Introduction

Galacto-oligosaccharides (GOS) are prebiotic compounds produced by the enzymatic transgalactosylation of lactose with β-galactosidase (E.C. 3.2.1.23) (Torres et al. 2010). Prebiotics have been defined as non-digestible food ingredients that selectively stimulate the proliferation and/or activity of beneficial microbial populations in the colon providing health benefits to the consumer (Crittenden and Playne 2009; Hutkins et al. 2016).

The use of non-digestible oligosaccharides (NDOs) as modulators of the intestinal motility and intestinal microbial composition goes back to the pioneering work of Petuely (1957), who evaluated the use of lactulose in infants to such purposes. Late in the 1970s, the use of GOS as a substitute for human milk-oligosaccharides (HMOS) in infant formulas for bottle-fed newborns was proposed in Japan with the purpose of promoting a healthy gut microflora, similar to that of breast-fed infants (Yazawa et al. 1978). A decade after, the first commercial product containing GOS was launched in Japan (Nauta et al. 2009) and soon after GOS were introduced in Europe in 1996 (Spherix Consulting and Inc. 2010). By 2007, the World production of GOS was estimated in 25,000 tons, representing a market value between 130 and 170 MMUSD (Paterson and Kellam 2009). In recent decades, GOS has gained considerable attention for several reasons: (a) it represents an attractive technology for whey upgrading, being whey a low-value byproduct and sometimes a waste form the cheese-making industry (Illanes 2011; Gänzle 2012); (b) different from other prebiotics, like inulin and fructo-oligosaccharides (FOS), GOS tolerate high temperatures and low pHs which allows its incorporation as additive in several food matrices (Sako et al. 1999; Wang 2009); (c) up to now GOS are the best substitute for HMOS, being it a unique property among prebiotics. GOS:FOS 9:1 mixtures have gained wide acceptance as a prebiotic supplement in infant formulas (Boehm et al. 2003; Boehm and Stahl 2007; Bode 2009) since such prebiotic mixture, comprising short-chain GOS and medium to long-chain FOS, allows obtaining a molecular size distribution similar to HMOS (Moro et al. 2002); (d) synthesis of GOS is catalyzed by β-galactosidases which are robust, cheap and readily available commercial enzymes with an extensive record of safe use in the dairy industry for the production of low-lactose milk and derived products, making them ideal catalysts for performing oligosaccharide synthesis (Illanes 2011). Even though GOS production is an established technology, there are several challenges related to their production which are mostly referred to the optimization of the enzymatic transgalactosylation reaction and the improvement in downstream operations. Both aspects will be analyzed in the following sections.

Enzymatic synthesis of GOS

GOS synthesis is a kinetically controlled reaction where hydrolysis and transgalactosylation of lactose occur simultaneously (Vera et al. 2011a). The mechanism of reaction involves two sequential steps: in the first one, a molecule of lactose binds to the β-galactosidase active site and forms the galactosyl-enzyme complex while liberating one molecule of glucose; in the second step, the transition intermediate reacts with another lactose molecule forming a gal–gal-glu trisaccharide (GOS-3), which in turn can act as acceptor of the galactosyl-enzyme complex forming a gal–gal-gal-glu tetrasaccharide (GOS-4) and so on to produce GOS with higher number of galactose units (i.e. GOS-5, GOS-6). In this way, lactose acts both as donor and acceptor of the galactose moiety. Water, being a nucleophile, can also act as acceptor of the galactosyl-enzyme complex in which case galactose will be liberated, the net result being the hydrolysis of lactose into its monosaccharide components glucose and galactose (see Fig. 1).

Fig. 1
figure 1

Schematic representation of the mechanism of GOS synthesis by transgalactosylation with β-galactosidases

Full size image

At the onset of reaction, the high concentration of lactose favors transgalactosylation, making GOS concentration to increase up to a maximum when the hydrolysis and transgalactosylation rates become equal; afterwards, the reaction of hydrolysis will prevail with the consequent reduction in GOS concentration (Playne and Crittenden 2009; Torres et al. 2010). Therefore the time at which the reaction must be stopped is critical and should be as close as possible to the one at which GOS yield is maximal. At this point, it is appropriate to define the two most used parameters for assessing the reaction performance: yield and productivity. There are several definitions of GOS yield in the literature, however the most frequently used is the ratio between the mass of GOS obtained and the initial mass of lactose; this definition will be the one employed in the present review. Typically, productivity is defined as the mass of GOS produced per unit of reaction volume and unit of time. Both parameters are usually evaluated at the point of maximum yield. The most relevant variables determining GOS yield are the initial concentration of lactose, temperature and enzyme origin whose effects will be shortly reviewed. The effects related to the GOS synthesis in non-conventional media and the reactor configuration will be also discussed.

