Development and Evaluation of a Low-Cost Triglyceride Quantification Enzymatic Biosensor Using an Arduino-Based Microfluidic System

Overweight and obesity promote diabetes and heart disease onset. Triglycerides are key biomarkers for cardiovascular disease, strokes, and other health issues. Scientists have devised methods and instruments for the detection of these molecules in liquid samples. In this study, an enzymatic biosensor was developed using an Arduino-based microfluidic platform, wherein a lipolytic enzyme was immobilized on an ethylene-vinyl acetate polymer through physical adsorption. This low-cost optical biosensor employed a spectrophotometric transducer and was assessed in liquid samples to indirectly detect triglycerides and fatty acids using p-nitrophenol as an indicator. The average triglyceride level detected in the conducted experiments was 47.727 mg/dL. The biosensor exhibited a percentage of recovery of 81.12% and a variation coefficient of 0.791%. Furthermore, the biosensor demonstrated the ability to detect triglyceride levels without the need for sample dilution, ranging from 7.6741 mg/dL to 58.835 mg/dL. This study successfully developed an efficient and affordable enzymatic biosensor prototype for triglyceride and fatty acid detection. The lipolytic enzyme immobilization on the polymer substrate provided a stable and reproducible detection system, rendering this biosensor an exciting option for the detection of these molecules.


Introduction
Triacylglycerols, commonly known as triglycerides (TGs), are composed of a glycerol backbone esterified with three fatty acids [1]. TG concentration measurement of samples has gained increased importance due to its essential role in cellular metabolism, as an energy source, and for lipoprotein formation for the transport of dietary lipids. Determining triglyceride concentration can also provide information about potential health risks associated with various diseases [2,3].
Typical TG blood levels range between 40 and 150 mg/dL; however, it is considered borderline high if the level falls between 151 and 199 mg/dL. On the other hand, a level of 200 to 499 mg/dL is classified as very high [4]. Individuals with high levels of fatty acids and TGs tend to present conditions such as obesity, lipid disorders, and insulin resistance, leading to various diseases, such as heart attacks, strokes, pancreas inflammation, and atherosclerosis [5][6][7][8]. When the level of TGs in serum exceeds 500 mg/dL, it is used as a biomarker for cardiovascular disease [9].
An alternative for measuring triglyceride levels is through the use of biosensors. This is a tool that can help in the diagnosis, monitoring, and control of diseases related to of polydiacetylenes and agarose, changes color from blue to red when exposed to spoiled almond milk.

Reagents and Materials
Chemicals used in this research include recombinant lipase B from Candida sp., expressed in Aspergillus niger, trioctanoin, p-nitrophenol (p-NP), p-nitrophenyl butyrate (p-NPB), Bradford Reagent, and bovine serum albumin, all of which were obtained from Sigma-Aldrich ® (St. Louis, MI, USA).
In addition, the experimental setup included the utilization of various materials. For instance, the Ethylene-vinyl acetate (EVA) polymer or XP500 sourced from PromaPlast was employed. In addition, solvents and reagents encompassed ethanol, isopropyl alcohol (2-propanol), 3-(N-morpholino) propane sulfonic acid (MOPS), Triton X100, and phosphate buffer (50 mM, pH 7.2). All other chemicals and solvents used were of analytical reagent grade and spectroscopic grade, respectively. Furthermore, the equipment configuration featured a violet LED (light-emitting diode), a transduced BH1750 sensor, a 1.3 OLED display using I2C communication, PLA filament for 3D printing, and an Arduino micro board.

Instruments Used
An LED and optical transducer BH1750 were used to measure absorbance in the biosensor. The absorbance signal from the reaction mixture in the biosensor was compared with a Thermo Scientific™ Multiskan SkyHigh spectrophotometer. Additionally, essential laboratory equipment, such as an AREC magnetic stirrer and a BIORAD UltraRocker lateral stirrer, were utilized. The microfluidic device was fabricated using Ultimaker s3.

Enzyme Activity and Triglyceride Quantification
In order to determine enzymatic and triglyceride quantification, two different methods were employed. The first methodology was developed by Kordel et al. [32] and uses a chromogenic substrate; this method has a high level of sensitivity to confirm enzyme immobilization and transducer applicability.
However, the Kordel method does not involve the use of TGs; for that reason, the methodology of Mateos et al. [33] was implemented, which allows for the use of a triglyceride and p-nitrophenol as a chromophore to establish the level of fatty acids. To determine the concentrations of TGs, Glasser et al. [34] devised a method that takes into account three fatty acids released to calculate the activity. In the case of the CalB enzyme, it releases a fatty acid at the sn-1 position of the triglyceride structure [35,36], generating a diacylglycerol [37].
Furthermore, the protein concentration in the enzymatic extract was quantified using the procedure described by Bradford, which involves the use of bovine serum albumin as a standard [38]. This allowed for the estimation of the specific activity of the lipase, expressed in U/mg.

