Solubility of DNP-amino acids and their partitioning in biodegradable ATPS: Experimental and ePC-SAFT modeling

Abstract

Predicting the behavior of dinitrophenylated amino acids (DNP-AA) in aqueous solutions requires an understanding and accurate description of interactions that can occur in such systems. In this work, some properties of DNP-AA (DNP-glycine, DNP-alanine, DNP-valine, DNP-leucine) have been determined experimentally. These were liquid densities obtained at T=298.15 – 318.15 K, p=1 bar, and pH-dependent solubility data measured at T=298.15 K, p=1 bar. It was observed that the solubility order for DNP-AA does not follow the same sequence as for aliphatic amino acids. The thermodynamic model ePC-SAFT has been applied to predict the properties density and solubility, and additionally to estimate partition coefficients of DNP-AA in PEG (PEG 4000, PEG 6000, PEG 8000) - organic salt (sodium citrate, potassium citrate, potassium sodium tartrate) aqueous two-phase systems (ATPS). ePC-SAFT pure-component and binary interaction parameters for neutral DNP-AA were acquired using the joinzt joint-parameter method, namely by combining the parameters for dinitrobenzene with parameters for amino acids (glycine, L-alanine, L-valine, L-leucine) from literature. The pure-component parameters for charged DNP-AA^- were inherited from their parent neutral DNP-AA. This work shows that ePC-SAFT allows predicting liquid densities and solubilities of neutral DNP-AA with good agreement to experimental results. Moreover, adjusting in sum six binary parameters between charged DNP-AA^- and phase-forming species allowed modeling partition coefficients of four DNP-AA in nine different PEG – organic salt ATPS, each at four different ATPS compositions. This can be considered an excellent modeling result and proofs the suitability of ePC-SAFT for such systems.

Introduction

Understanding the behavior of organic molecules in aqueous solutions is of importance for process design with special attention on separation processes. Separating a compound of interest between the two aqueous phases has been considered in the literature as an appropriate approach to evaluate interactions in water systems [1], [2], [3], [4], [5], [6], [7]. Willingness to unravel the phenomena which are behind the partitioning of very complex and large molecules such as proteins was the reason for introducing to research smaller compounds that share the same functional groups as macromolecules [1], [2], [3], [6]. Thus, dinitrophenyl derivatives of amino acids (DNP-AA) have been selected to study interactions of the amino acid side-chain groups in Aqueous Two-Phase Systems (ATPS) and the properties of ATPS phases [1], [5], [6], [8], [9], [10], [11].

In the literature, DNP-AA or their sodium salts were partitioned in ATPS which have usually contained two different polymers [6], [10], [11], or a combination of one polymer with an inorganic salt [5], [10], [12]. Partition coefficients (K) of reasonable magnitude measured for DNP-AA in polymer – inorganic salt systems have shown that these media can be very appropriate for efficient separation of ionizable organic components such as DNP-AA. However, high amounts of inorganic salt (e.g. phosphates, sulfates) needed to form two immiscible phases with polymer (e.g. polyethylene glycol – PEG), might be very inconvenient for the environment if such ATPS were applied at the industrial scale. For this reason, in recent studies, the replacement of inorganic salts by biodegradable organic compounds (e.g. citrates, tartrates) has been proposed [9], [13]. Such modification turned out to be a successful solution as very similar K values for DNP-AA could be obtained in the alternative PEG – sodium citrate ATPS if comparing with former results in PEG – sodium sulfate system [8], [9].

The partitioning in PEG – biodegradable salt ATPS can be still optimized by the change of several factors, e.g. the two-phase composition (tie-line length - TLL) or the molecular weight of PEG used [13]. However, measuring the partitioning at these moderating conditions is usually time- and cost-intensive. Thermodynamic predictive models that can describe the interactions in aqueous systems can reduce the number of experiments needed to determine K values. Thus, in literature at least several models have been proposed to correlate and describe the partitioning in PEG – organic salt ATPS, e.g. Pitzer model [14], extended NRTL [15], modified UNIQUAC [15]. Also predictive Statistical Association Fluid Theory (SAFT)-based models showed the ability to model the partitioning in both PEG – inorganic salt and PEG – organic salt ATPS [16], [17]. The quantitative modeling of K for five water-soluble vitamins in PEG (PEG 4000, PEG 6000) – organic sodium salt (Na₃Citrate, Na₂Tartrate) ATPS was achieved with electrolyte PC-SAFT (ePC-SAFT) [17] through activity coefficients. The success of ePC-SAFT with implemented Debye-Hückel term for this application relied on the precise description of interactions in ATPS, as well as on considering the presence of different ionic species at pH of ATPS. To the best of our knowledge, there has not been any work yet in which K for DNP-AA in PEG – organic salt ATPS were modeled.

In this work, ePC-SAFT has been at first used for predicting the liquid densities and solubilities of four N-(2,4-dinitrophenyl) amino acids, namely DNP-glycine, DNP-alanine, DNP-valine, DNP-leucine. The predictions were performed at T=298.15 – 318.15 K, p=1 bar (density), and T=298.15 K and p=1 bar (solubility). The ePC-SAFT pure-component parameters and the binary interaction parameters (k_ij) for DNP-AA were estimated with the joint -parameter method (J-PM). The J-PM that has been introduced in this work specifically for DNP-AA, origins from a strategy published for proteins [45]. The conception adopted in original method from literature was also used by Greinert et al. [3] for modeling solutions of 1,3-bisphospho-D-glycerate [44]. ePC-SAFT parameters for m-dinitrobenzene (DNB) fitted to vapor pressures and densities (data from the literature [18], [19]) have been joint with parameters for aliphatic amino acids (AA) established in the previous study [20]. Experimental data required for validation of density and solubility predictions of DNP-AA were not available in literature, so they needed to be measured in this work. The measurements were performed at the same conditions of pressure and temperature as the predictions. Since the solubility of ionizable DNP-AA strongly depends on the pH of a solution, solubility data were obtained at different pH values. Then, the intrinsic solubilities (neutral DNP-AA) were determined from pH-solubility profiles estimated using the Henderson-Hasselbalch (H-H) equation and dissociation constants (pK_a). Afterward, these solubilities were compared with ePC-SAFT modeling results. Solubilities measured at higher pH (pH≈6) in which DNP-AA are solely present as negatively charged species were used to access k_ij parameters between water and DNP-AA⁻. The method of fitting K_a with pH-dependent solubilities is established [21].

The second part of this work relied on ePC-SAFT modeling of DNP-AA partitioning in ATPS formed by PEG (PEG 1500, PEG 4000, PEG 6000) and organic salt (Na₃Citrate, K₃Citrate, KNaTartrate). The reliability of the modeling results was determined using experimental K values from the literature [9], [13]. For the partition-coefficient modeling, the pure-component parameters for phase forming components: PEG, Na+, K⁺, and anions (Tartrate²⁻, Citrate³⁻) were obtained from formerly reported studies [17], [22], [23], [24]. Almost all binary interaction parameters between ATPS phase formers could be found in the literature as well [17], [22], [24], [25]. In one case, the k_ij between PEG and K⁺ was fitted to tie-line data from the literature [26]. The interactions between K⁺ and salt anions (Tartrate²⁻, Citrate³⁻) also needed to be estimated in this study. This was done by fitting k_ij parameters to experimental osmotic coefficients [26], [51]. The k_ij between DNP-AA⁻ and ATPS phase formers (Na⁺, K⁺, PEG) were fitted just to one experimental K point for each DNP-AA⁻ partitioned in ATPS: PEG 6000 – Na₃Citrate, and PEG 6000 – K₃Citrate [9], [13].