Effect of the substrate concentration

It has been consistently proved that GOS yield increases at higher concentrations of lactose. Lactose solubility at moderate temperatures (20 °C to 60 °C) is in the range from 20 to 30 % w/w (Vera et al. 2014), which is rather low in comparison with other sugar. However, higher concentrations of lactose are attainable using supersaturated lactose solutions (Vera et al. 2012). For instance, Neri et al. (2009a) reported an increase in GOS yield from 11.2 to 26.1 % when increasing lactose initial concentration from 5 to 40 % w/w. However, at lactose concentrations close to 2.1 times saturation, precipitation occurred spontaneously producing a decrease in GOS yield. Vera et al. (2012) showed that increasing lactose initial concentration from 40 to 60 % w/w produced a reduction in GOS yield from 28 to 15 %, because of the reduction in soluble lactose concentration. Lactose initial concentration also affects the polymerization degree of GOS allowing the control of the proportion of GOS-3, GOS-4 and GOS-5 in the final product (Palai et al. 2012; Rodriguez-Colinas et al. 2014). On the other hand, most of the β-galactosidases are subjected to competitive inhibition by galactose (Gosling et al. 2010; Torres et al. 2010; Panesar et al. 2010), so high lactose concentrations also contribute to reduce the inhibitory effect of this monosaccharide. Galactose competitive inhibition may be severe in some cases; for instance, A. oryzae β-galactosidase has a Michaelis constant of 94 mM and an inhibition constant by galactose of 6 mM at 40 °C (Vera et al. 2011a). Albayrak and Yang (2002a) showed that a galactose concentration of 10 % w/w produced decreases of 15 and 85 % in GOS yield and productivity, respectively.

Effect of the temperature

Temperature is also a key variable affecting the reaction rates (hydrolysis and transgalactosylation), enzyme operational stability and lactose solubility altogether (Torres et al. 2010). Usually productivity of GOS synthesis increases with temperature (Neri et al. 2009b, 2011; Osman et al. 2010) because of the increase in reaction rate and in lactose solubility, which allows using higher lactose concentrations (Vera et al. 2012). However, temperature also increases enzyme inactivation rate: for instance, half-life of Kluyveromyces lactis β-galactosidase sharply decreased over 35 °C (Zhou and Chen 2001). In most cases, GOS yield presents a maximum value between 40 and 50 °C (Martínez-Villaluenga et al. 2008). For all the reasons stated above, the utilization of thermophilic β-galactosidases stands out as a technologically attractive alternative. Regrettably, these sources are not commercially available for industrial applications and are not sanitary certified for food and pharmaceutical applications. However, several thermophilic β-galactosidases and β-glucosidases have been evaluated in the synthesis of GOS (Torres et al. 2010; Panesar et al. 2010), showing dissimilar results. For instance, a yield of 19 % was reported using the β-galactosidase from Thermotoga maritima at 80 °C (Ji et al. 2005), while a yield close to 50 % was reported for Sulfolobus solfataricus β-galactosidase at temperatures between 75 to 80 °C (Park et al. 2008; Wu et al. 2013).

Effect of the enzyme origin

The origin of the β-galactosidase is probably the most important variable for GOS synthesis, determining yield, product composition and type of β-glycosidic bonds. Best values of operational variables, like pH and temperature, and kinetic behavior may vary significantly from one organism to another (Sako et al. 1999; Playne and Crittenden 2009; Torres et al. 2010). Table 1 lists some of the most reported microorganisms used for GOS synthesis, but the β-galactosidases from A. oryzae, B. circulans and K. lactis are the preferred ones being available as commercial preparations that have been customary used in the dairy industry (Sako et al. 1999; Playne and Crittenden 2009; Sanz-Valero 2009). The enzyme from A. oryzae is relatively cheap with a high specific activity and high transgalactosylation activity; pH optimum is around 4.5, operational temperature is in the range from 40 to 60 °C and maximum reported GOS yield is close to 30 % (Neri et al. 2009a; Sanz-Valero 2009; Vera et al. 2012). The commercial enzyme from B. circulans is actually a mixture of isoenzymes (Warmerdam et al. 2013); operational temperatures are also in the range from 40 to 60 °C and pH optimum is close to 6; maximum GOS yield is around 40 %, which is a definite advantage over the A. oryzae β-galactosidase and, despite being more expensive, complex and less stable, it has been a preferred choice for industrial production (Playne and Crittenden 2009; Torres et al. 2010). The enzyme from K. lactis has a pH optimum close to 6.5 and operational temperatures are in the range from 35 to 40 °C. GOS yields obtained are around 30 % but the product profile is quite different, with a high content of disaccharides whose prebiotic contribution has not been yet consistently proven (Martínez-Villaluenga et al. 2008; Sanz-Valero 2009; Fischer and Kleinschmidt 2015). Besides, yeast β-galactosidases have been shown to be more suitable for lactose hydrolysis than for oligosaccharide synthesis. The use of β-galactosidases from probiotic Bifidobateria and Lactobacilli is worth mentioning since it is to be expected that these enzymes may produce a stronger stimulatory effect on the intestinal healthy microbiota than those from unrelated organisms (Rabiu et al. 2001; Rastall and Maitin 2002). On the other hand, the β-galactosidases from thermophilic strains operate at temperatures around 80 °C and pH 6, reaching GOS yield over 50 % and high volumetric productivities (Park et al. 2008; Wu et al. 2013). Also, the β-galactosidase from Crypotoccus laurentii (permeabilized cells) is worth mentioning since currently Nissin Sugar Co., Ltd. is producing GOS (Cup-Oligo) with this enzyme .

