A Three-Step Process for the Bioconversion of Whey Permeate into a Glucose-Free D-Tagatose Syrup

We have developed a sustainable three-stage process for the revaluation of cheese whey permeate into D-tagatose, a rare sugar with functional properties used as sweetener. The experimental conditions (pH, temperature, cofactors, etc.) for each step were independently optimized. In the first step, concentrated whey containing 180–200 g/L of lactose was fully hydrolyzed by β-galactosidase from Bifidobacterium bifidum (Saphera®) in 3 h at 45 ◦C. Secondly, glucose was selectively removed by treatment with Pichia pastoris cells for 3 h at 30 ◦C. The best results were obtained with 350 mg of cells (previously grown for 16 h) per mL of solution. Finally, L-arabinose isomerase US100 from Bacillus stearothermophilus was employed to isomerize D-galactose into D-tagatose at pH 7.5 and 65 ◦C, in presence of 0.5 mM MnSO4. After 7 h, the concentration of D-tagatose was approximately 30 g/L (33.3% yield, referred to the initial D-galactose present in whey). The proposed integrated process takes place under mild conditions (neutral pH, moderate temperatures) in a short time (13 h), yielding a glucose-free syrup containing D-tagatose and galactose in a ratio 1:2 (w/w).


Introduction
D-Tagatose is one of the most promising low-calorie functional sweeteners [1]. It is a ketohexose, namely a C4 epimer of D-fructose. It is a rare sugar only found at small quantities in the gum exudates of the tropical tree Sterculia setigera and in several dairy products, e.g., in Ultra-High-Temperature (UHT) sterilized cow's milk [2]. It is heat tolerant, very stable at pH values between 2.0 to 7.0, and highly soluble in water (58% w/w at room temperature). It possesses a sucrose-like taste with 92% of its sweetness but contributing only 1.5 kcal/g (38% compared to sucrose) [3], with no cooling effect or aftertaste [4]. D-Tagatose was approved as a novel food in the European Union [3] and has obtained GRAS status by FDA in USA [5]. It is also approved in many other countries and is being widely used in foods, beverages, and dietary supplements [6,7].
Only 20% of the ingested D-tagatose is absorbed in the small intestine [8]. Since the remaining 80% seems to favourably modulate the composition of the gut microbiota [9], this sugar has been proposed as a potential prebiotic. In addition, D-tagatose is able to modulate lipid metabolism, minimize tooth It is well reported the effect of metal ions on the activity and stability of many enzymes, including those involved in the transformation of carbohydrates, and in particular in the two enzymes involved in the present work: β-galactosidase [40] and L-arabinose isomerase [41]. In particular, several members of these two families are activated by divalent cations such as Co 2+ , Mn 2+ and Mg 2+ [21,42]. For that reason, we measured by semiquantitative ICP-MS the composition of the concentrated whey permeate employed in this work. Table 1 summarizes the concentration of metal ions present at higher amounts than 0.1 ppm. As shown, the concentration of Mg 2+ was significant (98 mg/L) but the presence of Co 2+ and Mn 2+ was negligible.

Hydrolysis of Whey Permeate by β-Galactosidase from Bifidobacterium bifidus
The concentration of lactose in the cheese whey permeates employed in this work varied between 180-200 g/L depending on the batch, as determined by HPAEC-PAD. The pH of whey was adjusted to 6.8. We selected Bifidobacterium bifidum β-galactosidase (Saphera ® , Novozymes, Bagsvaerd, Denmark) for the hydrolysis of lactose in whey because we recently observed that at lactose concentrations lower than 200 g/L the main reaction catalyzed by this enzyme is the hydrolysis with negligible transgalactosylation [43]. In order to standardize the dose of enzyme, we measured the activity of Saphera with o-nitrophenyl-β-D-galactopyranoside (ONPG). The activity was 1506 ± 0.1 U/mL at 40 • C and pH 6.8, which are the typical preferred conditions for several β-galactosidases from Bifidobacteria [44,45].
First, we analyzed the effect of temperature on lactose hydrolysis in whey permeate. We assayed three different temperatures (30,40 and 45 • C) using 2.5 µL (3.75 ONPG units) per mL of whey. HPAEC-PAD showed that the fastest reaction was the one performed at 45 • C ( Table 2). Under these conditions, all the lactose is hydrolyzed in 3 h. We further increased the enzyme concentration up to 7.50 ONPG units per mL but the improvement in the hydrolysis process was not substantial. In order to minimize the cost for the process, we selected the lowest concentration of this enzyme.

