Byproduct from Triphala Extraction as Tannin and Rutin Sources for Production of Gallic Acid, Isoquercetin and Quercetin by Solid-State Fermentation

Byproduct from Triphala extraction process (BTP) was studied as a substrate for gallic acid, isoquercetin and quercetin production by Aspergillus niger fermentation in this research. The results showed that BTP was a good source of tannin and rutin. Neverthe­ less, the activity of A. niger on BTP as a sole substrate was very low. Supplementing nitro­ gen sources was found to be a key to enhancing conversion of tannin to gallic acid, and rutin to isoquercetin or isoquercetin and quercetin. BTP with 0.75 % sodium nitrate was sug­ gested to be an optimal supplemented nitrogen source for the production of gallic acid and isoquercetin in this research, which yielded the highest contents of 61.6±2.16 mg g –1DS and 3.27±0.29 mg g –1DS , respectively. In addition, the highest extraction yields of gallic acid, isoquercetin and quercetin were obtained by an ultrasound­assisted extraction using methanol as an extraction solvent as 12.24±2.12 mg g –1DS which was around 0.5 time higher than the one without ultrasound­assisted extraction (8.84±1.12 mg g –1DS ).


Introduction
Triphala is a wellknown traditional herbal medicine in India and Thailand with formulation of Terminalia bellirica (Bahera), Terminalia chebula (Myrobalan) and Phyllanthus emblica (amla) in the ratio 1:1:1 1,2 . Its properties come from a combina tion of antioxidant agents in the formula, for exam ple, gallic acid, ellagic acid, chebulic acid, tannin, and vitamin C 1 , which have all been reported for their many medicinal properties including antimuta genic 3 , anticancer 4 , stomach ulcer relieving 5 , rejuve nating 6 , and antibacterial and antiviral 7,8 effects. Thus, Triphala has been used in products of phar maceutical, food, and cosmetics industries. Hot wa ter is usually used for Triphala extraction; however, 35 % tannin still remains in it 9 . Therefore, it was interesting to use it as a material for tannin hydroly sis by chemical 10 and enzymatic reaction 11,12 to pro duce gallic acid. Enzymatic hydrolysis by a fungal fermentation has been recommended to produce gallic acid because of high yield and low operating cost. Some studies have reported that Aspergillus oryzae, A. awamori, and A. niger produced tannase, which was more efficient in gallic acid production by solidstate fermentation (SSF) than other spe cies [13][14][15] . Even though gallic acid is the primary con stituent in the leftover Triphala, some rutin is still present in the byproduct from Triphala extraction process (BTP) because of its low water solubility 16 . It can be hydrolyzed into isoquercetin (quercetin3 O-β-dglucoside) by the ability of A. niger to pro duce α-lrhamnosidase 17,18 , which is specific for re leasing isoquercetin from rutin by derhamnosylation. A previous report 19 showed that fermentation of BTP with A. niger increased both gallic acid iso quercetin content. This was interesting because iso quercetin is rarely found in nature and is difficult to isolate. It has been reported that it has higher phar macological activity than rutin 20,21 to improve blood flow and brain function, thus benefiting Alzheimer treatment 22 . It also helps maintain proper levels of blood sugar and lipid by improving pancreatic islets functions 23 . Since the advantages of highperfor mance thinlayer chromatography (HPTLC) are less solvent requirement and ability to simultaneously run multiple samples 24 , it was used to separate bio active compounds.
The content of gallic acid, quercetin, and iso quercetin produced from SSF of BTP were deter mined in this study by HPTLC, a fast, simple, rapid, and low cost method suitable for routine analysis 25 with small amounts of mobile phase 26 . The objec tive of this study was to take advantage of Triphala byproduct by using it as a source of tannin and rutin for gallic acid, quercetin, and isoquercetin produc tion [27][28][29] .

Chemicals
Gallic acid, rutin, quercetin, and isoquercetin (AR grade) purchased from SigmaAldrich were used for analytic standards. They were prepared at concentration of 1 g mL -1 in methanol and used im mediately. The stock solutions were diluted to 0.1 mg mL -1 to use as the working standard.

