Reliability factor for identification of amylolytic enzyme activity in the optimized starch-iodine assay

Amylolytic enzymes are a group of proteins degrading starch to its constitutional units. For high-throughput screening, simple yet accurate methods in addition to the reducing ends assays are required. In this article, the iodine assay, a photometric assay based on the intensely colored starch-iodine complex, was adapted to enable accurate and objective differentiation between enzyme and background activity using a newly introduced mathematical factor. The method was further improved by designing a simple setup for multiple time point detection and discussing the applicability of single wavelength measurements. One of the most important sources of carbohydrates for human diet is starch [1], a structure entirely built up of glucose units interlinked via α-1,4and α-1,6-glucosidic bonds, forming linear chains and branch points, respectively. This polysaccharide consists of the linear amylose and branched amylopectin in a highly varying ratio ranging from below 15% to about 40% amylose [2]. Starch is biochemically degraded by amylolytic enzymes such as αand β-amylases as well as amyloglucosidases [3]. Typically, quantitative determination of catalytic activity of amylolytic enzymes is conducted via estimation of the amount of formed reducing sugars (e.g. glucose or maltose) by using methods such as the Nelson-Somogyi [4] and dinitrosalicylic acid (DNS) assay [5]. Another frequently used method is the iodine assay [6,7]. This photometric method is often applied for quick determination of the activity of starch converting enzymes [6] and is based on the intensely colored complex between iodine and linear chains in starch long enough to form helices [8]. The wavelength of absorbance maximum (λmax) of the starch-iodine complex's color is governed by the chain length (DP) of the oligosaccharide and shifts from about 490 nm (DP 18) to around 600 nm (DP 72) [9]. This behavior not only results in a high diversity of λmax between different starches [10] but also causes a shift in λmax in response to the activity of amylolytic enzymes [11], causing a considerable error in estimated activity when not addressed. Although the iodine assay has been developed decades ago [6,11], optimizations are scarce and mostly designed for single-point measurements [7]. Due to its speed and specificity for long glucan chains, however, the iodine assay has great potential for being a screening method. The aim of this study was to design a method for the measurement of multiple time points that is fast (< 30 min), easy (stable stock solutions) and accurate. Further, challenges for the interpretation of the data regarding the shift in λmax caused by some amylolytic enzymes and the differentiation between background and enzyme activity were both addressed by proposing a procedure and mathematical formula, respectively. Firstly, the method was optimized regarding factors such sample volume, substrate concentration, and the aliquot volume. The developed method was conducted on two microtiter plates and using one row or column/sample. Appropriate amounts of enzyme was diluted to 150 μl with 50 mM sodium phosphate buffer, pH 6.0 in a microtiter plate and incubated in a water bath at 40 °C for optimal heat transfer. The enzyme reaction was started by adding 50 μl substrate solution (4 mg/ml, diluted with buffer) to the enzyme solution. Then, at every full min, 15 μl aliquots of the enzyme reactions were transferred to the analysis wells located on a plate at RT containing 100 μl freshly prepared iodine reagent (0.15% KI, 0.015% I2 from stock (26% KI, 2.6% I2 [12]) with 5 mM HCl), followed by a brief washing steps for the pipette tips using the washing wells (200 μl buffer). In between transfer, the plate containing the analysis wells was covered with a plastic lid to prevent color depletion. After transfer of the last aliquot, the absorbance of the samples in the analysis wells was detected at 610 nm and a spectrum was measured from 450 to 750 nm with 5 nm steps using a spectrophotometer (SpectraMax from Molecular Devices) and the enzyme activity was calculated as 1 U = −1 ABS/min. This activity unit was found suitable for quick determination of activity. For more accurate values, it is recommended to estimate the corresponding https://doi.org/10.1016/j.ab.2020.113696 Received 18 November 2019; Received in revised form 6 March 2020; Accepted 18 March 2020 Abbreviations: DMSO, dimethyl sulphoxide; λmax, absorbance maximum; FRel, Reliability Factor ∗ Corresponding author. E-mail addresses: a.l.o.gaenssle@rug.nl (A.L.O. Gaenssle), m.j.e.c.van.der.maarel@rug.nl (M.J.E.C. van der Maarel), e.jurak@rug.nl (E. Jurak). Analytical Biochemistry 597 (2020) 113696 Available online 19 March 2020 0003-2697/ © 2020 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/). T substrate concentration using a standard curve (Figure S1) [7]. Starches are typically challenging to dissolve in aqueous solutions and tend to have a rather limited stability [13]. Starch solutions in dimethyl sulphoxide (DMSO) on the other hand can be used as stock solutions as they are stable for several months [14], not only simplifying the