Partition of Tea Saponin with a Novel Recyclable Thermo-pH Aqueous Two-Phase Systems

Aqueous two-phase systems (ATPS) have the advantages of environmentally friendly, high mass transfer efficiency, and mild extraction conditions. However, it is difficult to recycle these polymers, which limits the large-scale application of ATPS. In this study, a novel recyclable ATPS was constructed with thermo-responsive polymer PN and pH-responsive polymer PADB4.78 for the partition of tea saponin. PN represents poly-(N-isopropylacrylamide), and PADB4.78 represents poly-(acrylic acid-dimethylamine ethyl methacrylate-butyl methacrylate), where 4.78 in the subscript indicate the isoelectric point of the polymer. The recoveries of PN and PADB4.78 were 95.36% and 93.48%, respectively, after two cycles. Meanwhile, the phase formation mechanism of ATPS was studied by surface tension and low-field nuclear magnetic resonance (LF-NMR). The effects of polymer concentration, pH, temperature, types and concentrations of salt were investigated on tea saponin partition. In the 1.5% (w/v) PN/3.5% (w/v) PADB4.78 ATPS, the optimal partition coefficient (K) of crude tea saponin were 0.15 in the presence of 1.5 mM KCl at pH 7.6 and 25 °C while the extraction recovery (ERb) reached 92.13%. The K and ERb of tea saponin from tea seeds were 0.12 and 94.50% with 7.5 mM LiBr at pH 8.0 and 25 °C, respectively.

Traditional ATPS are composed of polymer, salt, organic solvent, inorganic salt, and ionic liquid [8][9][10][11]; these phase components are difficult to reuse. In order to solve these problems, some polymers responsive to light, temperature, and pH have been synthesized in our laboratory, which can be conveniently recycled by adjusting environmental factors and used to construct novel recyclable ATPS [12][13][14]. Yang et al. [15] synthesized two pH-responsive polymers (P MDB3. 36 and P ADB4.05 ) with recovery higher than 95% in pilot scale and constructed recyclable pH-ATPS to study the partition behavior of demeclocycline and lincomycin. Xu et al. [16] used N-vinylcaprolactam (NVCL), butyl methacrylate (BMA), acrylamide (AM), and N-isopropylacrylamide (NIPA) to synthesize two high-recovery thermo-responsive polymers (P VBAm and P N ). These researches have proved that the novel ATPS provides a broad application prospect for the extraction and separation of biomolecules.
Tea saponin, a kind of pentacyclic triterpene glycoside compound, is from the seeds of Camellia oleifera [17], which has good emulsifying, dispersing, and wetting effects. It can be widely used in medicine, agriculture, feed, food, textile, and other fields [18][19][20]. Moreover, it plays an important role in antibacterial, anti-inflammatory, anti-tumor, and flame retardant materials and eliminates environmental pollution. Traditional extraction techniques of tea saponin mainly include water extraction, organic solvent extraction, and microwave-assisted extraction [21][22][23]. These extraction techniques have some disadvantages, for example, complex extraction process, low extraction rate, and environmental pollution. Therefore, the development of environmental-friendly, cost-saving extraction technology is very necessary for tea saponin.
In this study, thermo-responsive polymer (P N ) and pH-responsive polymer (P ADB4.78 ) were synthesized by random copolymerization. Meanwhile, thermo-pH responsive ATPS P N /P ADB4.78 were constructed to investigate the partition of tea saponin by optimizing the polymer concentration, system temperature, system pH, and different types and concentrations of salt. The phase formation mechanism was studied by surface tension and LF-NMR. The recyclable thermo-pH aqueous twophase systems showed potential applications in the seperation of bioproducts.

