1. Introduction
Muérdago (
Tristerix tetrandus) is a medicinal mistletoe species native to southern Argentina and central and southern Chile. This plant is a parasite of aspen (
Populus sp.), colliguay (
Colliguaya odorifera), maqui (
Aristotelia chilensis), willow (
Salix sp.), among other native Chilean species. It is commonly gathered by local collectors, dried, and sold in local markets. This plant has traditionally been used in alternative medicine as an anti-inflammatory, digestive [
1,
2], hemostatic, and hypocholesterolemic [
3] remedy and as an anxiolytic agent [
4].
Tristerix tetrandus contains a wide and important number of phenolics and anthocyanins in its fruits and leaves [
1,
2], and some other mistletoe plants have shown the presence of several amino acids [
5].
Several phenolics are bioactive compounds [
6], and they are widely distributed in fruits and vegetables, such as blueberries, blackberries, spinach, among others [
7,
8], and have the ability to protect against several human diseases [
8,
9,
10]. Phenolics such as quinic acid, rutin, quercetin, caffeoyl-glucose, p-coumaric acid, catechin, 5-O-caffeoylquinic acid (3-CQA), among others are contained in
Tristerix tetrandus [
1]. This plant possesses high concentrations of phenolic compounds, which have been found in studies in vitro and in vivo to possess a range of biological activities including anticancer and anti-platelet activities, as well as antioxidant properties [
11]. In addition, several amino acids have been found in several mistletoe species [
12]. Amino acids have regulatory roles in cell metabolism and function [
13]. Indispensable amino acids contained in food are needed to synthesize bodily proteins [
14].
The extraction of natural products, such as phenolics, is traditionally realized by using conventional procedures, including toxic organic solvents [
15], but the solvent extractions do not ensure that the liquid fraction obtained contains specific phenolic molecules according to their molecular size and/or molecular charge. Moreover, the driving force for conventional extraction methods (e.g., maceration, hydro distillation, water distillation, and steam distillation) is the application of heat mixing as well as toxic solvents. The problem with these methods is that they are expensive, time-consuming, have low extraction selectivity, cause thermal degradation of thermolabile compounds, among others. On the other hand, metabolites such as amino acids are hydrophilic and are, therefore, difficult compounds for conventional solvent extraction [
16]. Thus, there arises the need to study novel, effective, and green techniques for the extraction of bioactive compounds. The alternative extraction and isolation of bioactive molecules would find beneficial and specific applications in the food, pharmaceutical, and phytochemical industries.
Membrane technology has allowed the separation of bioactive compounds from a wide variety of plant and food solutions [
17]. High recovery or removal efficiency, low energy input, environmental safety, high selectivity, easy scale-up, low temperature processing, absence of phase transition, and versatile integration with other unit operations make membrane fractionation an appropriate technology for the treatment of organic and thermolabile solutions. The use of nanofiltration (NF) is an advantageous, green, and clean alternative for the purification of natural compounds by selecting membranes with a suitable molecular weight cut-off (MWCO) (150–1000 Da range) [
7,
18] and according to the target metabolites that are present in the treated solutions. Therefore, the application of NF appears as a novel alternative in the field of natural products to the conventionally used methods for the isolation, fractionation, and identification of pure compounds.
The separation and characterization of bioactive molecules contained in native plants are important for the preparation of nutraceuticals and food ingredients. Liquid chromatography (HPLC, UPLC, UHPLC) coupled to several mass spectrometers such as flight time (TOF or Q-TOF), quadrupole-orbitrap (Q or Q-OT), triple quadrupole (TQ), or quadrupole-electrospray ionization (Q-ESI) for metabolomic profiling and biological analysis in dietary supplements, plants, fruits, and vegetables has increased over the last few years [
1].
