Aqueous Biphasic Systems Comprising Natural Organic Acid-Derived Ionic Liquids

Abstract

Despite the progress achieved by aqueous biphasic systems (ABSs) comprising ionic liquids (ILs) in extracting valuable proteins, the quest for bio-based and protein-friendly ILs continues. To address this need, this work uses natural organic acids as precursors in the synthesis of four ILs, namely tetrabutylammonium formate ([N₄₄₄₄][HCOO]), tetrabutylammonium acetate ([N₄₄₄₄][CH₃COO]), tetrabutylphosphonium formate ([P₄₄₄₄][HCOO]), and tetrabutylphosphonium acetate ([P₄₄₄₄][CH₃COO]). It is shown that ABSs can be prepared using all four organic acid-derived ILs paired with the salts potassium phosphate dibasic (K₂HPO₄) and tripotassium citrate (C₆H₅K₃O₇). According to the ABSs phase diagrams, [P₄₄₄₄]-based ILs outperform their ammonium congeners in their ability to undergo liquid–liquid demixing in the presence of salts due to their lower hydrogen-bond acidity. However, deviations to the Hofmeister series were detected in the salts’ effect, which may be related to the high charge density of the studied IL anions. As a proof of concept for their extraction potential, these ABSs were evaluated in extracting human transferrin, allowing extraction efficiencies of 100% and recovery yields ranging between 86 and 100%. To further disclose the molecular-level mechanisms behind the extraction of human transferrin, molecular docking studies were performed. Overall, the salting-out exerted by the salt is the main mechanism responsible for the complete extraction of human transferrin toward the IL-rich phase, whereas the recovery yield and protein-friendly nature of these systems depend on specific “IL-transferrin” interactions.

1. Introduction

Aqueous biphasic systems (ABSs) are liquid–liquid systems used as extraction techniques, comprising water as the main constituent and (at least) two-phase splitting promoters such as two polymers or a polymer and a salt [1]. These systems may thus be formed free of volatile organic solvents. If their components are wisely chosen, these systems may be used as low-cost, simple, bio- and eco-friendly extraction platforms [2,3]. More recently, with the adoption of ionic liquids (ILs) as ABSs constituents, the limitations of the narrow hydrophobicity-hydrophilicity range of conventional polymeric systems have been defeated, significantly improving the efficiency and selectivity of many separation processes [4,5]. Known as “designer solvents” or “task-specific solvents”, ILs are organic salts that, through the combination of multiple cations and anions, can be designed to fit the purpose of ABS-mediated extraction processes [5,6]. Given the ILs’ ability as stabilizing media for several biomolecules [7,8,9], IL-based ABSs have been successfully applied in the domain of bioseparations, particularly involving proteins [10]. Imidazolium-based ILs, mostly those carrying fluorinated anions, are among the most widely adopted to formulate ABSs [5,10].

Although less investigated, bio-based ILs may help overcome the high cost of conventional imidazolium-based ILs while improving the green features of this class of solvents [11,12]. Examples of bio-based ILs comprise those based on at least one of the following precursors: choline [13,14], glycine-betaine [15,16], guanidine [17,18], purine and pyrimidine nucleobases [19], carbohydrates [20,21], amino acids [14,22], organic acids [13,23], fatty acids [24,25], and terpenes [26,27]. Thus far, ILs bearing a cholinium cation combined with natural acid-derived anions are the most studied and used as protein-friendly options in IL-based ABSs [28,29,30,31]. This work continues the pursuit for cheaper and bio-based ILs to form ABSs using formic and acetic acids as natural ILs’ anion precursors. To yield four distinct ILs incorporating organic acid-derived anions, tetrabutylphosphonium and tetrabutylammonium were used as the cations. These cations are among the less costly ILs’ cation sources available [32] and confer enough hydrophobicity to the IL to create ABSs with salts [33]. After attesting the ability of synthesized ILs to undergo liquid–liquid demixing in the presence of aqueous solutions of salts (K₂HPO₄ or C₆H₅K₃O₇), the resulting ABSs were investigated as simpler and faster paths to extracting valuable proteins. Human transferrin (TF), which plays a central role in iron metabolism, was adopted as a model protein in this work. Transferrin is a single unit glycoprotein comprising 679 amino acid residues and has a molecular weight of ca. 79 kDa [34]. Based on its amino acids sequence, transferrin consists of two homologous regions: the N-terminal (residues 1–336) and C-terminal (residues 337–679) domains, each carrying a metal-binding site [34]. Due to its outstanding biological properties, transferrin currently fits a wide array of applications, namely as an essential ingredient in serum-free cell culture media [35,36], drug and gene delivery [37,38], and as a potential biomarker for several diseases [39,40]. Although transferrin can be obtained from human sources by several methods, most are multi-step dependent or involve complex approaches [41,42,43,44]. For instance, methods including salt fractionation, gel filtration chromatography, ion-exchange chromatography, immunoaffinity chromatography, dialysis, and ultrafiltration may need to be combined and it may take up to about 4 days to obtain the purified protein [41,45]. Thereby, there is the need to simplify the purification schemes of transferrin, further reinforcing its relevance as a model protein in the current study.