Study of the Growth Time and Concentration of Pichia pastoris for Elimination of D-Glucose
Pichia pastoris is capable of using D-glucose, glycerol and methanol as carbon sources but not D-galactose [46]. In this context, Avila-Fernandez et al. successfully employed P. pastoris cells to selectively eliminate glucose and fructose in a syrup of fructooligosaccharides obtained by agave fructan hydrolysis [47]. Based on such background, the elimination of D-glucose using P. pastoris cells was postulated for the second step of the reaction.
We first developed a simple and fast test to evaluate the effect of P. pastoris growing time on the further consumption of glucose. Thus, samples from a P. pastoris culture were taken every two hours (after an initial lag phase of 6 h), and the absorbance at 600 nm (as turbidity measurement) of all samples was adjusted to the absorbance at 6 h, to have the same concentration of cells but with Catalysts 2020, 10, 647 4 of 14 different growing times. The P. pastoris cells were then mixed with a solution containing 1 g/L of both D-glucose and D-galactose, and the mixture was incubated for 10 min at 30 • C. The cells were separated by centrifugation and the concentration of both sugars was analyzed by HPAEC-PAD ( Figure 1). Our results showed that the most voracious cells were those corresponding to 12-16 h growing time. It is well reported that these times correspond the exponential phase of growth of P. pastoris [48].
Catalysts 2020, 8, x FOR PEER REVIEW 4 of 14 separated by centrifugation and the concentration of both sugars was analyzed by HPAEC-PAD ( Figure 1). Our results showed that the most voracious cells were those corresponding to 12-16 h growing time. It is well reported that these times correspond the exponential phase of growth of P. pastoris [48]. The next step was to determine the amount of cells (grown for 16 h) required to consume the glucose (90-100 g/L) released after lactose hydrolysis in whey permeate. Different amounts of Pichia pastoris cells (100-600 mg of wet weight) were added to 1 mL of a solution containing 100 g/L of Dglucose and 100 g/L D-galactose. The residual glucose was analyzed at 1, 2 and 3 h by HPAEC-PAD ( Figure 2). As illustrated, the glucose disappeared in one hour using 600 mg of P. pastoris cells per mL. However, adding 350 mg of cells per mL, the total disappearance of glucose took place in two hours, and the mass transfer limitations were less significant than with 600 mg cells per mL. On this basis, we selected 350 mg of cells per mL (3 h incubation) to assure the elimination of glucose in the integrated 3-step process for D-tagatose synthesis.

Effect of pH and Temperature on the Stability of L-Arabinose Isomerase
The L-arabinose isomerase (L-AI US100) from Bacillus stearothermophilus is a multimeric enzyme formed by four 56 kDa monomers [49]. Compared with other L-arabinose isomerases, L-AI US100 is relatively active at neutral and slightly acidic pH, and at moderate temperatures [50]. These properties are very valuable because at the typical conditions of alkaline pH and high temperature The next step was to determine the amount of cells (grown for 16 h) required to consume the glucose (90-100 g/L) released after lactose hydrolysis in whey permeate. Different amounts of Pichia pastoris cells (100-600 mg of wet weight) were added to 1 mL of a solution containing 100 g/L of D-glucose and 100 g/L D-galactose. The residual glucose was analyzed at 1, 2 and 3 h by HPAEC-PAD ( Figure 2). As illustrated, the glucose disappeared in one hour using 600 mg of P. pastoris cells per mL. However, adding 350 mg of cells per mL, the total disappearance of glucose took place in two hours, and the mass transfer limitations were less significant than with 600 mg cells per mL. On this basis, we selected 350 mg of cells per mL (3 h incubation) to assure the elimination of glucose in the integrated 3-step process for D-tagatose synthesis. separated by centrifugation and the concentration of both sugars was analyzed by HPAEC-PAD ( Figure 1). Our results showed that the most voracious cells were those corresponding to 12-16 h growing time. It is well reported that these times correspond the exponential phase of growth of P. pastoris [48]. The next step was to determine the amount of cells (grown for 16 h) required to consume the glucose (90-100 g/L) released after lactose hydrolysis in whey permeate. Different amounts of Pichia pastoris cells (100-600 mg of wet weight) were added to 1 mL of a solution containing 100 g/L of Dglucose and 100 g/L D-galactose. The residual glucose was analyzed at 1, 2 and 3 h by HPAEC-PAD ( Figure 2). As illustrated, the glucose disappeared in one hour using 600 mg of P. pastoris cells per mL. However, adding 350 mg of cells per mL, the total disappearance of glucose took place in two hours, and the mass transfer limitations were less significant than with 600 mg cells per mL. On this basis, we selected 350 mg of cells per mL (3 h incubation) to assure the elimination of glucose in the integrated 3-step process for D-tagatose synthesis.

