purification of recombinant proteins

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


INTRODUCTION:
The advent of biopharmaceuticals in modern medicine for the treatment of numerous diseases brought a significant impact on the improvement of global health. Proteins (including antibodies and/or other recombinant proteins) currently dominate the biopharmaceuticals market, with over 300 drug products approved and many more undergoing preclinical and clinical trials [1,2], and with global sales reaching $100 billion in 2017 [1]. Among these, interferon alpha-2b (INFα-2b) has become widely applied in the treatment of cancer, hepatitis, and hairy cell leukemia [3] [1]. Owing to the wide spectrum of biological activities displayed by INFα-2b, namely in terms of antiproliferation, immunomodulation and antiviral properties [3], there has been an increased demand for the development of cost-effective manufacturing processes of recombinant human INFα-2b.
The INFα-2b upstream phase has been typically accomplished through recombinant protein production in a host microorganism, such as Escherichia coli (E. coli) [4,5].
Within the upstream stage, the major advantage of protein expression strategies in the form of inclusion bodies is that they can be produced in high concentrations, so that the amount of generated product often outweighs the additional downstream steps, which can boost time/space yields for recombinant protein production [6]. Although inclusion bodies have been seen as inactive protein aggregates, a new term of non-classical inclusion bodies has emerged in more recent years, which are believed to contain reasonable amounts of correctly folded protein, while exceeding 40% of total recombinant protein produced in cells [7,8]. 5 The downstream processing of biopharmaceuticals comprises the removal of the process-and product-associated impurities, as well as other contaminants [9], allowing to obtain a purified protein sample. Currently, liquid chromatography is the key technique of most purification processes resorting to protein biopharmaceuticals [10], and has been applied to the purification of IFNα-2b comprising ion-exchange/size exclusion matrices [5,11] or immobilized metal-affinity matrices/reverse-phase HPLC [12]. However, chromatography may present some limitations in terms of the high cost derived from the resin's low stability [13]. To overcome such limitations, alternative downstream processes have been investigated, including non-chromatographic operations, namely membranebased procedures [14], high-performance tangential flow filtration [15], high gradient magnetic fishing [16], precipitation [17], crystallization [17] and aqueous two-phase systems (ATPS) [18]. Initially proposed by Albertsson as a separation technique [19], ATPS are liquid-liquid systems formed when two aqueous solutions of two polymers, or a polymer and a salt, are mixed above given concentrations [20]. Some of the advantages displayed by ATPS include their biocompatibility, mostly due to the high-water content in both phases. Polymer-polymer and polymer-salt ATPS have been applied in the separation, recovery, and purification of several (bio)molecules, including antibiotics, proteins, DNA, and enzymes [21]. In particular, ATPS formed by PEG 8000 and the dibasic potassium phosphate salt were investigated for the purification of inclusion bodies resulting from the overexpression of external genes in prokaryotes [22]. Most polymerpolymer-based ATPS investigated are composed of PEG and dextran [23][24][25]. However, due to the high cost of some dextran polymers, research has evolved towards the application of less expensive polymers, among which polypropylene glycol (PPG) [26].
In general, PEG and PPG present low toxicity and are approved by the Food and Drug 6 Administration (FDA), being both listed in the FDA's Inactive Ingredient Guide for uses in topical, oral and other formulations [27,28].
A common drawback associated to polymer-polymer-based ATPS is that they exhibit a small range of polarities between the two phases [20], thus hampering their effective application in the purification of target proteins from complex samples. To overcome such limitation, it has been proposed the functionalization of PEG with glutaric acid to improve the recovery yields of immunoglobulins [29,30]. More recently, and with the same goal of improving the ATPS separation performance, ionic liquids (ILs) were proposed as adjuvants in polymer-salt ATPS [31][32][33][34][35]. This strategy has been successfully applied to increase the target proteins purification performance of PEG-citrate salt [31,33,34], PEG-potassium phosphate [32], and PEG-ammonium sulphate [35] systems, although to the best of our knowledge was never attempted with polymer-polymer ATPS.
Up to date, ATPS composed of alcohol/salt [36] and PEG/salt [37] were investigated for the purification of IFNα-2b.
In addition to ATPS, three-phase partitioning (TPP) approaches based on ammonium sulfate aqueous solutions and tert-butanol can be used to concentrate and purify proteins, involving the recovery of the target protein as a precipitate at the interface of the two liquid phases [38]. In particular, TPP has been used for the refolding of proteins from urea-solubilized inclusion bodies [39]. However, TPP based on ATPS can be applied as well for the purification of target proteins by inducing their precipitation at the ATPS interface, while avoiding the use of tert-butanol. Recently, IL-based TPP have been proposed by Alvarez et al [40,41] to recover model food proteins at the ATPS interface.
Herein it is described the production and purification of IFNα-2b from E. coli BL21 (DE3) inclusion bodies using PEG -600-PPG-400-based ATPS/TPP with ILs as adjuvants (at 5 wt%). Previous reports [31][32][33][34][35] investigated the use of ILs as adjuvants in polymer-7 salt systems, which are composed of salts presenting a high salting-out ability and a high ionic strength [42]. Distinctly, the current work focuses on two polymers as phaseforming components, thus providing a low ionic strength, and hence a more amenable environment to address the relevance of the ILs chemical structure to tailor the phases characteristics, which is beneficial when envisioning the purification of a target protein from a complex medium rich in other proteins. The ATPS process was combined with TPP, in which the purification of the target protein was aimed at the PEG-rich phase with the simultaneous precipitation of the remaining proteins at the ATPS interface.

