THE ROLE OF POLYSACCHARIDES ON THE GRAPE MUST ULTRAFILTRATION PERFORMANCE

.


INTRODUCTION
Membrane processes like Microfiltration (MF), Ultrafiltration (UF), Reverse Osmosis (RO), Nanofiltration (NF) and Electrodialysis (ED) are widely used in wine industries, namely to clarify musts and wines, to concentrate them, to eliminate a part of the fermentable sugars, or to achieve tartaric stability.
Despite the fact that these techniques have become an alternative to conventional processes, the complex composition of grape musts and wines cause problems of membrane fouling that are not well understood. However, very often the fouling of ultrafiltration membranes is attributed to the retention of macromolecular components like polysaccharides, and this is very detrimental for the wine quality. The traditional approach that considers the ultrafiltration mainly governed by sieving mechanisms, has been taken over by an approach where simultaneously with these mechanisms, there are other ones resulting from the development of membrane/macromolecules interactions. In fact, taking into account the complex nature of these macromolecules, a multiplicity of complex interactions with the membrane matrix, can occur and lead not only to fouling but also to undesirable aspects like inhibition of their function as protective colloids (Gonçalves et al., 2001;De Pinho, 2010).
In grape must (unfermented juice) polysaccharides originate from the cell walls of grape berries, belonging two families. The largest one is rich in arabinose and galactose and contains arabinogalactanproteins (AGPs). An AGP of average molecular weight of 165 kDa (Saulnier and Brillouet, 1989) and another one of 110 kDa (Saulnier et al., 1992) were isolated from the pulp of grape berries. However, in other work, the average molecular weight of five arabinogalactan-proteins fractions separated from a red wine ranged from 180 to 260 kDa (Pellerin et al., 1995).
AGPs have a main chain of galactose units linked by β-(1,3) oside bonds, with short lateral chains of galactose bonded on β-(1,6). In C3, these are replaced by individual arabinose units. AGPs also have a few individual arabinose units on carbon 6 or 4 of the galactose in the main and secondary chains Pellerin et al., 1995). AGPs are linked to a protein (2 to 10% by weight) characterized by the presence of hydroxyproline. The AGPs are present in wine (Pellerin et al., 1995) at concentrations approximately of 100 to 250 mg/L for red wines and 50 to 150 mg/L for white wines (Pellerin and Cabanis, 1998;Doco et al., 2000). The pulp tissue from grape berry contains two-fold more soluble AGPs than the skin tissue (Vidal et al., 2001). AGPs represent a major proportion (40%) of the total wine polysaccharides, confirming the abundance of AGPs in red wine (Pellerin et al., 1995).
The elucidation of the composition and structure of the macromolecular fraction of the wines aged on lees, showed a depolymerization of the native polysaccharides of the grapes. This modification consisted of a 75% dearabinosylation of arabinogalactan-proteins (AGPs) and a total hydrolysis of the pectic arabinans .
The arabinans of the same family are only rich in arabinose, they are small polymers with a low molecular weight (6 kDa) and consist of short arabinose chains, bonded in α-(1,5), with ramifications of individual arabinose units on C3 (Villetaz et al., 1981). The concentration of arabinans is approximately 100 mg/L at the end of alcoholic fermentation, and is rarely present in mature wine (Pellerin and Cabanis, 1998;Doco et al., 2003). Nevertheless, an insoluble linear arabinan has been isolated from a red wine (Belleville et al., 1993).
The second family is rhamnogalacturonan type II (RG-II), identified in many plant tissues Ridley et al., 2001). It's a very complex pectic polysaccharide with a low molecular weight -5,3 kDa determined by high performance size-exclusion chromatography (HPSEC) and 9,8 kDa determined by low-angle laser-light-scattering (LALLS) (Doco et al., 1993) -and composed mainly of rhamnose and galacturonic acid. RG-II consists of a main chain of seven to nine galacturonic acids units with four lateral oligosaccharide chains (Doco et al., 1993;O'Neill et al., 2004). These chains contain arabinose, rhamnose, fucose, galactose, galacturonic and glucuronic acids and also rare sugars as 2-O-methyl-fucose, apiose, 2-O-methyl-xylose, KDO (2-Keto-3-deoxy-D-mannooctulosonic acid), DHA (3-deoxy-D-lyxohepyulosaric acid) and aceric acid (3-C-carboxy-5deoxy-L-xylose) (Doco et al., 1993;Pellerin et al., 1996;Ridley et al., 2001;O'Neill et al., 2004). RG-II is resistant to pectolytic enzymes and is found in fruit juice and wine obtained by total liquefaction with pectinases (Doco et al., 1997). RG-II represents about 20% of total red wine polysaccharides . The concentration of RG-II in white wines is smaller (ranging from 30 to 50 mg/L), than in red wines (from 100 to 150 mg/L) (Doco et al., 1997). It was shown that the content of RG-II per kilogram of berries is approximately 250 mg and its content in skin tissue is three-fold higher than that of pulp tissue (Vidal et al., 2001). This result is consistent with the fact that more grape polysaccharides are present in red wine than in white wine (Vidal et al., 2001).
Mannoproteins (MPs) are another major group of polysaccharides in wine and are originated from yeast cell walls. This polysaccharide leads to an enrichment of wine in mannose both in the beginning of alcoholic fermentation and during aging on lees by autolysis of yeasts (Saulnier et al., 1991;Waters et al., 1994;Feuillat, 2003). MPs have a wide range of molecular weights in wines, extending from 5 to more than 800 kDa (Saulnier et al., 1991;Gonçalves et al., 2002). A group of MPs are constituted by about 10% protein and 90% of mannose. The molecular structures of MPs consist of a peptide chain linked to D-mannose units in α-(1→6), α-(1→2) and α-(1→3) (Saulnier et al., 1991;Waters et al., 1994). Its concentration in wines is approximately 100 to 150 mg/L (Saulnier et al., 1991;Waters et al., 1994;Feuillat, 2003). MPs isolated from a white wine represent 32.2% of the total wine polysaccharides (Gonçalves et al., 2002).