Table 1 Synthesis of GOS with β-galactosidases from different origins
Full size table

Effect of non-conventional media and the reactor configuration

Since a low water activity depresses the hydrolytic of activity of β-galactosidases; high GOS yields have been observed in biphasic systems (Shin and Yang 1994; Bednarski and Kulikowska 2007). Several aqueous/organic solvent medium have been studied, being the best results attained by using a ratio cyclohexane to water of 95:5 (Shin and Yang 1994; Bednarski and Kulikowska 2007). Shin and Yang (1994) reported an increase of 19 % in GOS yield at the former condition with respect the aqueous system. Additionally, GOS synthesis by using reverse micelles was reported by Chen et al. (2001). In that work, an increase of the GOS yield from 31 to 55 % was achieved with AOT (dioctyl sodium sulfosuccinate)/isooctane reverse micelles system, in comparison to an aqueous media. However, these systems seem unsuitable for large-scale production of GOS: low water activities decrease the reaction rate as well as the carbohydrates solubility such as lactose (Cruz-Guerrero et al. 2006; Nikolovska-nedelkoska et al. 2009; Dong et al. 2015). Furthermore, organic solvents reduce the stability of β-galactosidase (Bednarski and Kulikowska 2007) and they must be removed from product, implying additional cost.

On the other hand, GOS synthesis is carried out mainly with soluble enzymes. However, this strategy has several disadvantages that can be overcome by using the enzyme in immobilized form. β-Galactosidase immobilization usually increases the enzyme stability, facilitates the enzyme reuse and its removal from the product, reduces the cost associated to the enzyme consumption and allows a more flexible reactor design (Albayrak and Yang 2002a; Panesar et al. 2010; Freitas et al. 2011). GOS synthesis has been studied using β-galactosidases immobilized by several methodologies. Among those, the immobilization in activated agarose (Huerta et al. 2011; Urrutia et al. 2013), in activated chitosan (Pan et al. 2009; Urrutia et al. 2014), in magnetic polysiloxane–polyvinyl alcohol beads (Neri et al. 2009a), in the form of self-supported cross-linked aggregates (Gaur et al. 2006), and in the form of whole permeabilized cells containing β-galactosidase (Sun et al. 2016) are prominent. The use of immobilized enzyme allows reactor operation according to the characteristics of the catalyst with the aim of maximizing GOS yield or productivity. Immobilized β-galactosidases have been used in sequential batch operation in stirred tank reactors and in continuous operation in packed-bed reactors (PBR) and to a lesser extent in continuously-operated stirred-tank reactors (CSTR). Benjamins et al. (2014) reported the synthesis of GOS in 15 sequential batches with B. circulans β-galactosidase immobilized in Eupergit C250L; the catalyst retained 60 % of its initial activity under the whole operation cycle, a GOS product of similar composition being obtained throughout the batches; productivity was 165 % higher than obtained with the soluble enzyme. Using a similar system, Huerta et al. (2011) conducted 10 sequential batches with A. oryzae β-galactosidase immobilized in glyoxyl agarose, with a 29 % conversion of lactose into GOS and invariable GOS composition in all batches; 75 % of its initial activity remained at the end of the tenth batch, after which a cumulative production of 130 g GOS/g catalyst was obtained, corresponding to 8500 g GOS/g of β-galactosidase. Extrapolating these results to catalyst replacement after 1 half-life, 23,000 g GOS/g of β-galactosidase could have been obtained in one cycle of reactor operation.

Continuous operation in PBR is a sound alternative which allows catalyst retention and prolonged use, but rather pure lactose solutions have to be used to prevent bed clogging (Nakkharat and Haltrich 2007). Using this type of reactor and B. circulans β-galactosidase immobilized in Eupergit C250L, Warmerdarm et al. (2014) observed that GOS yield obtained was similar than with the soluble enzyme in batch, but volumetric productivity was 6 times higher. Albayrak and Yang (Albayrak and Yang 2002a) used A. oryzae β-galactosidase immobilized in cotton cloth in a PBR obtaining high GOS volumetric productivities which increased with the increase in lactose concentration in the feed stream. Synthesis of GOS in CSTR with catalyst retention is not a preferred option because of kinetic considerations and has been seldom reported after the early work of (Mozaffar et al. 1986) where an increase in productivity from 20 to 40 % with respect to the soluble enzyme was reported. Spletchna et al. (2007) reported the synthesis of GOS in CSTR with an external crossflow ultrafiltration membrane using Lactobacillus reuteri β-galactosidase and compared it with conventional batch synthesis, but the focus was on the differences in GOS composition rather than in operational parameters. Ebrahimi et al. (2010) also reported the synthesis of GOS in a CSTR equipped with a ceramic membrane and highlighted the convenience of such system though recognizing that operation parameters still need to be optimized. More recently, Córdova et al. (2016a) designed and optimized an ultrafiltration membrane bioreactor (UF-MBR), using high lactose concentrations. The system allowed a significant increase in the amount of processed substrate, and a 2.44-fold increase in the amount of GOS produced per unit mass of catalyst was obtained with respect to a conventional batch system, concluding that the results can be improved further by tuning the membrane area/reaction volume ratio. These three cases represent a system where the enzyme is membrane-confined within the reactor rather than immobilized to a solid support. Synthesis of GOS in membrane reactors has also been reported by Botelho-Cunha et al. (2010) and Nath et al. (2013); in such cases the enzyme was attached to the membrane. This aspect will be analyzed in more detail within the context of GOS purification in Sect. 4.1. Finally, it should be considered that immobilized enzymes may be subjected to diffusional restrictions. Prenosil et al. (1987) observed a decrease in GOS yield from lactose (15 % w/w) as the result of the immobilization of Aspergillus niger β-galactosidase onto Duolite s-76. They suggested that the presence of internal restrictions in mass transfer may be responsible for the lower yields obtained, since values ranging from 5 to 10 were estimated for the Thièle modulus. However, in most cases the mass transfer limitations are negligible, because of the high substrate concentrations (>40 % w/w) employed in the synthesis (Albayrak and Yang 2002b; Neri et al. 2009a, b; Huerta et al. 2011). Table 2 presents a list with some illustrative examples of GOS synthesis with immobilized β-galactosidases. Detailed information about β-galactosidase immobilization can be found in comprehensive reviews on the subject by Panesar et al. (2010), Benjamins (2014) and Osman (2016).