Materials, strain, and media:
The plasmid pET-3a containing a codon-optimized version of the gene coding for human IFNα-2b (NCBI accession KY780371.1, full nucleotide sequence given in the Supplementary Information) -pET-3a_IFNα-2b -was acquired from Genscript

Biosynthesis and recovery of IFNα-2b:
Recombinant production of IFNα-2b was performed based on the protocol described by Ling and co-workers [36], yet with minor modifications: E. coli BL21 (DE3) was cultivated in shake-flasks containing SOB medium, and induction was performed with 1 mM IPTG when the optical density (600 nm) reached 0.2. Cell lysis was performed using glass beads according to the protocol previously reported by Passarinha and coworkers [43], while the subsequent recovery of IFNα-2b from inclusion bodies was achieved using a buffer containing urea in an alkaline pH, based on the works developed by Prazeres [4] and Mukherjee [5] research groups and general guidelines provided by Palmer and Wingfield [44]. Additional details on the recombinant production and recovery of IFNα-2b are given in the Supplementary Information.

Phase Diagrams:
The phase diagrams of ATPS composed of PEG 600 + PPG 400 + 10 mM Tris  Fig. 1. Each ATPS binodal curve was determined by the cloud point titration method at 25 (± 1) °C and atmospheric pressure, as previously described [45].

Purification of IFNα-2b in polymer-polymer aqueous two-phase systems:
The purification of IFNα-2b in PEG-PPG-based ATPS was evaluated in a common mixture point (30 wt% of PEG 600 + 30 wt% of PPG 400 + 5 wt% of IL). After careful weighing each ATPS component (including IL if present), systems were homogenized, 22.5 wt% of dialyzed inclusion body fraction was added (corresponding to a total protein mass of ca. 400 µg), and 10 mM Tris (pH 8) was added to make up 100 wt%. Subsequently, the mixture was stirred for 5 min at room temperature (25 (± 1) °C) and was then centrifuged (3500 rpm for 2 min at room temperature) to enhance phase separation. Phases (PEG-rich bottom and PPG-rich top phases) were isolated using a syringe and their weight determined. The precipitate was also isolated and dissolved in 10 mM Tris (pH 8). Both phases and the dissolved precipitate were analyzed for total proteins and IFNα-2b contents. Total protein concentration was determined by the Pierce BCA protein assay using convenient dilutions of ATPS phases, yielding typically concentrations below 3 wt% of each polymer that avoid polymer interferences in the determination of the proteins content, in agreement with the recommendations of Beutel and co-workers [46]. The migration pattern of IFNα-2b and remaining proteins between the two phases and precipitation at the interface were evaluated by SDS-PAGE; additionally, the levels of IFNα-2b were quantified by enzyme-linked immunosorbent assay (ELISA) (additional details are given in the Supplementary Information). Using the values obtained for the concentration of IFNα-2b and/or total proteins in each phase and at the precipitate, both the purification factor and extraction efficiency were calculated to evaluate the purification performance of the studied systems.
The overall purification performance of the ATPS was evaluated through the determination of the purification factor (PF) and equation 2: where [IFNα-2b] represents the concentration of IFNα-2b (µg/ml) in the PEG-rich phase and in the initial solubilized sample, and [total proteins] represents the concentration of total proteins (including IFNα-2b) in the PEG-rich phase and initial solubilized sample.