Effect of pH and Temperature on the Stability of L-Arabinose Isomerase
The L-arabinose isomerase (L-AI US100) from Bacillus stearothermophilus is a multimeric enzyme formed by four 56 kDa monomers [49]. Compared with other L-arabinose isomerases, L-AI US100 is relatively active at neutral and slightly acidic pH, and at moderate temperatures [50]. These properties are very valuable because at the typical conditions of alkaline pH and high temperature

Effect of pH and Temperature on the Stability of L-Arabinose Isomerase
The L-arabinose isomerase (L-AI US100) from Bacillus stearothermophilus is a multimeric enzyme formed by four 56 kDa monomers [49]. Compared with other L-arabinose isomerases, L-AI US100 is Catalysts 2020, 10, 647 5 of 14 relatively active at neutral and slightly acidic pH, and at moderate temperatures [50]. These properties are very valuable because at the typical conditions of alkaline pH and high temperature the formation of undesired byproducts in bioconversions of sugars is quite common [51]. For these reasons, L-AI US100 was selected for the transformation of whey permeate into D-tagatose.
In order to establish the optimal conditions for the biotransformation, the stability of L-AI US100 towards pH was evaluated by pre-incubation (at room temperature) of the enzyme for 1 h at pH values between 3.0 and 10.0 employing 0.1 M Britton & Robinson buffer. After that, the L-arabinose isomerase activity was measured following the standard activity assay (at 65 • C and pH 7.5, see Experimental Section). The results shown in Figure 3A represent the relative activity of pre-treated L-AI compared with the non-incubated enzyme. As illustrated, the enzyme is very unstable at pH values below 6.0 and is notably stable at neutral and moderately alkaline pH values. the formation of undesired byproducts in bioconversions of sugars is quite common [51]. For these reasons, L-AI US100 was selected for the transformation of whey permeate into D-tagatose. In order to establish the optimal conditions for the biotransformation, the stability of L-AI US100 towards pH was evaluated by pre-incubation (at room temperature) of the enzyme for 1 h at pH values between 3.0 and 10.0 employing 0.1 M Britton & Robinson buffer. After that, the L-arabinose isomerase activity was measured following the standard activity assay (at 65 °C and pH 7.5, see Experimental Section). The results shown in Figure 3A represent the relative activity of pre-treated L-AI compared with the non-incubated enzyme. As illustrated, the enzyme is very unstable at pH values below 6.0 and is notably stable at neutral and moderately alkaline pH values. The thermal stability of the enzyme was determined by pre-incubating the L-AI US100 for 1 h at temperatures between 4 and 80 °C in 100 mM MOPS buffer (pH 7.5). Following this incubation, the L-AI activity was measured by the standard activity assay (at 65 °C and pH 7.5). As shown in Figure  3B, the enzyme is very stable at temperatures up to 65 °C, and was fast inactivated at temperatures higher than 70 °C.

