A Novel Method for Extracting Glucomannan Using an Aqueous Two-Phase System

Materials

The main material was porang flour, which was obtained from a local supplier in Subang, West Java, Indonesia. The chemical reagents were technical grade ethanol, analytical grade K₂HPO₄ (Merck, Damstadt, Germany), and Na₂HPO₄ (Merck, Damstadt, Germany), chromatography grade water (Merck, Damstadt, Germany) and standard polyethylene oxide/polyethylene glycol (Agilent, Craven Arms, UK).

Process of glucomannan extraction using aqueous two-phase system

The glucomannan was extracted based on the schematic diagram of the operation procedure of an extraction using alcohol/salt aqueous two-phase system (27). ATPS was made by mixing K₂HPO₄ and Na₂HPO₄ solution at the mass fractions of 1, 2 and 3 % with 40 % of ethanol at a volume ratio of 1:1. The salt mass fractions were selected based on the binodal curve of ethanol/K₂HPO₄ and ethanol/Na₂HPO₄ obtained in a preliminary study (Fig. S1 andTable S1). A mass of 100 g porang flour that passes through a 40-mesh screen was added to the 250 mL ATPS solution and then it was extracted in a high-speed blender (8010BU Set; Waring Blender Laboratory, Torrington, CT, USA) at 18 000 rpm for 2 min (Table S2). The preliminary study was conducted to compare the low-speed and high-speed mixing processes. The low-speed mixing process was conducted at 400 rpm for 30 min using a magnetic stirrer (SMHS-3; Daihan Scientific, Gangwon-do, South Korea). High-speed mixing was selected based on a preliminary study on the effect of low and high-speed mixing on the glucomannan yield (Table S2). After resting for 30 min, the solution was separated into 3 layers, with the extracted glucomannan in the bottom layer. The glucomannan extracted using ATPS will be referred to as the treated glucomannan in the subsequent discussion. The control glucomannan sample was prepared by extracting porang flour with 40 % ethanol.

Separation of ATPS phases

Following the previous steps, porang flour in ethanol/K₂HPO₄ and ethanol/Na₂HPO₄ of ATPS was separated into three layers. The bottom layer (F1) and the middle layer (F2) were salt-rich phases and the top layer (F3) was ethanol-rich phase. The bottom layer and the middle layer were separated in a blender (Vitamax; Madato, Taipei, Taiwan) with a 60 mesh screen. The residue was mixed with 250 mL of 40 % ethanol, then mixed at a high speed mixer (8010BU set; Waring Blender Laboratory) and finally filtered again. This step was repeated with 70 % ethanol to wash the glucomannan fraction. After the washing step, the glucomannan fraction was dried at 50 °C for 12 h. Iodine test was used to confirm that starch component was separated from the glucomannan fraction. The filtrate obtained in the separation step was centrifuged (SL40R; Thermo Sciencetific, Osterode am Harz, Germany) at 2504×g for 30 min. The pellet was then dried to obtain starch fraction (F2). The yield of F1 and F2 was determined by dividing the mass of the dried F1 and F2 by the initial mass of porang flour. F3 was calculated by subtracting the percentage of F1 and F2 from 100 %. The control glucomannan sample was obtained according to the same protocol as the ATPS-treated samples.

Determination of proximate composition of glucomannan

The proximate composition of the glucomannan includes moisture, ash, protein and carbohydrate content. The moisture and ash contents were assayed using a gravimetric method based on the determination of moisture and ash content in animal feed (28,29). Nitrogen combustion method was used to measure protein content using a protein analyzer (Büchi Dumaster D480; Elementar Analysensysteme, Langenselbold, Germany). The calculation of protein content used a nitrogen conversion factor of 5.7 according to the standards of the European Commission (2). Glucomannan content was calculated as the percentage of carbohydrates, which was determined by subtracting the sum of the percentages of total ash, moisture and protein content from 100 % (3).