Enzyme Activity Using p-NPB as a Substrate
To determine enzymatic activity, the methodology of Kordel et al. [32] was used with some modifications. The Kordel method utilized p-nitrophenyl butyrate as a chromogenic substrate due to its exceptional sensitivity, quickness, and user-friendliness. For further details, please refer to Figure 1. Initially, a stock solution of p-nitrophenyl butyrate (p-NPB) at 10 mM in 2-propanol was prepared and kept frozen until use. Additionally, a phosphate buffer solution at 50 mM, pH 7.0, was prepared.
For each well of the microplate, 180 µL of buffer (18 parts), 10 µL of substrate (one part), and 10 µL of enzyme extract or immobilized enzyme (3 mg) were mixed. The reaction was carried out at a temperature of 30 • C for 9 min, with 30 s intervals between each reading at a wavelength of 415 nm. Immediately afterward, the molar extinction coefficient of the released p-nitrophenol p-NP ( ) was measured spectrophotometrically at 415 nm using a standard curve of p-NP, with concentrations ranging from 0 to 500 µM. One unit (U) of activity is defined as the amount of enzyme that will convert one µmol of substrate (p-NP) in one minute (µmol/min) [39][40][41]. Equation (1) shows how the enzymatic activity was calculated [42]: where: ∆A = kinetic slope (Abs/min); α = reaction volume (µL); β = sample volume (µL); = molar extinction coefficient (min/µmol); γ = appropriate sample dilution.

Enzymatic Activity Using TGs and p-Nitrophenol as a Reaction Indicator and Calculation Process of TG Quantification
The methodology of Mateos-Díaz et al. [33] was employed, with certain modifications; see Figure 2. A stock solution of trioctanoin at 25 mM in 2-propanol and a solution of p-NP 2.5 mM in 2-propanol were prepared. The buffer used was MOPS 2.5 mM pH 7.2, supplemented with 0.5% (w/v) Triton X100. The reaction mixture consisted of an emulsion containing 0.5 parts (5 µL) of triglyceride solution, 0.5 parts (5 µL) of p-NP solution, nine parts (90 µL) of buffer, and two parts of either enzyme extract or immobilized enzyme. A standard curve of different concentrations of octanoic acid from 0 to 10 mM in the presence of p-NP was measured spectrophotometrically at 415 nm under the same conditions as the reaction. One unit (U) of activity is defined as the amount of enzyme that will convert one µmol of substrate (octanoic acid) in one minute (µmol/min); see Equation (1). After calculating the enzyme activity, it was multiplied by the duration of triglyceride hydrolysis to determine the number of moles released. Similar to the calculation of TGs, the conversion to mg/dL was carried out [43]. For the validation of the biosensor, the release of octanoic/caprylic acid was used as a standard to determine the total amount of TGs (trioctanoin) [44]. This is based on the moles of fatty acids obtained from enzymatic activity in triglyceride calculations [34,45]. According to the literature, the CalB enzyme has a preference for the sn-1 bond in the triglyceride structure [35,36], resulting in the release of a single chain of fatty acids and 2-monoacylglycerol molecules [37]. Therefore, the calculation for releasing one mole of fatty acid is based on this specificity and multiplied by three to estimate the level of TGs.
Based on the described methodology, the final TG levels in the sample can be estimated using Equation (2), where the enzyme activity reported in µmol/min is multiplied by the reaction time, the three fatty acid chains, the molecular weight of the trioctanoin, and a conversion factor to convert from (µmoles, mL, g) to (mol, dL, mg) to obtain mg/dL of the TGs in the sample.

Enzyme Immobilization with Ethylene-Vinyl Acetate (EVA) Polymer by Physical Adsorption
The methodology of Manoel et al. [46] was used as a base with some modifications, using the physical adsorption method. The EVA pellets were cut to <1 mm, and 0.5 g of the input was weighed. Subsequently, the polymer was washed with 5 mL of ethanol for 30 min under constant agitation. After that, 10 mL of water was added, and agitation was continued for 30 min. The mixture was then filtered and dried at room temperature. Following the activation of the support, 0.1 g of pellets were placed in Eppendorf tubes.
For immobilization, eight different concentrations of the lipase enzyme and phosphate buffer were used, resulting in a final volume of 1000 µL in each Eppendorf tube ( Table 1). The software Statgraphics Centurion XV ® was utilized to improve the enzyme and polymer concentrations. The ideal conditions for immobilization were determined through a two-way ANOVA analysis, taking into account the size of the polymer (cut or whole) and the concentration of the enzyme (ranging from 10 µL to 700 µL; see Table 1) as factors. Before conducting the ANOVA analysis, a Box-Cox transformation was executed by using the square root of the enzyme activity to obtain a normal data distribution.

