Enzymatic Liquefaction of Cell-Walls from Kent and Tommy Atkins Mango Fruits

Tab le mango varieties discarded by the export market are generally not considered suitable for processing mainly because they yield too viscous fruit purées. The objective of this study was determining if appropriate enzymat ic treatment can overcome this barrier. Cell-wall polysaccharides from mesocarp and pericarp of fu lly ripe table mangoes were characterized analyzing the alcohol-insoluble residues (AIR). Content of celluloses, hemicellu loses, lignin and soluble and insoluble pectin were assessed after selective ext raction. After hydrolysis of main fractions, neutral sugars were determined by gas chromatography showing that xylose prevails (12-14 %) in the non-cellu losic fract ion of insoluble cell walls from mesocarp, indicating predominance with cellulose, of xilan-type polysaccharides in mango flesh. Water insoluble AIR (WAIR) was incubated with commercial preparat ions, characterized for their main enzymat ic activit ies and comprising balanced proportion of pectinases and cellulases with other different secondary activities. At equivalent 500 μl per kilogram of purée and at 45°C, solubilisation rates of uronids and neutral sugars reach respectively 100 and 90% only when xy lanase activities were present. Then, a commercial enzyme preparation containing pectinases, cellulases and high xylanase activities was applied to native mango purée varying enzyme concentration and incubation time according to a central composite rotatable design. It was shown that percent of final suspended insoluble solids and Bostwick consistency could be considerably reduced in minutes, using a relatively low amount of enzyme preparation (150 μl.L). Rheological properties of mango purées can be modulated easily according to incubation time and concentration of enzymatic solution to fit the needs of food industries.


Introduction
Global production of mango was forecasted to reach alm ost 32 million tonnes by 2010, fro m wh ich appro ximately 1,5 million were exported worldwide as fresh fruit [1]. Exported fresh fruits correspond to varieties that were selected for their firmness even at optimal maturity, as they must undergo considerable handling fro m field to international market, sometimes including hot-water treat ments to reduce anthracnose incidence or eradicate fruit flies as required by the US market [2]. The variet ies that can meet these technological requirements can be called table varieties, with Tommy Atkins and Kent among the main traded varieties.
The high standards of quality imp ly that large quantities of fru it must b e d is carded by p acking houses fo r being misshapen or having visual defects. Even in well managed plantations in Ecuador for instance, 20% to 30% of fruits are discarded in the fields and packing houses. These fruits are usually sold at low prices in local markets or even dumped in some cases. Although these mangoes have good internal properties, they are scarcely used by the juice industry because of poor technological properties such as low ju ice yield and highly viscous purée. Actually, other varieties are much mo re appropriate for juice production such as the Indian varieties Alphonso and Totapuri, which represent most of the international trade in mango purée [3,4]. These varieties give off better aro ma, have a h igher ju ice yield and lower content of suspended insoluble solids than table mango variet ies. Even so, the low value of the significant amount of discarded table mangos has become a major constraint to increasing the competitiveness of the export mango agro-industry. Special effort must therefore be made to process this fruit into a valuable product.
Renewed interest in fruit products rich in dietary fibbers may also represent an opportunity for adding value to products prepared with table mangos [5,6]. In fact, their technological characteristics, high viscosity and consistency, and low juice y ield can be improved by using appropriate enzy mes to degrade cell-wall polysaccharides, which are responsible for the purée's rheological properties. Characterizat ion of cell-wall polysaccharides has been studied by some authors for Alphonso [7,8], To mmy Atkins and other varieties [8][9][10][11], but emphasis was put on studying ripening and natural tissue softening of mango fruit. Although studies on enzymatic liquefaction of viscous fruit juice to achieve better concentration rates are yet to be generally performed, all of them have been carried out on mango varieties appropriate for ju ice ext raction and not on table mangoes [3,[12][13][14][15] except for Keitt variety [16]. Even these studies have been limited to testing commercial preparation, and little informat ion is availab le on the relevant enzy matic activ ities that should be involved to reach the targeted technological effect. Consequently, for the two main table varieties (To mmy Atkins and Kent), specific studies are likewise necessary. This paper deals with (1) the characterizat ion of cell-wall fruit polysaccharides of fully mature fruit , (2) the subsequent selection of an appropriate enzymat ic preparation that can better liquefy cell-walls polysaccharides, and (3) application to purées to obtain different Bostwick consistencies and extend the industrial uses of discarded fruits. Actually, Bostwick consistency index is of very high practical interest for many industrial processes, in various fruit industries [17].

