Food Chemistry 161 (2014) 208–215

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Analytical Methods

Assessing non-digestible compounds in apple cultivars and theirpotential as modulators of obese faecal microbiota in vitro

http://dx.doi.org/10.1016/j.foodchem.2014.03.1220308-8146/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Address: FSHN Bldg. 106, Pullman, WA 99164, USA. Tel.:+1 509 335 0382; fax: +1 509 335 4815.

E-mail address: giuliana.noratto@wsu.edu (G.D. Noratto).

Luis Condezo-Hoyos, Indira P. Mohanty, Giuliana D. Noratto ⇑School of Food Science, Washington State University/University of Idaho, USA

a r t i c l e i n f o

Article history:Received 17 June 2013Received in revised form 14 February 2014Accepted 26 March 2014Available online 3 April 2014

Keywords:ApplesDietary fibreGranny SmithMicrobiotaObesityPhenolics

a b s t r a c t

The health benefits of apple bioactive compounds have been extensively reported. However, only fewstudies have focused on bioactive compounds that are not absorbed and metabolised during gastrointes-tinal digestion and can induce changes in microbial populations of faeces. We have characterised Brae-burn, Fuji, Gala, Golden Delicious, Granny Smith, McIntosh and Red Delicious cultivars and foundsignificant differences for extractable phenolics (1.08–9.2 mg/g) non-extractable proanthocyanidins(3.28–5.7 mg/g), and dietary fibre (20.6–32.2%) among cultivars with Granny Smith having the highestcontents. Granny Smith was used after in vitro digestion for fermentation of faeces from diet-inducedobese mice. Results showed that relative abundances of Firmicutes, Bacteroidetes, Enterococcus, Entero-bacteriaceae, Escherichia coli, and Bifidobacterium in apple cultured faeces tended to resemble the abun-dance in faeces from lean mice with increased trend in the production of butyric acid. These resultssuggest that apple non-digestible compounds might help to re-establish a disturbed microbiota balancein obesity.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Apples, in general, have shown to protect against humanchronic diseases due to their content of fibre and phenolic com-pounds. These bioactive compounds have low availability andcan potentially reach to colon, modulate the balance of bacterialpopulations in the gut, and influence the host physiology(Delzenne, Neyrinck, & Cani, 2011; Moco, Martin, & Rezzi, 2012).The apple health benefits are, in part, due to the interaction of fibreand phenolics with gut microbiota that results in changes in phe-nolic bioavailability and activity, and the production of short chainfatty acids (SCFAs) after fibre fermentation (Tuohy, Conterno,Gasperotti, & Viola, 2012).

Dietary fibre in apples include mainly cellulose, hemicellulose,lignin and pectin (Yan & Kerr, 2012). Previous studies havereported that apple pectin can influence the intestinal microbiota,in part, due to its bacteriostatic effects (Shinohara, Ohashi,Kawasumi, Terada, & Fujisawa, 2010). Moreover, the utilisation ofapple pectin by health beneficial faecal bacteria in vitro and theeffects of daily apple consumption on improving the intestinalenvironment due to the greater production of SCFAs and decreased

ammonia and sulfide have been reported (Shinohara et al.,2010).

Extractable phenolic compounds in apples have also beenextensively investigated. A survey of seven apple cultivars grownin Turkey, including Granny Smith, reported that content of phen-olics vary according to the apple variety and parts (peel and flesh).The identified phenolics were flavan-3-ols ((+)-catechin, (�)-epi-catechin, and procyanidins B2), dihydrochalcone (phloridzin),flavonols (quercetin glycosides (rutin + isoquercitrin) determinedas quercetin), anthocyanidin (cyanidin (except for the green peelLutz Golden and Granny Smith)), and hydroxycinnamic acids(mainly chlorogenic acid) (Karaman, Tutem, Baskan, & Apak,2013). However, most of the studies have ignored the non-extract-able phenolics in apples, which are linked to the food matrix bycovalent bonds and might resist the gastrointestinal digestion toa greater extent than extractable phenolics, and therefore canreach the colon to be metabolised by intestinal bacteria (Arranz,Silvan, & Saura-Calixto, 2010). Non-extractable phenolics in applesare procyanidins mixtures, which are bound to the apple cell wall.The binding of procyanidins to the apple cell wall increases withthe degree of polymerisation and inhibits the enzymatic degrada-tion of cell walls. This strongly suggests that non-extractable phen-olics are available in the large intestine for bacterial fermentation(Saura-Calixto, 2011).

L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215 209

In general, apples are a good source of fibre and phenolic com-pounds, and increasing their consumption has been shown toincrease the proportion beneficial commensal bacteria (Tuohyet al., 2012). However, the content of these compounds can be sig-nificantly influenced by several factors including cultivar, soil,environment, harvest, storage conditions and processing as dem-onstrated by studies screening the phenolics in apple cultivars pro-duced in Turkey (Karaman et al., 2013) and Poland (Łata,Trampczynska, & Paczesna, 2009).

Our aim was to characterise seven apple cultivars produced inthe Pacific Northwest U.S. regarding their content of non-digestiblecompounds and to assess how these compounds can modulatebacterial populations in faeces from obese mice as a potentialunderlying mechanism to help re-establish a healthy balance andprotect from obesity-related disorders.

2. Materials and methods

2.1. Chemicals

Acetone, glycerol, hydrochloric acid, methanol, sodium potas-sium tartrate and Tris base were purchased from Fisher (NJ,USA). Acetic acid, propanol and sodium hydroxide were acquiredfrom J.T. Baker (Center Valley, PA, USA). Bacterial growth mediaand supplements were obtained from BD (Franklin Lakes, NJ,USA). Primers for quantitative real time PCR were acquired fromIntegrated DNA Technology (San Diego, CA, USA). The remainingchemicals and reagents were obtained from Sigma–Aldrich Co.Ltd. (Saint Louis, MO, USA).

2.2. Plant material

Apple commercial cultivars Braeburn, Fuji, Gala, Golden Deli-cious, Granny Smith, McIntosh and Red Delicious produced in theState of Washington were purchased from a local grocery marketin Pullman, WA. Apples were maintained at 4 �C to preserve chem-ical composition and maturity stage. Each sample was preparedout of three or more fruits, after removing the core, to study onlythe edible parts (flesh and peel).