Table 2 Synthesis of GOS with immobilized β-galactosidases
Full size table

Relationship between structure and functionality of GOS

Properties of GOS as active ingredients of well-established prebiotic condition in functional foods and as HMOS substitutes are mostly responsible for a rapidly expanding market (Crittenden and Playne 2009; Paterson and Kellam 2009). In 2013, GOS demand was supplied by four major companies: Friesland Foods Domo, Yakult Honsha, Nisin Sugar Manufacturing Co and Ingredion (Grand Review Research 2015). These companies offer GOS products with different degrees of purity and in different formats (concentrated syrup or solid powder). A list of most important GOS producers is presented in Table 3.

Table 3 Main GOS producing companies
Full size table

GOS are composed by variable number of galactose units and a terminal glucose residue, linked by β-glycosidic bonds (Gänzle 2012), mostly β-(1-4) and β-(1-6), and β-(1-3) to a lesser extent (Coulier et al. 2009). Prebiotic properties of GOS were initially attributed to the trisaccharides 4′-galactosyl-lactose and 6′-galactosyl-lactose (GOS-3) and the tetrasaccharide 6′-digalactosil-lactosa (GOS-4) (Gopal et al. 2001; Panesar et al. 2006). However, it has been recently reported that the disaccharides containing β-(1-6) bonds, this is, allolactose and galactobiose, present bifidogenic properties similar to GOS-3 (Rodriguez-Colinas et al. 2013, 2014). A list of GOS whose prebiotic condition has been demonstrated is listed in Table 4.

Table 4 Chemical structure of GOS reported as prebiotic
Full size table

Due to their β-glycosidic bonds, GOS are neither hydrolyzed nor absorbed in the upper intestinal tract, reaching intact the large intestine where they are selectively fermented by lactobacilli and bifidobacteria (Tzortzis and Vulevic 2009; Torres et al. 2010). Prebiotic effect of GOS is essentially indirect, since the stimulation of growth or activity of some microorganisms produces a healthy modification of the intestinal microbiota particularly in the colon (Rastall and Maitin 2002; Roberfroid 2007). So, it is not GOS per-se but the effect they produce that promotes health benefits to the consumer. There is abundant evidence that the increase in the bifidobacterial population stimulates the immune system, inhibits the growth of intestinal pathogens, enhances the production of vitamins B, reduces blood ammonia and cholesterol, and help in restoring the intestinal microbiota after antibiotic treatment (Gibson and Roberfroid 1995; Roberfroid 2007; Tzortzis and Vulevic 2009). It has been also reported that GOS promotes an increase in the concentration of lactobacilli, favors lactose digestion in lactose intolerant individuals, reduces constipation and infant diarrhea, help in preventing infections as those produced by Salmonella species, and alleviate irritable bowel syndrome (Manning and Gibson 2004; Wang 2009). In addition, other beneficial effects have been reported including protection against enteric diseases, increase in minerals absorption and immunomodulation in the prevention of allergies and intestinal inflammation, a trophic effect on the intestinal epithelium produced by the short-chain fatty acids (SCFA) resulting from GOS fermentation, an increase in fecal volume, and risk reduction of colon cancer by depressing toxigenic microbial metabolism (Chonan et al. 1996; Van den Heuvel et al. 2000; Sinclair et al. 2009; Vulevic et al. 2008; Nurmi et al. 2005; Macfarlane et al. 2008; Tzortzis and Vulevic 2009). Based on the understanding of the colon ecosystem, prebiotic effect has been traditionally defined in terms of the changes produced in the most representative intestinal microbial populations, namely bifodabacteria, bacteroides, lactobacilli and clostridia (Rycroft et al. 2001). On that basis, Palframan et al. (2003) proposed a parameter called “prebiotic index” (PI) which is a comparative relationship of the effect on growth of fecal beneficial (bifidobacteria and lactobacilli) and harmful bacteria (bacteroides and clostridia). A high PI means that the carbon source (GOS in this case) is selectively metabolized by beneficial bifidobacteria and lactobacilli and to a lesser extent by pathogenic bacteria, like clostridia (Rycroft et al. 2001; Sanz et al. 2005). This index has been thoroughly used as indicative of prebiotic condition (Palframan et al. 2003; Sanz et al. 2005; Ghoddusi et al. 2007; Vardakou et al. 2008). Li et al. (2015) reported a PI value of for 11.66 for GOS-3, much higher than the value of 5.05 for the fructo-oligosaccharide fructose–fructose glucose, and a much selective stimulation of bifidobacteria, as previously reported (Bouhnik et al. 2004; Depeint et al. 2008).