2. Materials and Methods

2.1. Materials

For the synthesis of the organic acid-derived ILs, the following precursors were used: tetrabutylphosphonium hydroxide ([P₄₄₄₄]OH, in water solution at 40 wt.%) and tetrabutylammonium hydroxide ([N₄₄₄₄]OH, in water solution at 40 wt.%), both from Sigma–Aldrich (Saint Louis, MO, USA); acetic acid glacial (PA) and formic acid (91 wt.%) from Fisher Chemical (Geel, Belgium) and Panreac (Barcelona, Spain), respectively. Methanol (HPLC grade) and acetonitrile (99.9 wt.%) were acquired from VWR (Radnor, PA, USA). The salts used in this work (cf. Figure 1) were potassium phosphate dibasic trihydrate (K₂HPO₄·3H₂O, extra pure) and tripotassium citrate monohydrate (C₆H₅K₃O₇·H₂O, 99 wt.%), purchased from GPR Rectapur (VWR, Radnor, PA, USA) and Scharlau (Barcelona, Spain), respectively. Human transferrin (TF) (>98.0 wt.% pure) was purchased from Sigma–Aldrich (Saint Louis, MO, USA). The water used was double distilled, passed across a reverse osmosis system, and further treated with a Milli-Q plus 185 water purification apparatus (Merck, Darmstadt, Germany).

2.2. Synthesis of Ionic Liquids

Four ILs were synthetized, namely tetrabutylammonium formate ([P₄₄₄₄][HCOO]), tetrabutylphosphonium acetate ([P₄₄₄₄][CH₃COO]), tetrabutylammonium formate ([N₄₄₄₄][HCOO]), and tetrabutylammonium acetate ([N₄₄₄₄][CH₃COO]). Figure 1 depicts their full name, acronym, and chemical structure. These ILs were synthesized by neutralizing [P₄₄₄₄]OH or [N₄₄₄₄]OH with the respective organic acids, namely formic and acetic acids. In a typical reaction, [P₄₄₄₄]OH or [N₄₄₄₄]OH (40 wt.% in water) and the respective organic acids are mixed at 1:1.1 molar ratio. The reaction mixture was kept at ambient temperature (25 °C) with continuous stirring for at least 24 h. Water was then removed under reduced pressure at 70 °C. The synthesized ILs were washed with acetonitrile/methanol (1:1, v:v) to remove unreacted acids. Excess solvent and water were then removed under reduced pressure using a rotary evaporator (70 °C, 2 h) and under high vacuum (70 °C, 72 h). The purity and structure of the synthesized ILs were verified by ¹H NMR spectroscopy (Supplementary Materials, Figures S1–S4).

