Technical field of the invention

The present invention relates to the field of in vitro digestion assays which are used to determine the digestibility of new dietary sugars. More specifically, the present invention discloses the usage of a mix of microbial enzymes which mimic the enzyme activity of mammalian intestinal brush border enzymes. As microbial enzymes are much more stable and affordable than their mammalian counterparts, the present invention is useful within the food industry.

Background art

As more and more sugars become available for application in the food industry, the need arises for a rapid, reproducible and relevant digestion assay. Indeed, before the new sugars can reach the market, their improved and/or altered properties must be verified. ¹ To compete with the current market, novel sugars must have appealing properties that are associated with a series of beneficial physiological effects, such as low-calorie (important in preventing obesity), low-glycemic (helpful in managing diabetes and cardiovascular disease), and low- digestible (helpful in reducing the intestinal transit time and in positively modulating the gut microbiota composition and activity). ²

The digestion of carbohydrates is a complex and multistage process, involving carbohydrolyases produced by three different organs. ² a-Amylase secreted by the salivary gland and pancreas can hydrolyse starch and maltodextrins. ³ In addition, three a-glucosidases and one p-glycosidase can be found in the intestinal tract. The a-glucosidases comprise the sucrase-isomaltase complex (SI), maltase-glucoamylase complex (MGAM) and trehalase, whereas lactase-phlorizin hydrolase (LPH) is the sole p-glycosidase. ⁴ - ⁵

Several in vitro models have been proposed for the initial screening of novel food products, generally encompassing simulations of oral, gastric and intestinal digestion processes. Remarkably, a missing feature in most of these models are the mucosal disaccharidases embedded in the small intestinal brush border membrane. ² However, with a-amylase only the digestion of starch can be evaluated, which certainly is not a complete representation of the in vivo situation. ³ To broaden the applicability, extracts from the small intestines of rats (containing brush border enzymes) have been added, which enabled the ranking of other oligosaccharides based on their digestibility. ¹ Regardless of the recent advances in the development of in vitro methods, a standardized batch gastrointestinal digestion method is needed that is based on physiologically relevant conditions. Clearly, the addition of the a- glucosidase activity located at the small intestine brush border needs to be considered besides the a-amylases. The rat intestinal extracts, however, suffer from poor reproducibility and are not readily available on a large scale. ²

The present invention relates to enzyme mixtures that can mimic the specificity of mammalian intestinal a-glucosidases and discloses the ratio of microbial enzymes needed to mimic the activity of mammalian brush border enzymes at any given concentration of the possible a- glucobiose substrates, i.e. trehalose (a- 1,1), kojibiose (a-1,2), nigerose (a-1,3), maltose (a-1,4) and isomaltose (a- 1,6).

Brief description of figures

Figure 1: Maltase-glucoamylase (MGAM) and sucrase-isomaltase (SI) protein complex structures of the small intestinal lumen linked to the transmembrane domain (TMD) via an O-glycosylated linkage (O-link) and their specificities towards different substrates. Hexagon = glucose, pentagon = fructose.

Figure 2: Human intestinal mimic, (a) Michaelis-Menten curve of the combined activities in the human intestine (bold line) and in the mimic (dashed line). The calculations were performed at 28 mM leading to a mixture of four enzymes (grey box), (b) Deviation (%) profile of the mimic compared to the theoretical human intestine activity.

Figure 3: Rat intestinal mimic, (a) Michaelis-Menten curve of the combined activities in the rat intestine (bold line) and in the mimic (dashed line). The calculations were performed at 28 mM leading to a mixture of four enzymes (grey box), (b) Deviation (%) profile of the mimic compared to the theoretical rat intestine activity.

Figure 4: Distribution of the activities of each substrate expressed as percentage of the total activity at 28 mM for the theoretical human intestine and the experimentally determined rat intestine and both mimics. Description of invention

The present invention relates in first instance to the usage of a mix of microbial enzymes to mimic the enzyme activity of mammalian intestinal brush border enzymes in vitro.

The present invention relates in particular to the usage of multiple microbial enzymes in a cocktail whose overall activity has been tailored to mimic the overall activity of the enzymes in the mammalian intestinal brush border in vitro, wherein said mammalian intestinal brush border enzymes is/are sucrase-isomaltase (SI), maltase-glucoamylase (MGAM) and/or trehalase.

The term 'to mimic' as used herein means that a particular mix of microbial enzymes has the same or similar enzymatic activity (deviation less than 20 %) or effect as the enzymatic activity or effect of a mix of particular mammalian intestinal brush border enzymes.

The term 'has been tailored to mimic' means 'has been specifically designed or chosen to mimic the overall activity of the enzymes SI, MGAM and/or trehalase'. The term 'overall activity' means the combined activity towards each substrate which is attributed by all or some of the enzymes SI, MGAM and/or trehalase.

The term 'a mix of microbial enzymes' or 'multiple microbial enzymes in a cocktail' as used herein means a combination of 2 or more (3, 4, 5, 6, 7, 8, 9, 10...) enzymes which can be found and are present in bacteria, yeast and/or algae.