Colour

A spectrophotometer (CM700D; Konica Minolta, Osaka, Japan) was used to measure the colour of the glucomannan sample. The colour parameter of a sample was read in a cuvette. The data were reported including the value of L* (lightness index), a* (red to green index) and b* (yellow to blue index). The change in the colour of the treated sample was compared to the control and calculated using the following equation:

/1/

Fourier transform infrared spectroscopy

Fourier transform infrared (FTIR) spectrophotometer (Alpha II; Bruker, Ettlingen, Germany) was used to identify the functional groups of the obtained glucomannan. The analysis was carried out in the infrared range of 400-4000 cm^-1 with a resolution of 4 cm^-1.

Nuclear magnetic resonance spectroscopy

The ¹H NMR spectra were read on a nuclear magnetic resonance spectrometer (JNM ECZR500; Jeol, Tokyo, Japan). The sample was prepared according to Tang et al. (30) with modifications. A total of 40 mg of sample was dissolved in D₂O (40 mg/mL) and mixed for 1 h, and the measurement was run at 25 °C and 500 MHz. Chemical shifts are expressed in ppm and tetramethylsilane was used as the reference standard. ¹³C CPMAS (solid NMR) of spectra were recorded on an NMR spectrometer (JNM-ECZ500R/S1 DPX200; Jeol) operating at a frequency of 125.76 MHz and used a solid-state probe equipped with 4 mm (o.d.) spinner. The spectra were recorded at 5000 scans, with relaxation delay 15 s and spin rate 10 kHz. The integration value of anomeric proton from ¹H-NMR spectra was used to calculate the ratio of mannose and glucose. The results of the integration of anomeric carbon area at a chemical shift of 105 ppm and methyl carbon at 21 ppm were used to calculate the degree of acetylation (DA) using the following equation (31):

/2/

where I_A is the integrated area of chemical shift at 105 ppm and I_Ac is the integrated area of chemical shift at 21 ppm.

Molecular mass

The molecular mass of the glucomannan sample was determined using a gel permeation chromatography/size exclusion chromatography (GPC/SEC) system (1260 Infinity II; Agilent Technologies, Wadbronn, Germany) with a column PL 2080-0700 and two detectors, a refractive index (RI) detector and a viscometer detector. The mobile phase consisted of water and 0.02 % NaN₃. The glucomannan sample was dissolved in water (1 mg/mL), stirred and filtered using 0.45-µL membrane filter (Agilent Captiva Econofilter; Agilent Technologies, Bejing, PR China). The flow rate of the eluent was 0.5 mL/min and the columns and detectors were maintained at 35 °C (15). Prior to use, the GPC/SEC was calibrated using standard substances of polyethylene oxide/polyethylene glycol (Agilent Technologies UK Ltd, Craven Arms, UK).

Thermal analysis

The thermal properties of the sample were assessed by a differential thermal analysis-thermogravimetry (DTA-TG) apparatus (DTG-60; Shimadzu, Kyoto, Japan). The sample of 5 mg was placed on the aluminium sample pan, sealed and heated from 25 to 450 °C with an average heating rate of 10 °C/min (7).

Morphological and residual mineral analysis using scanning electron microscopy with energy dispersive X-ray spectroscopy

The morphological properties of the glucomannan sample were observed by a scanning electron microscope (JSM-IT300LV; Jeol). Before analysis, the sample was sieved through an 80-mesh screen. The sample was placed on a metal stub that was previously covered with a double-sided adhesive tape. A rubber air blower pump dust cleaner was used to remove the excess sample from the metal stub. Prior to observation, the sample was coated with gold and it was examined with an accelerating voltage of 2 kV at magnification of 100, 500 and 1000 times.

Statistical analysis

A completely randomized design (CRD) with 2 factors including the salt (Na₂HPO₄ and K₂HPO₄) and salt mass fraction of 1, 2 and 3 % was used as the experimental design. The effects of extraction using ATPS including the separation, proximate composition of glucomannan, colour and thermal properties were observed. Multivariate analysis of variance (MANOVA) followed by a post-hoc Duncan’s test were carried out to test the significant differences between the mean values of each treatment (15). The statistical significance level was set at p=0.05. The statistical analysis was conducted using SPSS v. 26 (32).