Raw Materi als
Green mangos were randomly selected fro m five packing houses. They were physiologically mature, but discarded by the export market fo r physical defects. All fru its were harvested fro m co mmercial p lantations located in the Province of Guayas (Ecuador). Fruits were ripened at 22℃ and RH 75% for at least 8 days, and only coloured fruits (i.e. with mo re than 60% o f the surface red for the variety To mmy Atkins and yellow for Kent) were selected.
Fruits were manually peeled and mesocarp was separated and pulped with a processor, and then mash was sieved on a 2 mm diameter sieves to obtain native mango purée. Separated rind was also mashed under the same conditions.

Cell-Wall Materi als
Mashed mango or rind was suspended in five volu mes of hot 95% (v/v) ethanol. The slurry was washed with 80% ethanol until the elute responded negatively to anthrone reagent. The raw alcohol-insoluble residue was freeze-dried, ground in a hammer mill and sieved (mesh size < 500 µm). It was then suspended in cold water (4℃ ) and washed thoroughly to remove soluble materials. The lost weight corresponded to the total content of water-soluble pectin (WSP). An aliquot of the recovered washing water was dialysed against distilled water, concentrated and freeze-dried to obtain part of the WSP for subsequent analysis.
Starch and cytoplasmic proteins were partially removed, following the methodology presented by Brillouet, et. al. [18] Briefly, nonhydrosoluble residue was incubated with a protease (pronase) derived fro m Streptomyces griseus (and supplied by Boehringer-Mannheim, Germany), and then with a thermostable a-amy lase (Thermamy l Liquid 60® fro m Novo Industrials, Denmark) and an amyloglucosidase from Aspergillus niger (Merck, Germany). The slurry was subsequently deproteinated and destarched, and centrifuged. It was washed thoroughly with distilled water to remove soluble materials and thus, obtain residue that was insoluble in both water and alcohol (WAIR). The WAIR corresponded to purified insoluble, cell-wall polysaccharides was freeze-dried, milled and sieved (mesh size < 500 µm). The AIR corresponded to the sum of total contents of WSP and WAIR.

Chemical and Physical Anal ysis
Mango purée was analysed for pH, titratable acidity, dry matter, ash, soluble solids, total sugars and reducing sugars, using standard methods [19]. Vitamin C was assessed by the reflectometric method, using an ascorbic acid test (Merck's Reflectoquant® kit).
Cellu lose and hemicellulose were ext racted by selective fractionation of WAIR according to methods as described by Vo ragen et. al. [20]. Lignin was determined according to Effland.[21] Proteins were estimated, using the Kjeldahl method. Neutral sugars were analysed by gas chromatography (GC) (Shimad zu 14B, detector FID), featuring a Supelco SPB-225 co lu mn (30 m × 0.32 mm), after hydrolysing residues either by the Saeman (H 2 SO 2 72%) or the trifluoroacetic acid (2 M ) hydrolysis, and conversion of sugars to alditol acetates, according to the method adapted from Brillouet, et. al. [18] and Harris et. al. [22]. Only the higher content of neutral sugars obtained, whether by Saeman or TFA hydrolysis, were reported. Glucose liberated by TFA hydrolysis was reported as non-cellulosic g lucose (NC glu), whereas cellulosic glucose was estimated as the difference between the TFA and Saeman hydrolyses.
Total neutral sugars were measured, using the anthrone method [23] and deducing interference of galacturonic acid. Uronide material was determined by the m-hydroxydiphenyl spectrophotometric method. [24] Starch content was determined follo wing the enzy matic procedure described by Batey. [25] Methylation analysis of galacturonic acid was performed as described by Klavons et. al. [26] Suspended insoluble solids (SIS) were assessed as the percentage weight of the residue after centrifuging mango purée at 1000 g for 15 min. Consistency was assessed with a Bostwick consistometer, using 90 g of purée at a constant temperature (25±2℃ ), and measuring the distance (in cm) the pulp flowed under its own weight on a 15° slope for 30 seconds. Average of three consecutive tests was reported. Mesocarp firmness was assessed with a fruit pressure tester (Gullimex®), featuring a probe with a 0.8-mm d iameter. Two measures per fruit were taken at opposite sides after removing peel.