2.3. Proximal analysis

Apple samples were ground with dry ice, and kept at �20 �C formoisture, and ash analysis following the AOAC standard methods(AOAC, 2006). Protein was quantified using a Perkin-Elmer 2400Series II CHNS/O elemental analyzer (Perkin-Elmer, Waltham,MA, USA). A factor of 6.25 was used to convert the nitrogen contentto crude protein content.

2.4. Colour

Colour of apple flesh was quantified by image analysis as previ-ously reported (Medina, Skurtys, & Aguilera, 2010) with somemodifications. Briefly, images were taken with a Nikon Inc COOL-PIX L810 digital camera 16.1 Megapixels (Melville, NY, USA) fromtop view under constant distance (40 cm height) and uniform illu-mination of 2 bulbs (150 W each). The red (R), green (G) and blue(B) signals were obtained with the ImageJ program (http://rsb.in-fo.nih.gov/ij/) using colour histogram plugins from 5 differentregions of the picture. The colour of the apple samples were repre-sented in the hue (H) – saturation (S) – lightness (L) model. H, Sand L values were estimated as previously reported (Medinaet al., 2010) using the following equation: H = Arctan (

p3/

2(G � B))/(R � 0.5(G + B)), L = (R + G + B)/3, S = 1 � lowest value ofany ratio: R/L, G/L, or B/L. The RGB values were converted to

CIE-Lab space using EasyRGB calculator (http://www.easy-rgb.com/index.php?X=CALC#Result), to obtain L⁄, a⁄, and b⁄ values.

2.5. Soluble solids and titratable acidity

Apple samples were pressed to obtain 10 lL of juice, which wasanalysed with a hand refractometer (Extech model 2132, Japan) at20 �C to determine percentage of soluble solids (�Brix).

For titratable acidity, apple samples were homogenised with anUltra-Turrax model TR-10 (Tekman, OH, USA) (5 g/100 mL ultra-pure water), filtered through a Watman # 1 (Whatman Interna-tional Ltd, Maidstone, ME, UK) and analysed with 0.1 mol/LNaOH using phenolphthalein as an indicator. The results wereexpressed as g of malic acid equivalent/100 g sample wet weight(AOAC, 2006).

2.6. Total extractable phenolics

Apple samples (flesh and peel) were homogenised in an Ultra-Turrax with acidic methanol (HCl)/water solution (50:50 v/v, pH2) (200 g/L) and left for 1 h at room temperature with constantshaking, followed by centrifugation at 2100g for 10 min at 4 �C toobtain an acid methanolic extract in the supernatant. The precipi-tate was extracted with acetone/water solution (70:30 v/v) (1:5ratio, v/v) by agitation for 60 min, followed by centrifugation at2100g for 10 min at 4 �C. The combined supernatants were ana-lysed for total phenolic content using the Folin Ciocalteu micro-method (Abderrahim et al., 2012).

2.7. Non-extractable proanthocyanidins

Non-extractable polymeric proanthocyanidins were analysed aspreviously reported (Zurita, Diaz-Rubio, & Saura-Calixto, 2012)with some modifications. Briefly, 20 g of residue obtained aftersuccessive acidic methanolic and acetone extraction (as detailedin Section 2.6) was hydrolysed with 100 mL of HCl/butanol (5:95v/v) containing 0.7 g of FeCl3/L and incubated at 90 �C for 1 h. Thenthe sample was centrifuged at 15,000g for 3 min and the superna-tant absorbance read at 555 nm and 450 nm using a lQuant micro-plate reader (Biotek Instrument, Rochester, VT, USA).Proanthocyanidin concentrations were taken from the sum ofabsorbance at 550 and 450 nm and plotted against a standardcurve of polymeric proanthocyanidins (range 0–100 mg/L) andexpressed as mg proanthocyanidins/g apple dry weight.

2.8. Dietary fibre and available carbohydrates

2.8.1. Dietary fibreApple samples were analysed by the enzymatic–gravimetric

method as previously described (McCleary et al., 2012). Briefly,apple samples were blanched at 90 �C for 5 min, homogenised ina blender and freeze dried. The freeze dried samples were homog-enised (1 g/20 mL) with maleate buffer (50 mmol/L, pH 6.0 plus2 mmol/L CaCl2) containing 50 U/mL of porcine pancreatic a-amy-lase and 3.4 U/mL of amyloglucosidase. This mixture was incu-bated at 37 �C for 16 h under agitation at 150 rpm. Thereafter,0.75 mL Tris base (0.75 mol/L) was added to adjust pH to 8.2 andenzymes inactivated at 90 �C for 20 min. One millilitre of protease(8.8 U/mL) was added to the mixture and incubated at 60 �C for30 min under agitation. Thereafter, acetic acid (2 mol/L) was addedto the mixture to adjust pH to 4.5. Dietary fibre was precipitated byadding four volumes of absolute ethanol and 1 h incubation atroom temperature followed by centrifugation at 2000g for10 min. The pellet was filtered, dried at 105 �C and weighed toquantify dietary fibre. The supernatant was collected forquantification of available carbohydrates.

210 L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215

2.8.2. Available carbohydratesSupernatants from Section 2.8.1 were assessed for available car-

bohydrates by quantification of reducing sugars by the 3,5-dinitro-salicylic acid (DNS) assay (King, Donnelly, Bergstrom, Walker, &Gibson, 2009). Briefly, glucose standards 0.5, 1.0, 1.5 and 2.0 mg/mL were prepared in 0.1 mol/L sodium acetate buffer pH 5.5. A30 lL volume of the standards or diluted samples were added to60 lL 1% DNS reagent and the mixture was heated at 95 �C for5 min before being cooled to 20 �C in a T100 Thermal cycler (Bio-Rad, CA, USA). The reaction volume was transferred to 96-wellplate and absorbance measured at 540 nm in a lQuant microplatereader (Biotek Instrument, Rochester, VT, USA). The reducing sug-ars were expressed as g glucose equivalent per 100 g of freezedried sample.