Until now, no information regarding the NF of Muérdago fruit juice and the isolation of its bioactive compounds is available in the literature. Muérdago fruit juice contains valuable metabolites, which can be selectively fractionated through NF and allow the creation of specific fractions for use on food. In this way, the aim of this work was to evaluate the NF of a Muérdago fruit juice and to identify the fractionated metabolites (phenolics and amino acids) by using ESI-MS/MS. Specifically, three NF membranes were used, sequentially and separately, starting from the membrane with the smallest pore size and ending up by using the membrane with the biggest pore size. The process performance evolution, the fouling formation, a chemical cleaning procedure, and the specific molecule fractionation were evaluated. ESI-MS/MS analysis was used to identify several metabolites and bioactive compounds contained in the Muérdago fruit, the quantification of some phenolics, and their observation and evaluation during NF.
3. Materials and Methods
3.1. Muérdago Extract Solution Preparation
Muérdago (Chilean mistletoe,
Tristerix tetrandus) fruits collected in southern Chile were freeze-dried and stored at room temperature in complete darkness. Then, the freeze-dried fruits were milled into finely ground flour using a laboratory grinding machine (Polymix
® PX-MFC 90D, Kinematica AG, Malters, Switzerland), at 220 rpm, and stored hermetically in a freezer at −20 °C until used. Then, 10 g of the finely ground Muérdago flour was dissolved into three liters of distilled water (electrical conductivity (EC) < 4 µS/cm (pH = 6.5 ± 0.2)) during one hour at room temperature. Immediately afterwards, the Muérdago solution prepared was filtered twice through cotton with the aims of removing all the possible pectin material contained in it and protecting the NF membrane integrity. Then, the resulting solution was filtered twice, using two layers of gauze as a first step, and then subjected to three consecutive ultrasound baths (Ultrasonic TI-H 20; Elma Schmidbauer GmbH, Singen, Germany), treated with an ultrasonic power of 100% (250 W) under an ultrasonic frequency of 35 kHz for a period of 15 min each. After the ultrasound baths, the solution was newly filtered through filter paper under vacuum. This last procedure was repeated three times, until no more accumulated matter was observed on the filter paper material. The resulting liquid extract was reconstituted up to three liters using distilled water and immediately used as the feed solution for the membrane fractionation process. This last reconstitution was realized to obtain the solution volume lost during the previous filtrations. Muérdago fruit was chosen for the membrane treatments since its extracts possess a wide variety of interesting bioactive molecules, which molecular weights range from approximately 100 Da up to 700 Da, considering amino acids [
12] and phenolics. These molecules were the target fractionation materials.
3.2. Membrane Materials
Three different polyamide-TFC NF membranes with different molecular weight cut-offs (MWCOs) were used, which were purchased from Sterlitech Corporation, Auburn, WA, USA. The membranes were a DL membrane (Suez (GE)
TM) (pore size/MWCO: 150~300 Da), an NFW membrane (Synder
TM) (pore size/MWCO: 300~500 Da), and an NDX membrane (Synder
TM) (pore size/MWCO: 500~700 Da).
Table S1 (Supplementary Materials) displays the technical specifications of the mentioned membrane materials. The three membranes were selected according to the target bioactive molecules contained in the Muérdago fruit juice and the membranes’ MWCOs (
Table S1), which should be able to separate the treated biomolecules into different and profitable liquid fractions.
3.3. Protocol
The Muérdago fruit juice was fractionated using a crossflow membrane filtration system (CF042D membrane separation cell (Delrin Acetal)) (Sterlitech Corporation, Auburn, WA, USA). A Hydracell M03-S pump (positive displacement, diaphragm pump (Wanner Engineering, Minneapolis, MN, USA)) was used as a feed-flow pump to operate the CF042D crossflow cell unit and to pump the treated fluid through the entire fractionation system. The NF membranes were carefully cut and disposed into the membrane module, which had a 42 cm
2 effective area. Three commercial NF membranes (DL (Suez (GE)
TM), NFW NF membrane (Synder
TM), and NDX (Synder
TM)) were used during the consecutive fractionation trials. Each treatment was carried out separately and apart from the other treatments (
Figure 2). Specific technical properties of these NF membranes are described in
Table S1. Two pressure gauges (manometers) were connected to the inside and outside tubbing (SS-316) of the membrane separation cell with the aim of controlling the desired transmembrane pressure (TMP) value accurately. As the outlet for permeation was opened to the air (
Figure 2), the average value of these two pressure meters was assumed to be the TMP.