Enzymatic Treatment
Two co mmercial enzy me solutions, Rapidase® Carrot Juice and Rapidase® Pomaliq 2F, were purchased from DSM (Seclin, France). En zy matic hydrolysis of WAIR was performed as follows: 10 mg of WAIR was suspended in 10 ml of phosphate buffer at the p H of each fru it (p H 3.5 for Tommy Atkins and 4.7 for Kent). The en zy me concentrated between 5 and 50 µl per gram of WAIR, was added, and the whole incubated for 90 min between 30 and 60℃ , and gently agitated. The enzyme was inactivated by heating for 5 min at 80℃ , and the slurry centrifuged at 6000 g fo r 10 min . A blank without enzy me was also imp lemented. The supernatant was recovered for subsequent analysis of liberated sugars. All en zy matic p reparations used were previously assessed for their endogenous content in A GU and in neutral sugars to report only sugars liberated fro m WAIR. Assays on native mango purées were done by introducing 1ml of d iluted enzy me solution to 100 g of purée, incubating at 45℃ with gentle magnetic agitation. Then, purées were pasteurized at 80 ℃ during 5 min fo r subsequent analysis of SIS and Bostwick consistency.

Experi mental Designs
Two central co mposite rotatable designs (CCRD), , both with t wo variab les (X1 and X2) corresponding to (1) temperature and enzyme concentration to assess response pattern of WAIR solubilisation; and (2) time and en zy me concentration to estimates responses on native mango purée. Responses (Y) were co mputed, using statistical software (JMP, v. 9.02, SAS Institute) to fit a second-order polynomial equation: Y = a0 + a1X1 + a2X2 + a11X12 + a22X22 + a12X1X2 (Eq 2) Statistical analyses, like coefficient of regression R 2 , analysis of variance and P-value associated were calculated. Variance of experimental error was assessed by repeating experiment at least three times in the centre of the experimental field. A P-value, called P lof was also used to test if the variance of residual corresponding to the lack of fit was similar to experimental error. Finally, a plot of residual versus predicted value was analysed to check for randomized pattern. All these analyses were provided by the software JMP (v. 9.02, SAS Institute). Once the model has been validated, response surfaces were presented as contour plot graph using the software Sig maPlot V7, 2001.

General Characterization of Mango Fruits
At optimal maturity, both varieties Kent and Tommy Atkins differed slightly for several physico-chemical properties. Major differences were found for pH and acid ity, firmness, and soluble solids and vitamin C contents (Table 1). Kent appears to be sweeter and less acid, and to contain 50% more vitamin C than does Tommy Atkins. The pericarp always presented higher vitamin C content than did mesocarp. The flesh is firmer in To mmy Atkins than for Kent.