Effect of Cofactors, pH and Temperature on the Activity of L-Arabinose Isomerase
The effect of metal cofactors on the activity of L-arabinose isomerase (L-AI US100) from B. stearothermophilus at 65 °C and neutral pH was assessed. Four different reactions were performed in absence or presence of 0.5 mM MnSO4 and/or 0.1 mM CoCl2. The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer (pH 7.5) at 65 °C for 4 h. Results are displayed in Figure 4.  The thermal stability of the enzyme was determined by pre-incubating the L-AI US100 for 1 h at temperatures between 4 and 80 • C in 100 mM MOPS buffer (pH 7.5). Following this incubation, the L-AI activity was measured by the standard activity assay (at 65 • C and pH 7.5). As shown in Figure 3B, the enzyme is very stable at temperatures up to 65 • C, and was fast inactivated at temperatures higher than 70 • C.

Effect of Cofactors, pH and Temperature on the Activity of L-Arabinose Isomerase
The effect of metal cofactors on the activity of L-arabinose isomerase (L-AI US100) from B. stearothermophilus at 65 • C and neutral pH was assessed. Four different reactions were performed in absence or presence of 0.5 mM MnSO 4 and/or 0.1 mM CoCl 2 . The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer (pH 7.5) at 65 • C for 4 h. Results are displayed in Figure 4.
As shown, CoCl 2 had a negligible effect on the yield of D-tagatose under these conditions. However, the production of D-tagatose in the presence of 0.5 mM MnSO 4 was approximately 40% higher compared with the control.
These results correlate well with previous findings on the effect of metallic ions on the L-AI US100 activity and thermostability. The effect of divalent cations on L-AI US100 must be particularly considered at temperatures higher or equal to 65 • C [49]. In fact, at temperatures above 65 • C both Mn 2+ and Co 2+ exert a significant influence on the thermostability of L-AI US100. Since a temperature of 65 • C was selected for the isomerization reaction, and considering that the conversion from D-galactose to D-tagatose was higher in presence of MnSO 4 in the reaction medium, we decided to maintain MnSO 4 but not CoCl 2 in the integrated process from whey to D-tagatose.

Effect of Cofactors, pH and Temperature on the Activity of L-Arabinose Isomerase
The effect of metal cofactors on the activity of L-arabinose isomerase (L-AI US100) from B. stearothermophilus at 65 °C and neutral pH was assessed. Four different reactions were performed in absence or presence of 0.5 mM MnSO4 and/or 0.1 mM CoCl2. The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer (pH 7.5) at 65 °C for 4 h. Results are displayed in Figure 4.  Furthermore, we analysed the effect of pH and temperature on the production of D-tagatose. We selected the range of pH (7.0-9.5) and temperature (40-65 • C) at which the L-arabinose isomerase from B. stearothermophilus was stable (see Figure 3). The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer for 4 h, in presence of 0.5 mM MnSO 4 . Results are shown in Figure 5. It remains clear that temperature substantially affects the reaction course. The highest production of D-tagatose in 4 h (7-8 g/L, 35-40% yield) was achieved at 65 • C. The reaction was 6-7 fold faster than at 40 • C. Regarding the pH, its effect was very much lower than that of temperature ( Figure 5). As shown, CoCl2 had a negligible effect on the yield of D-tagatose under these conditions. However, the production of D-tagatose in the presence of 0.5 mM MnSO4 was approximately 40% higher compared with the control.
These results correlate well with previous findings on the effect of metallic ions on the L-AI US100 activity and thermostability. The effect of divalent cations on L-AI US100 must be particularly considered at temperatures higher or equal to 65 °C [49]. In fact, at temperatures above 65 °C both Mn 2+ and Co 2+ exert a significant influence on the thermostability of L-AI US100. Since a temperature of 65 °C was selected for the isomerization reaction, and considering that the conversion from Dgalactose to D-tagatose was higher in presence of MnSO4 in the reaction medium, we decided to maintain MnSO4 but not CoCl2 in the integrated process from whey to D-tagatose.
Furthermore, we analysed the effect of pH and temperature on the production of D-tagatose. We selected the range of pH (7.0-9.5) and temperature (40-65 °C) at which the L-arabinose isomerase from B. stearothermophilus was stable (see Figure 3). The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer for 4 h, in presence of 0.5 mM MnSO4. Results are shown in Figure  5. It remains clear that temperature substantially affects the reaction course. The highest production of D-tagatose in 4 h (7-8 g/L, 35-40% yield) was achieved at 65 °C. The reaction was 6-7 fold faster than at 40 °C. Regarding the pH, its effect was very much lower than that of temperature ( Figure 5). Considering the results on the effect of pH, temperature and metal cofactors on the stability and activity of L-AI US100, we selected pH 7.5, 65 °C and 0.5 mM MnSO4 as the optimal experimental conditions for the integrated process. These results represent a compromise between activity and stability of the enzyme and correlate well with previous studies reported with this enzyme by using fast spectrophotometric assays [49,50,52] rather than chromatographic analysis by HPAEC-PAD performed in this work.