The meaning of the terms 'the enzyme activity of mammalian intestinal brush border enzymes' is quite complex: disaccharidases are located in the (human) brush border membrane of the small intestinal mucosa. ⁵ Herein, a-glucosidase activity is attributed by a trehalase and two bi-enzymatic complexes, namely MGAM and SI complex. The complexes are linked to the brush border with the transmembrane domain (TMD) via an O-glycosylated linkage (O-link) and consist out of a C and N-terminal part (ntMGAM and ctMGAM versus ntSI and ctSI) (Figure l). ¹ - ⁵ The individual enzymes in both complexes exhibit a unique specificity profile on different substrates, besides the a-1,4 hydrolytic activity (EC 3.2.1.20), they all share, ⁶ which makes the human intestinal tract a quite complex mixture of a-glucosidase activity. ⁵ - ⁶ Each individual enzyme in the complex is classified and named after their signature specificity: maltase (ntMGAM, EC 3.2.1.20 ), glucoamylase (ctMGAM, EC 3.2.1.3), isomaltase (ntSI, EC 3.2.1.10) and sucrase (ctSI, EC 3.2.1.48) (Table l). ⁵ - ⁶ Exceptionally, trehalase activity (EC 3.2.1.28) is only found in the trehalase enzyme, which is solely specific towards the a-1,1- a bound. ^5,7 The term 'mammalian' refers to any mammal such a rat, pig, chimpanzee and human but specifically relates to rat or human (being).

More specifically, the present invention relates to the usage as described above wherein said mammal is a rat or a human being.

Furthermore, the present invention relates to the usage as described above wherein said intestinal brush border enzymes are a-glucosidases.

More specifically, the present invention relates to the usage as described above wherein said a-glucosidases comprise the sucrase-isomaltase (SI) complex, the maltase-glucoamylase (MGAM) complex and/or a trehalase.

More specifically the present invention relates to the usage as defined above wherein said microbial enzymes are a-glucosidases (EC 3.2.1.20) from family GH13/GH31 and/or trehalases (EC 3.2.1.28) from family GH37

Moreover, the present invention relates to the usage as described above wherein said microbial enzymes are bacterial and/or yeast enzymes.

More specifically, the present invention relates to the usage as described above wherein said bacterial and/or yeast enzymes are: an a-glucosidase from yeast type I belonging to the family GH13 and having as enzyme commission number 3.2.1.20, Thermus thermophilus HB27 belonging to the family GH13 and having as enzyme commission number 3.2.1.20, Thermus caldophilus GK24 belonging to the family GH13 and having as enzyme commission number 3.2.1.20 and Bacillus stearothermophilus belonging to the family GH13 and having as enzyme commission number 3.2.1.20.

More specifically, the present invention relates to the usage as described above wherein said bacterial and/or yeast enzymes are: an a-glucosidase from Aspergillus niger belonging to the family GH31 and having as enzyme commission number 3.2.1.20, Bacillus sp. AHU 2001 belonging to the family GH31 and having as enzyme commission number 3.2.1.20, Bacillus stearothermophilus belonging to the family GH13 and having as enzyme commission number 3.2.1. 20 and a trehalase from a prokaryote belonging to the family GH37 and having as enzyme commission number 3.2.1.28.

Furthermore, the present invention relates to the usage as described above wherein said enzyme activity is determined on the substrates maltose, nigerose, isomaltose, trehalose and/or kojibiose. The present invention also relates to an in vitro assay to assess the digestibility of disaccharides (digestion assay) comprising a mix of microbial enzymes which mimic the enzyme activity of mammalian intestinal brush border enzymes.

The present invention further relates to an in vitro digestion assay comprising a mix of microbial enzymes which mimic the enzyme activity of mammalian intestinal brush border enzymes wherein said mammalian intestinal brush border enzymes are alpha glucosidases comprising the sucrase-isomaltase (SI) complex, the maltase-glucoamylase (MGAM) complex and/or a trehalase. More specifically, the present invention relates to an in vitro digestion assay as described above wherein said mix of microbial enzymes which mimic the enzyme activity of human intestinal brush border enzymes comprise: an a-glucosidase from yeast type I belonging to the family GH13 and having as enzyme commission number 3.2.1.20 , Thermus thermophilus HB27 belonging to the family GH13 and having as enzyme commission number 3.2.1.20, Thermus caldophilus GK24 belonging to the family GH13 and having as enzyme commission number 3.2.1.20 and Bacillus stearothermophilus belonging to the family GH13 and having as enzyme commission number 3.2.1.20.

More specifically, the present invention relates to an in vitro digestion assay as described above wherein said mix of microbial enzymes which mimic the enzyme activity of rat intestinal brush border enzymes comprise: an a-glucosidase from Aspergillus niger belonging to the family GH31 and having as enzyme commission number 3.2.1.20, Bacillus sp. AHU 2001 belonging to the family GH31 and having as enzyme commission number 3.2.1.20, Bacillus stearothermophilus belonging to the family GH13 and having as enzyme commission number 3.2.1.20 and a trehalase from a prokaryote belonging to the family GH37 and having as enzyme commission number 3.2.1.28.