ATPS method for glucomannan extraction

The aqueous two-phase system (ATPS) composed of ethanol, salt and water, where the binodal curves (Fig. S1 andTable S3) were constructed to determine the compositions of ethanol/Na₂HPO₄ and ethanol/K₂HPO₄, was used in this experiment. It shows that the binodal curve limited the two ATPS zones, namely monophasic region (lower side) and biphasic region (upper side). For the extraction 40 % ethanol (22) was chosen and then the mass fractions of 1, 2 and 3 % of salts (18) were selected based on their amount in the monophasic region and near the critical point of the binodal curve. Based on these results, the composition of ATPS used in this study is given inTable S1.

Fig. S2 shows the proposed mechanism of glucomannan extraction using the ATPS based on the visual observation during extraction. The extraction of glucomannan from porang flour using ATPS consisting of ethanol/Na₂HPO₄ and ethanol/K₂HPO₄ resulted in three layers, namely the bottom (F1), the middle (F2) and the top layer (F3). The bottom and middle layers (F1 and F2) were rich in salt-water phases. Based on the iodine testing (5), the colour of the bottom (F1) and top layer (F3) did not change, while the middle layer (F2) turned dark blue. This result indicated that the bottom layer was rich in glucomannan and the middle layer had a high starch content. The top layer (F3) was the ethanol-rich phase that contained many simple sugars, dyes and other compounds. The formation of three layers after the extraction was greatly influenced by the salts in the composition of ATPS. The K₂HPO₄ and Na₂HPO₄ are dissociated into K⁺, Na⁺ and HPO₄^2- ions when they are dissolved in water. The ions can break the hydrogen bond between the water and the ethanol because of the hydration of ions (33). Hydration capacity of ions depends on the Gibbs energy of hydration ion (25). Furthermore, both salts have higher salting-out effect than other salts based on the ionic strength sequences of SO₄²⁻<H₂PO₄⁻<Cl⁻<NO₃⁻<ClO₄⁻<SCN⁻ and K⁺<Na⁺<H⁺<Mg²⁺<Ca²⁺<Al³⁺ (23,26). When the salts are added to the glucomannan solution, the ions compete with the glucomannan molecules to bind the water. This phenomenon reduces the glucomannan solubility in the solution and eventually the glucomannan molecules precipitate in the bottom layer. A similar result was reported during polysaccharide extraction of Lycium barbarum L. with ATPS (34). The F1 and F2 (Fig. S2), which consisted of polysaccharides, were separated into two layers due to the difference in their water absorption capacity and molecular mass. The water absorption capacity of glucomannan, starch and cellulose is 50-100, 0.884-0.951 and 40 g/g, respectively (5,35,36).

The results of ATPS separation are shown inTable 1. Results indicated that the yield of glucomannan (F1) extracted using ATPS did not significantly change but was lower than that of the control. However, the results of the middle (F2) and the top layer (ethanol) (F3) were higher than that of the control. These might be because some glucomannan molecules were not separated from the starch in the middle layer (F2) since they have almost similar molecular mass of 10⁶ Da. Other possibility is that there was an interaction between glucomannan and the other polysaccharide molecules through an ion bridging mechanism. Glucomannan interacts with xanthan gum via Na⁺ and Ca²⁺ (36). In our experiment, ATPS with K₂HPO₄ yielded higher F2 than that with Na₂HPO₄. The K⁺ cation has a bigger atomic size than Na⁺, so they can have different ionic strength when they interact with starch and protein molecules according to the lyotropic sequence (26). Results also showed that increasing salt mass fraction up to 3 % increased the yield of F1 and F2. This might be due to the increase in the salting-out effect of the salt. F3 of the ATPS was higher than that of the control (ethanol extraction). It contained non-polar components such as flavonoids, polyphenols and ethanol-soluble alkaloids, as also reported by Wan et al. (23) and Xi et al. (27). Moreover, F3 also contained other organic compounds such as carotenoids, oligosaccharides and monosaccharides that were soluble in the ethanol phase.