Simultaneous synthesis of GOS and other lactose-derived prebiotics

In general, the specificity of β-galactosidases with respect to the galactose acceptor molecule is low; therefore, sugars other than lactose, like fructose or sucrose, can act as acceptors (Kim et al. 2006; Li et al. 2009; Gänzle 2012; Guerrero et al. 2015). Not only carbohydrates can act as acceptors of the galactosyl group, but other hydroxyl-containing compounds, like alcohols, can as well (Stevenson et al. 1993; Klewicki 2000; Irazoqui et al. 2009). In this type of reactions, the product of synthesis contains the products of lactose hydrolysis, the GOS produced by transgalactosylation of lactose and the product of interest. Three valuable compounds can be obtained by this strategy, namely, lactulose, lactosucrose and lactitol. Lactulose is a well-established prebiotic while lactosucrose and lactitol are potential prebiotics with important applications as ingredients in functional foods and in non-food uses.

Lactulose (4-O-β-D-galactopyranosyl-d-fructose) is a synthetic disaccharide which is used mostly in the pharmaceutical field as a mild laxative for the treatment of constipation (Tamura et al. 1993), and also in the treatment of hepatic encephalopathy (Als-Nielsen et al. 2004). Lactulose, together with GOS, are the lactose-derived NDOs that are properly considered as prebiotics (Roberfroid 2007) so its use as a component in functional foods is promising. At present lactulose is produced exclusively by chemical catalysis. However, enzymatic synthesis is a more compatible option with green chemistry principles which is a powerful incentive for the gradual replacement of the chemical process. Lactulose can be produced by the enzymatic transgalactosylation of fructose with lactose catalyzed by β-galactosidase (Schuster-Wolff-Bühring et al. 2010; Guerrero et al. 2011). During the enzymatic synthesis of lactulose, GOS are synthesized along; however, it has been demonstrated that it is possible to drive the reaction to the synthesis of lactulose or GOS at will by varying the proportion of substrates (lactose and fructose) in the reaction medium (Guerrero et al. 2011; Guerrero et al. 2015). Production of pure lactulose is unattainable by this route because the fructose:lactose molar ratio required will be exceedingly high; however, being both compounds prebiotics, mixtures of them in defined proportions are quite interesting products. Furthermore, lactulose can also act as acceptor of transgalactosylated galactose so that oligosaccharides that have a terminal fructose instead of glucose, designated ad fructosyl-galacto-oligosaccharides (fGOS), can be produced (Martínez-Villaluenga et al. 2008; Rodriguez-Fernandez et al. 2011; Guerrero et al. 2013) which are interesting new generation potential prebiotics. Recently a quite appealing strategy to synthetize lactulose has appeared, which considers the direct isomerization of lactose into lactulose with cellobiose 2-epimerase from a genetically manipulated thermophilic strain of Caldicellulosiruptor saccharolyticus (Kim and Oh 2012) whose gene was cloned and expressed in Escherichia coli (Wang et al. 2015). This strategy may well turn competitive with chemical synthesis and adopted provided that the enzyme acquires status for safety use.

Lactosucrose (O-β-D-galactopyranosyl-(1,4)-O-α-glucopyranosyl-(1,2)-β-D-fructofuranoside) is a synthetic trisaccharide, being glucose, galactose and fructose its monosaccharide components. Despite having a bifidogenic effect and inhibiting Clostridia, its prebiotic effect is not well documented yet (Crittenden and Playne 2009), so for the moment it remains as a potential prebiotic candidate; however, its health-promoting effects have led to its industrial production in Japan as a functional food ingredient (Mu et al. 2013). Lactosucrose is mostly produced by lactose transfructosylation with β-fructofuranosidase (Pilgrim et al. 2001) or with levansucrase (Choi et al. 2004). However, it can also be produced by transgalactosylation of sucrose with lactose using β-galactosidase (Díez-Municio et al. 2014). During the synthesis with β-galactosidase, lactosucrose and GOS are produced simultaneously, the proportion of both being regulated by the ratio of the donor and the acceptor of the galactosyl group (Li et al. 2009).

Klewicki (2007a, b) studied the transgalactosylation of several polyhydroxyalcohols (sorbitol, xylitol, lactitol and erythritol). During these syntheses, GOS (mainly trisaccharides) are simultaneously produced (Klewicki 2000). Synthesis of galactosyl-sorbitol stands out because this compound is an epimer of lactitol (4-O-α-D-galactopyranosyl-D-sorbitol). Lactitol is a synthetic sugar alcohol obtained from lactose by catalytic hydrogenation of the glucose moiety in lactose (Soontornchai et al. 1999; Kuusisto et al. 2007), being a member of the family of polyhydroxyalcohols (mannitol, xylitol, arabitol and lactitol). Lactitol is a prebiotic candidate (Majumder et al. 2011) and its synbiotic effect with probiotics has been claimed by increasing the viability of the probiotic strains (Playne and Crittenden 2009), but current use in the food and pharmaceutical industries refers to its functional properties, being a low-calorie sweetener apt for diabetic persons and a mild laxative.

Downstream processing of GOS

One of the distinctive features of GOS synthesis is the rather low yields attained. GOS yields are usually between 20 and 45 % for the most used β-galactosidases (it may be higher in the case of thermophilic enzymes) corresponding to lactose conversions between 40 and 60 % (Torres et al. 2010). Efforts for substantially increasing GOS yield by biocatalyst or medium engineering have been for the most part unsuccessful. Therefore, purification of the product of synthesis (raw GOS) to remove the accompanying monosaccharides (glucose and galactose) and the unreacted lactose is a major concern in GOS production, since downstream processing costs represent a substantial fraction of the total production cost.