The involvement of polysaccharides in many oenological phenomena is due to their physicochemical properties, when in solution. For instance, their ability to interact with tannins to inhibit their aggregation, maintaining the stability of colour (Riou et al., 2002) and in red wines the increase of smoothness and body (Guadalupe et al., 2010). As already stated, another important property is their ability to act as protective colloids, their contribution to protein stability (inhibition of protein haze in white wines) (Waters et al., 1993) and tartaric stability (inhibition of hydrogen tartrate crystallization, increasing tartaric stability) (Gerbaud et al., 1997;Moine-Ledoux and Dubourdieu, 2002). Also, an important role of RGII, that is under investigation, and include nutritional benefits, is the complexation of divalent cations, to form specific coordination complexes with heavy metals Pellerin et al., 1997). Finally, the presence of polysaccharides is determinant in the persistence of bubbles in sparkling wines and in the interaction with aromatic compounds (Chalier et al., 2007).
In order to evaluate the role of these macromolecules in the fouling of filtration membranes, many studies have been done (Belleville et al., 1990(Belleville et al., , 1992Cameira dos Santos et al., 1994;Cameira dos Santos, 1995;Vernhet et al., 1997Vernhet et al., , 1999Vernhet and Moutounet, 2002;Vernhet et al., 2003;Ulbricht et al., 2009;El Rayess et al., 2011). Belleville et al. (1992Belleville et al. ( , 1993 identified the fouling of an inorganic tubular alumina membrane during red wines microfiltration and attributed it mainly to RG-II, that was strongly absorbed by the membrane and caused a sharp decrease of the permeation flux and a linear arabinan, also induced a decrease in permeate flux. Cameira dos Santos et al. (1994) studied wine crossflow microfiltration using polyethersulfone (PES) capillary membranes and concluded that macerated wine is more foulant, probably due to the higher levels of polysaccharides and polyphenols. They also suggested that differences in membrane fouling behaviour between wine samples is not related to its initial polyphenol and polysaccharide concentrations, but with the composition and structure of these foulant molecules. Cameira dos Santos et al. (1994) has revealed the involvement of wine polyphenols in the membrane fouling by washing the fouled membrane with acidified methanol and significant increases in permeability were obtained. This fact could be attributed to the elimination of the layers of phenolic compounds because the other wine constituents are insoluble in this solvent. Vernhet et al. (1999) isolated wine polysaccharides from a red wine, and microfiltered separately in a synthetic wine to determine their influence on the permeation fluxes of an hydrophilic organic microfiltration membrane (poliethersulfone). The permeation flux decline was differed according to the nature of the polysaccharide tested. According to these authors mannoproteins fractions (30-400 kDa) displayed the strongest decrease on wine filtrability. However, the fouling caused by mannoproteins was reduced in the presence of the pectic polysaccharides, which suggests that the performance of the membrane will depend on intermolecular interactions and interactions between membrane and foulant. Vernhet et al. (2002) also studied the effect of wine polysaccharides on three organic membranes, two of them made of PES with different amount of polyvinylpyrrolidone (PVP) and another one made from polyvinylchloride (PVC). They showed that the polysaccharides adsorption was negligible in static membranes for the three membranes. By filtering wines enriched with polysaccharides, no significant differences were observed between the three membranes in terms of amount and nature of the deposited polysaccharides. It was observed a deposition of AGP-II and MPs on the surface of the membrane. They concluded through observation of membrane surfaces by scanning electron microscopy that membrane fouling by polysaccharides is due to their accumulation on membranes, mainly at the pores entrance and on surface. Vernhet et al. (1997) demonstrated through adsorption experiments that the polarity of the membrane surface is a limiting factor in wine polysaccharides adsorption. Adsorbed polysaccharides were mainly grape arabinogalactanproteins and yeast mannoproteins. According to these authors when the polarity of the membrane surface increased polysaccharides adsorption decreased, due to unfavourable polar interactions between hydrophilic polysaccharides and the material. Contrary to polysaccharides, the maximum tannins adsorption occurred on the most polar membrane. The aim of one study by Ulbricht et al. (2009) was to pursue the hypothesis that polypropylene (PP) and PES membranes exhibit different levels of adsorption of typical foulants in wine such as polyphenols and polysaccharides, to link the level of adsorption to polymer characteristics and to correlate membrane fluxes with these findings. It was found that polyphenols and polysaccharides are only marginally adsorbed by PP membranes that presents hydrophobic character, but strongly adsorbed by PES membranes that has hydrophilic character. The low adsorption tendency of wine ingredients to PP membranes results in higher fluxes and longer service life of the respective filtration modules in wine clarification.
The present work investigates the ultrafiltration of grape must with two UF cellulose acetate membranes, of different molecular weight cut-offs, for the separation of polysaccharide macromolecules into two streams, that are further processed by the dialysis (allowing the liberation of small molecules like monosaccharides, salts, organic acids and small polyphenols) and identification of its composition in terms of neutral monosaccharides. Cellulose acetate membranes have many advantages. They combine high rejection to solutes with high hydraulic permeability due to its asymmetrical structure and have a good behaviour against fouling, because of the hydrophilic character of cellulose acetate.
Therefore the main objective of this study is to understand how the different grape must polysaccharides interact with UF membranes, due to their specific monosaccharide composition, and how that determines the retention and permeation of each polysaccharide group. Winemakers must also be aware of the risk of eliminating high molecular weight polysaccharides that are not only an integral part of a wine's composition but also of its organoleptic characteristics.