2.3. Determination of ABSs Phase Diagrams

The cloud point titration method was used to determine the IL-based ABSs binodal (solubility) curves at (25 ± 1 °C) and atmospheric pressure, following a well-established procedure [46]. Firstly, aqueous solutions of ILs and each salt were prepared gravimetrically at predefined concentrations (i.e., ILs at ca. 50–80 wt.%, K₂HPO₄ at ca. 40 wt.%, and C₆H₅K₃O₇ at ca. 50 wt.%). The aqueous solution of salt was repetitively added drop by drop to the aqueous solution of IL until a cloudy (biphasic) mixture was detected. Water was then repetitively added drop by drop until a clear and limpid (monophasic) mixture was obtained. These steps were alternately conducted under continuous agitation. After each solution addition, the ternary system compositions of the phase diagram were quantified by weight (within ±10⁻⁴ g). Salt hydrates, namely K₂HPO₄·3H₂O and C₆H₅K₃O₇·H₂O, were used. Therefore, the complexed water was regularly considered a portion of the total water in each mixture.

The experimental binodal curves were correlated by the equation proposed by Merchuk et al. [47]. The tie-lines (TLs) and their respective lengths (TLLs) were determined by a gravimetric method developed by the same authors [47]. TLs determination began with the gravimetric preparation of a ternary mixture belonging to the biphasic area of the phase diagram, which was further stirred and centrifuged (30 min, room temperature), allowing for phase separation. Each phase was then separated and weighed, with each TL calculated by the lever-arm rule through the correlation between the IL-rich aqueous phase weight and whole system composition. A more detailed description of the binodal curves correlation and TLs calculation is provided in the literature [46].

2.4. Extraction of Human Transferrin

During the extraction studies, an aqueous solution of human transferrin was prepared at 0.5 g·L⁻¹ and used as the water component in the ternary system. A biphasic composition comprising 25 wt.% of IL, 28 wt.% of salt, and 47 wt.% of the transferrin aqueous solution was used. Extraction studies were performed at a pH of 7.4 ± 1, controlled by adding the respective organic acid precursor (acetic or formic acid) to prepare the IL solutions. The pH of each aqueous phase was measured at 25 °C using an HI 9321 Microprocessor pH meter (HANNA instruments). In both the top and bottom phases, transferrin quantification was conducted by size-exclusion high-performance liquid chromatography (SE-HPLC). A Chromaster HPLC system (VWR Hitachi) equipped with a binary pump, column oven, temperature-controlled auto-sampler, and diode array detector (DAD) was used. The analytical column was a Shodex Protein KW-802.5 (8 mm × 300 mm). Both the column and autosampler worked at a controlled temperature of 25 °C. The mobile phase was made of a 50 mM phosphate buffer at pH 7.0 with 0.3 M NaCl, which ran in isocratic mode at a flow rate of 0.5 mL·min⁻¹, and the injection volume was 25 µL. The DAD was set at a wavelength of 280 nm. Before injection in the SE-HPLC, each phase was diluted at a 1:10 (v/v) ratio in phosphate buffer (50 mM).

Equations 1 and 2 were used to calculate the extraction efficiencies (EE_TF, %) and recovery yields (Y_TF, %) of transferrin, where $w_{\mathrm{TF}}^{\mathrm{IL}}$, $w_{\mathrm{TF} }^{\mathrm{Salt} }$ and $w_{\mathrm{TF} }^{\mathrm{Initial} }$ represent the total weight of transferrin in the IL-rich phase, salt-rich phase, and initial mixture, respectively.

$$E{E}_{TF}, \%=\frac{w_{\mathrm{TF}}^{\mathrm{IL}}}{w_{\mathrm{TF} }^{\mathrm{IL} }{}^{ }+w_{\mathrm{TF} }^{\mathrm{Salt} }} \times 100$$

(1)

$$Y_{TF}, \%=\frac{w_{\mathrm{TF}}^{\mathrm{IL}}}{w_{\mathrm{TF}}^{\mathrm{Initial} }{}^{ }} \times 100$$

(2)

