Adaptation of a microfluidic qPCR system for enzyme kinetics studies

Microfluidic platforms offer a drastic increase in throughput while minimizing sample usage and hands-on time, which makes them important tools for large-scale biological studies. A range of such systems have been developed for enzyme activity studies, although their complexity largely hinders their application by a wider scientific community. Here we present adaptation of an easy-to-use commercial microfluidic qPCR system for performing enzyme kinetics studies. We demonstrate functionality of the Fluidigm Biomark HD system (the Fluidigm system) by determining kinetic properties of three oxidases in a resorufin-based fluorescent assay. The results obtained in the microfluidic system proved reproducible and comparable to the ones obtained in a standard microplate-based assay. With a wide range of easy-to-use, off-the-shelf components, the microfluidic system presents itself as a simple and customizable platform for high-throughput enzyme activity studies. Graphical abstract


Introduction
In the past decade we have been witnessing a drastic increase of genomic data deposited in public databases, which provides the potential for discovery of new enzymatic activities and deeper understanding of the known ones [1][2][3] . However, current experimental methods cannot keep up with characterisation of novel genes, resulting in a chronic lack of validated data in public databases 4 . Currently, the most common methods for in-depth enzyme characterisation rely on manual preparation of enzyme-substrate mixtures in a microplate, which is time-consuming and requires large amounts of protein and substrate. This usually results in a limited range of conditions and substrates being tested and hinders the exploration of enzymatic potential of a protein. Number of high-throughput methods have been developed for both end-point and kinetic measurements of enzyme activity [5][6][7] , however, their adaptation by a wider scientific community proves difficult, as it requires specialized knowledge and access to microfluidic manufacturing facilities 8 . A simple method for performing enzyme kinetic studies in readily available high-throughput off-the-shelf devices would help to address this problem.
Although no high-throughput microfluidic devices for measuring enzyme kinetics are commercially available, such instruments were developed for real-time qPCR applications 9,10 .
One such system was established by Fluidigm Corporation, where gene expression analysis is carried in integrated fluidic circuit chips, with a setup that offers an increase in throughput and reduction in sample volumes, yet remains user-friendly 11 . In the Fluidigm system samples and reagents are first pressure-loaded into nanoliter-sized reaction chambers of the chip, which is then transferred to a real-time PCR instrument designed to thermal cycle the microfluidic chips and image the data in real time. The Fluidigm ecosystem, including a number of different chips and a range of available fluorescent filters (Supplementary table 1), ind icates its potential for novel applications, in addition to qPCR. In this work we show that a qPCR microfluidic system can be easily adapted for enzyme kinetic studies with improved throughput and decreased sample usage over conventional methods.

Results
Characterisation of experimental platform: microfluidics qPCR device for enzymatic assays.
To assess the suitability of Fluidigm system for enzyme activity screening, we chose a FlexSix Gene Expression IFC chip which contains six partitions, each with 12 wells on the assay side and 12 wells on the sample side each ( Figure 1A). The chip provides a medium range of throughput: from 144 up to 864 reactions per chip, depending on how samples are routed in the chip. As a model enzyme we selected the hydrogen peroxide producing enzyme lactate oxidase (LOX, EC 1.1.3.2). Activity of lactate oxidase can be detected by a simple coupled fluorescent assay, in which a non-fluorescent probe reacts with hydrogen peroxide to produce a fluorescent product, resorufin (Supplementary figure 1).
We first tested over what range res orufin fluorescence was linear in the Fluidigm system by measuring a range of resorufin concentrations. Our results indicated that it is possible to capture linear resp onse for up to 5 µM of resorufin, which is comparable to data obtained in a microplate reader (Supplementary figure 2). Next, we aske d whether enzyme activity could be captured in the system, and measured the increase of fluorescence over time for different enzyme concentrations with 5 mM lactate. To lower adhesion of protein to the chip's channels and minimize protein precipitation, we tested two concentrations of nonionic detergents in a buffer, in presence or absence of BSA. Additionally, we used fluorescein dye as a loading control to inspect whether the mixing of the sample is consistent throughout the chip. The results showed that not only were we able to detect the activity of the enzyme, but also to capture its initial reaction rates ( Figure 1B). Additi on of BSA did not influence the results, while higher concentration of detergents provided a more c onsistent signal readouts throughout  Measurement of enzyme kinetics in the Fluidigm microfluidic system Next, we asked whether the Fluidigm system is reliable for studying enzymatic Michaelis-Menten kinetics. To test this, we measured initial reaction rates of three hydrogen peroxide producing enzymes: lactate oxidase, glucose oxidase (EC 1.1.3.4) and glutamate oxidase (EC 1.4.3.11). For each of the three oxidases, five different enzyme concentrations were placed in duplicates on the assay side of the FlexSix chip, and 11 substrate concentrations were placed on the sample side ( Figure 1C). During the sample loading, in each partition, all enzyme samples were separately mixed with all substrate samples, creating an enzyme-substrate gradient ( Figure 1DE). The use of different enzyme concentrations enabled capturing kinetic information in one run only, without the need of previous knowledge of the enzyme activity. Final enzyme concentrations were between 9.6 ng/ml and 6 µg/ml, and final substrate concentrations were between 1.7 µM and 100 mM. To enable rapid analysis of this data we developed R-scripts which automatically identifies the enzyme concentration for which initial rates are linear, fits a linear regression model to those data points, and extracts the slope.
When establishing a new method it is of key importance to validate it. We therefore set out to test the systems reproducibility as well as accuracy. In order to test reproducibility of the system, we repeated the assay for each enzyme three times using three different FlexSix chips.
The obtained kinetic parameters of the three replicates were highly similar (Table 1