1 Material and methods

1.1 Materials

Kojibiose and nigerose were produced in-house according to Beerens et al. ⁸ and Franceus et al. ⁹ Trehalose was kindly provided by Cargill. Isomaltose was sourced from Megazyme, maltose and sucrose from Sigma Aldrich. Enzymes were sourced from Megazyme (a- glucosidase from Aspergilles niger (E-TRNGL) and Bacillus stearothermophilus (E-TSAGL), sucrase from Saccharomyces cerevisiae (E-SUCR) and trehalase from a prokaryote (E-TREH)), NZYTech (trehalase from Escherichia coli (CZ0703) and Mycobacterium smegmatis (CZ0673)) and Sigma Aldrich (rat intestinal extract powders (I1630-10G), a-glucosidase from rice type V (G9259-100UN) and Saccharomyces cerevisiae type I (G5003-1KU)). All chemicals and reagents were purchased from Sigma-Aldrich unless noted. Solutions were prepared in distilled water.

1.2 Enzyme expression and purification

Plasmids (pET21a) containing ampicillin resistance and the a-glucosidases originating from Bacillus sp. AHU 2001 (Uniprot ID: A0A0B6VK42_9BACI, GenBank accession number: BAQ19546) ¹⁰ , Thermus calclophilus strain GK24 (UniProt ID: Q9RA62_THECAGenBank accession number: AF096282) ¹¹ and Thermus thermophilus strain HB27 (Uniprot ID: Q72LF2_THET2, GenBank accession number: WP_011172564) ^1:L were ordered via GeneArt (Thermo Fisher scientific) with a C-terminal Hise-tag. The enzymes were recombinantly expressed in E. coli LOBSTR-BL21 (DE3). Cells were grown in 5 mL Lysogeny Broth (LB) containing 100 pg.mL ¹ of ampicillin at 200 rpm at 37°C overnight. Subsequently, the cells were inoculated in 250 mL of LB containing 100 pg.mL ¹ of ampicillin and expression was induced with isopropyl p-D-l-thiogalactopyranoside (IPTG) after approximately 2 hours of growth at 37°C, when the ODeoo reached 0.6. From then on growth continued at 20°C for 21 hours and cells were harvested via centrifugation (9000 rpm, 15 min, 4°C, Sorval RC 6+ centrifuge) and stored at -20°C. Cell were lysed, sonicated, and centrifuged. The supernatant was purified by means of His-tag Ni-affinity chromatography. A column containing 1 mL of bed volume HisPur™ Ni-NTA resin (Thermo Fisher) was equilibrated with 6 mL of equilibrium buffer (10 mM imidazole, 50 mM Na2H2PO4 and 300 mM NaCI; pH 7.4). The enzyme supernatant was added and the column was washed two times with 4 mL of wash buffer (30 mM imidazole, 50 mM NaH2PO4 and 300 mM NaCI; pH 7.4). Then, the enzyme was eluted using 8 mL of elution buffer (250 mM imidazole, 50 mM NaH2PO4 and 300 mM NaCI; pH 7.4). To wash away the denaturing imidazole and alter the buffer to sodium phosphate (100 mM, pH 7), the elution fraction was collected in a 30-kDa cutoff Amicon centrifugal filter unit (Millipore) for buffer exchange. The enzymes were then stored at -20°C in an Eppendorf tube. 1.3 Human intestinal theoretical dataset

In this study, one unit of disaccharidase is expressed as the activity hydrolyzing 1 pmole of substrate per minute. The calculations were performed in Excel (Table 1). Input data are the amount of maltase activity (U) at 28 mM from Aurecchio et al. ⁷ ’ ¹² and the kinetic data ( M and Vmax) of each separate enzyme (ntMGAM, ctMGAM, ntSI and ctSI) from Lee et al. ⁶ Some assumptions had to be made, a fixed protein content of 0.01 mg was used (Table 1). The distribution of the maltase activity was put at an average of 75% attributed by SI and 25% by MG AM. ^13,14 The molar amount of each enzyme in the complexes is equal, thus n(ntMGAM) = n(ctMGAM) (Variable 1) and n(ntSI) = n(ctSI) (Variable 2). Table 1: Parameters (input, variables and aused to calculate the human intestinal Michaelis- Menten kinetics.

Input

Literature Experimental Remarks

Maltase activity 246 U.g- ¹ protein / 0.246 U.mg ¹

(jejunum, average of 15 adults) ¹²

Trehalase activity 39 U.g ¹ (jejunum, / 0.039 U.mg ¹ average of 15 adults) ¹² ntMGAM, ctMGAM, Kinetic data maltose, / Units adjusted to the ntSI, ctSI kojibiose, nigerose amount of enzyme and isomaltose required to hydrolyse 1

Molecular weight pmole of substrate at the given conditions

Trehalase (porcine Molecular weight Kinetic data kidney) Assumptions

0.01 mg of protein

Substrate 28 mM concentration

Maltase activity ratio 25/75 MGAM/SI

Variables

1: molar amount of MGAM

2: molar amount of SI

3: molar amount of trehalase

Using the kinetic data of each separate enzyme and the Michaelis-Menten equation (Equation 1), the maltase activity at 28 mM could be calculated (U.mg ^-1 ). Multiplying this amount with the amount of protein (mg), retrieved from the molar amount and molecular weight of each enzyme, lead to the activity expressed in units (U). By using the solver add-in in Excel, the difference between the activity (U) in the human tissue and the sum of the activities (U) of the two separate enzymes in the complex was minimalized (Equation 2 and 3). Trehalase activity is a little special as it had a separate calculation, but it was similar, again the amount of moles is determined. The amount of trehalose activity in human tissue had to be equal to the number of units calculated at 28 mM using the kinetic data from porcine kidney trehalase.