Preparations of spore suspension and Triphala byproduct
A. niger ATCC 16888 was purchased from Mi croBiologics Inc., USA. It was inoculated to potato dextrose agar (PDA), and incubated for 3 days at temperature 30 °C. Its spores on the media surface were collected by adding sterilized water, and gen tly scraped to obtain a spore suspension. The con centration of suspension was determined by count ing in a Neubauer chamber. It was adjusted to 1·10 5 spores mL -1 using sterilizeddistilled water. BTP (N mass fraction = 0.32) was collected from the Institute of Thai Traditional Medicine after the hotwater extraction process to use as a solid substrate. To control the properties of BTP, BTP used in this research was collected from 10 ex traction batches. Each batch was washed and dried at 60 °C for 1 day. All batches were then mixed and ground by hammer mill to obtain particle size of 600 microns for the SSF study. Initial tannin and rutin content in BTP substrate were also analyzed by gravimetric method and HPTLC, respectively.

Solid-state fermentation (SSF)
Solid substrate was prepared for SSF by mix ing BTP with different types and concentrations of nitrogen sources, including yeast extract (N mass fraction = 0.96), sodium nitrate (NaNO 3 , N mass fraction = 0.16), and ammonium sulfate (NH 4 ) 2 SO 4 , N mass fraction = 0.21), as shown in Table 1. The portion of solid substrate determined the initial mois ture content by drying in hotair oven at 105 °C for 3 h 30 . The moisture of 30 g of BTP was then adjust ed to 55 % (wet basis) with an amount of sterilized distilled water calculated by Eq. (1), and mixed with A. niger 5·10 5 spores per g of dry substrate. The solid substrate was placed in 125mL Erlen meyer flasks. The flasks were plugged with cotton stoppers, and incubated at 30 °C for 5 days. They were then dried at 60 °C for 24 h, and milled into a powder for analysis.
( ) where, W 1 = weight of the dry substrate, W 2 = initial moisture content of substrate, W 3 = amount of sterilized distilled water needed to adjust the moisture content of substrate to 55 %.