Sequential Biotransformation of Whey Permeate into D-Tagatose
We propose a three-step process for D-tagatose synthesis from whey permeate based on the experimental conditions that were previously optimized. In the first stage, the lactose (180-200 g/L) contained in whey is hydrolyzed by β-galactosidase from Bifidobacterium bifidum (7.5 U/mL). Total Considering the results on the effect of pH, temperature and metal cofactors on the stability and activity of L-AI US100, we selected pH 7.5, 65 • C and 0.5 mM MnSO 4 as the optimal experimental conditions for the integrated process. These results represent a compromise between activity and stability of the enzyme and correlate well with previous studies reported with this enzyme by using fast spectrophotometric assays [49,50,52] rather than chromatographic analysis by HPAEC-PAD performed in this work.

Sequential Biotransformation of Whey Permeate into D-Tagatose
We propose a three-step process for D-tagatose synthesis from whey permeate based on the experimental conditions that were previously optimized. In the first stage, the lactose (180-200 g/L) contained in whey is hydrolyzed by β-galactosidase from Bifidobacterium bifidum (7.5 U/mL). Total hydrolysis is achieved in 3 h at 45 • C. Figure 6 shows a HPAEC-PAD chromatogram illustrating the complete disappearance of lactose in the reaction medium. In the second step, the hydrolyzate is treated with Pichia pastoris cells (350 mg of cells per mL, obtained as described elsewhere) during 3 h at 30 °C. As shown in Figure 6, all the glucose is metabolized by the yeast and the amount of D-galactose remains intact. After centrifugation to remove the cells, the pH of supernatant is adjusted to 7.5 (because the medium is acidified during the treatment with P. pastoris) and MnSO4 is added to reach a final concentration of 0.5 mM. These adjustments prepare the mixture for the third step.
Finally, L-arabinose isomerase L-AI US100 is added (0.65 U per mL in the mixture, measured by the standard assay). The reaction is incubated at 65 °C and different aliquots are taken to follow the progress of the reaction. As depicted in Figure 6, D-tagatose is formed and the final product contains a mixture of D-tagatose and D-galactose. Figure 7 shows the progress of D-tagatose formation over time. The concentration of D-tagatose increases rapidly during the first four hours and then tends to stabilize in approximately 30 g/L. Starting of a whey permeate with 180 g/L lactose, the yield of D-tagatose referred to lactose was 16.7%, and 33.3% referred to the D-galactose present in whey. We believe that this yield could be improved by using a higher concentration of L-AI thus minimizing the enzyme inactivation effects (see Figure  3). In the second step, the hydrolyzate is treated with Pichia pastoris cells (350 mg of cells per mL, obtained as described elsewhere) during 3 h at 30 • C. As shown in Figure 6, all the glucose is metabolized by the yeast and the amount of D-galactose remains intact. After centrifugation to remove the cells, the pH of supernatant is adjusted to 7.5 (because the medium is acidified during the treatment with P. pastoris) and MnSO 4 is added to reach a final concentration of 0.5 mM. These adjustments prepare the mixture for the third step.
Finally, L-arabinose isomerase L-AI US100 is added (0.65 U per mL in the mixture, measured by the standard assay). The reaction is incubated at 65 • C and different aliquots are taken to follow the progress of the reaction. As depicted in Figure 6, D-tagatose is formed and the final product contains a mixture of D-tagatose and D-galactose. Figure 7 shows the progress of D-tagatose formation over time. The concentration of D-tagatose increases rapidly during the first four hours and then tends to stabilize in approximately 30 g/L. Starting of a whey permeate with 180 g/L lactose, the yield of D-tagatose referred to lactose was 16.7%, and 33.3% referred to the D-galactose present in whey. We believe that this yield could be improved by using a higher concentration of L-AI thus minimizing the enzyme inactivation effects (see Figure 3).