2.9. Enzymatic browning

The enzymatic browning was assessed according to the proce-dure described by Quintero Ruiz, Demarchi, Massolo, Rodoni, andGiner (2012) with some modifications. Briefly, a composite of 4or more blanched apple samples were mashed with a mortar,and pictures were taken with a Nikon Inc COOLPIX L810 digitalcamera 16.1 Megapixels (Melville, NY, USA) under controlled illu-mination, every 5 min for 55 min. The red (R), green (G) and blue(B) signals were obtained from blanched samples and from non-blanched controls (t = 0 min). The colour histogram plugin fromImageJ program (http://rsb.info.nih.gov/ij/) was used to calculatethe browning index (BI) as follows: BI = R/(R + G + B). The kineticand order of the enzymatic reaction were calculated by powerlaw model (Quevedo, Jaramillo, Díaz, Pedreschi, & Aguilera, 2009).

2.10. Optimisation of blanching

Apple slices were immersed in distilled water (1/10, w/v) at80 �C for 10, 15 and 20 min or 90 �C for 5, 10 and 15 min. The bestcombination time–temperature was selected as a function of avail-able carbohydrates (as detailed in Section 2.8.2), extractable phen-olics (as detailed in Section 2.6) and enzymatic browning (asdetailed in Section 2.9).

2.11. In vitro digestion (IVD)

Apple samples (flesh and peels) were blanched, blended, freezedried and successively incubated with digestive enzymes to simu-late salivary, gastric and duodenal digestion at 37 �C in a waterbath under agitation (200 strokes/min agitation using a shakingwater bath (VWR, Radnor, PA, USA) (Hollebeeck, Borlon,Schneider, Larondelle, & Rogez, 2013). Briefly, 10 g of dried appleswere homogenised with 20 mL of Tris-maleate buffer (0.1 mol/L,pH 6.9) and salivary digestion mimicked with 3.9 U of amylase/mL for 5 min. After centrifugation at 3200g for 10 min at roomtemperature, the gastric step was mimicked by mixing the precip-itate with 20 mL of KCl–HCl buffer (0.2 mol/L, pH 1.5) containing71.2 U of pepsin/mL for 90 min followed by centrifugation at3200g for 10 min at room temperature. The precipitate was recov-ered and mixed with 20 mL of phosphate buffer (0.1 mol/L, pH 7.5)containing 9.2 mg/mL of pancreatin and 55.2 mg/mL of bile extractfor 150 min to mimic the duodenal step followed by centrifugationat 3200g for 10 min at room temperature to obtain a pellet of applenon-digestible compounds. The pellet was freeze dried and storedat �20 �C for faecal in vitro fermentation and analysis of extract-able phenolics and non-extractable proanthocyanidins as detailedin Sections 2.6 and 2.7.

2.12. In vitro faecal fermentation

Faecal samples were collected from one year old diet-inducedobese mice (n = 3), body mass index (BMI) = 5.7 ± 0.7, and agematched lean mice (n = 3), BMI = 3.6 ± 0.4. Obese mice were fed ahigh fat diet #D12451 containing 24% protein, 41% carbohydrate,24% fat, 5.8% cellulose and 5.2% minerals (45% kcal from fat)(Research diets, Inc., New Brunswick, NJ). Lean mice were fed astandard diet #2018 containing 18% protein, 44.2% carbohydrate,6.2% fat, 20% fibre and 5.3% ash (18% kcal from fat) (Teklad Diets,Madison WI). Animals were housed in a temperature-controlledroom with a 12 h light and 12 h darkness cycle and in compliancewith the WSU Institutional Animal Care and Use Committee. Fae-ces from obese mice were cultured in batches from each animaldonor per apple treatment (1% and 2%, w/v), i.e. no pooled faecalsamples. For obese and lean controls, a composite of faecesobtained from the obese (n = 3) or lean mice (n = 3) were preparedand fermentations were carried out in duplicates. For the faecalculture experiments, an anaerobic atmosphere was generated byGasPak EZ™ Gas Systems (BD Diagnostic Laboratory, NJ, USA). Apre-culture was prepared with fresh faeces diluted with anaerobicphosphate buffer 1:10 (w/v) (1 mol/L, pH 7.2), homogenised toproduce faecal slurries and used for inoculation of pre-culture(10%, v/v) in a pre-reduced sterile medium at pH 7.0 (peptone(2 g/L), yeast extract (2 g/L), NaCl (0.1 g/L), K2HPO4 (0.04 g/L), KH2-

PO4 (0.04 g/L), NaHCO3 (2 g/L), MgSO4�7H2O (0.01 g/L), CaCl2�6H2O(0.01 g/L), Tween 80 (2 mL/L), hemin (50 mg/L), vitamin K (10 lL/L), L-cysteine (0.5 g/L), bile salts (0.5 g/L), resazurin (1 mg/L), anddistilled water) (Sanchez-Patan et al., 2012) and anaerobic incuba-tion for 12 h at 37 �C under agitation (200 rpm). Cultures were pre-pared with 5% (v/v) of pre-culture in modified pre-reducedmedium containing the dried apples after IVD at 1% and 2% (w/v)or starch at 2% (w/v) for the obese and lean controls. After 24 hincubation at 37 �C and 200 rpm, culture fermentation was leftwithout agitation for 10 min to separate apple compounds foranalysis of phenolics as detailed in Sections 2.6 and 2.7; then theculture was centrifuged at 15,000g for 3 min at 4 �C to recover pel-lets containing faecal bacteria and supernatants containing SCFAs.Pellets and supernatants were stored at �80 �C for further analysis.

2.13. Analysis of bacterial populations

Relative abundances of bacterial populations were analysed byquantitative real-time PCR (qPCR). Briefly, DNA was extracted frompellets recovered after faecal fermentation using QIAamp DNAStool Mini Kit (QIAGEN Inc, CA, USA) according to the manufac-turer’s protocol. The concentrations of DNA and absorbance ratio260/280 nm as measure of purity were assessed using a NanoDrop2000 spectrophotometer (Thermo Scientific, NC, USA). PCR reac-tion mixtures (total of 10 lL) contained 5 lL of SsoAdvanced™SYBR� Green Supermix (Bio-Rad, Hercules, CA, USA), 0.4 lL of pri-mer mix (forward and reverse, final concentration 200 nmol/L),2 lL of adjusted (2 ng/lL) template DNA or non-template control,and 2.6 lL nuclease-free water (Qiagen GmbH, Hilden, Germany).PCR conditions were as follows: 95 �C for 30 s and 40 cycles at95 �C for 5 s and 55 �C for 30 s. A melt curve analysis was per-formed to verify the specificity of the primers using the followingconditions: 65–95 �C at 0.5 �C increments. Relative bacterial popu-lation was quantified by the Livak Method (2�DDCT) using, as a ref-erence, the universal primer CT values. Data was normalised toobese controls fermented with starch (2% w/v) as a carbon source.Primer sequences were obtained from Bergstrom et al. (2012). Thepairs of forward and reverse primers were purchased from Inte-grated DNA Technologies, Inc. (San Diego, CA, USA) and sequenceswere:

L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215 211

Universal F: 50-ACTCCTACGGGAGGCAGCAGT-30, Universal R: 50-GTATTACCGCGGCTGCTGGCAC-30; Firmicutes F: 50-TGAAACTYA-AAGGAATTGACG-30, Firmicutes R: 50-ACCATGCACCACCTGTC-30;Enterococcus spp. F: 50-CCCTTATTGTTAGTTGCCATCATT-30, Entero-coccus spp. R: 50-ACTCGTTGTACTTCCCATTGT-30; Bacteroidetes F:50-GGARCATGTGGTTTAATTCGATGAT-30, Bacteroidetes R: 50-AGCTGACGACAACCATGCAG-30; Enterobacteriaceae F: 50-CATTGACGTTACCCGCAGAAGAAGC-30, Enterobacteriaceae R: 50-CTCTACGAGACTCAAGCTTGC-30; Bifidobacterium spp. F: 50-GCGTGCTTAACACATGCAAGTC-30, Bifidobacterium spp. R: 50-CACCCGTTTCCAGGAGCTATT-30; E. coli F: 50-CATGCCGCGTGTATGAAGAA-30, E. coli R: 50-CGGGTAACGTCAATGAGCAAA-30.

2.14. Short-chain fatty acid analysis

Propionic, butyric, and acetic acids (SCFAs) were quantified insupernatants from faecal fermentation as previously reported(Campos et al., 2012). Briefly, supernatants filtered through a0.22 lm PVDF filter (Thermo Scientific, TN, USA) were analysedby HPLC using the Aminex HPX-87H column (8% cross-linked resinhydrogen ionic form, particle size 9 lm, pH range 1–3, size300 � 7.8 mm i.d.) and 30 � 4.6 mm, hydrogen form, pH range1–3 guard column (Bio-Rad, Hercules, CA, USA) in a Waters 600SSeparation Module equipped with an autoinjector 717 Plus, a pho-todiode array detector (999 PAD) (Waters, Milford, MA, USA) and aMillennium software for HPLC data analysis. A sample of 20 lL waseluted with H2SO4 (0.016 mol/L) at 0.6 mL/min and 35 �C. SCFAswere identified and quantified by comparing retention times,UV–visible spectral data, and pick areas to known standards.

2.15. Statistical analysis

Quantitative data represent mean values with the respectivestandard deviation (SD) or standard error of the mean (SE) of threeor more replicates. One-way analysis of variance (ANOVA) fol-lowed by a Bonferroni post-hoc test were performed with Graph-Pad Prism (GraphPad Software, Inc., San Diego, CA). A value ofp < 0.05 was considered statistically significant. Principal compo-nent analysis was assessed using SPSS version 15.0 (SPSS Inc., Chi-cago, IL).

3. Results and discussion

3.1. Physicochemical characterisation of apples cultivars

Results from physicochemical characteristics of the seven applecultivars used for our study are presented in Table 1. Moisture val-ues ranged from 84.7 ± 1.3% to 89.1 ± 0.1% for Golden Delicious andGala cultivars, respectively. The moisture of other cultivars wasbetween these values. In general, these values are in agreementwith those previously reported for cultivars grown in Brazilincluded Fuji Suprema (80.45–84.8%) (Vieira et al., 2009) and

Table 1Physicochemical characteristic of apple cultivars. Values are mean of three or more sample

Apple cultivars Moisture (%) Titratable acidity(g malic acid/100 g wet basis)

Braeburn 86.6 ± 0.4a 0.18 ± 0.01aFuji 85.4 ± 0.3a 0.16 ± 0.01aGala 89.1 ± 0.1b 0.59 ± 0.04bGolden Delicious 84.7 ± 1.3a 0.18 ± 0.01aGranny Smith 86.9 ± 0.2a 0.25 ± 0.01cMcintosh 86.8 ± 0.2a 0.34 ± 0.03cRed delicious 84.9 ± 0.3c 0.30 ± 0.02c

88.1% for Granny Smith produced in Spain (Picouet, Landl,Abadias, Castellari, & Viñas, 2009).

The values of titratable acidity ranged from 0.16 to 0.59 g malicacid/100 g of apple wet weight for Fuji and Gala, respectively. Thetitratable acidity of other cultivars was between these values andare comparable to those previously reported for Golden Deliciousand Gala apples (Hoehn, Gasser, Guggenbühl, & Künsch, 2003) andfor Granny Smith and Fuji cultivars (Drogoudi, Michailidis, &Pantelidis, 2008). Results for soluble solids ranged from 9.1 to14.5 �Brix for Gala and Red Delicious, respectively. These values werecomparable to those previously reported values for Golden Deliciousand Granny Smith cultivars (Drogoudi et al., 2008; Hoehn et al.,2003; Vieira et al., 2009). The content of protein in fruits is, in gen-eral, low and apple has around 0.26% dry weight. Results from pro-tein analysis ranged from 0.53 ± 0.23% to 1.85 ± 0.01% (dry basis) forRed Delicious and McIntosh, respectively.

Results from colour analysis are presented in Fig. 1 and in Sup-plementary material Table 1. Values of H were from 1.9 ± 0.6 to21.4 ± 2.5 for the red apple cultivars (Braeburn, Fuji, Gala, McIntoshand Red Delicious), and from 47.7 ± 0.9 to 60.7 ± 0.6 for GoldenDelicious and Granny smith, respectively. The L and S of Fuji weresubstantially different from the other red cultivars (Fig. 1A and B).When RGB was converted to CIE-Lab space, the values obtained forGolden Delicious (L⁄ = 55, a⁄ = �18 and b⁄ = 30) and Granny Smith(L⁄ = 62, a⁄ = �12 and b⁄ = 28) were comparable to those previouslyreported for the same cultivars (Drogoudi et al., 2008).