Synthesis of P N and P ADB4.78
In this work, ATPS was constructed by polymers P N and P ADB4.78 , which were demonstrated in Fig. 1. The P N was polymerized using the monomer NIPA with AIBN as initiator. The specific synthesis method was detailed according to Xu [16], where the reaction tempertaure was 62°C. P ADB4.78 was synthesized by 0.047 mol DMAEMA, 0.081 mol AA, and 0.003 mol BMA in 120 mL deionized water under nitrogen protection. The initiators were 0.24 g NaHSO 3 and 0.24 g APS, and the reaction was carried out under the following conditions: rotating speed 200 rpm, temperature 60°C, and time 12 h. The production was precipitated with ethanol and dried at 55°C. The process of tea saponin extraction in P N /P ADB4.78 ATPS is shown in Fig. 2.

Polymer Characterization
The chemical structures of the polymers were characterized with Fourier transform infrared spectroscopy (FT-IR, Nicolet MagnalR550 infrared equipment, Thermo, USA) [24]. The viscosity-average molecular weight and intrinsic viscosity of the polymers were determined using the Ubbelohde viscometer (Liang Jing Glass Company, Shanghai) [25]. Varied concentrations of polymer solutions were prepared. Then, the time for deionized water and polymer solutions to flow through the capillary were recorded as t and t 0 at 25°C, respectively. Dynamic light scattering (DLS) was used to measure weight-average molecular weight and size of P N and P ADB4.78 at room temperature. P N (0.1% (w/v)) and 1% (w/v) P ADB4.78 were prepared and carried out by DynaPro NanoStar equipment (Wyatt Technology Corporation, USA).

Recovery of Polymers
The thermo-responsive polymer could be recycled by changing the temperature to exceed its lower critical solution temperature (LCST). The LCST represented the temperature at which polymer solutions initially become turbid. P N (2% (w/v)) solution was placed in a constant temperature water bath, and the temperature was increased by 0.1°C every 10 min. In addition, a series of salt including KCl, NaCl, Na 2 SO 4 , (NH 4 ) 2 SO 4 , K 2 HPO 4 , and KH 2 PO 4 (10 mM) were used to study the effect on polymer recovery. The pH-responsive polymer could be recycled by changing the pH of solution to its isoelectric point (pI). The pI of P ADB4.78 was determined by zeta potentials. The corresponding zeta potential value is also zero when the net charge of the polymer solution is zero, and the pH of the solution is the isoelectric point of the polymer.
The recovery of the P N /P ADB4.78 ATPS was measured by the percentage of the dry weight to that of initial weight. Firstly, the pH of the ATPS was adjusted to 4.78 to precipitate P ADB4.78 . Then, the temperature of the ATPS was adjusted to 40°C to precipitate P N . Finally, these polymers were centrifuged at 5000 rpm for 10 min and dried at 55°C. Two polymers were dissolved again to form ATPS. Then, these polymers were precipitated at same conditions to calculate the second recovery.
Preparation of P N /P ADB4.78 ATPS Firstly, P N and P ADB4.78 were, respectively, dissolved in deionized water and 0.15% NaOH solution to form 2-5% (w/v) and 5-10% (w/v) polymer solutions. Then, two solutions in equal volume were mixed in graduated centrifuge tube. Finally, these tubes were placed in the water bath until a clear phase interface was observed.

Phase Diagram
The cloud point curve was used to determine the phase diagrams of ATPS [24]. Firstly, 10% (w/v) P N and 10% (w/v) P ADB4.78 were prepared and adjusted to the same Fig. 2 The process of tea saponin extraction in P N /P ADB4.78 ATPS pH. One milliliter P ADB4.78 solution was added to the centrifuge tube, and then, 100 μL P N was added dropwise until the mixed solution changed from clear to turbid, which represented a cloud point on the binodal curve. Then, the deionized water of same pH was used to dilute the mixed solution until the solution was observed to become clear. Finally, dozens of points were obtained by repeating these steps. In addition, the binodal curves of the ATPS at different pH (6.5, 7.0, 7.5) and temperatures (20°C, 25°C, 30°C) were determined.