Characterization of Cell-Wall Pol ysacchari des
The AIR, after deproteination and destarching, represented 1.7% and 1.45%, respectively, of the mesocarp fro m To mmy Atkins and Kent, and 7.2% and 4.1% of the pericarp (Tab le 2). The value for the mesocarp fro m To mmy Atkins is slightly higher than the 1.1 % found by other authors [10]. However, d ifferences may be exp lained by the different ext raction method followed. High content of AIR is responsible for the higher firmness of table mangos that are selected as better able to withstand mechanical stresses. The higher content of AIR in To mmy Atkins's flesh explains the higher firmness observed previously. Both varieties yield almost the same amount of flesh (mesocarp). Although Kent has a thicker skin, it is counterbalanced by having a s maller seed that represents only 4% of the whole fruit, co mpared with 11 % for To mmy Atkins.
Cellu lose is the predo minant co mponent of AIR in the mesocarp, being around 37% for To mmy Atkins and 45% for Kent. Cellu lose content is lower in the pericarp where soluble pectin is the major co mponent (30% for To mmy Atkins and 40% for Kent). In the flesh (mesocarp), soluble pectin (WSP) represents only 32% of the AIR for To mmy Atkins and 25% for Kent. The content of hemicellulose is similar for both varieties. The amount of lignin in the Tommy Atkins mesocarp exceeded that in Kent, data, which can be also correlated with greater firmness.  Results of the complete analysis of sugars in the hydrosoluble pectin (WSP) and in the insoluble cell walls (WAIR) are presented in Table 3. Galacturonic acid is the predominant sugar in WSP ranging from 40% (pericarp fro m Tommy Atkins) to almost 74% (mesocarp fro m Kent). Soluble pectin was highly esterified, being about 78% for the mesocarp of both varieties. Highly esterified pectin appears to be a characteristic of mango fruits in concordance with results obtained by [11,28]. Predo minant neutral sugars in the WSP are arabinose, non-cellulosic glucose, xy lose and galactose in the mesocarp, showing a probable predominance of arabino-g lucans, arabino-xylan or arabino-galactan polysaccharids in the side chains of the soluble pectins. These results are in agreement with those find by other authors who studied more deeply the linkages between neutral sugars [29]. In the soluble pectin from the pericarp, the contents of galactose and arabinose are much higher indicating predominance of arabino-galactan-type polysaccharides. It can be also noted that soluble pectin (WSP) fro m Kent variety has a higher content of uronids that for To mmy Atkins.
Analysis of sugars in the WAIR shows that glucose proceeding fro m cellulose was the prevailing sugar, followed by uronides and xy lose. Predominance of xy lose among non-cellulosic neutral sugars has been reported previously for ripe To mmy Atkins [10] and other mango varieties [13] , [28]. The part icularly h igh content of xy lose in the WAIR fro m the mesocarp shows a presumable predominance of xylan -type polysaccharides in the hemicellulosic fraction. In the pericarp, the predominant non-cellulosic neutral sugar is arabinose, followed by xy lose and galactose.

Selecting an Enzymatic Preparati on
The importance of cellulose and insoluble pectin in mango cell walls implies the need to use an enzymat ic preparation able to hydroly ze at least both substrates. Nonetheless, the importance of other sugar polymers also may imp ly the use of other side enzymes, specifically, xy lanases and arabinases. Seven commercial preparations were tested (data not shown), and among them, Rapidase® Carrot Juice and Rap idase® Pomaliq were able to solubilize the h ighest amount of insoluble cell walls fro m mango mesocarp (WAIR). The main spectrum of enzy mat ic activit ies of both commercial preparations has been published previously [18,27]. They contain a high and balance amount of pectinases and cellu lases, as well as other side activ ities, contrary to the other preparation tested (data not shown). The central composite rotatable design implemented, allowed testing different temperatures and enzyme concentrations, ranging respectively from 30 to 60 º C and from the equivalent of 50 to 500 µl per liter of mango purée (the mango purée being studied contains 10 ± 0.015 g o f WAIR per kilogram of purée, as calculated from Table 2 fo r both varieties). The second-degree models obtained after co mputing the results of the experimental design gave rise to excellent statistical test values ( r 2 > 0.95; P≤0,01; P lof >10%). Corresponding response surfaces are presented as contour plot graphs in Fig.  1 and 2. Figure 1 shows, that for both preparations, at an equivalent of 500 µl per kilogram of purée, 100% of uron ides (AGU) is released in the same way for both varieties as values around 12% w/w of WAIR are inclusively slightly h igher than the ones obtained by chemical hydrolysis (Table 2). Figure 1 also shows that no significant differences between both preparations were found for the rate at which uronic material was released. The similar activity of the pectin lyase in both preparations may exp lain this result, as mango pectin is highly esterified. In fact, h igher polygalacturonase and pectinesterase activities in Rap idase® Carrot Ju ice preparation [27] d id not seem to improve uronides release. Figure 1 shows also that optimu m temperature for pectinase activity is between 45 and 50℃ .
Enzyme concentration µl.kg -1 of mango puree For both varieties, Rapidase® Pomaliq appears to be more efficient in releasing neutral sugars (Fig. 2) than the other enzy matic preparation at the same concentration. In fact, taking into account total neutral sugars content in WAIR endocarp (Table 3), Rapidase® Po maliq is able to release up to 90% w/ w of neutral sugars contained in WAIR fro m both varieties, whereas Rapidase Carrot Juice ® releases less than 70% at 45℃ and at a concentration level equivalent to 500 µl per kilogram of mango purée.
Even so, both preparations appear to have had very similar content of cellulase enzy mes, differing only slightly for the ratio endoglucanase to cellobiohydrolase (1.9 for Rap idase® Pomaliq and 1.5 for Rap idase® Carrot Ju ice) and for xy lanase and exo arabinase (table 4). A higher exo-arab inase activity in Rapidase® Carrot Juice does not seem to make a difference, probably because this enzyme acts mainly on the side chain of the previously solubilised pectin. In contrast, higher xy lanase activity appears to be a factor that may explain the better solubilization rate observed, as xy lose prevails in the noncellu losic fraction of neutral sugar fro m WAIR mesocarp. Neutral sugars fro m Kent's WAIR, appear to be more easily released than for Tommy Atkins. A higher content of cellulose in Kent, less lignin content and a natural pH closer to the optimal pH of endocellu lase [30] may explain this result.