Grape must samples
Grape must used as source of polysaccharides in UF experiments was prepared from matured grapes of Trincadeira Preta (a red cultivar of Vitis vinifera), harvested in 2010 in INIAV vineyard (Oeiras, Portugal), used in Carcavelos DOC wine production. The samples were kept refrigerated and decanted, under 6ºC. Sulfur dioxide was added (300mg/L) to conservation and the grape must was brought to room temperature for the experimental runs.

Preparation of the must and of the must ultrafiltration streams, previous to polysaccharide analysis
Prior to polysaccharide analysis, samples of grape must and aliquots of ultrafiltration streams, were extensively dialyzed against distilled water, concentrated and freeze-dried.
The total colloids are first extracted from must and from aliquots of UF streams by dialysis using a 3500 Da cut-off membrane purchased at Spectrum Laboratories, Inc from CA, USA. This treatment allowed the removal of small molecules such as monomeric sugars, salts, organic acids, low size polyphenols and monomeric anthocyanins. Dialysis was carried on till the permeate conductivity was lowered up to 40 µS/cm (the initial must conductivity is 3260 µS/cm). The dialysis retentate was then concentrated under vacuum at low temperature (40ºC) and freeze-dried to give the fraction called total colloids.

Membranes
Two flat-sheet laboratory made membrane were prepared according to the phase inversion method, of Kunst and Sourirajan (1974), using cellulose acetate (CA) with 39.8%, acetyl content, supplied by Eastman-Kodak (Rochester, NY, USA). Table I displays the casting solution composition and the casting conditions of the membranes designated by CA-400-28 and CA-400-32 with different content of formamide to yield an increasing hydraulic permeability.