11
The migration pattern or precipitation of total proteins in each sample was analyzed by SDS-PAGE according to the method of Laemmli [47]. To remove interferences caused by polymers and/or ILs, samples were precipitated using acetone (1:4 (v/v)) and the mixture was incubated at -20 °C overnight, followed by centrifugation at 13000 rpm (4 °C for 15 min) [48]. The resulting pellet from each ATPS phase was then and scanning speed, 50 nm/min.

Partition coefficients of ionic liquids:
The partition coefficients of each IL (K IL ) in ATPS were determined to evaluate their influence on the IFNα-2b partitioning between the two phases following the protocol previously reported by Coutinho and co-workers [45].

Production and recovery of IFNα-2b:
The purification of IFNα-2b from recombinant bacterial lysates is herein addressed due to this protein potential as a major biopharmaceutical currently employed in the treatment of several human diseases [1]. Considering that E. coli continues to be a major host for manufacturing biopharmaceutical proteins [1], and the high product titers delivered by E. coli inclusion bodies [7], it is of utmost importance to develop protein purification strategies from inclusion bodies.
The manufacture of IFNα-2b is commonly performed using E. coli, for which distinct operational parameters such as medium composition, inducer concentration and period of induction influencing the production of IFNα-2b should be initially optimized.
After testing different culture media, it was observed that the production of monomeric  [4] and Mukherjee [5] research groups on the solubilization of IFNα-2b from inclusion bodies, and general guidelines provided by Palmer and Wingfield [44], an optimized strategy towards the efficient solubilization of IFNα-2b was here adopted. The designed protocol includes two