Phase Formation Mechanism
Surface Tension Measurement The platinum plate method was used to measure surface tension at constant temperature (25°C) by surface tension meter (DCAT11, Dataphysics Co., Ltd.). Firstly, 3% (w/v) P N and 7% (w/v) P ADB4.78 were prepared. Then, 2 mL of solution from the top and bottom phases was taken out for measurement. All measurements were repeated three times to obtain the average value of surface tension.

Low-field Nuclear Magnetic Resonance Relaxation Time Measurement
In this study, the measurement of transverse relaxation time (T 2 ) were conducted on a 21-MHz NMR analyzer (PQ001, Niumag Electric Company, Shanghai, China). Carr-Purcell-Meiboom-Gill (CPMG) sequences, a multipulse sequence applied to protons, was used to measure the spin-spin relaxation time T 2 [26]. These data were analyzed by a multiexponential model of MultiExp Inv Analysis software using the inverse Laplace transform algorithm. The relaxation time reflected the interaction between polymers and water. In addition, 3% (w/v) P N and 7% (w/v) P ADB4.78 solutions were prepared to measure the T 2 at 25°C.

Partition of Tea Saponin
Crude tea saponin and tea saponin from tea seeds were chosen to determine the partition process in P N /P ADB4.78 ATPS. The concentration of P N (1-2.5%, w/v), P ADB4.78 (2.5-4%, w/v), pH 6.0-8.0, and temperature 15-30°C were selected to determine the effect on tea saponin partition in the P N /P ADB4.78 ATPS. Moreover, eight different types and concentrations of salt including KCl, LiCl, LiBr, NH 4 Cl, NH 4 Br, (NH 4 ) 2 SO 4 , Na 2 SO 4 , and MgSO 4 were selected to study the influence on tea saponin partition. Then, 100 μL solutions from each of the two phases were used to measure the absorbance value at 552 nm with an ultraviolet spectrophotometer and calculated the concentration of tea saponin by using the calibration curve. The tea saponin partitioning behavior was discussed by partition coefficient (K) and extraction recovery (ER). These parameters were calculated as Eqs. (1)-(4): where C was the concentration of tea saponin, V indicated the volume of the phase, and the subscripts "top" and "bottom" represented the top phase and bottom phase.

Results and Discussion
Characterization of P N and P ADB4.78

Intrinsic Viscosity of Polymers
The intrinsic viscosity was calculated by Eqs. 5 and 6 depending on the extrapolation method [16].
where η r and η sp represented relative and specific viscosity, respectively. Relative viscosity (η r ) and specific viscosity (η sp ) can be calculated by η r = t / t 0 and η sp = η r − 1. C was the concentration of polymer; α and β were the constant parameters. The viscosity-average molecular weight (M η ) was calculated by viscometer using Mark-Houwink equation as Eq. (7): K and α were empirical constants depending on temperature and solvent. The K and α of P N were 1.45 × 10 −1 mL/g and 0.65. For P ADB4.78 , the K and α were 1.78 × 10 −1 mL/g and 0.56, respectively. The calculation results of [η] and Mη are presented in Fig. 3b, c and Table 1. The intrinsic viscosity was proportional to the viscosity average molecular weight. The intrinsic viscosity of P N was higher than P ADB4.78 . The high concentration of P N was not helpful to the formation of ATPS due to its high intrinsic viscosity.

The Weight-Average Molecular Weight (M w ) and Particle Size of Polymers
The M w and particle size of P N and P ADB4.78 were measured by DLS ( Table 2). The M w of P N and P ADB4.78 were 92 kDa and 32 kDa, respectively. The particle sizes of P N and P ADB4.78 were 4.103 nm and 2.634 nm, respectively. The monomers and initiators used in the synthesis of the two polymers were different, so weight-average molecular weight and particle size of polymers were also different.