[Equation

[Equation 2] v _humanMCAM [U. mg ^x ] ■ m _humanMCAM [mg] ■ 0.25 - ( _ctMGAM [U. mg ^x ]

’ ^m ctMGAM [mg] + ^v ntMGAM ’ ^m ntMGAM [mg]) 0 [U]

[Equation 3] v _{human SI} [U. mg ^x ] ■ m _{human SI} [mg] ■ 0.75

- (v _ctsI [U. mg ^-1 ] ■ m _ctsI [mg] + v _ntsI [U. mg ^-1 ] ■ m _ntsI [mg]) -+ 0 [U]

Next step was to determine the overall kinetic parameters of all selected substrates, like it would occur in the human intestine. Over the substrate concentration range (0.1 - 100 mM) the combined activity, meaning the sum of the separate activities (U) (Equation 4). The activity of each complex for a certain concentration can be calculated in U.mg ¹ (Equation 1), multiplying this with the calculated amount (mg) of each complex, leads to an activity in units. Dividing this again by the 0.01 mg of protein a total of U.mg ¹ for each substrate concentration can be obtained. Via Excel and the solver add-in the KM (mM) and v _m ax (U.mg ¹ ) was determined for this combined activity. All calculation can be found in the Excel File HumanlntestineKinetics.xls. ru 4--

[Equation enzyme = ntMGAM, ctMGAM, ntSI and ctSI

1.4 Rat intestinal experimental dataset

Rat intestinal extract was prepared by dissolving 200 mg rat intestinal acetone powder in 6 mL sodium phosphate buffer (100 mM, pH 7.0). The mixture was vortexed and then homogenized by sonication in ice cold water for 10 min (5 times 2 minutes). After centrifugation (9000 rpm, 30 min, 4°C), the supernatant was used in the assay. Reactions were performed in Eppendorfs. Four substrate concentrations were used ranging between 1 and 100 mM in duplicate. An appropriate dilution of disaccharide stock solution (50 or 500 mM) and the volume was adjusted to 90 pL with sodium phosphate buffer solution (100 mM, pH 7.0). To start the reactions 10 pL of an appropriate dilution of the rat intestinal extract was added to the reaction mixture to have a final volume of 100 pL. Samples were taken during the time course of 14 min every 2 min and were inactivated by heat at 95°C for 5 min. 10 pL of sample and the glucose standards were mixed with the GOD-POD ¹⁵ assay solution (200 pL) and incubated for 30 min at 37°C. The optical density was measured at 420 nm. The amount of enzyme in the extract was determined by the Pierce BCA assay kit. Michaelis-Menten kinetics were determined via the Hanes-Woolf plot.

1.5 Alternative enzymes experimental dataset

In a similar way as for the rat intestinal extract, reactions were performed under the same conditions in sodium phosphate buffer (100 mM, pH 7.0) at 37°C. A series of substrate concentrations (4) ranging from 1-100 mM were incubated with the appropriate enzyme dilution. 10 pL of this enzyme solution was added to 90 pL of substrate and sodium phosphate buffer solution. Every 2 min samples were taken during 14 min, which were inactivated by heat at 95°C for 5 min and analysed via GOD-POD ¹⁵ at 420 nm. Michaelis-Menten kinetics were determined via the Hanes-Woolf plot.

1.6 Calculations of the mimic

The obtained data for the human and rat intestine can be compared to the data of the alternative enzymes. The general approach is again getting the activity (U) at a fixed concentration equal, by altering the amount of protein (Excel File MixDigest.xls). The activity (U) for each substrate was calculated for both the intestinal data and each alternative enzyme at 28 mM (Equation 5). The combined activity of the alternative enzymes is equal to the sum of the activities (U) of all enzymes. To minimalize the difference between the intestinal activity and the alternative enzymes, the square of the differences for each substrate was calculated. The sum of these squares is used for the solver add-in and minimalized, by changing the amount of alternative enzymes (mg) (variables) (Equation 6). From this equation and solver run, the amount of enzyme was found, some equaled zero or were low. So to optimize, these were left out, i.e. removing their kinetic data from the file, and all enzyme concentration were put at zero again to rerun the solver add-in. These steps were done until removal of an enzyme leads to an unwanted result. In the file, the activity (U) is also simulated over the whole substrate range for each substrate with the Michaelis-Menten kinetics of the enzymes in the mixture (Equation 7).

[Equation

[Equation 7] v _substrate [U] = v _substrate [U.mg ^x ] ■ m _enzyme [mg] for 0.1 - 100 mM 2 Results

In order to find an alternative in vitro assay for disaccharide evaluation, several glycoside hydrolase (GH) families were explored to find enzymes with interesting properties, i.e. a- glucosidase activity, similarity to intestinal enzymes, commercial availability and microbial origin. Two important datasets had to be created. A first dataset consisted out of the reference material, namely the kinetic data from human or rat origins retrieved from literature or own experiments, respectively. A second dataset was created based on the selected alternative enzymes and their kinetic data obtained from experiments. Finally, both datasets were compared with mathematics (as described above) and a calculation was made (as described above) to obtain a mixture of enzymes which can mimic the activity of -as a non-limiting example- rat intestinal extract and of -as a second non-limiting example- human intestinal enzymes (Table 2).