First, three liters of the prepared Muérdago fruit juice was disposed into the system feed tank at room temperature (≈20 °C) and was treated using a DL membrane (DL treatment) (
Figure 2) at a TMP of 10 bar and a constant crossflow velocity of 2.85 (L/min) for three hours. The final volumes obtained of both the permeate and the feed solutions were recounted. Then, the feed solution obtained from the DL treatments was reconstituted with distilled water until reaching again a volume of three liters. This reconstituted fruit juice was processed using an NFW membrane (NFW treatment) (
Figure 2) at a TMP of 30 bar and a constant crossflow velocity of 2.85 (L/min) for three hours. The final volumes obtained of both the feed and permeate solutions were recounted, and the resultant feed solution (obtained from the NFW treatment) was newly reconstituted with distilled water (pH = 6.5 ± 0.2) until reaching a three-liter volume. Immediately afterwards, this reconstituted solution was processed using an NDX membrane (NDX treatment) (
Figure 2) at a TMP of 25 bar and a constant crossflow velocity of 2.85 (L/min) for three hours. Finally, the resultant volumes of the permeate and feed solutions were recounted. Samples of 1.5 mL were taken along each processing trial, at the processing times 5 and 180 min, from the feed and the permeate streams. All the taken samples were immediately refrigerated and kept at 4 °C until rapid analysis. All the fractionation experiments were carried out in a concentration mode. After each membrane treatment, the respective obtained permeate volumes (
Figure 2) were stored and not reused during the next consecutive treatments.
The high TMP values used with the membranes NFW and NDX (30 bar and 25 bar, respectively) were chosen according to previous preliminary tests and also because of these membranes having reported to perform better at high TMPs, even close to the membranes’ burst pressures (41 bar) [
34]. The concentration of the treated Muérdago extract was kept low in solution (10 g of dry Muérdago powder initially dissolved into 3 L of distilled water). This would allow to observe the fractionation process more clearly, even while membrane fouling appears (forming not excessively thick layers), and the impact on the migration rates and the permeate flux. Membrane fouling by phenolics has been previously observed [
35].
The parameters monitored during the fractionation process were solution pH, solution electrical conductivity, and temperature. These parameters helped to understand the mass transfer process and were repeatedly measured during NF [
36,
37]. The solution was recirculated within a closed stainless steel 316 system, and the permeate was continuously collected into a 250 mL test tube in order to determine the permeate flux along the processing. In addition, each run was carried out in triplicate for all the different treatments realized, and the average value of each parameter measured was a final result. Afterwards, the fouled membranes were evaluated in relation to the permeate flux achieved and then cleaned by washing them up with a cleaning solution of 1% Ultrasil 11 (membrane alkaline detergent) (Henkel, Ecolab, Saint Paul, MN, USA) (pH = 12.0) for at least 1 h, at a slightly elevated temperature (around 40 °C), and at a TMP of 3.5 bar. Finally, the whole system (CF042D cell, pump, SS-316 tubbing) was rinsed several times with deionized water (EC < 4 µS/cm) until the total Ultrasil 11 was removed from the circuit. A test measuring pH and electrical conductivity was then performed in the rinsing water in order to corroborate a clean membrane circuit. Then, the permeate flux of distilled water was determined with the clean membrane material.