Wanarska et al. reported a 30% yield of D-tagatose (referred to D-galactose) using a strain of Pichia pastoris that co-expressed L-AI and β-galactosidase, starting of a whey permeate containing 110 g/L lactose [37]. Zheng et al. employed a whey with 110 g/L lactose to get 43.6% yield of D-tagatose (referred to D-galactose) using a genetically engineered strain of Escherichia coli that expressed L-AI [32]. Similar results (40.4% yield) were reported by Xu et al. employing E. coli that co-expressed a β-galactosidase from Thermus thermophilus and L-arabinose isomerase from Lactobacillus fermentum [38,39]. Jayamuthunagai et al. reported 38% yield of D-tagatose (referred to D-galactose) from a hydrolyzed whey permeate containing 300 g/L D-galactose, using permeabilized and alginate-entrapped cells of Lactobacillus plantarum [36]. Shen et al. expressed xylose isomerase and L-AI in a strain of Corynebacterium glutamicum able to metabolize lactose. Starting from whey containing 98 g/L lactose, they obtained 20.4 g/L D-tagatose (44% yield referred to D-galactose) [33].
Finally, L-arabinose isomerase L-AI US100 is added (0.65 U per mL in the mixture, measured by the standard assay). The reaction is incubated at 65 °C and different aliquots are taken to follow the progress of the reaction. As depicted in Figure 6, D-tagatose is formed and the final product contains a mixture of D-tagatose and D-galactose. Figure 7 shows the progress of D-tagatose formation over time. The concentration of D-tagatose increases rapidly during the first four hours and then tends to stabilize in approximately 30 g/L. Starting of a whey permeate with 180 g/L lactose, the yield of D-tagatose referred to lactose was 16.7%, and 33.3% referred to the D-galactose present in whey. We believe that this yield could be improved by using a higher concentration of L-AI thus minimizing the enzyme inactivation effects (see Figure  3). A scheme summarizing the different steps of the integrated process is presented in Figure 8. Wanarska et al. reported a 30% yield of D-tagatose (referred to D-galactose) using a strain of Pichia pastoris that co-expressed L-AI and β-galactosidase, starting of a whey permeate containing 110 g/L lactose [37]. Zheng et al. employed a whey with 110 g/L lactose to get 43.6% yield of D-tagatose (referred to D-galactose) using a genetically engineered strain of Escherichia coli that expressed L-AI [32]. Similar results (40.4% yield) were reported by Xu et al. employing E. coli that co-expressed a βgalactosidase from Thermus thermophilus and L-arabinose isomerase from Lactobacillus fermentum [38,39]. Jayamuthunagai et al. reported 38% yield of D-tagatose (referred to D-galactose) from a hydrolyzed whey permeate containing 300 g/L D-galactose, using permeabilized and alginateentrapped cells of Lactobacillus plantarum [36]. Shen et al. expressed xylose isomerase and L-AI in a strain of Corynebacterium glutamicum able to metabolize lactose. Starting from whey containing 98 g/L lactose, they obtained 20.4 g/L D-tagatose (44% yield referred to D-galactose) [33].
A scheme summarizing the different steps of the integrated process is presented in Figure 8.

Elemental Analysis of Whey Permeate using Inductive Coupled Plasma Mass Spectrometry
Semiquantitative analysis of metals in whey permeate was performed on a NexION 300X Inductively Coupled Plasma-Mass Spectrometer (ICP-MS, PerkinElmer, Waltham, MA, USA) equipped with Universal Cell Technology, as previously described [54].

β-Galactosidase Activity Assay
The assay of β-galactosidase activity was performed using o-nitrophenyl-β-D-galactopyranoside (ONPG) as substrate. The activity was measured at 40 • C following o-nitrophenol (ONP) release at 405 nm using a microplate reader (Versamax, Molecular Devices, San Jose, CA, USA). The reaction was started by adding 10 µL of the enzyme solution (properly diluted) to 190 µL of 15 mM ONPG in 0.1 M sodium phosphate buffer (pH 6.8). The increase of absorbance at 405 nm was followed continuously at 40 • C during 5 min. The molar extinction coefficient of o-nitrophenol at pH 6.8 was 1627 M −1 cm −1 . One unit of activity (U) was defined as the corresponding to the hydrolysis of 1 µmol of ONPG per min under the above specified conditions.