Construction of Optical TG Biosensor and Response Measurement
Regarding the transducer, optical sensor BH1750 was used as a receiver, coupled with a violet LED and an Arduino micro board equipped with a 1.3 OLED display I2C module with a resolution of 128 × 64 and an SH1106 driver as output construction, similar to the setup described by Nandiyanto et al. [27].
Arduino code was developed, along with the corresponding electrical circuit, to trigger the emission of a light beam from the LED upon pressing a button. The sensor captures the signal transmitted through the camera and automatically calculates the necessary data for determining the concentration of TGs. To minimize light reflection, the device was coated with a dark-colored material. Prior to sampling, the device and the light intensity were calibrated using distilled water.

Characterization Immobilization of the CalB Lipase in EVA by Physic Adsorption
The ANOVA analysis, conducted at a 95% confidence level, did not reveal any statistically significant difference in the tested polymer sizes (p-value = 0.2324 > 0.05). Additionally, there was no interaction between polymer size and enzyme concentration (p-value = 0.7839 > 0.05), indicating that the effects of enzyme concentrations are independent of polymer size.
On the other hand, the enzyme concentration had a statistically significant effect on the enzyme activity in the immobilized samples (p-value = 0.0369 < 0.05). Different amounts of the enzyme were added to the immobilized solution to enhance the process ( Table 1). The graph in Figure 3 shows the correlation between the enzyme volume in the polymer and its enzyme activity (U/ml) and specific enzyme activity (U/mg). The lipase activity escalated exponentially in the initial four samples (10 µL, 50 µL, 100 µL, and 200 µL), reaching its maximum at 200 µL of the CalB enzyme, with an activity of 5.6118 U/mL. Enzymatic activity remains within the range of 5 to 6 U/mL when the enzyme quantity ranged from 200 µL to 500 µL. Therefore, the immobilized product was selected with 200 µL of the CalB enzyme, as higher concentrations would result in unnecessary enzyme waste. It is essential to mention that, according to this immobilization optimization, 200 µL of the enzyme was used for each 100 mg of EVA, and only 9 mg to 54 mg of immobilized enzyme were placed in the biosensor. The physicochemical properties of EVA (melting temperature, crystallinity, flexibility, adhesion, among others) vary depending on the vinyl acetate (VA) content, expanding the uses for this copolymer [47][48][49][50].
Pedersen and Eigtved [51] mentioned that the support should have a reduced pore size of no less than 100 Å. According to Cai et al. [52], the immobilization of the CalB enzyme was evaluated using different pore diameters (6.6, 8.1, and 12.5 nm), finding the highest enzymatic activity with a pore size of 12.5 nm (125 Å). Therefore, the EVA polymer, with a variable pore size ranging from 5 µm to 9 µm (depending on the %VA) [53], is considered a suitable support for the immobilization of the CalB enzyme.

TG Biosensor Fabrication
The device has a final dimension of 35 × 36 × 17 mm in width, height, and depth (W × H × D), respectively. It consists of two interconnected sections for sample transport and analysis through channels. Figure 4A shows the 3D printed microfluidic device with a cross-section providing a view of the internal parts with detailed sections. Figure 4B shows two sections. The first section includes a first chamber with dimensions of 8.5 × 8.5 × 5.75 mm (W × H × D) for the initial process involving the immobilized enzyme. The second section incorporates an LED light source, Gy-302 BH1750 light intensity sensor for spectrophotometric measurements, and a second chamber measuring 9 × 9 × 4.55 mm (W × H × D) to analyze the sample in a manner similar to the "stopped flow" methodology in a sequential injection analysis system [42].
The channels used for fluid transportation have a diameter of 1 mm. In the second section, two compartments are specifically designed for the light emitter and receiver. The location for the LED is cylindrical, with a radius of 3.25 mm and a width of 12 mm. Additionally, there is a circular opening of 2 mm for the passage of light into the chamber where the sample is placed. On the other hand, the BH1750 sensor section has dimensions of 11 × 19.5 × 1.5 mm (W × H × D) and includes a cylindrical opening of 2 mm positioned near the receiving sensor area. A potentiometer was used to regulate the intensity of emitted light in the spectrophotometer. Additionally, a violet-hued light-emitting diode (LED) was employed; this is the color absorbed in the wavelength range of 390-435 nm, reflecting the greenish yellow color observed in the enzymatic reaction [54].
A BH1750 light sensor was used to measure transmitted light. The biosensor was controlled using an Arduino Nano V3 board, programmed with Arduino software version 1.8.16. The OLED display I2C was connected as the output. Refer to the connection diagram in Figure 5A for more details. Moreover, in Figure 5B, the 3D printed device integrated with the electronic components on a breadboard can be observed.
To prevent light reflection, the device was painted in a dark color. Prior to sampling, the device and light intensity were calibrated using water as a reference substance.
It is important to acknowledge that using a low-cost sensor like the BH1750 and an LED light source may result in reduced sensitivity compared to commercial laboratory spectrophotometers, where it is possible to select a specific wavelength. The present transducer cost less than USD 1 (code 840, ISO 3166-1), and the Arduino Nano board cost less than USD 6 (code 840, ISO 3166-1), which makes this technology easily accessible for developing countries. The biosensor also has integrated microfluidic technology built on an easily accessible 3D printer and cost-effective electronics. The transducer and Arduino board have been validated in both chemical and biomedical contexts [27,28]. Commercial laboratory spectrophotometers, which were used for validation in this study, employ a xenon light source and a temperature control system, enabling measurements with high accuracy and precision, but at higher costs. In contrast, the device implemented in this work lacks a temperature control system and utilizes an LED light source, enabling measurements in a wavelength range of 390-435 nm approximately. Nonetheless, it offers the advantages of small size, portability, lower cost, and in situ monitoring.