Membrane characterization
The membranes were characterized in terms of hydraulic permeability (Lp) and molecular weight cut-off. Membranes were first compacted through the recirculation of pure water (conductivity <1 µS/cm) at a transmembrane pressure 20% higher than the operating pressure for 3h, to avoid pressure effects on membrane structure throughout subsequent experiments.
The membrane molecular weight cut-off was determined through the permeation of reference solutes, polyethylenglycols (PEG) supplied by Merck and dextrans supplied by Pharmacia. The rejection coefficients to reference organic solutes are displayed in Table II,  The MWCO value is obtained by the intersection of the 91% rejection line with the linear relationship of log (f/(1-f)) versus solute molecular weight. This procedure is represented in Figure 1 for the CA-400-32 membrane and in Figure 2 for the CA-400-28. The membrane MWCO is defined by the molecular weight of a given macromolecule whose rejection coefficient is higher than 91%. The pure water permeate flux was measured at operating pressures in the ranging from 1 x 10 5 Pa to 3 x 10 5 Pa to yield the hydraulic permeability (L p ) of the two membranes. The membrane hydraulic permeability is given by the slope of the straight line of the pure water permeate flux versus transmembrane pressure. The results obtained are shown in Table III.

Ultrafiltration experiments in concentration mode to achieve polysaccharide separation
Polysaccharides of the grape must were fractionated by UF in a Lab-unit ("CELFA"), type P-28 ( Figure  3). The experiments were run in concentration mode for which the concentrate was recirculated to the feed tank. The membrane surface area was 25.52 x 10 -4 m 2 and the feed temperature was kept at 25ºC for all experiments.

Diagrama esquemático da unidade de filtração (tipo P-28 Lab-Unit -CELFA)
The permeation of reference solutions for membrane characterization, as well as permeation of grape must, was performed at a transmembrane pressure of 3 x 10 5 Pa, at a feed flow velocity of 0.67 m/s and at a temperature of 25°C. The permeate samples were collected after the stabilization time. The stabilization time for each experimental run was 10 min. After each essay, membranes were washed with deionized water and the pure water flux was measured for the verification of any irreversible fouling or damage.
Samples of the feed reference solutions were taken at the beginning and at the end of each experimental run and the average concentration of these two samples were considered the feed concentration. Samples of the feed grape must and of the UF permeates, in the essay with CA-400-32 membrane were taken at the beginning, at the 34 th min, at the 49 th min, at the 64 th min and at the end. Samples of the feed grape must and of the UF permeates in the essay with CA-400-28 membrane were taken at the beginning, at the 60 th min, at the 130 th min, at the 190 th min and at the end.

Membrane cleaning
To recover the initial fluxes, membranes were cleaned, after each essay, first with circulation of deionized water at the temperature of 20ºC for 1h and after with circulation of water at the temperature of 50ºC for 2h. The pure water flux was measured for the verification of any irreversible fouling or damage.

Analytical methods
The feed and the permeate reference solutions concentrations were determined in terms of Total Organic Carbon (TOC). TOC measurements were performed in a Dohrmann Carbon Analyser DC-85A (Dohrmann, Santa Clara, CA, USA) calibrated with potassium hydrogen phthalate solution (2000 ppm).
The neutral monosaccharide composition of the feed and permeate of the grape must were determined by Gas Cromatography (GC) after acid hydrolysis with 2M trifluoracetic acid (120ºC, 75 min), reduction, acetylation and extraction of the acetylated alditols (Harris et al., 1984). The separation of alditol acetates was achieved using a fused silica DB-225 (210ºC) capillary column (30 m x 0.25 mm i.d., 0.25 µm film) with H 2 as the carrier gas at a flow rate of 1.3 mL min -1 on a Varian Chrompack CP-3800 Gas Chromatograph, equipped with a flame ionization detector (FID) and a capillary split/splitless inlet. Samples were injected in the pulsed split mode with a split ratio of 60:1. The injector and the FID were operated at 250ºC.
Grape must sample was characterized in terms of density, total soluble solids, probable alcohol content, pH, total acidity, total polysaccharides, total polyphenols, conductivity, turbidity and viscosity. Density, total soluble solids, probable alcohol content and conductivity were determined according to Reg. (CEE) 2676/90; pH was determined according to method unique of International Organization of Vine and Wine (OIV); total acidity was determined according to NP-2139NP- /1987. Total soluble solids was carried out by using an Abbe refractometer Atago NAR-1T (Atago Co., Ltd., Tokyo, Japan) and expressed as ºBrix. The pH was measured at 20ºC with a Crison micropH 2002 (Crison, Barcelona, Spain). The total acidity was determined by titration with 0.1N NaOH and results were expressed as g/L tartaric acid. Total polysaccharides of the feed and permeate grape must were determined according to Segarra et al. (1995): 4 mL of grape must were precipitated with 20 mL of ethanol; the precipitate was dried and dissolved in 4 mL of water; phenolsulphuric method, according to Dubois et al. (1956), was used to analyze total polysaccharide content and the absorbance is read at 490 nm by using a spectrophotometer; the results were expressed in mg/L of glucose. A calibration curve was previously done. Total polyphenols were determined by total phenolic index, that is a spectrophotometric method (Shimadzu UV, model UV-1700) by measuring the UV absorption at 280 nm, under 1 mm optical path (Somers and Evans 1977) and the results were expressed as mg/L of gallic acid (product of absorbance at 280 nm by the dilution factor of the sample of grape must and using a standard curve). Conductivity was measured at 25ºC with a Crison conductimeter, model 525. Colour intensity was determined by the sum of absorbance at 420, 520 and 620 nm under 1 mm optical path (Sudraud, 1958;Glories, 1984) with a Shimadzu UV, model UV-1700. The tint was measured as the ratio of absorbances A 420 /A 520 (Somers and Evans, 1977). Clarity was measured by transmittance at 625 nm using a Shimadzu UV, model UV-1700. Turbidity was determined with a Turbiquant 3000 IR turbidimeter and expressed in nephelometric turbidity units (NTU). Viscosity was measured by using a Brookfield DV-I viscometer (Brookfield Engineering Laboratories Inc., Stoughton, MA, USA).