Recovery of P N and P ADB4.78
The zeta potential of polymer solutions is determined and presented in Fig. 4a. The polymer was precipitated from solution. When pH was adjusted to its pI, the surface charge of P ADB4.78 was completely neutralized. The recovery of P ADB4.78 is shown in Fig. 4b. The maximum recovery of P ADB4.78 was 97.32% at pI of P ADB4.78, which was 4.78. The LCST of P N was approximately 33.0°C. The P N could be easy to precipitate from solution above its LCST. It was mainly due to that the macromolecular chain of P N has both hydrophilic amido groups and hydrophobic isopropyl groups [27], which makes P N aqueous solutions exhibit temperature sensitive characteristics. When the temperature rised, the polymer molecular chain separated from water, leading to the precipitation of the polymer. The result is shown in Fig. 4c. The maximum recovery of P N was 98.71% when 10 mM Na 2 SO 4 was added. It may be that the addition of salt ions changed the solubility of P N . As shown in Fig. 4d, the recoveries of P N and P ADB4.78 were 95.36% and 93.48% after two cycles. Thus, the P N and P ADB4.78 could be effectively recycled.

Phase Diagram
The formation conditions and quantitative relationship of ATPS could be illustrated by the phase diagram. The influence of pH and temperature on the phase diagram were investigated, and the results are shown in Fig. 5a, b. In the P N /P ADB4.78 ATPS, P N primarily existed in the top phase and P ADB4.78 primarily existed in the bottom phase. This area below the binodal curve Table 2 The weight-average molecular weights and particle sizes of P N and P ADB4.78 Particle size (nm) Mw (kDa) P N 4.103 92 P ADB4. 78 2.634 32 Fig. 4 The recovery of polymers. a Zeta potentials of P ADB4.78 at different pH. b Recovery of P ADB4.78 at different pH. c Recovery of P N at different salts. d The recoveries of P N and P ADB4.78 after two cycles was single-phase region, and the rest was two-phase region. The binodal curves were fitted by Eq. (8) [28]. The heterogeneous region decreased with the increasing system pH, which may be the ionic state of the carboxyl group in AA when the system pH was higher [29]. The closer the temperature to LCST of P N , the more hydrophobic. Therefore, the two-phase region was easier to form.
where B was the concentration of P N and A was the concentration of P ADB4.78 . The α, δ, μ, and Φ represented fitting parameters.

Surface Tension
The measurement of surface tension was helpful to understand the phase formation mechanism of two incompatible polymers. The difference in surface tension between polymers was one of the factors driving the formation of phase interfaces [30]. The surface tension of P N and P ADB4.78 solutions alone and the top and bottom phases after phase formation were measured. The results are shown in Fig. 6. The surface tension values of P N and P ADB4.78 were 49.31 mN/ m and 32.27 mN/m at the same temperature, respectively. The surface tensions of P N -riched phase and P ADB4.78 -riched phase after phase formation were 38.92 mN/m and 38.58 mN/m. Before the phase formation, the surface tension of the two polymers were quite different. It was precisely that two polymer solutions can quickly form aqueous two-phase systems after mixing because of the difference in surface tension [27].

LF-NMR T 2 Relaxation Time Measurement
In this experiment, the T 2 relaxation time helped to further understand the mechanism of phase formation of P N /P ADB4.78 ATPS through the water mobility. The results are shown in Fig. 7a, b. T 2 relaxation curve of polymers represented a multiexponential distribution with two states of water. The peaks in the range of 10-100 ms (T 21 ) and 1000-5000 ms (T 22 ) represented tightly bound water and weakly bound water, respectively [31]. Water with a shorter relaxation time was more strongly bound to molecules than water with a longer relaxation time. Therefore, weakly bound water (T 22 ) was chosen to study the interaction between water and polymer molecules. As shown in Fig. 7a, the T 22 of P ADB4.78 was shorter than P N , which could indicate that the water mobility of P ADB4.78 was weaker than P N and the water-binding capability of P ADB4.78 was stronger than P N . The T 22 relaxation time of each phase after phase formation is shown in Fig. 7b. The peak widths of T 22 in top and bottom phases were the same. The peak widths on the time axis indicated that the P N /P ADB4.78 ATPS reached equilibrium [16]. Due to the presence of a small amount of P ADB4.78 in the top phase, compared with the T 22 relaxation time of P N alone, the T 22 relaxation time of the top phase was shifted to the left. The relaxation time of the bottom phase was higher than the T 22 relaxation time of P ADB4.78 alone due to the presence of P N in the bottom phase. Such difference among the water-binding capability of polymers led to the occurrence of repulsion, reflecting the feasibility in forming aqueous twophase systems [6,32].