Physicochemical properties

Results of proximate analysis (Table 2) indicated that the protein mass fraction of glucomannan extracted using ATPS was significantly lower than that of the control glucomannan (p<0.05). Cheng et al. (37) reported that protein could be denatured in the aqueous organic solvents and precipitated in the aqueous phase. Furthermore, the salting-out effect of K₂HPO₄ and Na₂HPO₄ caused the decrease of solubility of the protein in water and it was precipitated in the middle layer (Table 3). Therefore, the protein mass fraction of glucomannan obtained from the ATPS was lower than that obtained by the conventional extraction method (control). These results were similar to the report of Antunes et al. (25) where the separation of polysaccharides using ATPS resulted in the lower protein content. The ATPS extraction effectively reduced the protein content up to <0.5 %. Thus, the protein mass fraction obtained from this study was lower than that obtained by ethanol extraction (3.8-4.4 %) (13), freeze-thaw cycle pre-treatment (1.4-2.3 %) (15) and microwave-assisted extraction (0.82 %) (38).

Results (Table 2) also showed that the ash mass fraction of glucomannan extracted by ethanol/Na₂HPO₄ was lower than that obtained using ethanol/K₂HPO₄. This result indicated that glucomannan contained a residue of potassium and phosphate from the ATPS. The results obtained by scanning electron microscopy combined with energy dispersive X-ray analysis (SEM-EDX) showed that the mass fractions of potassium and phosphorus in the glucomannan obtained by ethanol/K₂HPO₄ extraction (Table 3) were higher than those obtained by ethanol extraction. The K⁺ may have higher ionic interaction with glucomannan molecules than the Na⁺ because the atomic size of K⁺ is larger than of Na⁺. The degree of ionic binding is directly related to the nuclear charge effect that depends on the size and charge of the dissolved ions (39).

Table 4 shows that glucomannan extracted using ATPS was brighter in colour than that of the control (p<0.05). Results indicated that the lightness values of glucomannan obtained by ATPS increased significantly compared to those of the control. Moreover, the total colour difference (ΔE) showed that the treated samples were different from the control. Results also showed that higher salt percentages of ATPS caused lower lightness values of glucomannan. Among the treated samples, glucomannan produced using 2 % Na₂HPO₄ had the highest lightness value and the lowest values of a* and b* parameters and the highest ΔE value. ATPS can inhibit browning reactions by inhibiting mechanisms of the activity of oxidizing enzymes (22). Moreover, the addition of salt affected the increasing polarity of the bottom layer, which resulted in an increase in the solubility of organic compounds in the ethanol-rich layer (top layer) including the carotenoid compounds. The colour of glucomannan is influenced by the natural yellow-orange colour characteristics of porang tuber (5). Porang tuber contains organic compounds such as carotenoids, polyphenols and other colour compounds that are susceptible to oxidation reactions. The oxidizing reaction occurs more intensively during the processing stages of porang chip and flour, particularly when the sliced porang tubers are exposed to air (40). Therefore, the colour of glucomannan obtained by ATPS had higher lightness and lower a* and b* values.

Structural properties

Results showed that extraction using ATPS produced glucomannan with the average molecular mass (M_w) ranging from 1.55·10⁶ to 2.9·10⁶ g/mol and the number of average molecular mass (M_n) was 5.42·10⁵-9.95·10⁵ (Table 4). This result indicated that the molecular mass of glucomannan was not affected by the ATPS extraction method (p>0.05). Jiang et al. (41) reported that salt did not affect the degradation of molecular mass of the polysaccharides. The molar mass (M) of glucomannan depends on the species of Amorphophallus. For instance, the molar mass of glucomannan isolated from Amorphophallus paeoniifolius, Amorphophallus panomensis, Amorphophallus tonkinensis and Amorphophallus konjac is 1.115·10⁶, 1.023·10⁶, 1.043·10⁶ and 9.1·10⁵ g/mol, respectively (7,13).

Table 4 shows that the isolated glucomannan has a polydispersity index (PDI) ranging from 2.27 to 3.35, which was similar to previous studies (5,42). This result showed that isolated glucomannan has a broad molecular mass distribution. It also indicated that the synthesis of glucomannan occurred by an uncontrolled reaction mechanism, namely chain reaction. Due to chain reaction mechanism, polymer chains are formed with widely varying molecular mass indicated by PDI between 1.5 and 20 (43). Qi et al. (44) reported that the biosynthesis pathway of glucomannan in plant occurred by enzymatic mechanisms that produce glucomannan molecules with varying chain lengths.