Extraction of gallic acid, quercetin, and isoquercetin
Control sample (C) was used in the extraction study. The fermented samples were extracted by maceration with and without sonication using wa ter, ethanol, or methanol as extraction solvents. The ratio of sample and solvent was 1:10 (w/v). They were extracted for 5 days at 30 °C to obtain the crude extract solution. In the ultrasoundassisted ex traction experiments, the ultrasonication was per formed at 40 kHz for 60 min once a day. The solu tion obtained from extraction was filtered and The method was modified from Makkar et al. 31 The extracts were prepared by adding 200 mg of BTP in 10mL mixture of acetone and water (7:3) in a tube and centrifuge with Allegra X15R (Beck man Coulter, USA) at 10,000 rpm at 4 °C for 10 min. The process was repeated until 110 mL of su pernatant was collected and stabilized by keeping in ice at 0 °C for 4 h before analysis. The extracts were divided into 65mL and 45mL portions. The first portion (65 mL) was diluted with distilled wa ter by 1:1 (v/v) before adding 6.5 g of polyvinylpyr rolidone (PVP) and stirring for 15 min at 4 °C. The supernatant was collected by centrifugation at 10,000 rpm at 4 °C for 10 min. 20 mL aliquots were taken from the PVPtreated portion, and 10 mL ali quots from the untreated PVP (45 mL) portion. The aliquots were transferred into separate aluminum weighing dishes, and ovendried at 100 °C until constant weight. The difference in weight of the two portions represented tannin weight and was re ported as percent tannins on dry weight basis. This was done in triplicate for data analysis.
High-performance thin-layer chromatography (HPTLC) analysis HPTLC was used to determine gallic acid and flavonoids, including rutin, quercetin, and isoquer cetin. Four mobile phases were preliminarily tested for this study, including the mixtures of 1) formic acid: water: methyl ethyl ketone: ethyl acetate (10:10:30 32 , 2) formic acid: water: ethyl acetate (9:9:82 v/v/v) 33 , 3) formic acid: water: ethyl acetate (1:1 34 , and 4) methanol: formic acid: ethyl acetate: toluene (1:4:15 [35][36][37] . The latter showed the best separation, and was used as a mobile phase for HPTLC analysis in this research. An amount of 0.1 mg dry crude extracts sample was mixed with 1 mL methanol (selected from the highest extraction yield from the section: extraction of gallic acid, quercetin, and isoquercetin), and the suspension was sonicated for 30 min and then cen trifuged at 2,000 pm for 3 min. The supernatant was used for HPTLC analysis. 200 μm thickness HPTLC aluminum plates (Merck, Germany) coated with 60F 254 silica gel were the stationary phase (10 × 10 cm for standards, 10 × 20 cm for samples). The plates were soaked in methanol and dried at tem perature 110 °C for 15 min before chromatography. The samples were spotted in 8mm wide bands (CAMAG 100 microliter syringe with CAMAG Li nomat 5). The mobile phase was prepared by a mix ture of methanol: formic acid: ethyl acetate: toluene (1:4:15:15 v/v/v/v). A twin trough glass tank (20 ×10 cm) was presaturated with mobile phase at room temperature for 60 min before placing HPTLC sample plates to perform linear ascending. The chromatogram height was 80 mm with 20-25 min developing time. A CAMAG TLC 3 scanner with winCATS software was used for densitometric scanning at 254 and 366 nm. The scanning speed was 20 mm s -1 with 6.00 × 0.45 mm slit dimension. The retention factors (R f ) of rutin, quercetin, iso quercetin, and gallic acid were 0.08, 0.15, 0.40, and 0.55, respectively. The scanning was done in tripli cate for data analysis.
A linearity study of HPTLC analysis was per formed. Five concentration levels of individually prepared standard solution in methanol (20,40,60,80, and 100 μg mL -1 ) were used to determine the linearity. The solutions were spotted on a HPTLC plate with calibration range of 100-500 ng per spot. The calibration graphs were plotted as the peak area versus the standard concentration. The R 2 of the standards of gallic acid, isoquercetin, rutin, and quercetin standards were 0.996, 0.991, 0.994, and 0.993, respectively.