2.1 Dataset 1: Reference

Initially, a dataset with reference material had to be created. As no experimental data could be obtained from humans, literature data was used. As rat intestinal extract is often used to evaluate novel sugars, it was also considered, by creating an experimental dataset. In this way, we can obtain a mixture able to mimic the rat intestinal extract and thus eliminate the use of animals and the variability of this assay due to difference is the sample.

Table 2: a-Glucosidase activity found in the human body.

2.1.1 Human a-glucosidase activity

Three different types of data can be found in literature: (1) an activity profile of all enzymes together ⁷ - ¹³ - ¹⁶ - ¹⁷ (2) each complex together ¹⁸ or (3) each enzyme separately ⁶ . Ideally, for this study a uniform dataset is needed with the kinetic data of all the human digestive enzymes together towards all the different a-glucobioses. However, this is not the case for every disaccharide and/or enzyme combination. The first obstacle that occurs is obtaining a suitable human sample and getting the enzyme active out of this sample. Dahlqvist and co-workers have obtained a huge amount of data on this subject and worked with human intestinal samples, which report different gut localized samples. ¹³ - ¹⁴ - ¹⁶ - ¹⁹ - ²⁰

By combining data specific activities provided by Auricchio et al. (1966) ¹² and the Michaelis- Menten parameters for all substrates, their overall kinetic profile could be determined (Table 3). The human a-glucosidase brush border kinetic profile of maltose, kojibiose, nigerose and isomaltose could be simulated to the data available in Lee et al. (2016) ⁶ . Trehalase is a more special case as no human trehalase kinetic parameters, both KM and v _m ax, can be found in literature. They can, however, be derived from porcine kidney, which was experimentally determined.

The most complex kinetic profiles belong to maltose, kojibiose, nigerose and isomaltose, as the activities are attributed by multiple enzymes found in the two complexes, MGAM and SI. First, a molar ratio had to be found between the two complexes which would result in a relevant combined maltase activity, namely 0.246 U.mg ¹ protein at 28 mM of maltose. ¹² - ¹⁶ As described by Dahlqvist et al. on average 75% of the maltase activity is assigned to the SI complex and the remaining 25% is attributed by MGAM. ¹³ - ¹⁴ In this way we can split the total maltase activity over the MGAM and SI complex. Each complex consists out of two enzymes, which can both contribute to the maltase activity. By using the KM and v _mO x ⁶ and a variable amount of protein (mg) derived from the molar concentration of each complex, the maltase activity (U) at 28 mM of each separate enzyme could be determined. This activity of each complex should be equal to their respective part (75 vs. 25% for SI and MGAM, respectively) at a fixed protein content of 0.01 mg. By using the solver add-in of Excel, the difference between the two activities for both MGAM and SI was minimized by changing the variable, namely the molar concentration of each complex. Next a simulation of the activity profile in the substrate range of 0.1 - 100 mM was performed combining the obtained amount of the separate enzymes (mg) and their reported kinetic parameters ⁶ . From this activity profile, the combined kinetic parameters were determined (Table 3), which will be further used to compare the mimic. To validate the obtained data, the activity at 28 mM was compared to the total maltase and isomaltase activity of 0.246 and 0.074 U.mg ¹ in the jejunal, which was as good as equal to the ones derived from the calculated parameters: 0.246 and 0.095 U.mg ^-1 .

In a similar manner a more simplistic calculation could be performed to simulate the kinetic data of trehalase, as this activity is only assigned to one enzyme. The trehalase activity at 28 mM was derived from a human intestinal sample, namely 0.039 U.mg ¹ protein. ¹² No complete set of kinetic parameters of human trehalase can be found in literature, however a trehalase originating from a pig is available, and it is expected to have a similar kinetic profile as in humans. The KM value equals 3.8 mM and v _m ax is 1.55 U.mg ^-1 . In a similar way, the activity at 28 mM can be calculated from these kinetic parameters multiplied with the varying amount of enzyme (mg), which needed to be equal to the trehalase activity (U) at 0.01 mg of protein.

2.1.2 Rat intestinal extract a-glucosidase activity

Besides the hypothetical human enzyme activities, the kinetics of rat intestinal extract were experimentally determined for maltose, kojibiose, nigerose, isomaltose and trehalose with the GOD-POD assay ¹⁵ (Table 3). As this extract is often used as an alternative to a-glucosidases of human origin, due to its lower cost and expected resemblance to the human intestine. ²¹ The intestinal extracts are solubilized in sodium phosphate buffer (pH 7, 100 mM) and homogenized through sonication before use. The extract was incubated with different concentrations of the substrates in duplicate, leading to the determination of the extract's kinetic parameters (KM and v _m a _X ). Table 3: The combined theoretical determined kinetic parameters of all enzymes present in the human intestine and the experimental kinetic data obtained from rat intestinal extract in 100 mM of phosphate buffer (pH 7) at 37°C. When looking at the turnover number k _ca t and the specificity constant k _ca t/KM, both appear to be quite small, this is attributed due to impurities incorporated in the total amount of protein.