Effect of Growth Time and Concentration of Pichia pastoris on the Elimination of D-Glucose
Pichia pastoris preculture was grown in 5 mL of 1% yeast extract, 1% peptone and 2% dextrose (all w/v) liquid medium (YEPD) at 30 • C and 1200 rpm for 24 h. After this time, the absorbance was measured, and 1 mL of preculture was added to a 250 mL flask containing 25 mL of YEPD, which was left growing at 30 • C and 1200 rpm. The first sample was taken at 6 h. The optical density (OD) was measured at 600 nm in a UV-1800 spectrophotometer (Shimadzu, Kioto, Japan) taking this OD as the reference value. The cells were centrifuged and washed three times with distilled water to remove the remaining medium from the cells. Samples were taken every 2 h until 24 h. Before centrifugation, all the samples were diluted to reach the reference value of absorbance, with the aim of having the same amount of cells in the experiments but with different growing times. Reactions with P. pastoris cells were performed adding 500 µL of a solution containing 1 g/L D-glucose and 1 g/L D-galactose to the cells. Reactions were incubated in a tube rotator (Argos Technologies Inc., Vernon Hills, IL, USA) for 10 min at 30 • C, then samples were centrifuged, the supernatant was inactivated at 95 • C for 10 min and analyzed by HPAEC-PAD. To determine the amount of cells required to consume the glucose expected in the sequential process for production of D-tagatose, different amounts of P. pastoris cells (300, 400, 500 and 600 mg of wet weight) were added to 1 mL of a solution containing 100 g/L D-glucose and 100 g/L D-galactose. Aliquots were withdrawn at 1, 2 and 3 h. After each extraction, the samples were centrifuged at 10,000 rpm for 3 min and the concentration of glucose and galactose in the supernatant was analyzed by HPAEC-PAD.

L-Arabinose Isomerase Activity Assay
The L-arabinose isomerase (L-AI) activity was determined adding the enzyme to a solution containing 20 g/L of D-galactose in 100 mM MOPS buffer (pH 7.5) containing 0.5 mM MnSO 4 and 0.1 mM CoCl 2 . The mixture was incubated at 65 • C for 4 h, and the reaction stopped in a water bath at Catalysts 2020, 10, 647 10 of 14 90 • C for 10 min. The concentration of D-tagatose was measured by HPAEC-PAD. One unit of L-AI activity (U) was defined as the corresponding to the formation of 1 µmol of D-tagatose per minute, under the conditions specified above.

Stability of L-Arabinose Isomerase
The stability of L-arabinose isomerase (L-AI US 100) from B. stearothermophilus towards pH was assessed pre-incubating the enzyme at different pH values (3.0 to 10.0) in 0.1 M Britton & Robinson buffer [55] at room temperature for 1 h. The thermal stability of the enzyme was determined by pre-incubating the L-AI at 4, 20, 30, 40, 50, 65, 70 and 80 • C in 100 mM MOPS buffer (pH 7.5) for 1 h. After incubation, the remaining L-arabinose isomerase activity was measured following the standard activity assay (Section 3.6). The activity of pre-treated L-AI US100 was compared with the activity of non-pre-incubated enzyme that was taken as control (100%).

Effect of Metals, pH and Temperature on the Activity of L-Arabinose Isomerase
To analyze the effect of metal cofactors on the activity of L-arabinose isomerase (L-AI US 100), different reactions were carried out in presence of 0.5 mM MnSO 4 and/or 0.1 mM CoCl 2 . The reactions were carried out with 20 g/L galactose in 100 mM MOPS buffer (pH 7.5) at 65 • C for 4 h, using 0.65 U/mL L-AI, and the concentration of D-tagatose was determined by HPAEC-PAD. A control reaction without cofactors was also performed. To determine the best reaction conditions for the enzyme, different values of pH (7, 7.5, 8, 8.5, 9, 9.5) and temperature (40,50,60 and 65 • C) were tested, in a reaction containing 20 g/L of D-galactose, 0.5 mM of MnSO 4 in 100 mM MOPS buffer (pH 7.5) and 0.65 U/mL L-AI. After 4 h, the reaction mixtures were analyzed by HPAEC-PAD.