Results and discussions
Solid-state fermentation of Triphala waste by Aspergillus niger The chromatograms of constituents in the fer mented BTP were visualized under different wave length. The chromatogram of isoquercetin of BTP was difficult to detect at UV illumination at 254 nm, but it was more visible at 366 nm and derivat izing with 10 % H 2 SO 4 helped obtain a clearer band for gallic acid.
The concentration of gallic acid in the fer mented BTP calculated from the band intensities is plotted in Fig. 1, together with the contents of tan nin as a substrate for gallic acid production of A. niger. The results showed that supplementing nitro gen sources affected the fungal tannin utilization. The highest conversion of tannin to gallic acid was obtained in the SSF of BTP supplemented with 0.75 % nitrate (SN075). This was indicated by the high est reduction of tannin from the initial content of 133.8 mg g -1 DS (156.4 mg g -1 DS · (100-0.75) %) -21.0 mg g -1 DS ), and the largest gallic acid produc tion of 61.6±2.16 mg g -1 DS . The effect of supple menting yeast extract and ammonium sulfate were (2) quite similar, especially in the experiments denoted with YE050 and AM050; YE125 and AM125; YE150 and AM150, despite the fact that nitrogen in yeast extract was in organic form, and that in am monium sulfate was in inorganic form.
To obtain gallic acid from BTP, the production of fungal tannase was required to hydrolyze tannins in BTP. Nitrogen sources are essential not only for a fungal growth, but also for promoting fungal en zyme activity, depending on their species 38 because nitrogen is a precursor for nucleic acid and protein syntheses in fungal cells 39 . Nitrate assimilation pro cess is effective in which nitrogen is consumed 40 by nitrate and nitrite reductases, and then converted to ammonia, which further reacts with glutamic acid to produce amino acids and glutamine. Nitrate re ductase is also active in producing NAD+ (Nicotin amide Adenine Dinucleotide) needed for cell respi ration. Thus, it promotes the growth of A. niger and results in an increase in tannase production. Beniw al et al. 41 found that 0.2 % (w/v) sodium nitrate was suitable for tannase and gallic acid production of A. heteromorphus MTCC8818. On the other hand, 0.3 % (w/v) diammonium hydrogen phosphate, (NH 4 ) 2 HPO 4 and 1 % (w/v) potassium nitrate, KNO 3 , were suitable nitrogen sources for promot ing tannase and gallic acid production of P. atramentosum 42 and A. niger ATCC 16620 43 , respective ly. Nevertheless, the imbalance of the C/N ratio in substrate caused osmotic stress, which suppressed enzyme activities 14 .
The concentrations of isoquercetin and querce tin calculated from the band intensity of BTP after fermenting for 5 days with different nitrogen sup plement sources were plotted together with the re maining content of rutin, as shown in Fig. 2 production of isoquercetin and quercetin is related to the content of rutin in substrate due to ability of A. niger to produce naringinase, a multienzyme consisting of α-l-rhamnosidase and β-glucosidase. α-lrhamnosidase hydrolyzes rutin to isoquercetin, and β-glucosidase hydrolyzes isoquercetin to quer cetin 44 . Compared with the rutin content from BTP in the control sample (1.43±0.06 mg g -1 DS ), the highest decrease was detected in the SN075 sample (0.59±0.08 mg g -1 DS ), which was not significantly different from YE050, SN050, AM050, YE075, and SN125 samples. Of the aforementioned samples, the highest isoquercetin content (3.27±0.29 mg g -1 DS ) was found together with the lowest quercetin con tent (0.21±0.12 mg g -1 DS ) with the SN075 sample. When compared with control sample, it was 57.5 and 9.6 times higher on gallic acid, followed by iso quercetin content. This suggested that supplement ing BTP with sodium nitrate 0.75 % (SN075) en hanced only the conversion of rutin to isoquercetin, but decreased that of isoquercetin to quercetin. In other words, using SN075 condition raised α-l-rhamnosidase activity but reduced that of β-glu cosidase. On the other hand, supplementing nitro gen sources as in the conditions of SN050, AM050, YE100, AM100, YE125, and SN125 samples pro vided highest quercetin content (~ 2.03-2.37 mg g -1 DS ) with the lowest isoquercetin content (~ 0.27-0.43 mg g -1 DS ). Largely different conversions of ru tin to isoquercetin, and isoquercetin to quercetin were found with different concentrations and types of nitrogen supplement sources. Moreover, the ob tained isoquercetin contents in experiments with high nitrogen sources (AM125, YE150, SN150 and AM150) were found to be 3.5, 4, 2.4, and 8.3 times greater than in the control sample, even when the reductions in rutin level as its precursor were not significantly different. This showed that the addi tion of high content of nitrogen sources enhanced the utilization of rutin in the isoquercetin produc tion only, but adding lower nitrogen content en hanced both the consumption of rutin, and conver sion of rutin to isoquercetin. At the same time, using high nitrogen supplement levels depressed the con version of isoquercetin to quercetin, except in the AM150 sample. Obviously, the types and concen trations of nitrogen sources used affected the β-glu cosidase production as well as the observed querce tin content 45,46 .