Partition of Tea Saponin
Partitioning of biomolecules between two phases depends on the physical and chemical properties of the systems and biomolecules, such as the isoelectric point, the hydrophobicity of the surface, pH, temperature, hydrophobic interaction, charge interaction, Vander Waals forces, and hydrogen bonds. The partition of biomolecules is the result of the mutual influence of various parameters [16,33].

Effect of Different Polymer Concentrations
The polymer concentration has a great influence on the phase diagram of the ATPS and the difference between the two phases. In the P N /P ADB4.78 ATPS, tea saponin was mainly partitioned to the bottom phase; the reason may be explained in terms of hydrophilicity and hydrophobicity, owing to tea saponin having strong hydrophilicity, and the LF-NMR results of the ATPS showed that the bottom phase contained more water. The effect of polymer concentration on the partition of tea saponin is shown in Table 3. All of the K were less than 1; the reason was that the viscosity and surface tension of the polymer increased as the polymer concentration increased, which made the mass transfer efficiency of the tea saponin lower in ATPS [24]. The optimal K of crude tea saponin and tea saponin from tea seeds reached 0.46 and 0.42 in 1.5% P N /3.5% P ADB4.78 ATPS, which was used to study the partition of tea saponin.

Effect of pH
When HCl or NaOH was added to the ATPS, the change of pH influenced the hydrophilicity and hydrophobicity of the pH-responsive P ADB4.78 [13]. Low pH was hard to form two phases. Thus, the effect of pH (6.0-8.0) was investigated at room temperature. The K and ER b of crude tea saponin are observed in Fig. 8a, and those of tea saponin from tea seeds are shown in Fig.  8b. The K decreased slowly with the pH increased. The K and ER b of crude tea saponin were 0.57 and 88.01% when the pH was 7.6. When the system pH was 8.0, the K and ER b of tea saponin from tea seeds reached 0.49 and 82.43%. It could be the change of the pH, which changed the ionization balance of chemical groups such as carboxyl groups and tertiary amine groups on the P ADB4.78 surface [6]. In the studied pH range, the hydrophilicity of the P ADB4.78 increased with the increase of pH, and tea saponin is also hydrophilic, which caused tea saponin transferring from the top phase to the bottom phase.

Effect of Temperature
Temperature influenced the hydrophilicity and hydrophobicity of the P N and the phase behavior of ATPS [34]. In this study, the influence of temperature on the partition of tea saponin in the P N / P ADB4.78 ATPS was investigated, and the range of temperature was from 15 to 30°C at optimized pH. Moreover, thermodynamic parameters (ΔH, ΔS, ΔG) as Eqs. (9)-(11) were used to understand the effect of temperature. The result is shown in Fig. 8c, d and Table 4. ΔG and ΔH were less than zero, which indicated that the selective partition of tea saponin in P N /P ADB4.78 ATPS was a spontaneous process and reflected that low temperature was beneficial to the formation of ATPS [35]. The K of tea saponin decreased with the temperature increased; the reason was that the temperature approached to the LCST of P N , the hydrophobic of P N increased, so tea saponin tended to transfer to the bottom phase [16]. Compared with the K at 25°C, the phase interface of the system was not clear and the ER b of tea saponin decreased when the temperature was 30°C. Therefore, 25°C was used for the further study of tea saponin. where ΔH and ΔS were obtained from slope and intercept of regressed line. R was universal gas constant, and temperature was in Kelvin; K was partition coefficient of tea saponin.