2.5. Molecular Docking

The interactions of transferrin with IL cations ([N₄₄₄₄]⁺ and [P₄₄₄₄]⁺) and anions ([CH₃COO]^– and [HCOO]^–) were calculated using AutoDock Vina 1.1.2 [48]. The crystal structure of transferrin (PDB: 3v83) was used in the molecular docking analysis as a receptor. The PDB file was downloaded from the Protein Data bank and uploaded in ProteinPrepare [49]. The pKa calculation was performed at pH 6.5 and pH 8 (average pH of experimentally studied ABSs) without water molecules. AutoDockTools (ADT) [50] was used to prepare the protonated proteins input files. Ligands (ILs ions) 3D atomic coordinates were created using Discovery Studio, v20 (Accelrys, San Diego, CA, USA), and applied to Chem3D-MM2 protocol for energy minimization (Chem3D Ultra, CambridgeSoft Co., Cambridge, MA, USA, 02140). Ligands rigid roots were generated using AutoDockTools (ADT), setting all possible rotatable bonds defined as active by torsions. The grid center at the center of mass (x-, y-, and z-axes) for transferrin was −52.411 × 17.914 × −30.166, respectively. The grid dimension was 98 Å × 126 Å × 112 Å. The binding model with the lowest binding free energy was sought out from nine different conformers for each ligand (IL cations and anions).

3. Results and Discussion

3.1. ABSs Phase Diagrams

The potential for natural organic acid-derived ILs to create ABSs was initially evaluated by determining their phase diagrams with C₆H₅K₃O₇ and K₂HPO₄ as the salting-out agents. Four ILs were synthesized (viz. [P₄₄₄₄][HCOO], [P₄₄₄₄][CH₃COO], [N₄₄₄₄][HCOO] and [N₄₄₄₄][CH₃COO]) and the respective binodal curves were determined at 25 °C and atmospheric pressure, and further correlated using the equation proposed by Merchuk et al. [47]. All related data, as well as TLs and TLLs, are provided in Supplementary Materials (Tables S1–S4). Despite the importance of weight percent representations for projecting extraction processes more directly, molality units were used to plot Figure 2 and Figure 3. Molality representations are useful in discussing ABSs formation trends more accurately by excluding any effects derived from the distinct molecular weights of the ILs and salts studied. The experimental binodal data in weight percent reported in Tables S1 and S2 in Supplementary Materials was converted to molality units (mol·kg⁻¹, moles of IL per kg of salt + water or moles of salt per kg of IL + water) according to the following equation:

$$molalit{y}_{\mathrm{IL}/\mathrm{salt}}, \mathrm{mol}·{\mathrm{kg}}^{-1}=\frac{\frac{\mathrm{wt}.{\%}_{\mathrm{IL}/\mathrm{salt}}\times 0.01}{M{W}_{\mathrm{IL}/\mathrm{salt}}}}{\left(100-\mathrm{wt}.{\%}_{\mathrm{IL}/\mathrm{salt}}\right)\times 0.01}\times 1000$$

(3)

Figure 2 shows the impact of the IL chemical structure on the formation of ABSs. All [N₄₄₄₄]- and [P₄₄₄₄]-based ILs can undergo liquid−liquid demixing in the presence of both salts, exhibiting a similar predisposition to form ABSs. Although binodal curves are somewhat overlaying in the case of C₆H₅K₃O₇, a more defined trend of ILs’ aptitude to form ABSs can be established in the presence of K₂HPO₄ at 1.0 mol·kg⁻¹: [P₄₄₄₄][HCOO] > [P₄₄₄₄][CH₃COO] > [N₄₄₄₄][HCOO] > [N₄₄₄₄][CH₃COO]. Overall, [P₄₄₄₄]-based ILs are stronger phase splitting promoters than their [N₄₄₄₄]-based congeners in the presence of salts, which is a well-documented phenomenon [46]. In turn, ILs bearing the [HCOO]^– anion appear more susceptible to undergoing two-phase separation than those comprising the [CH₃COO]^– anion.