Evaluation of maceration and ultrasoundassisted extraction with different solvents
Yields of crude extracts obtained from the mac eration and the ultrasoundmaceration extractions of the fermented BTP are shown in Fig. 3. The maxi mum extraction rates, estimated from the slope of curve, were found in the first 24 h for all extraction experiments of 0.06±0.03, 0.09±0.04, 0.12±0.04, 0.24±0.07, 0.20±0.06, and 0.32±0.06 mg g -1 DS h -1 on WM, WUM, EM, EUM, MM, and MUM, re spectively. This showed that ultrasonication in creased the extraction rates by 1.5-2 times on water and ethanol extractions, but slightly on that of methanol. Regarding the extraction solvent in the maceration method, the highest extraction yield was found at 24 h with methanol as a solvent (MM, 8.84±1.12 mg g -1 DS ), followed by that of ethanol (EM, 5.56±1.10 mg g -1 DS ), and water (WM, 2.23±0.29 mg g -1 DS ). The results clearly showed that the ultrasonication increased the extraction yield to the values of 12.24±2. 12, 9.13±2.14, and 3.97±1.21 mg g -1 DS in methanol, ethanol, and water solvents, respectively. After maceration at 120 h, it was found that MM reached 1.7 and 3.3 times higher crude ex traction compared to EM, followed by WM. For the ultrasoundmaceration, the MUM achieved 1.2 and 4.9 times higher crude extraction compared to EM, followed by WM.
Most constituents in the fermented BTP had an affinity to methanol, the polarity (relative polarity 0.762) of which was between that of ethanol (rela tive polarity 0.654) and water (relative polarity 1.0) 47 . Moreover, ultrasonication significantly in creased ethanol and methanol extraction yields (EUM, MUM), while it increased water extraction yield only slightly (WUM). The ultrasonication en hanced the extraction performance by generating cavitation bubbles in the suspension. Microjets pro duced by implosion of the cavitation bubbles im proved the penetration of solvent into pores of BTP substrate, and increased the contacting surface area between BTP and solvent [48][49][50] .
The content of target products, including quer cetin, isoquercetin, and gallic acid in the fermented BTP (without nitrogen supplement) quantified with HPTLC are shown in Fig. 4. The highest yields were obtained by methanol extraction with ultra soundassisted maceration (MUM), and amounted to 0.028±0.001 mg g -1 DS , 0.017±0.001 mg g -1 DS , and 0.033±0.002 mg g -1 DS , for quercetin, isoquercetin, and gallic acid, respectively. All those components exhibited higher solubility in methanol than in eth anol and water, both for extractions with and with out ultrasonic assistance. When compared among each other, the MM was 1.3, 0.9, 3.4 times higher than EM in quercetin, isoquercetin, and gallic acid content, respectively, and 67.8, 1.37, and 24.8 times higher than WM in quercetin, isoquercetin, and gal lic acid content, respectively. For the ultra soundmaceration, the MUM was 1.1, 0.9, and 2.1 times higher than EUM in quercetin, isoquercetin, and gallic acid content, respectively, and 45.7, 2.4, and 15.9 times higher than WUM in quercetin, iso quercetin, and gallic acid content, respectively. Even though methanol and ethanol maceration slightly differed in the quercetin and isoquercetin extraction, the superiority on gallic acid content in methanol made it a preferred solvent for the ex traction in this study. The results contrast those of the works of Chebil et al. 51 and Valentová et al. 21 , where higher solubility was achieved in water than in ethanol and methanol. Also, the finding that quer cetin was more soluble in methanol than ethanol was inconsistent with the results of Idris et al. 52 The results of this study imply that MUM technique can be applied to the BTP with supplemental nitrogen to enhance higher productivity of gallic acid and iso quercetin due to their better solubility in methanol, while sonication makes their extraction more effi cient.

Conclusions
Tannin and rutin leftovers in the byproduct of Triphala extraction process were detected as possi ble substrates for gallic acid, isoquercetin, and quer cetin production by solidstate fermentation with Aspergillus niger. However, the fungus required a supplemental nitrogen source to enhance the pro ductions. The addition of 0.75 % sodium nitrate was found appropriate for production of gallic acid and isoquercetin, while supplementing 1.00 % sodi um nitrate or 1.00 % yeast extract or 1.00 % ammo nium sulfate enhanced quercetin production. Ultra soundassisted maceration was found to be a technique that increased productivity of gallic acid and isoquercetin from the nitrogensupplemented BTP substrate.

FUnDIng STATEMEnT
The authors would like to thank for the financial support from the national Research Council of Thailand, which contributed to King Mo ngkut's Institute of Technology Ladkrabang

ConFLICT oF InTEREST
The authors declare that they have no conflict of interest.