ATPS phase diagrams:
To use ATPS as an extraction/purification strategy it is of utmost importance to define their monophasic and biphasic regions by the determination of the respective phase diagrams. Considering that the addition of ILs as adjuvants in ATPS can influence not only the partition of proteins but also the extent of the biphasic region, the binodal curves for quaternary (PEG 600 + PPG 400 + 10mM Tris pH 8 + 5 wt% IL) and ternary (same phase-forming components but without IL) systems were first determined at 25 (± 1) °C and atmospheric pressure. The respective phase diagrams in an orthogonal representation are shown in Fig. 1. Detailed experimental data and the corresponding regression parameters obtained by Eq. (1) are reported in Table S1 in the Supplementary Information.
All mixture compositions above each binodal curve shown in Fig. 1 [45,49,50], which are grouped in increasing order of hydrophilicity. According to the data shown in Fig. 1A, there is a good correlation between the ability of each IL cation to donate protons and to establish hydrogen-bonds and their ability to induce the formation of two phases. On the other hand, the predicted hydrogen-bond basicity (β) of [C 4 mim + ]-based ILs paired with bromide, chloride, and acetate anions are, respectively, 0.87, 0.95 and 1.20 [50]. By being the most hydrophilic IL, the acetate-based IL is the one with the highest impact in the formation of two phases.
As verified with the IL cation effect, the higher the ability of the IL anion to accept protons and establish hydrogen-bonds, the higher the IL salting-out ability and capacity to create two-phase systems (Fig. 1B). Although with a different anion, the predicted hydrogen bond acidity (α) of bis(trifluoromethylsulfonyl)imide-([NTf2] -)-based ILs paired with 1ethyl-3-methylimidazolium-, 1-butyl-3-methylimidazolium-and 1-hexyl-3methylimidazolium-cations was reported to be, respectively, 0.750, 0.692 and 0.659 [49], allowing to address the IL cation effect to donate protons. As discussed above, also when analysing the IL cation alkyl side chain length, there is a common trend between the IL cation hydrogen-bond acidity and ability to create ABS (Fig. 1C).
Based on the exposed, the effect of ILs on the PEG-PPG-water binodal curves follows the IL ions capacity to accept or donate protons (measured as a function of α and β values), and thus to establish hydrogen-bonds, which may occur with water or the -OH terminal groups of both polymers. An increase of the ILs' capacity to accept or donate protons also correlates with the ILs partitioning to the PEG-rich phase, the phase where ILs are preferentially enriched (data given in the Fig. S3 in Supplementary Information). As a consequence, the differences in hydrophobicity of both ATPS phases are enhanced by using ILs with higher hydrogen-bond donor and acceptor characteristics, resulting in increased phase separation of the PEG-PPG systems herein investigated.
Previously, it was reported that ILs are preferentially partitioned towards the PEG-rich phase when applied as adjuvants in PEG 600-Na 2 SO 4 [51] and PEG 400-(NH 4 ) 2 SO 4 ATPS [45], the most hydrophobic phase in such systems. In these reports, an increase of the ILs' hydrophobic nature leads to a lower affinity for water, thus enhancing phase separation [45,51]. The results obtained in this work are in good agreement with previous reports since ILs preferentially partition to the PEG-rich phase, and the higher the partition extent the higher is the phase separation ability. However, in the current work, the PEG-rich phase is the most hydrophilic (when compared to the PPG-rich phase and not with a salt-rich phase as shown in the literature [45,51] It has been additionally shown that PEGs with a molecular weight lower than 3400 g mol -1 are not able to form two-phases at 25 °C with protic ILs [56]. Overall, the results and trends reported in the literature point out that at the concentrations of ILs applied (5 wt%) in the current work, such compounds will not form two-phases with PPG 400 or PEG 600.

3.3.
IFNα-2b purification using polymer-polymer aqueous two-phase systems using ILs as adjuvants: PEG and PPG were investigated as phase-forming components of ATPS aiming the purification of IFNα-2b from E. coli inclusion bodies. The biocompatible behaviour displayed by such polymers coupled with their low molecular weight (600 and 400 g mol -1 , respectively for PEG and PPG) allows to have relatively low viscous phases. On the other hand, the use of ILs as adjuvants in polymer-polymer ATPS represents a more amenable environment in which the effect of the ILs is less masked by salting-out effects occurring in the previously studied salt-polymer-based ATPS [31][32][33][34][35].
The purification of IFNα-2b was investigated using a common mixture point in ATPS: PEG 600 at 30 wt% + PPG at 30 wt% + IL at 5 wt%. In general, an interfacial precipitate corresponding to precipitated proteins was formed in all systems (Fig. S4 in the Supplementary Information), fitting within the three-phase partitioning approach previously applied for the purification of proteins [57] and antibodies [58]. Contrarily to these reports, the interfacial precipitate in the current work is not preferentially enriched in the target protein, but instead in the remaining proteins.
The migration patterns and precipitation of the target protein and protein impurities were evaluated by SDS-PAGE, whose results are shown in Fig. 3 and Figs. S4 and S5 in the Supplementary information. In general, no proteins are found in the top PPG-rich phase, which is the most hydrophobic ATPS phase. Instead, both IFNα-2b and protein impurities preferentially migrate to the bottom PEG-rich phase or precipitate at the interface, in which the latter is most enriched in protein impurities. The SDS-PAGE profile of the initial sample (Fig. 3) shows that IFNα-2b appears as a 19 kDa band (identity confirmed by western-blot in Fig. 2). Moreover, multiple bands with higher molecular weights corresponding to E. coli endogenous proteins demonstrate the evident complexity of the starting sample and the inherent difficulty to separate IFNα-2b from proteins with similar properties.
Without the addition of an adjuvant, IFNα-2b migrates to the PEG-rich phase (Fig.   3), although some target protein precipitates at the interface (Fig. S4 in the Supplementary   Information). When ILs are added as adjuvants, it is seen that more hydrophobic and higher volume ILs, such as [N 4444 ]Cl, lead to a lower recovery yield of IFNα-2b in the bottom phase (Fig. 3). However, good recoveries of IFNα-2b in the PEG-rich phase were observed using the investigated imidazolium-, ammonium-, piperidinium-, and pyrrolidinium-based ILs, outperforming the system without the addition of adjuvant (Fig.   3). Overall and in all systems, the PEG-rich phase was identified as the ATPS phase promoting the highest recoveries of IFNα-2b in a more purified form.