3.2. Bioactive compounds in apple cultivars

Previous studies have reported that apples are a good source ofdietary fibre, and phenolic compounds are variable among culti-vars (Karaman et al., 2013; Tsao, Yang, Christopher, Zhu, & Zhu,2003). To our knowledge, this is the first study assessing the vari-ability in the content of bioactive compounds in apple cultivarsproduced in the U.S. Pacific Northwest. Our results showed that,among the seven apple cultivars analysed, Granny Smith had thehighest content of dietary fibre and phenolics (extractable andnon-extractable) (Table 2). Red Delicious and Granny Smith repre-sented the lowest and highest extremes regarding the contents ofextractable phenolics (1.08 ± 0.04 to 9.2 ± 1.8 mg GAE/g dryweight, respectively). Results were consistent with values of phen-olics reported for Fuji, Gala, Golden Delicious, Granny Smith, andMcIntosh (Łata et al., 2009). However, in another study of applesgrown in Ontario, Canada, the content of phenolics in peel andflesh in Red delicious was higher than in Golden delicious andMcIntosh (Tsao et al., 2003). Discrepancies with our findings mightbe due the influence of maturity stage and environmentalconditions.

The content of non-extractable phenolics ranged from3.28 ± 0.02 to 5.70 ± 0.9 mg proanthocyanidins/g dry weights forGolden Delicious and Granny Smith, respectively. The other culti-vars were between these values. Likewise, content of fibre rangedfrom 20.6% to 32.2% dry weight for Fuji and Granny Smith,

s ± SD, different letters within same column indicate statistical significance (p < 0.05).

Soluble solids (�Brix) Protein (% dry basis) Ash (% dry basis)

11.6 ± 0.3a 1.70 ± 0.46a 1.2 ± 0.1a12.0 ± 0.0a 1.28 ± 0.02a 1.7 ± 0.4a

9.1 ± 0.1b 1.79 ± 0.03a 2.5 ± 1.5a10.1 ± 0.2b 1.52 ± 0.47a 2.1 ± 0.5a13.1 ± 0.1c 1.31 ± 0.23a 3.1 ± 0.5a14.3 ± 0.6d 1.85 ± 0.01a 2.0 ± 0.3a14.5 ± 0.1d 0.53 ± 0.23b 2.9 ± 0.7a

Fig. 1. Colour parameters lightness, saturation (%) and hue (�) of apple cultivars. Values are mean ± SD (n = 3).

Table 2Bioactive compounds and available carbohydrates in apple cultivars. Values are mean of three or more samples ± SD, different letters within same column indicate statisticalsignificance (p < 0.05).

Apple cultivars Total extractable phenolics(mg GAE/g of dry weight)

Non-extractable proanthocyanidins(mg proanthocyanidins/g dry weight)

Dietary fibre(% dry weight)

Availability carbohydrate(g glucose/100 g lyophilised)

Braeburn 5.91 ± 0.13a 5.23 ± 0.96a 24.9 ± 1.4a 7.3 ± 0.7aFuji 5.78 ± 0.83a 5.30 ± 1.19a 20.6 ± 2.4a 7.1 ± 0.7aGala 8.04 ± 0.27a 4.47 ± 1.30a 26.7 ± 1.4a 8.1 ± 1.1aGolden Delicious 3.93 ± 0.24a 3.28 ± 0.02a 23.1 ± 2.1a 7.3 ± 0.6aGranny Smith 9.18 ± 1.79b 5.70 ± 0.90a 32.2 ± 1.3b 7.3 ± 2.3aMcintosh 7.23 ± 0.39a 5.65 ± 0.87a 21.5 ± 2.1a 7.5 ± 1.3aRed delicious 1.08 ± 0.04c 4.15 ± 0.25a 24.3 ± 3.3a 6.6 ± 0.9a

212 L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215

respectively, with other cultivars between these values. Fruits andvegetables supply a major proportion of the non-extractable poly-phenols to the human diet (Arranz et al., 2010; Saura-Calixto,2011). Results showed that non-extractable proanthocyanidinsranged from 3.28 ± 0.02 to 5.7 ± 0.9 mg of proanthocyanidins/gdry weight for Golden Delicious and Granny Smith, respectively(Table 2). These values were comparable to those previouslyreported for Golden Delicious (3.2 ± 0.07 mg of proanthocyani-dins/g dry weight) (Arranz et al., 2010; Zurita et al., 2012). In gen-eral, non-extractable proanthocyanidins were similar among theapple cultivars and variability in the ratio of extractable phenolicsto non-extractable proanthocyanidins were determined by thecontents of extractable phenolics.

Dietary fibre has been consistently found to be inversely associ-ated with chronic human diseases such as cancer, obesity and

obesity-related diseases (Saura-Calixto, 2011). Recently, dietaryprebiotics have been defined as an ingredient that is selectivelyfermented, resulting in specific changes in composition and/oractivity of gastrointestinal microbiota, thus conferring benefits tothe host health (Tuohy et al., 2012). Results of dietary fibre con-tents ranged from 20.6 ± 2.4% to 32.2 ± 1.3% in dry weight, withGranny Smith being the highest, whereas not statistical differencewas found in the contents of available carbohydrates (Table 2). Pre-vious studies have reported values of 12.4% and 48% in dry weightfor dietary fibre of freeze dried apple flesh and pomace, respec-tively (Yan & Kerr, 2012), and effects of apple pectin on improvingthe faecal environment have been reported (Shinohara et al., 2010).

Finally, a principal component analysis of extractable phenolic,non-extractable proanthocyanidins, dietary fibre, available carbo-hydrate, and titratable acidity allowed us to position apple

Fig. 2. Principal component analysis of bioactive compounds and available carbo-hydrates in apple cultivars. Values of dietary fibre (DF), available carbohydrate(CHO), extractable phenolics (EP), non-extractable proanthocyanidins (NE-P) andtitratable acidity (TA) were analysed by SPSS using the dimension reduction factoranalysis.

L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215 213

cultivars based on their similarity and identify those withenhanced content of dietary fibre and low content of available car-bohydrates (Fig. 2). In general, we identified Granny Smith as a cul-tivar with enhanced content of bioactive compounds (dietary fibre,extractable phenolics, and low in available carbohydrates) and

Fig. 3. Granny Smith non-digestible bioactive compounds change relative abundances offaecal slurries from each animal donor (no pooled faecal samples) in pre-reduced modifieand lean controls for 24 h under anaerobiosis and constant agitation as outlined in Sectiobacterial populations by quantitative real-time PCR (qPCR). Values are mean of DNA extra(p < 0.05).

based on these results, Granny Smith was selected for futurestudies.