Sequential Biotransformation of Whey Permeate into D-Tagatose
The biosynthesis of D-tagatose syrup from whey permeate was performed in three steps. Initially, the pH of whey was adjusted to 6.8. In the first stage, β-galactosidase from Bifidobacterium bifidum (5 µL, 7.5 U/mL measured with ONPG) was added to 2 mL of concentrated whey permeate (containing 180-200 g/L lactose, depending on the batch). The mixture was incubated for 3 h in an Envirogenie orbital stirrer (Scientific Industries Inc., Bohemia, NY, USA) at 45 • C. The second step involved the treatment of the hydrolyzate with 700 mg (wet weight) of Pichia pastoris cells (previously grown for 16 h) and the mixture was incubated in an orbital shaker (Orbitron, Infors HT, Surrey, UK) for 3 h at 30 • C. Then, the cells were removed by centrifugation (model 5810, Eppendorf) at 5000 rpm for 20 min at 4 • C. The supernatant was then separated from the cells and inactivated in a water bath at 90 • C for 10 min. In the third stage, the pH of supernatant was adjusted to 7.5 and MnSO 4 was added up to a final concentration of 0.5 mM. Then, 1.3 units of L-arabinose isomerase L-AI US100 were added (0.65 U/mL in the mixture). The reaction was incubated in a thermoshaker (model TS-100, Biosan, Riga, Latvia) at 65 • C and 1500 rpm. Aliquots were taken every 30 min, filtered on UltraFree centrifugal filters (0.45 µm, Millipore, Burlington, MA, USA) and analysed by HPAEC-PAD.

HPAEC-PAD Analysis
Sugar (D-galactose, D-glucose, D-tagatose, D-fructose and D-lactose) analysis was carried out by high performance anion-exchange chromatography coupled with pulsed amperometric detection (HPAEC-PAD) on an ICS3000 system (Dionex, Thermo Fischer Scientific Inc., Waltham, MA, USA) consisting of a SP gradient pump, an electrochemical detector with a gold working electrode and Ag/AgCl as reference electrode, and an autosampler (model AS-HV). All eluents were degassed by flushing with helium. A pellicular anion-exchange 4 × 250 mm Carbo-Pack PA-1 column (Dionex) connected to a 4 × 50 mm CarboPac PA-1 guard column was used at 30 • C. Eluent preparation was performed with MilliQ water, 50% (w/v) NaOH and sodium acetate trihydrate. The compounds were eluted by an isocratic method in which the mobile phase contained 10 mM NaOH and 2 mM sodium acetate, using a flow rate of 1 mL/min for 25 min. The peaks were analyzed using the Chromeleon software. Identification of the different carbohydrates was carried out employing commercially available standards.

Conclusions
The present work describes an environmentally friendly process for the bioconversion of cheese whey, a lactose-rich byproduct of the food industry, into D-tagatose, a rare sugar that has become one of the most promising low-calorie sweeteners in the market due to its functional properties. The methodology is based in the complete hydrolysis of lactose by a bifidobacterial β-galactosidase, followed by the selective removal of glucose with Pichia pastoris cells, and finally the isomerization of the remaining D-galactose into D-tagatose by L-arabinose isomerase from Bacillus stearothermophilus. The three steps were optimized independently in such a way that the integrated process is carried out in a short time (13 h) yielding 33.3% of D-tagatose (referred to the initial D-galactose). The total time could be even reduced by using a higher concentration of L-AI in the third step. One of the main advantages of the proposed project is its sustainability, as the three steps take place under mild conditions of pH (6.8-7.5), moderate temperatures (30-65 • C) and with completely biodegradable catalysts. The final syrup is free of glucose, and contains D-tagatose and D-galactose in a ratio 1:2 (w/w). This work opens new possibilities for the synthesis of D-tagatose and for the development of different reaction engineering strategies including enzyme immobilization.