It was previously shown that the formation of IL-salt-based ABSs correlates well with ILs’ hydrogen-bond acidity and basicity [51,52]. That is, ILs’ aptitude to behave as hydrogen-bond donors or acceptors governs the molecular mechanisms behind ABS formation involving salts and water. Phosphonium-based ILs are more efficient phase splitting promoters than their ammonium-based counterparts due to their lower hydrogen-bond acidity; thus, they are less prone to form hydrogen bonds with water and more easily salted-out by the salt [52]. On the other hand, ILs bearing anions with high hydrogen-bond basicity, such as the [CH₃COO]^– anion, are less disposed to form ABSs in the presence of salts [51]. No hydrogen-bond basicity data was found for [HCOO]-based ILs [51]. Even so, the presence of an extra –CH₃ moiety in [CH₃COO]^– was expected to enlarge the biphasic zone [53], contrasting the trend observed here. Intricate tendencies in the ABS formation ability of other organic acid-derived ILs have been reported elsewhere [29,54].

Figure 3 shows the impact of salt on the formation of ABSs. K₂HPO₄ induces the ABS formation more easily in the presence of synthesized ILs than C₆H₅K₃O₇. This trend is unanticipated considering the Hofmeister series and salt ions’ molar entropy of hydration [55]. Accordingly, salts bearing ions of higher valence (e.g., C₆H₅O₇^3– vs. HPO₄^2–) are more easily hydrated, thus promoting the generation of IL-based ABSs better [55]. Deviations to the Hofmeister series have also been identified in previous works [51,56], particularly concerning polyvalent ions [57]. In this work, by addressing IL anions with high charge density, it seems that there are competing interactions between the salt anions and IL anions to be hydrated and thus to induce the salting-out or to be salted-out.

3.2. Extraction of Transferrin

The ABSs proposed herein were investigated as a streamlined extraction platform using human transferrin. To perform the extraction studies, a common mixture point falling within the biphasic region of all ABSs (i.e., 25 wt.% IL + 28 wt.% salt + 47 wt.% aqueous solution containing transferrin) was chosen. Since this method aimed at extracting a labile molecule, namely the protein transferrin, the pH of the extraction medium was controlled at 7.4 ± 1 by initially preparing IL stock solutions with the corresponding precursor organic acid. Detailed pH values are provided in Table S5 of the Supplementary Materials.

Figure 4 shows the extraction efficiencies (EE_TF, %) and recovery yields (Y_TF, %) of transferrin obtained (refer to the Supplementary Materials for numerical data—Table S5). Complete extraction of transferrin toward the IL-rich phase was achieved in one step (EE_TF = 100%) with minor losses of protein resulting from precipitation or denaturation phenomena (Y_TF ≥ 86.19 ± 0.02%) in all investigated systems. The transferrin partition to the IL-rich phase can be justified based on the salting-out effect produced by the salts; however, differences in the recovery yield may be related to specific “protein-IL” interactions, as discussed below.

Molecular docking analysis was performed to shed light on possible “protein-IL” interactions. The docking affinity was calculated for the IL ions ([N₄₄₄₄]⁺, [P₄₄₄₄]⁺, [CH₃COO]^– and [HCOO]^–) with transferrin at two protonation states by considering the experimentally measured pH of the systems (Table S5 in Supplementary Materials). The bind poses with the lowest absolute value of affinity (kcal·mol⁻¹) for transferrin and IL cations/anions are displayed in Figure 5 and Figure 6. The docking bind energies of each IL ion with transferrin, interacting amino acids residues, and type of interaction and geometry distance (Å) are detailed in Tables S6 and S7 of the Supplementary Materials.