3.3. Optimisation of processing to stabilise bioactive compounds in theselected apple cultivar

Blanching optimisation aimed to protect apples from enzymaticbrowning and oxidation of phenolic compounds while decreasingthe amount of reducing sugars. Granny Smith apples wereblanched at 80 �C for 10–20 min or 90 �C for 5–15 min and resultsshowed that inactivation of enzymatic browning was similar for allblanching conditions tested with not statistical difference onkinetic of enzymatic browning (0.0011 6 K 6 0.0139), while K forthe non-blanched control was 0.15 ± 0.02 with a reaction order(n) of 0.3 ± 0.1 (Supplementary material, Table 2). Simultaneously,the blanching conditions required to significantly decrease reduc-ing sugars (approximately 40% of control) were 80 �C for 20 minor 90 �C for 5 min (Supplementary material, Fig. 1) while none ofthe blanching temperature–time combinations decreased extract-able phenolics (data not shown). Based on these results, blanchingat 90 �C for 5 min was used for further experiments due to lowestcontent of available carbohydrates.

3.4. Modulation of faecal bacteria and production of SCFAs by non-digestible compounds in Granny Smith

The beneficial effects of apple consumption and apple pectin inpromoting the growth of Bifidobacterium and Lactobacillus and

bacteria in faeces from obese mice. In vitro faecal fermentation was prepared usingd medium containing apple after in vitro digestion (1–2%, w/v) or 2% starch for obesen 2. DNA was extracted from faecal material and analysed for relative abundances ofcted and analysed by duplicate ± SE, different letters indicate statistical significance

Fig. 4. Principal component analysis. Relative DNA abundances of bacterialpopulations in faeces from obese mice were positioned closer to the lean controlswhen fermented with apple compounds. Data was analysed by SPSS using thedimension reduction.

214 L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215

improving intestinal and faecal environment has been reported(Shinohara et al., 2010). However, to our knowledge, this is the firststudy assessing the effects of apple non-digestible compounds con-taining dietary fibre and phenolic compounds on faecal bacterialchanges in faeces from obese mice. Extractable phenolics werereduced in apple after IVD to �48% of initial content (from6.3 ± 0.3 to 3.05 ± 0.23 mg GAE/g dry weight). Similar losses inextractable phenolics have been previously reported for GoldenDelicious and Mutzu apples (from 36.3–51.4 mg/100 g to 18.0–27.4 mg/100 g) (Bouayed, Deußer, Hoffmann, & Bohn, 2012).Proanthocyanidins, instead, resisted the IVD almost completely(Supplementary material, Fig. 2), due to their covalent link to die-tary fibre, which makes possible for them to reach the colon and tobe metabolised by faecal bacteria (Arranz et al., 2010).

Altered proportions of the two dominant bacteria phylum Bac-teroidetes and Firmicutes in the human gut have been reported inobese state (Lyra, Lahtinen, Tiihonen, & Ouwehand, 2010). Ourresults, although not statistically significant (p = 0.06), showed atrend to increase the relative DNA abundance of Firmicutes in leancontrols compared to obese controls (from 1.00 ± 0.03-fold to1.91 ± 0.11-fold), and this trend increased in faeces from obesemice cultured with 1–2% apple compounds (to 2.3 ± 0.6 and2.6 ± 0.2 of obese controls) (Fig. 3A). The relative abundance of Bac-teroidetes DNA, instead, decreased significantly (p < 0.05) in leancontrols to 0.35 ± 0.01-fold of obese controls. When faeces fromobese mice were cultured with 1–2% apple compounds, Bacteroi-detes DNA abundance also decreased significantly (p < 0.05), to0.22 ± 0.04- and 0.23 ± 0.1-fold of obese controls, respectively(Fig. 3A). Previous studies have reported that relative abundanceof the two phyla Firmicutes and Bacteroidetes differ among leanand obese mice, with obese mice having a higher proportion of Fir-micutes to Bacteroidetes and similar results were observed inobese humans compared to lean subjects (Kallus & Brandt, 2012).However, our results showed an inverse trend between the propor-tions of Firmicutes and Bacteroidetes in faeces from obese and leanmice and apple compounds, which tended to increase Firmicutes(p = 0.06), and decrease Bacteroidetes significantly (p < 0.05). Thiswas consistent with a previous study showing that consumptionof apple pectin increased the population of Firmicutes anddecreased Bacteroidetes in the rat gut (Licht et al., 2010). The dis-crepancies regarding the association of bacterial groups with obes-ity might be due to the complexity of the microbial ecosystem withhigh subject specificity (Lyra et al., 2010).

Similarly, relative DNA levels of Enterococcus, which belong tothe phylum of Firmicutes were assessed and results showed thatEnterococcus DNA in lean controls tended to be higher (p = 0.07),compared to the obese controls (1.84 ± 0.84-fold). EnterococcusDNA levels also tended to increase in faeces from obese mice cul-tured with 1–2% apple compounds (3.55 ± 0.37- and 1.84 ± 0.84-fold of obese control, respectively) (Fig. 3B).

We also found that relative DNA proportions of Enterobacteria-ceae, which belong to the phylum of Proteobacteria, were higher infaeces from lean controls (5.82 ± 0.16-fold of obese controls). Sim-ilarly, Enterobacteriaceae DNA levels tended to increase in faecesfrom obese mice when cultured with 1–2% apple compounds (to2.4 ± 0.65- and 2.29 ± 0.36-fold of obese control, respectively)(p = 0.1025) (Fig. 3C). Escherichia coli belongs to the family of Enter-obacteriaceae and, accordingly, the relative DNA abundances ofE. coli were consistent with the trend found for Enterobacteriaceae,with lean controls having 2.15 ± 0.95-fold of obese controls, and anincreasing trend for rising levels in faeces from obese mice whencultured with 1–2% apple compounds (4.04 ± 0.94 and2.77 ± 0.65-fold of obese controls, respectively) (p = 0.1025)(Fig. 3C). Most of E. coli strains are harmless and part of the normalflora of the gut and can benefit the host by preventing pathogenicbacteria from establishing within the intestine. A recent study has

reported a negative correlation between E. coli and BMI (Millionet al., 2013).