Once the maximum GOS concentration is attained, reaction is stopped by enzyme inactivation (in the case of soluble enzyme) or removal from the reaction medium (in the case of immobilized enzyme). Enzyme inactivation can be done by a drastic reduction in pH (Environ 2007; Soni and Associates 2014) or by an increase in temperature (GTC Nutrition 2009; Spherix Consulting and Inc. 2010; International Food Focus Ltd 2013). When using immobilized enzymes reaction is stopped by simply removing the solid catalyst which can be done using basket-type reactors or by rapid filtration of the reacted medium; this is simpler when scaling-up to production level, being a definite advantage of using immobilized enzymes, besides the obvious advantage of catalyst reuse in sequential batch operation. Once the immobilized enzyme (or the inactivated enzyme when used in soluble form) has been removed, usually by filtration, the product is subjected to color removal by adsorption in activated carbon, demineralization in ion-exchange columns and sterilization by ultraviolet radiation or by microfiltration. The refined product is then concentrated to produce a 70-75°Brix syrup with water activity around 0.75 (Environ 2007; Spherix Consulting and Inc. 2010; Clasado 2013), or spray-dried to obtain a product in powder form (Clasado 2013; GTC Nutrition 2009; Soni and Associates 2014). In the latter case a carrier like maltodextrin needs to be added for proper drying (Clasado 2013; GTC Nutrition 2009; Spherix Consulting and Inc. 2010; Soni and Associates 2014).

Removal of monosaccharide and residual lactose will be in most cases necessary, especially when high purity GOS is required for special uses. Removal of lactose is important for avoiding crystallization (Gänzle 2012) and allowing consumption by lactose intolerant people (Mattar et al. 2012). On the other hand, monosaccharides contribute with an undesirable caloric content, are cariogenic and make the product inadequate for diabetic people (Hernández et al. 2009). Purification of raw GOS is then a central aspect of the production technology. Ongoing technology for GOS purification is simulated moving bed (SMB) chromatography, which is expensive and complex (Vanneste et al. 2011; Kovács et al. 2014); therefore, considerable effort has been spent in developing alternative technologies for GOS purification that may become soon the leading edge for competitiveness. Most promising strategies for GOS purification are reviewed in the next sections.

GOS purification by membrane technology

Not being a mature technology yet, the use of membrane nanofiltration (NF) appears as a sound technological alternative for GOS purification (Gosling et al. 2010). The intended purpose is the retention of the oligosaccharides in the concentrate and the removal of monosaccharides, and hopefully residual lactose, in the permeate stream (Goulas et al. 2002). The high osmotic pressure and viscosity of highly concentrated carbohydrate solutions are the main constraints of the separation process (Schäfer et al. 2005; Córdova et al. 2016b). Also, flux decay due to membrane fouling limits its economic viability (Ruby Figueroa et al. 2011). Critical operational variables, whose effect on performance need to be carefully evaluated are: effective transmembrane pressure (TMPe), temperature, flow regime, materials and type of configuration used (Pinelo et al. 2009). Also, the molecular weight cut-off value (MWCO), the pore size distribution and the morphology of the membrane will directly affect the performance of the NF system (Wang and Chung 2005; Carvalho et al. 2011; Michelon et al. 2014). The separation of compounds of similar molecular weight, like lactose and GOS-3, is a big challenge when a significant purification is required (Pinelo et al. 2009). To such purpose, NF membranes may not be selective enough considering that GOS-3 is the predominant GOS species in raw GOS (Frenzel et al. 2015). Therefore, a reasonable task should be the effective removal of monosaccharides. Since the origin of the β-galactosidase is a very important variable for GOS synthesis regarding to product yield, this also affects the initial composition of the raw GOS obtained. Therefore, the type of β-galactosidase used during the synthesis has a direct impact in the purification factor (PF) that can be achieved at the end of the operation, and must be considered for designing the operation of NF (Torres et al. 2010).

Table 5 summarizes the most relevant results reported in the last decade and the pioneering work of Goulas et al. (2002), who evaluated the purification of diluted commercial GOS by NF in a TMP range from 7 to 27 bar. As expected, the increase in TMP allowed a flux increase but such increase resulted in more solute compaction in the membrane, reducing the effective pore size with the consequent increase in oligosaccharides retention, but also of disaccharides and monosaccharides, with a stronger effect on the lower molecular weight compounds so altering the membrane selectivity (Goulas et al. 2002; Ren et al. 2015). Even so, 38 % of the initial lactose was removed. Such results were later improved by Feng et al. (2009) using continuous diafiltration and 52.5 % of the lactose and 90.5 % of the monosaccharides were removed.