2.2 Dataset 2: Alternative enzymes for the mimic mixture

Ideally, alternative enzymes would occur in the same family as the reference enzymes, namely in GH31 and GH37 family (Table 2) and have a microbial origin. However, a-glucosidases (EC

3.2.1.20) from bacteria, yeast (Saccharomyces cerivisiae) and insects mainly belong to GH13 and those from plants, animals, fungi and yeast (Schizosaccharomyces pombe) to GH31. ¹⁰ The former family is a large family containing various amylolytic enzymes such as a-amylase (EC 3.2.1.1), cyclodextrin glucotransferase (EC 2.4.1.19) and a-glucosidase (http://www.cazy.org/GH13_characterized.html). ¹⁰ - ²² Only a handful of GH31 a-glucosidases originating from bacteria have been found (http://www.cazy.org/GH31_characterized.html). ^10,23-26 Interestingly, the catalytic domains of both families are formed by (P/a)s barrel-folds and their catalytic nucleophiles are located at the same position, i.e. C-terminal end of the fourth p-strand. ¹⁰ The general acid/base catalyst of GH13 and GH31 enzymes are at the C-termini of the fifth and sixth p-strands of the catalytic domain, respectively. ¹⁰ Besides GH31 and GH13, a-glucosidases (EC 3.2.1.20) can be found in family GH4 . Glucoamylases (EC 3.2.1.3) are classified in the GH31, GH97 and GH15 families. Higher eukaryotic species are found in the GH31 family, fungal and yeast species are classified within the GH15 family. For isomaltases (EC 3.2.1.10), a similar trend as to maltases can be seen, most higher species are classified in GH31 and bacterial and yeast into GH13. Noteworthy, sucrases (EC 3.2.1.48) are only classified in GH31 and most enzymes originate from eukaryotic species, most bacterial enzymes are classified within Lactobacillus sp. Trehalases (EC 3.2.1.28) are found in three different families: GH37 (neutral trehalase), GH65 (acid trehalases) and GH15 (mycobacterial trehalase).

For our mixture, both GH31 and GH13 families were explored. We searched for enzymes of commercial suppliers and possible interesting enzymes from literature (Table 4). Twelve enzymes were selected, six (a-glucosidase from Aspergilles niger, rice type V, and Bacillus stearothermophilus, Saccharomyces cerevisiae type I, sucrase (maltase) from Saccharomyces cerevisiae and trehalase from a prokaryote and E. coli) of them were purchased from the respective suppliers. Three a-glucosidases (from Bacillus sp. AHU 2001 ¹⁰ , Thermus thermophilus HB27 and Thermus calclophilus GK24 ¹¹ ) were retrieved from literature due to their more special profile in comparison to the commercially available enzymes. Their genes were ordered via GeneArt (plasmid: pET21a) and they were expressed and purified by means of nickel affinity chromatography. a-Glucosidase from Bacillus sp. AHU2001 is a bacterial enzyme in the GH31 family and is able to hydrolyse maltose, maltooligosaccharides, kojibiose, nigerose, neotrehalose and in a very low amount isomaltose. ¹⁰ Two a-glucosidase from Thermus thermophilus from strain HB27 and Thermus calclophilus GK24 were selected and are classified in the GH13 family. Interestingly, the former one can hydrolyse also trehalose and the specific activity towards isomaltose is higher than maltose. ¹¹ Table 4: Overview of the selected alternative enzymes with their respective GH family, EC number and source (supplier or literature).

In a similar experimental set-up as for the rat intestinal extract, the kinetic parameters of each selected enzyme were determined in duplicate (Table 5). From these tests, rice glucosidase did not seem to be active at the tested conditions. Furthermore, trehalase from Mycobacterium smegmatis seemed to interfere with the GOD-POD assay. These two enzymes were thus excluded in the following steps as no kinetic parameters were determined. Clearly, most of the enzymes showed hydrolytic activity towards only a couple of the substrates. Remarkably, only one enzyme could be found showing activity towards all five substrates (i.e. maltose, kojibiose, nigerose, isomaltose and trehalose), namely a-glucosidase from Thermus thermophilus HB27 (enzyme 6). However, mimicking mammalian brush border enzyme activity with this enzyme is rather impossible. The enzyme showed a different activity profile towards the five substrates (Table 6 and 7). Therefore, it became clear that a more complex mixture of the selected enzymes would be needed (see section 2.3).

Table 5: Overview of the kinetic parameters of the alternative enzymes (experimentally determined). Enzyme number is equal to numbers listed in Table 4.

2.3 Calculation of the mimic

To find a combination of alternative enzymes that could mimic the activity of the human enzymes and the rat intestinal extracts, the activity (U) must be equal. To achieve this, some assumptions had to be made, i.e. a fixed substrate concentration (mM) and protein amount

(mg). A concentration of 28 mM was opted, as it is a reoccurring concentration in several studies and thus relevant for comparison. ⁷ - ¹³ - ¹⁹ By selecting an appropriate protein amount

(0.01 mg for human intestine and 0.025 mg for rats), the activity (U.mg ^-1 ) at 28 mM for each substrate could be derived from the kinetic parameters of the intestinal activities. This activity needed to be equal to the combined activities for the selected commercial enzymes, by using the Excel solver add-in the difference (rat) or deviation (human) was minimized, leading to a protein amount for each enzyme needed for the cocktail. After several rounds an optimal mixture to mimic the human and rat intestine was found.