Finally, relative abundances of Bifidobacterium spp., whichbelong to the Actinobacteria phylum tended to increase in faecesfrom lean controls (2.22 ± 0.99-fold of obese control) (p = 0.58);this trend was also observed when faeces from obese mice werecultured with 1–2% apple compounds (1.84 ± 0.56- and1.6 ± 0.33-fold of obese controls) (Fig. 3D). Increases in the num-bers of Bifidobacterium and Enterococcus have been found in caecalcontents and faeces of mice fed with high fat diet containing fer-mentable carbohydrates, which prevented body weight gain(Tuohy et al., 2012). Moreover, Bifidobacterium is a genus knownas probiotic beneficial bacteria inversely associated with BMI(Million et al., 2013).

Shinohara et al. (2010), demonstrated that Bifidobacterium andEnterococcus increased in faeces cultured with apple pectin andin faeces from healthy adults after apple consumption. Our resultsare consistent with this study, which attributed major changes ingut microbiota to pectin. However, we found that phenolics arealso being metabolised by faecal bacteria since the concentrationsof extractable phenolics and non-extractable proanthocyanidinsdecreased after faecal fermentation by 74% and 38%, respectively.This is in agreement with studies reporting the effects of phenoliccompounds on the intestinal environment due to modulation ofthe intestinal bacterial populations and implications in the mainte-nance of gastrointestinal health (Moco et al., 2012). Overall, thenon-extractable proanthocyanidins resisted the IVD and were fur-ther metabolised by the faecal bacteria (Supplementary material,Fig. 2). Therefore, they might have contributed to changes in faecalbacterial populations. This is consistent with studies reporting thatpart of the health benefits of non-extractable phenolics are relatedto their effects on modulation of intestinal bacteria population(Saura-Calixto, 2011). Finally, principal component analysis of rel-ative microbiota populations showed that faeces cultured withGranny Smith non-digestible compounds shifted the profile of bac-terial DNA abundances of obese faecal material towards the profileof faeces from lean mice (Fig. 4).

L. Condezo-Hoyos et al. / Food Chemistry 161 (2014) 208–215 215

Results from SCFAs analysis showed that butyric acid in culturesupernatants were 90.15 ± 54 lmol/g starch in obese + starch con-trol, 141.8 ± 78 lmol/g, and 154.28 ± 77 lmol/g apple in obese + di-gested apple (1% and 2%, respectively). Butyric acid in faeces fromobese mice cultured with 1–2% apple compounds was 1.57 ± 0.86-and 1.71 ± 0.85-fold of starch control, respectively (Supplementarymaterial, Fig. 3). These values show that apple compounds promotethe production of butyric acid, even though the increase was notstatistically significant due to the high variability (p = 0.53). TheSCFAs, particularly butyric acid, are beneficial for the host coloniccells because these nutrients help increase colon barrier integrityand protect against pathogens and blood endotoxemia, the presenceof endotoxins in the blood (Cani & Delzenne, 2011). Levels of aceticacid instead, were lower compared to butyric acid levels (from 2.3 to11 lmol/g), and apple compounds decreased acetic acid concentra-tions (Supplementary material, Fig. 3). Levels of propionic acid werebelow the limits of detection.

4. Conclusions

Granny Smith apples were identified as a cultivar withenhanced content of bioactive compounds including dietary fibre,extractable and non-extractable phenolics; these compounds canpotentially reach the colon. After faecal fermentation of GrannySmith non-digestible compounds, we demonstrated that relativeabundances of bacterial populations in faeces from obese micetended to be similar to the lean controls. These results suggest thatdietary fibre and phenolic compounds remaining in apple after IVDmight help to prevent metabolic disorders driven by an alteredmicrobiota in obesity, and potentially protect from an obesity-disturbed balance of microbiota.

Acknowledgments

This study was funded with an Emerging Research Issues Inter-nal Competitive Grant from Agricultural Research Center at Wash-ington State University, College of Agricultural, Human, andNatural Resource Sciences, Grant No 10A-3057-3668.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2014.03.122.

References

Abderrahim, F., Huanatico, E., Repo-Carrasco-Valencia, R., Arribas, S. M., Gonzalez,M. C., & Condezo-Hoyos, L. (2012). Effect of germination on total phenoliccompounds, total antioxidant capacity, Maillard reaction products andoxidative stress markers in canihua (Chenopodium pallidicaule). Journal ofCereal Science, 56(2), 410–417.

AOAC (2006). Official analytical chemist. In Official methods of analysis, pp. 2000).Gaithersburg. Md.: AOAC Intl.. 2012/04/06 ed..

Arranz, S., Silvan, J. M., & Saura-Calixto, F. (2010). Nonextractable polyphenols,usually ignored, are the major part of dietary polyphenols: A study on theSpanish diet. Molecular Nutrition & Food Research, 54(11), 1646–1658.

Bergstrom, A., Licht, T. R., Wilcks, A., Andersen, J. B., Schmidt, L. R., Gronlund, H. A.,Vigsnaes, L. K., Michaelsen, K. F., & Bahl, M. I. (2012). Introducing GUt Low-Density Array (GULDA) – A validated approach for qPCR-based intestinalmicrobial community analysis. FEMS Microbiology Letters, 337(1), 38–47.

Bouayed, J., Deußer, H., Hoffmann, L., & Bohn, T. (2012). Bioaccessible and dialysablepolyphenols in selected apple varieties following in vitro digestion vs. theirnative patterns. Food Chemistry, 131(4), 1466–1472.

Campos, D., Betalleluz-Pallardel, I., Chirinos, R., Aguilar-Galvez, A., Noratto, G., &Pedreschi, R. (2012). Prebiotic effects of yacon (Smallanthus sonchifolius Poepp.& Endl), a source of fructooligosaccharides and phenolic compounds withantioxidant activity. Food Chemistry, 135(3), 1592–1599.

Cani, P. D., & Delzenne, N. M. (2011). The gut microbiome as therapeutic target.Pharmacology & Therapeutics, 130(2), 202–212.

Delzenne, N. M., Neyrinck, A. M., & Cani, P. D. (2011). Modulation of thegut microbiota by nutrients with prebiotic properties: Consequences forhost health in the context of obesity and metabolic syndrome. MicrobialCell Factories, 10(Suppl. 1), S10.