method but also increasing the comparability between separate experiments. Starches were either found to be soluble at 100 mg/ ml in pure DMSO (high amylose content) or at 25 mg/ml in 90% DMSO (low amylose content) and were thus concentrated enough to prevent inhibition of enzyme activity by DMSO [13–15]. Regarding the optimal wavelength for measurement, four starch substrates (potato amylose, Hylon VII, potato and rice starch) were tested. The estimated λmax of their complex with iodine ranged from potato amylose (λmax = 641 ± 8 nm), over Hylon VII (λmax = 590 ± 10 nm) and potato starch (λmax = 588 ± 12 nm) to rice starch (λmax = 562 ± 16 nm), being in good agreement with previous data [10]. Therefore, the wavelength of 610 nm was selected as it enabled detection of all substrates with a single wavelength (< 3% difference in the range 580–640 nm). The activity of amylolytic enzymes on starch was found to cause a blue-shift in λmax of varying strength, ranging from β-amylases (~25 nm/ABS on potato amylose, Fig. 1B) to α-amylases (~100 nm/ABS, Fig. 1A, for the full list see Figure S2 and S3). Fig. 1C shows the underestimation of the absolute absorbance compared to the relative absorbance measured at 610 nm of all ten studied amylolytic enzymes on potato amylose. Above a λmax of about 570 nm, the difference between the absolute and relative absorbance values was minimal (below 10%). The underestimation of absorbance led to an overestimation of activity due to a predicted faster decrease in absorbance. If the λmax fell below 570 nm, the activity overestimation was up to 25% which was observed for e.g. α-amylase on amylose below approx. 0.6 ABS at 610 nm. In this case it was recommended to estimate the enzyme activity based on the absorbance values at the λmax of each time point. For all experiments in which all data points exhibited a λmax within the given range, however, the overestimation was below 10%, providing a sufficiently narrow window for accurate detection of enzyme activity at a single wavelength. The assay was designed for detection of the enzyme reaction at multiple time points as the enzyme activity was not always found to be identical throughout the entire assay time frame, rendering single-point analysis inaccurate, especially for low enzyme activity or non-linear trends (deceleration). Those two types of reactions are often difficult to differentiate from background activity. Therefore, investigations were conducted to establish a factor that could reliably distinguish between enzyme and background activity. Since the method was aimed for an array of amylolytic enzymes on a series of starches, the factor should be applicable to a range of initial absorbance values, length of incubation, and enzyme type. Analysis of a large dataset of ten amylolytic enzymes and negative controls on amylose (n = 97, see Supplementary Data) revealed that the main differences between background and enzyme activity were the linearity and level of decrease in absorbance. The linearity of the decrease was determined by the coefficient of determination (R) as it shows the percentage of data points of an experiment described by the estimated linear slope [16]. The level of decrease was calculated by using the ratio of the first and last absorbance value (ABSStart and ABSEnd, respectively) as it was less distorted by different starting absorbance values and non-linear decreases. In general, high enzyme activity was found to be indicated by a sharp, highly linear decrease in absorbance over time (< -0.075 ABS/ min, R > 0.95) that could conclude in a deceleration of absorbance decrease due to substrate depletion and thus a lower R value. Background activity, on the other hand, resulted in a series of absorbance values that exhibited small, random fluctuations (about 7.5% of initial absorbance value), almost no decrease over time (> -0.007 ABS/ min) and a low linearity (R < 0.4). Low enzyme activity generally showed a very linear decrease, however, the fluctuations in absorbance were more visible than for higher enzyme activities. Detection of low enzyme activity was further impeded by the fact that sometimes the deviations in absorbance of samples with no enzyme activity may seem non-random, giving the appearance of a trend and thus enzyme activity. Taken together, basing the modeled Reliability Factor (FRel) on both linearity and level of decrease in absorbance enabled detection of both high and low enzyme activity. Additionally, cutoff values were introduced based on the careful manual identification on presence or absence of enzyme activity of each sample in the dataset (Figure S4). The two components contributing to the FRel were further weighted similarly to ensure a positive result

One of the most important sources of carbohydrates for human diet is starch [1], a structure entirely built up of glucose units interlinked via α-1,4and α-1,6-glucosidic bonds, forming linear chains and branch points, respectively. This polysaccharide consists of the linear amylose and branched amylopectin in a highly varying ratio ranging from below 15% to about 40% amylose [2]. Starch is biochemically degraded by amylolytic enzymes such as αand β-amylases as well as amyloglucosidases [3].