3.4. Membrane Filtration Assessment
The filtration assessment was performed on the new, the used, and the chemically washed membranes. After the membrane compaction pretreatments, trials using deionized water were carried out on each membrane sample. This was performed before the treatments of the Muérdago fruit solutions and in order to have a record of their respective filtration performances as new membrane material. These trials were also carried out on the fouled membranes after cleaning them with Ultrasil 11. This procedure assessed the effect of fouling formation on the membrane material integrity and the respective performances after the chemical cleaning was realized. The treatments were realized using a crossflow velocity of 2.85 (L/min), at 20 °C, and using TMP values of 5, 10, 15, 20, and 25 bar. At each tested TMP value, the permeate flux was recorded in triplicate considering a filtrated volume of 10 mL for each time-lapse measurement. The TMP value increased slowly and gradually when carrying out the mentioned tests and starting from the lowest up to the highest TMP value, keeping constant each of them while registering the time elapsed. The membrane resistances were calculated according to Equations (1) [
38] and (2) [
20], considering the slope of each generated curve and the viscosity of water at 20 °C (0.001 Pa*s). The
RM was calculated on the new and on the chemically washed membranes after fouling. This parameter was determined on all the tested membranes (DL, NFW, and NDX). Equation (1) is presented as follows:
where
J is the permeate flux,
TMP is the transmembrane pressure,
RM is the membrane resistance,
RF is the fouling resistance,
RCP is the concentration polarization resistance,
µ is the viscosity, and Δ
π is the osmotic pressure. When the solution is pure water, as in the case of this study,
RF and
RCP become zero and only
RM exists. In that case,
RM can be obtained according to Darcy’s law, as shown in Equation (2):
where
J is the permeate flux of sample solution during NF processing (m
3*m
−2s
−1), Δ
P is the TMP (bar),
μ is the solution viscosity (bar*s), and
RM is the hydraulic membrane resistance (m
−1). The membrane resistances were determined for the new, the fouled, and the washed (Ultrasil 11 washing) membranes (DL, NFW, and NDX). The hydraulic MR indicates the membrane integrity and performance.
3.5. Retention Percentage (RP)
A particular process variable used to evaluate the membrane fractionation was the retention percentage (
RP). The
RP allowed to observe the selective passage of some of the studied molecules through the tested DL, NFW, and NDX membranes. The
RP was calculated using Equation (3) [
39]:
where
CP and
CF represent the respective solute concentrations (mg/L or ppm) in the permeate and the feed streams, respectively, at a determined processing time.
3.6. Permeation Percentage (PP) of Total Solids
The permeation percentage of total solids was determined during each of the treatments realized (for the membranes DL, NFW, NDX) at the processing times 5 and 180 min. It can be defined as the recovery of total compounds in the permeate during the concentration process [
40] and was calculated using Equation (4):
3.7. Electrical Conductivity, pH, and Temperature Measurements
Measurements of electrical conductivity, pH, and temperature were made in the feed, the concentrate, and the permeate streams, during all the treatments realized, using an HI 991,301 pH/EC/TDS/temperature meter (Hanna Instruments, Cluj-Napoca, Cluj, Romania).
3.8. Visual Membrane Inspection and Characterization
Digital camera photographs were taken in order to identify and to characterize the aspect of the original membrane material and to compare it with the presence of fouling layers on each of the used membranes DL, NFW, and NDX. This was made in order to visually inspect the active surfaces of the original and of the used active layers of every membrane used and to the detect the presence of membrane fouling in each case. Membrane fouling disturbs the process performance and membrane lifetime.