Drogoudi, P. D., Michailidis, Z., & Pantelidis, G. (2008). Peel and flesh antioxidantcontent and harvest quality characteristics of seven apple cultivars. ScientiaHorticulturae, 115(2), 149–153.

Hoehn, E., Gasser, F., Guggenbühl, B., & Künsch, U. (2003). Efficacy of instrumentalmeasurements for determination of minimum requirements of firmness,soluble solids, and acidity of several apple varieties in comparison toconsumer expectations. Postharvest Biology and Technology, 27(1), 27–37.

Hollebeeck, S., Borlon, F., Schneider, Y.-J., Larondelle, Y., & Rogez, H. (2013).Development of a standardised human in vitro digestion protocol based onmacronutrient digestion using response surface methodology. Food Chemistry,138(2–3), 1936–1944.

Kallus, S. J., & Brandt, L. J. (2012). The intestinal microbiota and obesity. Journal ofClinical Gastroenterology, 46(1), 16–24.

Karaman, S., Tutem, E., Baskan, K. S., & Apak, R. (2013). Comparison of antioxidantcapacity and phenolic composition of peel and flesh of some apple varieties.Journal of the Science of Food and Agriculture, 93(4), 867–875.

King, B. C., Donnelly, M. K., Bergstrom, G. C., Walker, L. P., & Gibson, D. M. (2009). Anoptimized microplate assay system for quantitative evaluation of plant cellwall–degrading enzyme activity of fungal culture extracts. Biotechnology andBioengineering, 102(4), 1033–1044.

Łata, B., Trampczynska, A., & Paczesna, J. (2009). Cultivar variation in apple peel andwhole fruit phenolic composition. Scientia Horticulturae, 121(2), 176–181.

Licht, T. R., Hansen, M., Bergstrom, A., Poulsen, M., Krath, B. N., Markowski, J.,Dragsted, L. O., & Wilcks, A. (2010). Effects of apples and specific applecomponents on the cecal environment of conventional rats: Role of applepectin. BMC Microbiology, 10(9), 13.

Lyra, A., Lahtinen, S., Tiihonen, K., & Ouwehand, A. C. (2010). Intestinal microbiotaand overweight. Beneficial Microbes, 1(4), 407–421.

McCleary, B. V., DeVries, J. W., Rader, J. I., Cohen, G., Prosky, L., Mugford, D. C., &Okuma, K. (2012). Determination of insoluble, soluble, and total dietary fiber(CODEX definition) by enzymatic-gravimetric method and liquidchromatography: Collaborative study. Journal of AOAC International, 95(3),824–844.

Medina, W., Skurtys, O., & Aguilera, J. M. (2010). Study on image analysis applicationfor identification Quinoa seeds (Chenopodium quinoa Willd) geographicalprovenance. LWT – Food Science and Technology, 43(2), 238–246.

Million, M., Angelakis, E., Maraninchi, M., Henry, M., Giorgi, R., Valero, R.,Vialettes, B., & Raoult, D. (2013). Correlation between body mass index andgut concentrations of Lactobacillus reuteri, Bifidobacterium animalis,Methanobrevibacter smithii and Escherichia coli. International Journal ofObesity (London).

Moco, S., Martin, F. P., & Rezzi, S. (2012). Metabolomics view on gut microbiomemodulation by polyphenol-rich foods. Journal of Proteome Research, 11(10),4781–4790.

Picouet, P. A., Landl, A., Abadias, M., Castellari, M., & Viñas, I. (2009). Minimalprocessing of a Granny Smith apple purée by microwave heating. InnovativeFood Science & Emerging Technologies, 10(4), 545–550.

Quevedo, R., Jaramillo, M., Díaz, O., Pedreschi, F., & Aguilera, J. M. (2009).Quantification of enzymatic browning in apple slices applying the fractaltexture Fourier image. Journal of Food Engineering, 95(2), 285–290.

Quintero Ruiz, N. A., Demarchi, S. M., Massolo, J. F., Rodoni, L. M., & Giner, S. A.(2012). Evaluation of quality during storage of apple leather. LWT – Food Scienceand Technology, 47(2), 485–492.

Sanchez-Patan, F., Cueva, C., Monagas, M., Walton, G. E., Gibson, G. R.,Quintanilla-Lopez, J. E., Lebron-Aguilar, R., Martin-Alvarez, P. J.,Moreno-Arribas, M. V., & Bartolome, B. (2012). In vitro fermentation of a redwine extract by human gut microbiota: Changes in microbial groups andformation of phenolic metabolites. Journal of Agriculture and Food Chemistry,60(9), 2136–2147.

Saura-Calixto, F. (2011). Dietary fiber as a carrier of dietary antioxidants: Anessential physiological function. Journal of Agriculture and Food Chemistry, 59(1),43–49.

Shinohara, K., Ohashi, Y., Kawasumi, K., Terada, A., & Fujisawa, T. (2010). Effect ofapple intake on fecal microbiota and metabolites in humans. Anaerobe,16(5), 510–515.

Tsao, R., Yang, R., Christopher, J., Zhu, Y., & Zhu, H. H. (2003). Polyphenolic profiles ineight apple cultivars using high-performance liquid chromatography(HPLC). Journal of Agricultural and Food Chemistry, 51(21), 6347–6353.

Tuohy, K. M., Conterno, L., Gasperotti, M., & Viola, R. (2012). Up-regulating thehuman intestinal microbiome using whole plant foods, polyphenols, and/orfiber. Journal of Agriculture and Food Chemistry, 60(36), 8776–8782.

Vieira, F. G. K., Borges, G. D. S. C., Copetti, C., Amboni, R. D. D. M. C., Denardi, F., &Fett, R. (2009). Physico-chemical and antioxidant properties of six applecultivars (Malus domestica Borkh) grown in southern Brazil. ScientiaHorticulturae, 122(3), 421–425.

Yan, H., & Kerr, W. L. (2012). Total phenolics content, anthocyanins, and dietaryfiber content of apple pomace powders produced by vacuum-belt drying.Journal of the Science of Food and Agriculture, 19(10).

Zurita, J., Diaz-Rubio, M. E., & Saura-Calixto, F. (2012). Improved procedure todetermine non-extractable polymeric proanthocyanidins in plant foods.International Journal of Food Sciences and Nutrition, 63(8), 936–939.