Typically, quantitative determination of catalytic activity of amylolytic enzymes is conducted via estimation of the amount of formed reducing sugars (e.g. glucose or maltose) by using methods such as the Nelson-Somogyi [4] and dinitrosalicylic acid (DNS) assay [5]. Another frequently used method is the iodine assay [6,7]. This photometric method is often applied for quick determination of the activity of starch converting enzymes [6] and is based on the intensely colored complex between iodine and linear chains in starch long enough to form helices [8]. The wavelength of absorbance maximum (λ max ) of the starch-iodine complex's color is governed by the chain length (DP) of the oligosaccharide and shifts from about 490 nm (DP 18) to around 600 nm (DP 72) [9]. This behavior not only results in a high diversity of λ max between different starches [10] but also causes a shift in λ max in response to the activity of amylolytic enzymes [11], causing a considerable error in estimated activity when not addressed. Although the iodine assay has been developed decades ago [6,11], optimizations are scarce and mostly designed for single-point measurements [7]. Due to its speed and specificity for long glucan chains, however, the iodine assay has great potential for being a screening method. The aim of this study was to design a method for the measurement of multiple time points that is fast (< 30 min), easy (stable stock solutions) and accurate. Further, challenges for the interpretation of the data regarding the shift in λ max caused by some amylolytic enzymes and the differentiation between background and enzyme activity were both addressed by proposing a procedure and mathematical formula, respectively.
Firstly, the method was optimized regarding factors such sample volume, substrate concentration, and the aliquot volume. The developed method was conducted on two microtiter plates and using one row or column/sample. Appropriate amounts of enzyme was diluted to 150 μl with 50 mM sodium phosphate buffer, pH 6.0 in a microtiter plate and incubated in a water bath at 40°C for optimal heat transfer. The enzyme reaction was started by adding 50 μl substrate solution (4 mg/ml, diluted with buffer) to the enzyme solution. Then, at every full min, 15 μl aliquots of the enzyme reactions were transferred to the analysis wells located on a plate at RT containing 100 μl freshly prepared iodine reagent (0.15% KI, 0.015% I 2 from stock (26% KI, 2.6% I 2 [12]) with 5 mM HCl), followed by a brief washing steps for the pipette tips using the washing wells (200 μl buffer). In between transfer, the plate containing the analysis wells was covered with a plastic lid to prevent color depletion. After transfer of the last aliquot, the absorbance of the samples in the analysis wells was detected at 610 nm and a spectrum was measured from 450 to 750 nm with 5 nm steps using a spectrophotometer (SpectraMax from Molecular Devices) and the enzyme activity was calculated as 1 U = −1 ABS/min. This activity unit was found suitable for quick determination of activity. T substrate concentration using a standard curve ( Figure S1) [7].
Starches are typically challenging to dissolve in aqueous solutions and tend to have a rather limited stability [13]. Starch solutions in dimethyl sulphoxide (DMSO) on the other hand can be used as stock solutions as they are stable for several months [14], not only simplifying the method but also increasing the comparability between separate experiments. Starches were either found to be soluble at 100 mg/ ml in pure DMSO (high amylose content) or at 25 mg/ml in 90% DMSO (low amylose content) and were thus concentrated enough to prevent inhibition of enzyme activity by DMSO [13][14][15].
Regarding the optimal wavelength for measurement, four starch substrates (potato amylose, Hylon VII, potato and rice starch) were tested. The estimated λ max of their complex with iodine ranged from potato amylose (λ max = 641 ± 8 nm), over Hylon VII (λ max = 590 ± 10 nm) and potato starch (λ max = 588 ± 12 nm) to rice starch (λ max = 562 ± 16 nm), being in good agreement with previous data [10]. Therefore, the wavelength of 610 nm was selected as it enabled detection of all substrates with a single wavelength (< 3% difference in the range 580-640 nm). The activity of amylolytic enzymes on starch was found to cause a blue-shift in λ max of varying strength, ranging from β-amylases (~25 nm/ABS on potato amylose, Fig. 1B) to α-amylases (~100 nm/ABS, Fig. 1A, for the full list see Figure S2 and S3). Fig. 1C shows the underestimation of the absolute absorbance compared to the relative absorbance measured at 610 nm of all ten studied amylolytic enzymes on potato amylose. Above a λ max of about 570 nm, the difference between the absolute and relative absorbance values was minimal (below 10%). The underestimation of absorbance led to an overestimation of activity due to a predicted faster decrease in absorbance. If the λ max fell below 570 nm, the activity overestimation was up to 25% which was observed for e.g. α-amylase on amylose below approx. 0.6 ABS at 610 nm. In this case it was recommended to estimate the enzyme activity based on the absorbance values at the λ max of each time point. For all experiments in which all data points exhibited a λ max within the given range, however, the overestimation was below 10%, providing a sufficiently narrow window for accurate detection of enzyme activity at a single wavelength.