22
The SDS-PAGE analysis carried out provided useful insights about the purification performance of the investigated PEG-PPG systems. However, the need of an acetone precipitation step and subsequent dissolution in a specific buffer may not guarantee that all proteins are equally precipitated, hence preventing an accurate evaluation of the proteins partition behavior and precipitation. Therefore, ELISA was used to quantify the levels of biologically active IFNα-2b in each ATPS phase and BCA was used to determine the total proteins content in all samples. Based on these assays, the  Supplementary Information). Accordingly, in addition to the quantity of IL in the bottom phase which seems to play a relevant effect as appraised by the IL partition coefficients discussed above, the chemical structure of each IL seems to play a more significant role to improve the studied ATPS separation performance.
All ILs have a positive charge center, allowing the establishment of electrostatic interactions with the target protein, which is negatively charged at pH (8) (IFNα-2b pI = 5.9 [5]). Although it has been reported that NaCl increases the purification factor of IFNα-2b in alcohol-salt [36] and polymer-salt [37] ATPS, it should be remarked that ILs additionally allow a range of non-covalent interactions to be established with the target protein, including electrostatic, hydrophobic, π-π, cation-π, and hydrogen-bonding interactions that can be exploited towards the development of enhanced IFNα-2b 24 purification processes. Taking into account the results obtained and ILs chemical structures, and in addition to electrostatic interactions common to all ILs, the enrichment of IFNα-2b in the PEG-phase containing ILs as adjuvants seems to be driven by aromatic interactions between the aromatic residues of IFNα-2b and aromatic IL cations and hydrogen bonding interactions established between IL anions and IFNα-2b.

CONCLUSIONS:
The purification of IFNα-2b as a major protein biopharmaceutical using polymerpolymer ATPS and ILs as adjuvants was attempted in this work. The IFNα-2b upstream stage consisted of E. coli BL21 (DE3) cultivation with the target protein accumulated as inclusion bodies, followed by E. coli lysis with glass beads and inclusion bodies solubilization with a buffer containing urea and at alkaline pH. After the phase diagrams determination, ATPS were then investigated for IFNα-2b purification by enriching IFNα-2b in the PEG-rich phase, with the simultaneous precipitation of the remaining proteins at the interface (fitting within the three-phase partitioning approach). Overall, ILs present in higher quantities in the PEG-rich phase, being those composed of aromatic cations and anions with high hydrogen-bond basicity, allow to obtain higher IFNα-2b purification factors. Accordingly, electrostatic, π-‧‧‧π and hydrogen-bond interactions seem to be the main forces ruling the target protein selective partitioning to the PEG-rich phase, where ILs are preferentially partitioned. The IFNα-2b recovered in the bottom PEG-rich phase was shown to be immunologically active, with its secondary structure preserved. In summary, this work reinforces the potential effect brought by the use of ILs in small concentrations as adjuvants in PEG-PPG ATPS, by enhancing the purity of high-value proteins such as IFNα-2b.

ACKNOWLEDGEMENTS:
This work was developed within the scope of the project CICECO-Aveiro