3.9. Tentative Identification (ESI-MS/MS) and Quantification (UHPLC-ESI-MS/MS) of Metabolites in the Muérdago Fruit
Liquid samples of processed Muérdago extract obtained from the feed and from the permeate streams, at the processing times 5 and 180 min, respectively, were filtered and injected in the UHPLC-ESI-MS/MS equipment. Nylon filters (Iso-disc 0.45 µm; Millex-HN, Millex®, Merck KGaA, Darmstadt, Germany) were used to filter the final extract before injection. All the samples were analyzed by UHPLC-MS/MS in an Ekspert UltraLC 100-XL ultra-high-pressure liquid chromatograph coupled to an electrospray (ESI) ABSciex Triple Quad 4500 triple quadrupole mass spectrometer. A PhenomenexSynergi™ Fusion-RP 80 Å (50 mm × 2.0 mm, 4 μm) column was employed, and the mobile phase was prepared from 0.1% v/v formic acid in water (eluent A) and acetonitrile (eluent B). All the constituents of the mobile phases were HPLC-grade. The gradient was programmed as follows: (time, min/%B) 1/5%, 12/50%, 13/50%, 14/5%, and 15/5%. The mass spectrometer parameters were gas 1 N2 (40 psi); gas 2 N2 (50 psi); ion spray voltage, 3500 V; ion source temperature, 650 °C; curtain gas N2 (25 psi); flow 0.3 mL/min; and scan mode MRM with both positive and negative polarity. The UHPLC-MS/MS system was controlled with Analyst 1.6.2, and the data were processed with Multiquant 3.0. Calibration curves were built for each compound in the 0.1–0.8 μg/mL range. The high resolution and accurate mass via orbitrap (HESI orbitrap HR-MS) used in this study enabled the identification and tentative characterization of compounds including phenolics and amino acids. Some of the identified phenolics from all the detected ones in the extract were directly identified, without using references, and quantified by UHPLC-MS, which are presented later on.
3.10. Statistical Analysis
Data obtained were subjected to one-way ANOVA using software Sigmaplot (Sigmaplot 14.0, Systat Software Inc. San Jose, CA, USA) in order to compare the mean values of the calculated membrane resistances and some other parameters. Two-way ANOVA was used to compare the mean values of the calculated permeation percentages of the total solids in solution. Two-way repeated measures (RMs) ANOVA was used to evaluate the evolution of certain parameters measured in the solution samples through the processing time and the influence of two independent variables on the obtained data. All the experiments were carried out in triplicate. The power of all the performed tests was the standard criteria for significance (α = 0.0500). Values of p < 0.0500 were considered as denoting a significant statistical difference among average parameter values.
4. Conclusions
The protocol realized and the membranes tested allowed an interesting fractionation of bioactive compounds that are present in the Muérdago fruit and generate considerable permeate flux. Some phenolics were highly retained by the membranes tested, while some amino acids permeated them progressively and selectively (aspartic acid, proline, tryptophan, valine, leucine, isoleucine, methionine, arginine), as the solution was continuously reconstituted. Tryptophan was found only in the DL permeate fractions. These DL permeate fractions contained also the phenolics quinic acid, 3-CQA, cryptochlorogenic acid, chlorogenic acid, and p-coumaric acid. On the other hand, whereas leucine and isoleucine permeated only the membrane NFW, the amino acids methionine and arginine were found only in the NDX permeate fractions. The NFW permeates contain the phenolics quinic acid, 3-CQA, ferulic acid, cryptochlorogenic acid, chlorogenic acid, and coumaric acid. The NDX permeates contain mainly the phenolics 3-CQA and ferulic acid.
Membrane fouling was observed during each membrane treatment carried out, but it was successfully removed from each membrane after a chemical cleaning treatment and recovered the initial performance. It was observed that the DL membrane allows higher permeate flux at lower TMP values than the NFW and the NDX ones. The DL membrane also allows the more important permeation amounts of total compounds, followed by the NFW membrane and then by the NDX membrane. Permeate fractions of great interest to the food and pharmaceutical industries were obtained and are suitable for process optimization and scale-up. Membrane technology shows promising applicability for the extraction of metabolites from numerous fruit and vegetable juices. Nevertheless, each different process must be carefully studied and optimized in terms of performance and membrane fouling formation.