Discussion
In this work we show that a microfluidic qPCR system can be applied in biochemistry without modification and is suitable for determining enzyme kinetic parameters. Although we have only tested one microfluidic system in this work, the method can likely be adapted to systems from other manufacturers. The Fluidigm system offers the well-known advantages of microfluidic devices over conventional methods: drastic reduction in both reagent volumes used, as well as time and manual handling involved in performing experiments 5 . Like many microfluidic devices, the Fluidigm system used here comes with the drawback of only being suitable for fluorescent-based assays, which limits the type of enzyme activities that could potentially be monitored. This, however, is becoming less of an issue with an increase of new fluorescent enzyme screens being developed 12 . The majority of microfluidic approaches for measuring enzyme kinetics rely on creating concentration gradients of enzyme and substrate inside a device 7 , which is not the case with the system tested here, where some degree of hands-on work is still required in the preparation of sample dilutions. However, the system presented in this work provides a clear advantage over the previously described methods: low system complexity, which allows users to operate it without fluid handling expertise or infrastructure required for microfluidic device fabrication. This is an important advantage, as the lack of easy to use devices is a frequent cause for low adaptation of microfluidics by non-specialists like biochemists 8 .
A further advantage of the system is that it allows parallelization of many enzyme kinetic measurements at the same time, with the possibility of testing different enzymes, substrates and conditions in a single chip. Availability of many easy to use, off-the-shelf chips makes the For capturing enzyme initial rates (Figure 1), five concentrations of lactate oxidase (1 mU/ml, 5 mU/ml, 10 mU/ml, 20 mU/ml, 50 mU/ml) were assayed with 5 mM lactate, in the buffer listed above as well as buffers containing no BSA and 0.1% of the detergents Tween20 and

Data analysis
Data obtained from the Fluidigm instrument was analyzed as follows. After a finished run, corners of the chip were adjusted manually in Fluidigm's Real-Time PCR Analysis Software and the quantification of fluorescent intensities in the ROX and FAM channels were carried automatically by the software. The run data containing fluorescent intensity values were exported in a *.csv format. Two input files describing sample layout on both "assay" and "sample" sides were manually created. Custom scripts in R version 3.4.4 ( www.r-project.org ) were used to analyze data. Briefly, the resorufin fluorescence (ROX channel) was background-subtracted and converted to product concentration by using a standard curve.
Regression models were fitted to the linear range of the data and the slope was calculated to obtain resorufin production per minute ( µM min -1 ). The enzyme specific activity ( µ mol mg -1 min -1 ) was finally calculated using a reaction volume of 8.87 nl and the protein concentration used in each reaction.
Data obtained from the FLUOstar Omega plate reader was analyzed in the same manner as data from the Fluidigm instrument. The only difference being that fluorescence intensities did not have to be extracted and that the reaction volumes were 20 µl.