2.3.1 Mimic of the human intestine

As mentioned above, only one enzyme (6) showed hydrolytic activity towards all five substrates. To evaluate its potential to mimic human brush border activity, the excel file was adjusted to solve the activity equation with only enzyme 6 (Mix 1). At a substrate concentration of 28 mM, large deviations can be found, however, a fit can be found for isomaltose and trehalose. Subsequently, all enzymes were combined in a 1:1 ratio (Mix 2). Clearly, a large deviation could be seen for trehalose in this mixture. Therefore, both trehalases were excluded and all other enzymes were used in a 1:1 ratio in Mix 3. The deviation for trehalose activity is now at 0%. However, for the other substrates, i.e., kojibiose, nigerose and isomaltose, deviations are larger than 20%. Mixture 4 used both GH31 enzymes (1 and 3) in combination with enzyme 6, which was included for isomaltose and trehalose activity (see Mix 1). Again, 1:1 ratio leads to large deviations. In a similar way, all GH13 enzymes were combined leading to the same results. Some enzymes seemed to be less relevant, i.e., lower enzyme concentration and/or activity contribution. The removal of these, namely enzymes 8, 1 and 3 had a limited effect (Mix 6). Finally, the solver-add in was used to optimize the mixture to obtain a deviation of less than 20% for all substrates at 28 mM (Mix 7). Remarkably, for every substrate a deviation below 0.01% could be found.

Thus, starting from the nine selected enzymes a final mixture containing four enzymes (a- glucosidase from yeast type I, Thermus thermophilus HB27, Thermus caldophilus GK24 and Bacillus stearothermophilus in a ratio of 68/18/1/1) was able to mimic the human intestinal brush border enzyme activity (Figure 2a). The a-glucosidase originating from Thermus thermophilus HB27 is required to have a good resemblance to the human intestine data. However, this enzyme also has trehalase activity, which is in the calculated mixture is higher than wanted. Interestingly, the /C/w-values of both human trehalase and Thermus thermophilus HB27 a-glucosidase are quite similar. If the mixture would be used in a 1.4-fold dilution, the mixture could be used as a mimic for trehalase activity. This enzyme ratio was also used to compare the activity over the whole substrate range, which showed that for all substrates the deviation is lower than 20% for a substrate concentration of 7 mM or higher (Figure 2b). Kojibiose and trehalose perform best, as their interval of less than 20% deviation covers the full substrate range (0.1 to 100 mM). The interval (deviation below 20%) for maltose, nigerose and isomaltose starts at 2, 7 and 4 mM up to 100 mM. The biggest differences noted for maltose and nigerose are roughly 15% (Figure 2b). So overall, the hydrolytic profile of the theoretical human intestine and the mimic should remain equal.

Table 6: Overview of the mixtures to mimic human intestinal brush border activity.

Mix l Mix 2 Mix 3 Mix 4 Mix 5 Mix 6 Mix 7

Enzymes ratio ratio ratio ratio ratio ratio ratio *deviation at 28 mM

2.3.2 Mimic of the rat intestine

In a similar way, the same exercise was performed to mimic the rat intestinal activities (Table 7). Firstly, the potential of enzyme 6 was evaluated (Mix 1). Again, large deviations were found. When looking at the activities of all the enzymes (Mix 2), trehalose differed from the other substrates. Therefore, different combinations were made based on the different families (GH31 vs. GH13) and all contained trehalase activity (enzyme 6 vs. 9 and 10), which lead to mixtures 4-6. Which did not result in a sufficient mixture. Only after optimization of the mixture via our Excel calculations a mixture (Mix 7) consisting out of four different enzymes (a-glucosidase from Aspergillus niger, Bacillus sp. AHU 2001 and Bacillus stearothermophilus and trehalase from a prokaryote in a ratio of 45/73/4/1) was found.

Table 7: Overview of the mixtures to mimic rat intestinal brush border activity.

Mix l Mix 2 Mix 3 Mix 4 Mix 5 Mix 6 Mix 7

Enzymes ratio ratio ratio ratio ratio ratio ratio

Deviation (%)* Maltose 98 24 85 40 40 31

Kojibiose 95 78 85 93 93 84

Nigerose 91 63 83 81 81 74

Isomaltose 27 29 73 56 56 72

Trehalose 78 -1384 -1379 95 -1384 -1040

Absolute sum 389 1578 1705 365 1659 1301 *deviation at 28 mM

When simulating this ratio over the complete substrate range a quite similar profile can be found for the substrates (Figure 3a). For maltose and kojibiose the difference to rat intestinal extract is lower than 10% over the whole substrate range (Figure 3b). For nigerose, concentrations below 16 mM deviate more than 20%, up to roughly 86% (Figure 3b). For isomaltose, an interval (deviation less than 20%) between 5.5 and 50 mM can be noticed

(Figure 3b). In the case of trehalose, a substrate concentration higher than 13 mM gives a difference below 20%, lower than 2 mM the difference is higher than 100%, due to the difference in /C/wof the rat intestine and trehalase from a prokaryote (Figure 3b). References

1. Hodoniczky, J., Morris, C. A. & Rae, A. L. Oral and intestinal digestion of oligosaccharides as potential sweeteners: a systematic evaluation. Food Chem. 132, 1951-1958 (2012).