Table 5 Comparison of operation conditions and performance in the purification of raw GOS by nanofiltration
Full size table

NF of GOS has been usually conducted at solute concentrations lower than 100 g/L. However, GOS synthesis must be performed at concentrations higher than 300 g/L of lactose for depressing the competing reactions of hydrolysis (Vera et al. 2012), which implies the undesirable dilution of the raw GOS. Considering this, Córdova et al. (2016b) evaluated the performance of different commercial NF membranes using concentrated raw GOS (~460 g/L), obtaining a high and stable flux at 53.5 °C; higher temperatures allowed reducing the viscosity, increasing flux and reducing membrane fouling, as well microbial contamination (Bucs et al. 2014; Pruksasri et al. 2015). However, this was obtained at the expense of an increase in membrane pore size so that a significant portion of GOS was also permeated reducing the global performance of purification (Pruksasri et al. 2015). Some strategies worthwhile exploring for improving purification at productive level are the use of membrane cascade systems (Patil et al. 2014) and NF at temperatures as low as 5 °C (Pruksasri et al. 2015). Considering the existence of a sieving effect of temperature on the NF membranes (Sharma et al. 2003), the use of low temperatures during NF allows increasing the rejection coefficient of GOS, improving the membrane selectivity towards a better separation from lactose and monosaccharides, as well as reducing the likelihood of microbial contamination; however, the operation time should be increased significantly with respect to those filtrations conducted in the range from 30 °C to 60 °C, because of the low flux obtained at such low temperature condition. In this sense, NF operations must be optimized for each specific case in order to solve the compromise between the maximization of the permeate flux and the desired rejection of GOS and removal of monosaccharides and lactose.

A quite interesting strategy has been proposed consisting in the simultaneous synthesis and purification of GOS using membrane bioreactors (Botelho-Cunha et al. 2010), where the enzyme is immobilized in the NF membrane so that the in situ synthesized GOS are selectively retained while the contaminant sugars are permeated. This also allows some increase in GOS yield because of the removal of galactose, which is a reputed β-galactosidase inhibitor (Botelho-Cunha et al. 2010; Nath et al. 2013); however, mass transfer limitations and a significant reduction of membrane permeability are the drawbacks of this system (Palai et al. 2012). Separation of the catalyst from the reacting mixture can be done in membrane ultrafiltration bioreactors (UF-MBR) with the soluble enzyme under recirculation (Córdova et al. 2016a); this strategy allows increasing GOS yield but the raw GOS still needs to be purified, which can be attained by coupling a subsequent NF stage (Ren et al. 2015). Even though GOS purity attainable by NF does not exceed 65 %, this is within the purity range of several commercial GOS products (Gosling et al. 2010), so this is a simple, mild and fast purification process compatible with biologically active molecules that may be considered as a valid alternative to chromatographic purification (Drioli 2004; Nordvang et al. 2014).

Selective bioconversion

Selective fermentation (bioconversion) has been used for GOS purification using mostly yeast strains from the genera Kluyveromyces and Saccharomyces, being also a plausible technological alternative (Cheng et al. 2006; Li et al. 2008; Guerrero et al. 2014). The basis of this strategy is the selective removal of the metabolizable sugars (monosaccharides plus lactose, or monosaccharides only) from raw GOS by yeast fermentation (bioconversion).

Saccharomyces cerevisiae

Yoon et al. (2003) evaluated the specificity of S. cerevisiae in the removal of the different carbohydrates present in raw GOS during 24 h of fermentation, pointing out to the removal of unwanted monosaccharides (glucose and galactose) by anaerobic glycolysis, obtaining ethanol and glycerol as side products. Goulas et al. (2007) evaluated the removal of monosaccharides with the same yeast of a commercial GOS product (VIVINAL®) and a raw GOS with 450 g/L of total carbohydrates produced by lactose transgalactosylation with Bifidobacterium bifidum NCIMB 41171 cells. GOS and lactose content remained unaltered, while glucose was completely metabolized to ethanol and CO2; galactose was metabolized to a much lower extent than glucose, which is different from the results reported by Hernández et al. (2009) where galactose was completely metabolized after 10 h of fermentation, but they used a lower carbohydrate concentration.

Li et al. (2008) using 20 % w/w GOS synthesized with Penicillium expansum β-galactosidase and calcium alginate immobilized S. cerevisiae cells, reported the total consumption of glucose after 1 h and galactose after 4 h of reaction, increasing GOS purity from 28.7 % to 39.4 % with the advantage that immobilized cells could be reused for 19 sequential batches. Recently, Aburto et al. (2016) reported the simultaneous synthesis and purification of GOS using A. oryzae β-galactosidase and S. cerevisiae cells, obtaining a product with 39.7 % GOS, 0.7 % glucose, 19.2 % galactose and 40.3 % lactose after 24 h of reaction; purity attained was similar than obtained by Li et al. (2008); however, GOS yield was higher than obtained in the conventional synthesis of GOS with A. oryzae β-galactosidase (Vera et al. 2012). Even though the purity obtained with S. cerevisiae is modest since lactose is not consumed, it has the advantage of being commercially available at low cost (bakers’ yeast and even spent brewers’ yeast can be used), having GRAS or equivalent safe status, and good tolerance to low pH (which reduces microbial contamination without the addition of acidulants) and inhibitory metabolites from fermentation (Borodina and Nielsen 2014).

Purity of GOS can be significantly increased if lactose is partially removed by a pre-hydrolysis step that can be conveniently done using β-galactosidases from Kluyveromyces strains which are reputedly good hydrolytic enzyme. Santibáñez et al. (2016) reported a 70 % reduction of lactose in crude GOS by a pre-hydrolysis step using a commercial K. lactis β-galactosidase; this operation can be coupled to the bioconversion with S. cerevisiae cells or NF to obtain a highly purified GOS.