The FTIR spectra of glucomannan obtained by ATPS (Fig. 1) show the groups of glucomannan structure. The wide peak at 2900-3600 cm^-1 is a typical peak of the OH group originating from glucomannan monomers, both glucose and mannose. In addition, the broad peak indicates a large number of hydrogen bonds or bound water molecules. The -CH- aliphatic, C=O of the acetyl, C-H bending and C-O-C group appear at 2800-2900, 1724, 1200-1400 and 1000-1100 cm^-1, respectively. The protein content, as another component of the glucomannan, was detected by the presence of the amide group peak -CONH- at a wavenumber of 1640 cm^-1. The FTIR spectra of the obtained glucomannan showed the same pattern as those of the glucomannan obtained by extraction with ethanol (45). The bound water in the glucomannan was identified by the presence of peaks at 1611 and 1411 cm^-1 (30). The peaks at 878 and 800 cm^-1 were attributed to β-glucosidic and β-mannosidic linkages, respectively. This result was in agreement with the glucomannan extracted from Amorphophallus konjac (13).

Fig. 1 Fourier transform infrared spectra of glucomannan: a) glucomannan obtained by ethanol/Na2HPO4 extraction, and b) glucomannan obtained by ethanol/K2HPO4 extraction. 1 to 3 %=mass fractions of salt in the solution

NMR spectra of glucomannan showed the chemical shift (δ) of the proton and carbon of the glucomannan. The proton and carbon spectrum patterns of glucomannan were identical to those of control glucomannan (Fig. 2). The difference was in the appearance of a proton from the salt residue (Na₂HPO₄ or K₂HPO₄) at a δ=5.703 ppm. The chemical shift of anomeric proton (H1) of glucomannan control was observed at δ=5.128 ppm for H1 mannose and 5.213 ppm for H1 glucose (Fig. 2a). The protons (H2-H6) were at δ=3.822-4.580 ppm. Anomeric proton of the treated glucomannan (Fig. 2b) showed the chemical shift at δ=5.053 and 5.268-5.280 ppm for H1-mannose and glucose, respectively. Meanwhile, the proton shift from H2 to H6 (mannose/glucose) was 3.933-4.637 ppm and the proton of methyl group

Fig. 2 NMR spectra of glucomannan: a) 1H NMR of control, b) 1H NMR of glucomannan obtained by aqueous two-phase system (ATPS) extraction, c) 13C-NMR of control glucomannan, and d) 13C-NMR of glucomannan obtained by ATPS extraction

(-CH₃) of acetyl appeared at δ=2.7 ppm. Enomoto-Rogers et al. (46) reported that 1.9, 2.0 and 2.1 ppm are the chemical shift values for the proton of the acetyl group and the chemical shift was 3.3-4.1 ppm for H2-H6 protons from polysaccharides (47). The ¹H NMR spectra of control and treated glucomannan were similar to a previous study (48).

The ratio of glucose and mannose in the glucomannan was calculated by the ratio of the integration of H1 of glucose and mannose. The control and treated glucomannan had mannose/glucose ratios of 1.09:1.00 and 0.71:1.0, respectively. Meanwhile, the ratio of H1 mannose/glucose of glucomannan extracted from Amorphopallus panomensis in Vietnam and Amorphophallus konjac was 1.00:0.13 and 1.6:1.0, respectively (49,50). These results show that the Amorphophallus species influenced the chemical structure of glucomannan (42).