Obtaini ng Mang o Purées of Gi ven Bostwick Consistencies
Native mango purées obtained immediately after sieving are extremely rich in insoluble suspended solids (SIS; 84% for To mmy Atkins and 73% for Kent), giv ing rise to low Bostwick consistency (8 and 12 cm.30s -1 respectively for Tommy Atkins and Kent) ( Table 5, run 1) only comparab le to concentrated banana purée [17]. Purée fro m Kent variety appears to be less consistent than Tommy Atkins which is in agreement with previous results.
As it can be observed in Fig. 3, enzy mat ic treat ment with Rapidase® Po maliq significantly decreases SIS and Bostwick consistency. By using relatively low concentrations of enzy mes (150 ml per kilogram of purée) for 60 min of incubation, we could reduce SIS content by almost 30% and increase flow velocity on the Bostwick consistometer by appro ximately the same amount in both purées. For a very similar result on SIS, other authors [14] used much mo re en zy me solution (up to 2 ml kg -1 ), containing mainly pectinesterase and polygalacturonase activities.   Additionally, by modulating incubation time and enzyme concentration, we could obtain purées with different Bostwick consistency. SIS content and Bostwick consistency can be predicted by using mathematical models obtained through computing CCRD results (table 5). Mathematical models fit well, as r 2 was always superior to 0,92 and P≤0,01, P lof >10%. We can see, in concordance with previous results, that Tommy Atkins purée appears to be less easily liquefied than Kent. Even so, we could obtain fro m both varieties mango purée that was consistent enough for ju ices blends, smoothies or baby foods.

Conclusions
Mango table varieties are characterized by their firmness to endure mechanical stress during handling and transport of fresh fru its. Actually, their higher content of insoluble and soluble pectin, cellu lose and xylan -type polysaccharides strengthens the structure of cell-walls, mainly in the flesh. As a consequence, purée resulting fro m fruits discarded by the export market was generally not considered suitable for the fruit ju ice industry. Nonetheless, we have shown that the appropriate choice of an en zy matic p reparation can change this perspective. We have shown that the application of pectinases and cellulases is not enough and presence of xy lanase, wh ich is considered generally as a secondary activity in commercial preparation, was crucial to increase significantly the release of neutral sugars from isolated cell-walls fro m mango mesocarp. Xy lanase may act on xy lan-type polysaccharides but the effect on cell-wall solubilization is probably indirect, by facilitating access of pectinases and cellulases to their substrate. As a result, a commercial en zy me that presents high pectinase, cellulase and xylanase activities can reduce drastically content of suspended insoluble solids yielding in minutes a purée with rheological characteristics that can be modulated to fit better the needs of food industries. Additionally, the content of soluble and insoluble fibber of these purées could be high with subsequent health benefits for consumers. Therefore, although, purées of table mango varieties may be less flavorfu l than purées fro m other mango variet ies, fresh table mangoes discarded from the export market only for visual defects, should be considered as highly valuable raw material for being used by food industries as a basis for Tilme of incubation (min) a) b) different juice blends and smoothies with increasing market world wide.