Kluyveromyces strains

Kluyveromyces cells have the definite advantage of metabolizing monosaccharides and lactose as well since they possess the LAC/GAL regulon (Rubio-Texeira 2006). In the presence of glucose, this regulon is catabolite-repressed while in its absence LAC: LAC12 and LAC4 genes involved in lactose uptake and hydrolysis are expressed. Cheng et al. (2006) reported GOS purities of 97-98 % when using 20 % w/w raw GOS (synthesized with Bacillus sp. β-galactosidase) supplemented with yeast extract as medium for Kluyveromyces marxianus fermentation. Lactose consumption was observed after 12 h of reaction once the monosaccharides were almost completely consumed; GOS consumption was observed after 30 h. Similar results were reported by Li et al. (2008) using 20 % w/w raw GOS (synthesized with P. expansum F3) as medium for fermentation with alginate-immobilized K. lactis cells. Glucose was consumed after 6 h of reaction while lactose and galactose consumption proceeded up to 18 h, reaching a GOS purity of 97.5 %; cells could be recycled once without affecting product purity. On the other hand, Guerrero et al. (2014) reported a GOS purity of 95 % using undiluted and unsupplemented raw GOS (50 % w/w) as medium for bioconversion with K. marxianus cells, and a purity close to 100 % when using raw GOS diluted to 20 % w/w; reactions were conducted for 24 h.

Table 6 summarizes the results obtained in terms of GOS purity with both types of yeast. Even though high purity is attainable, selective fermentation (bioconversion) has some important drawbacks, namely, the high biomass to carbohydrate mass ratio required, the dilution of the raw GOS (Cheng et al. 2006; Li et al. 2008; Guerrero et al. 2014) and the concomitant production of metabolites like ethanol, acetic acid and glycerol that reduces the biomass yield and product purity (Beudeker et al. 1990; Goulas et al. 2007).

Table 6 Operating conditions and final purity obtained by selective fermentation (bioconversion) of raw GOS with yeast cells
Full size table

Other technologies for GOS purification

Simulated moving bed (SMB) chromatography has been utilized by decades and up to now in the sugar, pharmaceutical and petrochemical industries (Nicoud 2000; Wiśniewski et al. 2013), being one of the most promising techniques for carbohydrate fractionation since it allows continuous operation and increased productivity with reduced eluent consumption with respect to conventional chromatography. Its main drawback is its high cost and conceptual complexity (Nicoud 2000; Vanková et al. 2008; Vanneste et al. 2011). Wiśniewski et al. (2013) reported that SMB chromatography is capable of providing a raffinate with a GOS purity of 99.9 % and an extract containing exclusively lactose and monosaccharides. These results demonstrate that SMB technology not only allows high product purity but also high process yields. However, flow velocities ranging from 0.5 to 4 cm/min are needed so that quite big chromatographic columns are required at production level. Other technologies, like selective adsorption in activated carbon (Hernández et al. 2009) and selective precipitation with ethanol (Sen et al. 2011) have also been proposed for GOS purification; for the moment their scale-up to industrial level seems doubtful since considerable dilution of the raw GOS is required to obtain acceptable purification levels.

Concluding remarks

Among high added value products obtained from whey lactose, GOS outstands for being a well reputed prebiotic with unique properties as HMOS substitute. GOS are hardly synthesized by chemical catalysis, so that enzymatic synthesis is a must from a technological perspective, but also for compliance with green chemistry principles.

Several aspects of GOS production still need improvement which makes research in this area quite exciting and challenging. The most important refers to GOS yield which is low due to the subtleties of this kinetically-controlled reaction. Most efforts for increasing GOS yields have been unsuccessful, but protein engineering techniques may help in obtaining mutant enzymes with a better transglycosylation to hydrolysis profile. GOS yields are higher with thermophilic β-galactosidases, but the problem remains of isolating thermophilic strains accepted as safe for use in food and medicine; thermophilic genes may be cloned and expressed in safe strains and this is also an avenue for further improvement.

Enzymes currently used for GOS synthesis are rather low-price hydrolases so that industrial processes used them in soluble form; immobilized enzymes, despite their obvious advantages have for the moment little relevance. This situation may change if better but highly priced enzymes are developed, and if competition gets stronger as new partners appear.

Purification will still be a key issue in GOS production as long as modest yield are obtained. Strategies for raw GOS purification challenging current chromatographic operations should be considered with caution. Purification of raw GOS by membrane fractionation is a complex task due to the narrow range of molecular size between the product (GOS) and the contaminants (mostly lactose and monosaccharides). Effective removal of monosaccharides is a reasonable goal but removal of lactose will require an enzymatic pre-hydrolysis step with the consequent reduction in productivity and increase in cost. Fermentation (bioconversion) with yeast strains is quite appealing, but a high yeast biomass to lactose ratio is required for efficient removal of GOS contaminants. An economic evaluation of these strategies in particular situations will shed light as to their real industrial potential.

Cost of lactose-containing raw materials will remain a most important component of GOS production cost and this may vary considerably in time and place. Purified lactose, whey permeate or whole whey are options as raw materials for GOS synthesis whose pros and cons should be analyzed in each particular situation. When a highly purified product is required the use of purified lactose will be the best option since the industrial lactose-containing byproducts will need to be purified very much as in the industrial production of lactose so that the economy of scale will probably play against their use. On the other hand, the use of surplus whey may represent an incentive, especially for medium to small scale cheese factories where whey drying and commercialization may not be an option, so that an adequate disposal of it becomes a problem that may be solved by whey upgrading.