Effect of Inorganic Salt Types and Concentrations
The potential difference between two phases was one of the driving forces for the partition of target molecules [36]. When salt ions were added, the phase equilibrium of ATPS and potential difference were changed. In this study, eight different types and concentrations of salt were added into the ATPS in the range from 1.5 to 9 mM. As shown in Fig. 9a, b, the optimal K and ER b of crude tea saponin were 0.15 and 92.13%, respectively, when the 1.5 mM KCl was added to the ATPS. As shown in Fig. 9c, d, the K and ER b of tea saponin from tea seeds reached 0.12 and 94.50% when adding 7.5 mM LiBr. The reason may be that the addition of KCl and LiBr changed the potential difference between two phases. Therefore, the  . 9 Effect of salts on tea saponin partition. a, c The partition coefficient of crude tea saponin and tea saponin from tea seeds. b, d The extraction recovery of crude tea saponin and tea saponin from tea seeds influence of KCl concentration and LiBr concentration on the potential difference between two phases were investigated. When the positive and negative ions of salt have different affinities for the two phases, that was, when K Z − A and K Z þ B are not equal, potential difference between two phases would generate [24]. The potential difference between two phases was calculated by Eq. (12): where Δφ was the potential difference between two phases, U represented the potential, T was the absolute temperature, R was the universal gas constant, and Z + and Z − represented the positive and negative ion valances of salt, respectively. F was the Faraday constant. K was the partition coefficient; A and B represented the negative and positive ion in salt solution, respectively. As shown in Fig. 10a, b, the corresponding 1/K of tea saponin from tea seeds was the largest when the LiBr concentration was 7.5 mM, and the potential difference between two phases was also the largest. The potential difference between two phases and the 1/K were consistent with the changing trend of the KCl and LiBr concentrations. Therefore, the influence of salt ions on the partition of tea saponin was mainly attributed to the effect of salt ions on the potential difference between two phases, which resulted in more tea saponin transferring from the top phase to the bottom phase.

Conclusion
The recyclable thermo-responsive polymer (P N ) and pH-responsive polymer (P ADB4.78 ) were synthesized and characterized in this study. Furthermore, the phase formation mechanism was studied by measuring the surface tension and transverse relaxation time of polymer solutions. These results demonstrated that the difference in water-binding capacity between the two polymers was an important factor in the formation of ATPS. The recoveries of the two polymers were above 93% after two cycles. The optimal K and highest ER b of tea saponin from tea seeds were 0.12 and 94.50% in the presence of 7.5 mM LiBr at pH 8.0 and 25°C, which showed that tea saponin from tea seeds could be effectively separated. Compared with traditional tea saponin extraction methods, such as water extraction and organic solvent Fig. 10 The relationship between the interphase potential difference and KCl and LiBr concentrations. a KCl concentrations at 25°C, pH 7.6. b LiBr concentrations at 25°C, pH 8.0 extraction, the P N /P ADB4.78 ATPS have the advantages of being a cost-saving, environmentalfriendly, and simple extraction process. Moreover, it provided wide application prospects for the separation of industrialized biomolecules.
Author Contribution Yanli Wei carried on the data collection and wrote the manuscript. Xi Chen carried on the data collection. Ting Yang conducted the data analysis. Junfen Wan conducted the data analysis, writing and revision of the manuscript. Xuejun Cao conducted the data analysis, writing and revision of the manuscript. All authors have read and approved the published version of the manuscript.Data Availability Not applicable.

Declarations
Ethics Approval Not applicable.
Consent to Participate Not applicable.

Consent for Publication
All authors consent to publish the manuscript.

Competing Interests
The authors declare no conflict of interest.