The ¹³C NMR spectra of the anomeric carbon of control and treated glucomannan were 102.941-105.147 ppm (Fig. 2c) and 103.554-105.392 ppm (Fig. 2d). The chemical shift of C2-C5 is 50.489-82.843 ppm with overlapping peaks indicating that the carbon atoms of the pyranose ring, namely glucose and mannose, have almost the same properties (50). C6 has a chemical shift at 62.070 ppm for control glucomannan and 61.396-62.254 ppm for the treated glucomannan. The acetyl group appeared as C=O at chemical shift at 170 ppm and CH₃ at 21 ppm. The α- and β-glucose and mannose configuration can be determined using the chemical shift of anomeric proton (H1)/carbon (C1) at 90–110 and 4.5–5.5 ppm (30). Chemical shift of anomeric carbon at 98-108 and 101-105 ppm indicated the α-glycoside or β-glycoside bonds, respectively (48). According to the chemical shift of anomeric carbon, the β-glycoside bonds contructed both glucomannan structures. The involvement of C4 in the formation of glycosidic bonds is shown at a chemical shift of 79.24 ppm (49,51). Therefore, based on the proton and carbon shift values, the α-glycoside bonds formed are β(1→4)-glycoside and β(1→6)-glycoside. These bonds indicate that the structure of glucomannan has a straight chain as a backbone and a branched structure (48).

The peaks for acetyl carbon -CH₃ and C=O at chemical shifts of 21 and 170 ppm, respectively, were of low intensity. The ratio of peak areas at 21 and 105 ppm indicated the degree of acetylation (DA) of glucomannan (31). The results showed that the DA of control glucomannan and the treated glucomannan was 4.46 and 1.88. The reducing value of DA of the treated glucomannan due to deacetylation process was due to the interaction with Na₂HPO₄ salt (52).

Thermal properties

Results indicated that glucomannan extraction using ATPS influenced the thermal properties of the obtained glucomannan (Fig. 3). The thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) thermograms of all samples showed consistent thermal degradation patterns of glucomannan in which the first and second peaks show dehydration and degradation patterns, respectively. The dehydration process of the control glucomannan required higher energy and occurred at higher temperature than that of the treated samples (Table S4).Fig. 3a,Fig. 3b andTable S4 show that the onset temperature of degradation of the control glucomannan was higher than that of the glucomannan from the ATPS. This indicated that the glucomannan extracted using ATPS was easier to degrade than the control glucomannan at the starting point. The protein content of the control glucomannan was higher than that of the treated samples. Protein and glucomannan can interact through hydrogen bond between the hydroxyl group (-OH) and the amine group (-NH-) (53), an interaction that can enhance the thermal stability of protein.

Fig. 3 Determination of thermal properties: a) differential scanning calorimetry (DSC) thermogram of control glucomannan and the glucomannan obtained by ethanol/Na2HPO4 extraction, b) DSC thermogram of control glucomannan and glucomannan obtained by ethanol/K2HPO4 extraction, c) thermogravimetric analysis (TGA) of control glucomannan and glucomannan obtained by ethanol/Na2HPO4 extraction, and d) TGA of control glucomannan and glucomannan obtained by ethanol extraction/K2HPO4. 1 to 3 %=mass fractions of salt in the solution

Thermogravimetric analysis (Fig. 3c,Fig. 3d andTable S5) showed that the mass loss of the control glucomannan was higher than of the treated samples. This result indicated that the phosphate residue may act as the stabiliser of glucomannan molecules, which is similar to the previous results in that there was an effect of salt in potato starch and iota-carrageenan solutions (54). Deng et al. (55) reported that the leftover phosphate in the glucomannan sample prevented mass loss during the degradation step.

Morphological properties

Fig. 4 shows the morphological surface of glucomannan particles. The morphological surface of the particles of the control glucomannan was different from that of the treated samples. The particles of the treated samples were relatively uniform in size compared to the particles of the control glucomannan. The morphological surface of control glucomannan was similar to that of the purified konjac glucomannan as reported by Yanuriati et al. (5). However, there were no significant differences among the particles of the treated samples. This result emphasised that the extraction using ATPS was able to produce uniform glucomannan particles and the extraction method did not destroy them. The surface of the glucomannan particles was wrinkled. Some impurities can be trapped on the surface, including starch, cellulose, protein and soluble sugar (56). The presence of the phosphate group might result in glucomannan particles with rougher surfaces and larger size (55).

Fig. 4 Morphological surface of the control glucomannan and glucomannan extracted using aqueous two-phase system ATPS) observed with 100× magnification. 1 to 3 %=mass fractions of salt in the solution