Quantification of polysaccharides adsorbed on the membrane
A known mass of each anhydrous membrane was cut, weighed, and directly hydrolyzed on a glass tube with 500 µL of TFA. After hydrolysis, the acid was eliminated under a current of air, and the monosaccharide released were reduced, acetylated and separated under the conditions described before.

Characterization of the grape must
The physico-chemical composition of grape must is shown in Table IV.

TABLE IV
Physico-chemical composition of grape must.

Composição físico-química do mosto.
The composition of the polysaccharides of grape must expressed in terms of the seven major neutral monosaccharides is summarized in Table V. The arabinose and galactose shown in Table V confirm the presence of arabinogalactan-proteins that are released after the crushing and pressing of the grapes . Quantitatively AGPs (Arabinose -25.21 mg/L and galactose -26.5 mg/L) are the major grape polysaccharide present in the grape must . The arabinose to galactose ratio is 0.95, which is close to 1.0 that is the usually ratio in AGPs .
The content of rhamnose in the grape must, 16.03 mg/L, confirm the presence of RG-II. The composition of RG-II contains fucose and in the grape must small amounts of fucose (4.22 mg/L) were also found. Xilose is a typical constituent of cell walls in the Vegetal Kingdom (Hall et al., 1976) and it is present in grapes and must in the amount of 6.30 mg/L. With the data available, we chose four monosaccharides to assess the main structure of grape must polysaccharides. The identification of 4 constituent neutral sugars was selected to act as markers of macromolecules. Rhamnose acts as a marker of RG-II, arabinose and galactose act as a marker of AGPs, while mannose acts as a marker of mannoproteins.

Permeate flux during ultrafiltration
The variation of permeation flux (J p ) as a function of the volume reduction factor (VRF) is displayed in Figures 4 and 5. These Figures show a clear decline of Jp, with the increase in VRF, due to increasing fouling effect on the membrane, because of concentration polarization and adsorption on membrane material. The initial permeate flux of 10.65 L/h/m 2 decreased to about 3.5 L/h/m 2 , after 278 min of operation, which corresponds to a VRF of 1.13 ( Figure 4). This permeate flux relative to the initial flux presents a flux reduction of 68%. In Figure 5 the initial permeate flux was 8.22 L/h/m 2 and decreased up to 0.75 L/h/m 2 , where the flux reaches a plateau, after 313 min of operation, which corresponds to a value of VRF=1.04. The permeate flux relative to initial flux presents a flux reduction of 91%. It can be observed that these curves can be divided into three periods: an initial phase with a rapid decrease of the permeate flux; a second phase with a smaller decrease of the permeate flux that begins around VRF = 1.05 ( Figure 4) and VRF = 1.02 ( Figure 5); a third phase with a very small decrease in permeate flux up to steady-state conditions, that occurred after VRF = 1.08 ( Figure 4) and VRF = 1.033 ( Figure 5).
The reduction of the flux is probably due to the adsorption of macromolecules in the matrix of the membranes, which is reported in the literature to be caused by polysaccharides and polyphenols (Belleville et al., 1990(Belleville et al., , 1992Cameira dos Santos et al., 1994;Cameira dos Santos, 1995;Vernhet et al., 1997Vernhet et al., , 1999Vernhet and Moutounet, 2002 al., 2003;Ulbricht et al., 2009;El Rayess et al., 2011).