Figure 5 shows the bind pose of each IL ion in the transferrin surface at pH 6.5. In C₆H₅K₃O₇-based and K₂HPO₄-based systems, the protein is mostly negatively charged (transferrin isoelectric point is around 6 [58]). Therefore, the cation binding affinity to transferrin surfaces derived from electrostatic forces should be higher than the anions studied, as further corroborated by the results obtained (cf. Tables S6 and S7 in Supplementary Materials). The [P₄₄₄₄]⁺ cation has a lower bind energy (−3.5 kcal·mol⁻¹) than its ammonium congener (−4.0 kcal·mol⁻¹). In turn, IL anions have binding affinities with transferrin of −3.4 and −3.0 kcal·mol⁻¹, for [CH₃COO]^– and [HCOO]^–, respectively. According to the bind energy score, the affinity of ILs to interact with transferrin can be rated as follows: [N₄₄₄₄][CH₃COO] > [N₄₄₄₄][HCOO] > [P₄₄₄₄][CH₃COO] > [P₄₄₄₄][HCOO]. The highest recovery yield (Y_TF = 98.73 ± 1.59%) afforded by [N₄₄₄₄][CH₃COO] in the presence of C₆H₅K₃O₇ can be attributed to its higher binding energy. Both cations, [N₄₄₄₄]⁺ and [P₄₄₄₄]⁺, can also interact through hydrophobic interactions with the protein. Hydrophobic interactions are established with Alanine 658 and Lysine 657 for the former and with Aspartic acid 477 for the latter. In addition, the [P₄₄₄₄]⁺ cation is adjacent to Arginine 475, interacting through electrostatic interactions. Both the anions studied are prone to participate in hydrogen-bonding interactions. [CH₃COO]^– interacts with Arginine 644, Asparagine 411, Glutamic acid 423, and Asparagine 584, whereas [HCOO]^– interacts with Arginine 456, Glycine 459, and Tyrosine 517. In addition, [CH₃COO]^– and Arginine 644 establish electrostatic interactions.

Figure 6 shows the bind poses of each IL ion in the transferrin surface at pH 8. The order of protein-IL docking affinities observed at pH 6.5 also applies for the anions (−3.3 kcal·mol⁻¹ for [CH₃COO]^– and −3.1 kcal·mol⁻¹ for [HCOO]⁻). However, it is reversed regarding the cations (−3.8 kcal·mol⁻¹ for [N₄₄₄₄]⁺ and −4.0 kcal·mol⁻¹ for [P₄₄₄₄]⁺). Accordingly, the ability of ILs to interact with transferrin at pH 8 follows the order: [P₄₄₄₄][CH₃COO] > [P₄₄₄₄][HCOO] = [N₄₄₄₄][CH₃COO] > [N₄₄₄₄][HCOO]. Showing the lowest interaction ability with transferrin, [N₄₄₄₄][HCOO] is the IL leading to the lower recovery yield in ABSs containing K₂HPO₄ (Y_TF = 86.73 ± 1.57%). Although transferrin amino acid residues (Phenylalanine 186) only interact with the [P₄₄₄₄]⁺ cation through hydrophobic effects, with [N₄₄₄₄]⁺, a multitude of interactions occur (viz., hydrophobic effect with Proline 508 and Phenylalanine 395 as well as hydrogen bonds and electrostatic interactions with Glutamic acid 375). [HCOO]^– binds to transferrin interacting with Alanine 458, Glycine 459, Tyrosine 517, and Threonine 452 through hydrogen bonds. On the other hand, [CH₃COO]^– is susceptible to interaction by both electrostatic interactions and hydrogen bonding, presenting the same binding sites described at pH 6.5.

Based on the extraction efficiencies obtained (100% for all ABSs studied), it is clear that the partition of transferrin to the IL-rich phase is ruled by a salting-out effect exerted by the salt. However, higher differences were observed in the recovery yields (ranging between 86 and 100%), which seem to result from specific “IL-transferrin” interactions, as unveiled by computational studies, that may reinforce the design of ABSs with a protein-friendly nature.

In a study focused on IL-based ABSs for extracting transferrin, Yee et al. [59] used an ABS comprising 1-butyl-3-methylimidazolium tetrafluoroborate ([C₄C₁im][BF₄]) and sodium phosphate to extract transferrin for point-of-care diagnostic purposes. Although the authors achieved a 20-fold enrichment factor, no extraction efficiencies or recovery yields were reported; therefore, a direct comparison was not possible. Even so, [BF₄]^– anion is prone to undergoing hydrolysis yielding hydrofluoridric acid [60], being here demonstrated that other IL options exist to extract transferrin.