The assay was designed for detection of the enzyme reaction at multiple time points as the enzyme activity was not always found to be identical throughout the entire assay time frame, rendering single-point analysis inaccurate, especially for low enzyme activity or non-linear trends (deceleration). Those two types of reactions are often difficult to differentiate from background activity. Therefore, investigations were conducted to establish a factor that could reliably distinguish between enzyme and background activity. Since the method was aimed for an array of amylolytic enzymes on a series of starches, the factor should be applicable to a range of initial absorbance values, length of incubation, and enzyme type. Analysis of a large dataset of ten amylolytic enzymes and negative controls on amylose (n = 97, see Supplementary Data) revealed that the main differences between background and enzyme activity were the linearity and level of decrease in absorbance. The linearity of the decrease was determined by the coefficient of determination (R 2 ) as it shows the percentage of data points of an experiment described by the estimated linear slope [16]. The level of decrease was calculated by using the ratio of the first and last absorbance value (ABS Start and ABS End , respectively) as it was less distorted by different starting absorbance values and non-linear decreases.
In general, high enzyme activity was found to be indicated by a sharp, highly linear decrease in absorbance over time (< -0.075 ABS/ min, R 2 > 0.95) that could conclude in a deceleration of absorbance decrease due to substrate depletion and thus a lower R 2 value. Background activity, on the other hand, resulted in a series of absorbance values that exhibited small, random fluctuations (about 7.5% of initial absorbance value), almost no decrease over time (> -0.007 ABS/ min) and a low linearity (R 2 < 0.4). Low enzyme activity generally showed a very linear decrease, however, the fluctuations in absorbance were more visible than for higher enzyme activities. Detection of low enzyme activity was further impeded by the fact that sometimes the deviations in absorbance of samples with no enzyme activity may seem non-random, giving the appearance of a trend and thus enzyme activity.
Taken together, basing the modeled Reliability Factor (F Rel ) on both linearity and level of decrease in absorbance enabled detection of both high and low enzyme activity. Additionally, cutoff values were introduced based on the careful manual identification on presence or absence of enzyme activity of each sample in the dataset ( Figure S4). The two components contributing to the F Rel were further weighted similarly to ensure a positive result whenever one of the factors was slightly negative but the other was sufficiently positive (see Equation (1)).
It should be noted that the F Rel provided only information about the reliability of presence or absence of enzyme activity and not the level of activity itself. The F Rel can, however, be used to estimate reliably if the derived activity rates can be trusted.
After establishment of the equation, the model was tested on a control experiment with three amylolytic enzymes as positive controls and two negative controls on four different starchy substrates. Fig. 2A shows the obtained raw data while Fig. 2B presents the estimated F Rel values. Both raw data and F Rel values showed a clear distinction between the positive and negative controls on all substrates. Rice starch exhibited the lowest absorbance values and decreases and thus the smallest F Rel values. β-Amylase showed only very low decreases and F Rel values with two of them even being slightly negative. This result, however, was not surprising as almost no decrease was visible in the raw data. Further, the observed non-linear trend for β-amylase indicating early substrate depletion was in agreement with its inability to bypass or cleave α-1,6-linkages [11,17]. Notably, statistical results (Pearson correlation, Table S1) estimated a significant correlation (linear regression) between time and absorbance values for all positive controls but none of the negative controls at a confidence interval (CI) of 99.9%, suggesting presence of enzyme activity even for β-amylase on rice starch. However, there was also a significant correlation between negative enzyme and potato starch at a CI of 99.5% (p = 0.002). Therefore, the parameters for the F Rel were not altered accordingly to prevent false positives. Additionally, F Rel values very close to zero indicate the requirement for a repetition of the experiment to obtain a more defined result.
In summary, a new procedure for the iodine assay has been proposed to conduct fast and easy experiments with multiple time points. All data points could be measured at a single wavelength (610 nm) whenever their absorbance maxima were within the range of 570-650 nm, providing accurate activity profiles of an array of amylolytic enzymes. Furthermore, a novel variable, the Reliability Factor, has been designed to provide an objective and simple way to distinguish between enzyme activity and background activity and its application was verified on both positive and negative controls on four starchy substrates. . There were three positive controls (0.2 μg/ml α-amylase from A. oryzae, 6 μg/ml amyloglucosidase from A. niger and 4 μg/ml β-amylase from B. cereus; all purchased from Megazyme) and two negative controls (63 μg/ml xylanase from A. niger (Megazyme) and negative enzyme) incubated with 1 mg/ml substrate (potato amylose, potato and rice starch from Sigma-Aldrich and Hylon® VII from Ingredion). Data are presented as mean ± stdev (n = 3).