Quantification of polysaccharides adsorbed on the membrane
The results of the analysis of the membranes as described in the paragraph quantification of polysaccharides adsorbed on the membrane of the Material and Methods section are presented in Table  VI and show that mannose and glucose were adsorbed in the CA-400-32 membrane matrix. In fact, Table VI shows the composition in neutral sugars of the membrane matrix (first line), and of the organic matter adsorbed on it (second and third lines). First it was determined the yield of an unused membrane in sugars, to know which glucidic constituents could be liberated (after acid hydrolysis) from the cellulose acetate membrane matrix (CA-400). Secondly it was analyzed the matrix of the CA-400-32 and CA-400-28 membranes after being used in the ultrafiltration of grape must.
The observation of glucose and mannose in the CA-400-32 membrane indicates the adsorption of mannoproteins. The results show that the adsorbed amount of glucose was 17.35 mg/25.52 cm 2 (after removing the 25.78 mg/25.52 cm 2 constituent of the membrane matrix) and 2.04 mg/25.52 cm 2 of mannose. In contrast, the membrane CA-400-28 did not show adsorption of polysaccharides.

Membrane cleaning
The recovery of the pure water permeation fluxes after water flushing of the membranes (description in Membrane cleaning of Material and Methods) was 64% for CA-400-32 and 80% for CA-400-28 membranes. The higher value of 80% of pure water flux recovery for the membrane of lower MWCO is in accordance with the previous result that did not show adsorption of polysaccharides.   The results in Table VII (membrane CA-400-32) show a decrease of the rhamnose content in the retentate to the point of total disappearance, while the content of other neutral sugars in the retentate remain almost constant. Furthermore the content of neutral sugars in the polysaccharides of the permeate during the UF have a decrease, verifying the complete disappearance of rhamnose in the fractions of permeate collected. So, we can say that rhamnose crosses the membrane and is not adsorbed. Despite the fact that mannose also crosses the membrane, the behaviour of this sugar is different. Mannose was found in the permeate only up to 49 min, but in small quantities, which shows retention and adsorption of the mannoproteins in membrane matrix. In fact, as already discussed, the analysis of the membrane matrix described in Table VI, give evidence that mannose was adsorbed on the matrix, which has originated, in some extension, the clogging of the pores. The presence of mannose in the permeate stream can be due to the permeation of MPs of low molecular weight, while the retention in the retentate stream due to MPs of high molecular weight. The contents of arabinose and galactose in the permeate stream are lower than those present in the original grape must, and this fact also show the retention of AGP.
In Table VIII, the neutral sugars of the retentate and permeate stream indicate that all the polysaccharides of grape must (RG-II, AGP and MPs) are retained by the membrane CA-400-28. In fact, only small quantities of mannose, galactose and glucose were found in the permeate at the first 20 minutes of the experiment.

CONCLUSIONS
It was concluded that the grape must studied contains various fractions of RG-II, and some of them crossed the CA-400-32 membrane easily. However the disappearance of the fraction with the lowest molecular weight (RG-II) in the retentate and permeate stream of the CA-400-32 membrane, may be due to its deposition on the surface of the membrane, and in the inside of pores. This latter phenomenon, called in literature as "clogging", mainly happens at the entrance of the pores and on its surface, as found in a previous study of polysaccharides by Vernhet and Moutounet (2002).
Due to the fact that we are dealing with polyssacharide fractions of low molecular weigth (5kDa), we can also admit the hypothesis that this accumulation of RG-II on membrane surface is probably caused by the previous adsorption of mannoproteins on the membrane matrix. This hypothesis is supported by the fact that the macromolecules with highest molecular weight (AGPs and MPs), were found in the permeate streams of the CA-400-32 membrane, which means that this membrane don't retain big macromolecules.
By the other hand, it was found that in UF experiments with the CA-400-28 membrane, RG-II and the majority of AGPs and MPs remained in the retentate, which means that this membrane largely retains big macromolecules.