By focusing on other proteins, Li et al. [29] showed that ABSs comprising organic acid-derived cholinium-based IL and a polymer allow recovery yields between 86.4 and 99.9% of bovine serum albumin, trypsin, papain, and lysozyme. Zafarani-Moatar et al. [31] used ABSs comprising amino acid-derived cholinium-based ILs and polymers to show that extraction efficiencies of bovine serum albumin as high as 85.91% are achievable. With ABSs formed by cholinium-based ILs bearing natural acids as the anions and a polymer, recovery yields of 20 to 100% were obtained for immunoglobulin G by Ramalho et al. [28] and Mondal et al. [30]. Despite their remarkable performance [28,29,30,31], most cholinium-based ILs show little to no ability to undergo phase-separation in the presence of most salts [54,61,62], thus limiting their widespread use. In this work, by using [P₄₄₄₄]⁺ or [N₄₄₄₄]⁺ cations in conjugation with organic acid-derived anions, which are more easily salted-out by conventional salts, the toolbox of protein-friendly ABSs is broadened while offering a competitive performance in recovery yield and extraction efficiency.

Finally, it is also worth noting that the synthesis of bio-based ILs comprising organic acids has largely resorted to precursors produced petrochemically rather than naturally (e.g., via fermentation). This strategy is valid to expedite the development of bio-based ILs and their application in separation processes with theoretically transposable outcomes if, for instance, acetic acid—easily produced by the microbial oxidation of ethanol—is considered [63]. Moreover, petrochemically derived organic acids are more easily obtained than their naturally derived equivalents, whose purification is a difficult task [64]. If mixtures of organic acids are used instead of pure compounds, both the IL synthetic pathway and ABS-mediated extraction process may need to be revisited.

4. Conclusions

This work expanded the options available to create protein-friendly IL-based ABSs by using ILs formed by organic acid-derived anions paired with [N₄₄₄₄]⁺ and [P₄₄₄₄]⁺ cations and two salts. Four ILs, viz. [N₄₄₄₄][HCOO], [N₄₄₄₄][CH₃COO], [P₄₄₄₄][HCOO] and [P₄₄₄₄][CH₃COO], were synthesized via a neutralization reaction and further employed in the development of ABSs. ABSs phase diagrams were established in the presence of C₆H₅K₃O₇ and K₂HPO₄, allowing to understand the impact of the ILs constituent ions and the salt effect on the two-phase formation. Of all ILs, [P₄₄₄₄][HCOO] and [N₄₄₄₄][CH₃COO] were identified as the strongest and the weakest phase splitting promoters, respectively. Although the cation effect was governed by the decreased acidity and consequent weaker propensity to establish hydrogen bonds with water, the same does not apply to the impact of the anion, with [HCOO]^– outperforming [CH₃COO]^–. Regarding the salts’ effect, the formation of ABSs benefited more from the presence of K₂HPO₄ than C₆H₅K₃O₇, following an opposite trend to the Hofmeister series. This trend may be due to the high charge density of the studied anions, thus improving the competition between the salt and IL anions to be hydrated and to salt-out the competitive ionic species.

Complete extraction (EE_TF = 100%) of human transferrin toward the IL-rich phase was achieved with all ABSs, with no major losses of protein (86.19 ≤ RY_TF ≤ 100%). The salting-out effect appear to be mainly responsible for the complete transferrin extraction to the IL-rich phase. By contrast, differences in the recovery yield appear to depend on specific “IL-transferrin” interactions, which in turn provide a protein-friendly environment, as complemented by molecular docking studies.

Overall, the organic acid-derived ILs proposed in this work bestow ABSs with protein-friendly credentials while allowing for high extraction efficiencies and recovery yields. As shown with human transferrin, these systems are expected to contribute significantly to developing streamlined and efficient separation processes for highly valuable proteins. Moreover, to foster the application of ILs derived from natural organic acids in bioseparations, studies involving microbially obtained precursors may be subject to future work.