2. Hernandez-Hernandez, O., Olano, A., Rastall, R. A. & Moreno, F. J. In vitro Digestibility of Dietary Carbohydrates: Toward a Standardized Methodology Beyond Amylolytic and Microbial Enzymes. Front. Nutr. 6, 1-5 (2019).

3. Lin, A. H. M., Lee, B. H. & Chang, W. J. Small intestine mucosal a-glucosidase: A missing feature of in vitro starch digestibility. Food Hydrocoll. 53, 163-171 (2016).

4. Semenza, G. Summary for Policymakers, in Climate Change 2013 - The Physical Science Basis (ed. Intergovernmental Panel on Climate Change) 3, 1-30 (Cambridge University Press, 1976).

5. Elferink, H., Bruekers, J. P. J., Veeneman, G. H. & Boltje, T. J. A comprehensive overview of substrate specificity of glycoside hydrolases and transporters in the small intestine: "A gut feeling". Cell. Mol. Life Sci. 77, 4799-4826 (2020).

6. Lee, B. H. et al. Contribution of the Individual Small Intestinal a-Glucosidases to Digestion of Unusual a-Linked Glycemic Disaccharides. J. Agric. Food Chem. 64, 6487- 6494 (2016).

7. Auricchio, S. et al. Disaccharidase Activities in Human Intestinal Mucosa. Enzymol. Biol. Clin. (Basel). 74, 193-208 (1963).

8. Beerens, K. et al. Biocatalytic synthesis of the rare sugar kojibiose: process scale-up and application testing. J. Agric. Food Chem. 65, 6030-6041 (2017).

9. Franceus, J. et al. Rational design of an improved transglucosylase for production of the rare sugar nigerose. Chem. Commun. 55, 4531-4533 (2019).

10. Saburi, W., Okuyama, M., Kumagai, Y., Kimura, A. & Mori, H. Biochemical properties and substrate recognition mechanism of GH31 a-glucosidase from Bacillus sp. AHU 2001 with broad substrate specificity. Biochimie 108, 140-148 (2015).

11. Alarico, S., Da Costa, M. S. & Empadinhas, N. Molecular and physiological role of the trehalose-hydrolyzing a-glucosidase from Thermus thermophilus HB27. J. Bacterial. 190, 2298-2305 (2008).

12. Auricchio, S., Rubino, A. & Murset, G. Intestinal glycosidase activities in the human embryo, fetus and newborn. Pediatrics 35, 944-954 (1965). 13. Dahlqvist, A. [99] Intestinal disaccharidases. in Complex Carbohydrates 8, 584-591 (Academic Press, 1966).

14. Dahlqvist, A. Specificity of the human intestinal disaccharidases and implications for hereditary disaccharide intolerance. J. Clin. Invest. 41, 463-470 (1962).

15. Kapustka, L. A., Annala, A. E. & Swanson, W. C. The peroxidase-glucose oxidase system: A new method to determine glucose liberated by carbohydrate degrading soil enzymes. Plant Soil 63, 487-490 (1981).

16. Dahlqvist, A., Auricchio, S., Semenza, G. & Prader, A. Human intestinal disaccharidases and hereditary disaccharide intolerance. The hydrolysis of sucrose, isomaltose, palatinose (isomaltulose) and 1,6-alpha-oligosaccharide (isomalto-oligosaccharide) preparation *. J. Clin. Invest. 42, 6 (1963).

17. Dahlqvist, A. & Thomson', D. L. The Digestion and Absorption of Maltose and Trehalose by the Intact Rat. Acta Physiol. Scand. 59, 111-125 (1963).

18. Amiri, M. & Naim, H. Characterization of Mucosal Disaccharidases from Human Intestine. Nutrients 9, 1106 (2017).

19. Dahlqvist, A. & Semenza, G. Disaccharidases of small-intestinal mucosa. J. Pediatr. Gastroenterol. Nutr. 4, 857-865 (1985).

20. Dahlqvist, A. & Telenius, U. Column chromatography of human small-intestinal maltase, isomaltase and invertase activities. Biochem. J. Ill, 139-146 (1969).

21. Oku, Tanabe, K., Ogawa, S., Sadamori, N. & Nakamura, S. Similarity of hydrolyzing activity of human and rat small intestinal disaccharidases. Clin. Exp. Gastroenterol. 4, 155 (2011).

22. Okuyama, M., Saburi, W., Mori, H. & Kimura, A. a-Glucosidases and a-l,4-glucan lyases: structures, functions, and physiological actions. Cell. Mol. Life Sci. 73, 2727-2751 (2016).

23. Tsujimoto, Y. et al. Molecular determinants of substrate recognition in thermostable a- glucosidases belonging to glycoside hydrolase family 13. J. Biochem. 142, 87-93 (2007).

24. Kang, M. S., Okuyama, M., Mori, H. & Kimura, A. The first a-l,3-glucosidase from bacterial origin belonging to glycoside hydrolase family 31. Biochimie 91, 1434-1442 (2009).

25. Wang, Y. H., Jiang, Y., Duan, Z. Y., Shao, W. L. & Li, H. Z. Expression and characterization of an a-glucosidase from Thermoanaerobacter ethanolicus JW200 with potential for industrial application. Biologia (Bratisl). 64, 1053-1057 (2009). Tan, K. et al. Novel a-glucosidase from human gut microbiome: Substrate specificities and their switch. FASEBJ. 24, 3939-3949 (2010).