Response of the TG Optical Biosensor
Once the biological and electronic components of the device were integrated, tests were carried out using the immobilized material, to compare the kinetics in the spectrophotometer with the kinetics in the biosensor.
According to the results, the biosensor exhibited an enzyme activity of 5.4176 U/mL, while the spectrophotometer measured a slightly higher activity of 6.3647 U/mL. This difference in measurements can be attributed to the sensitivity of the BH1750 sensor, which may be slightly affected at low levels of received light [55]. Additionally, the lack of a temperature control system in the device can lead to a decrease in the enzymatic kinetic reaction rate.
Based on the data provided, the Kordel method adapted to the biosensor demonstrated an accuracy rate of 85.12%, with a coefficient of variation of 1.51%.
In this study, trioctanoin was used as stock solution, with a concentration of 25 mM or 58.835 mg/dL to estimate the concentration of TGs. Dilutions were performed to avoid saturation, and the absorbance was adjusted to 1.4 at the beginning of the reaction. The enzyme's reaction rate was measured from this point.
Based on the described methodology, the measurement of trioctanoin was performed using 54 mg of immobilized enzyme, as shown in Figure 6B. The resulting slope was calculated as 0.0622. The enzyme CalB exhibited an enzymatic activity of 0.0338 U/mL over a reaction time of 10 minutes. After calculating the enzyme activity, it was multiplied by the duration of triglyceride hydrolysis to determine the number of moles released. These results can be used to estimate the final TG levels in the sample based on Equation (2). Therefore, the amount of triglyceride calculated is 47.7270 mg/dL. Comparing this result with the TG standard concentration of 25 mM (1.25 µmol/mL or 58.835 mg/dL), the obtained percentage of recovery of added trioctanoin in the sample was 81.12%.  Table 2 provides a summary that compares the analytical attributes of different TG biosensors. This comparison encompasses optical, amperometric, potentiometric, impedimetric, and deoxymetric biosensors, detailing the enzyme type, immobilization method, optimal temperature, optimal pH, minimum detection limit, recovery, precision, response time, and accuracy.
Detection Limit. According to the analysis of enzyme kinetics, the maximum achievable concentration of TGs is 58.835 mg/dL without dilution. The minimum enzymatic activity is 0.005435 U/mL, which corresponds to 2.3512 mg/dL of fatty acids and 7.6741 mg/dL or 0.1630 mM of trioctanoin. The minimum detection limit is lower than that of the hydrogel pectin-based triglyceride optical biosensor (1.2387 mM [13]) and higher than the nanocomposite biosensor (0.0113 mM [29]). The minimum detection limit reported in some biosensors falls within the range of amperometric biosensors (10 −7 mM to 14 mM), potentiometric biosensors (9 × 10 −5 mM to 5 mM), and DO metric biosensors (0.0005 mM) [9].
Precision and response time. The within-batch coefficients of variation (CV) for TG determination in samples by the present sensor were 0.79% (n = 5), which is comparable to some colorimetric-based biosensors (2.5% and 0.0089% in [13] and [29], respectively) and lower than other electrochemical transducer biosensors (4.14% in [25]). The response time in this biosensor is 600 s, allowing for comparison with the behavior of the enzyme and the triglyceride with some colorimetric-based biosensors (300 s and 360-420 s in [13] and [29], respectively). On the other hand, certain amperometric, potentiometric, and DO metric biosensors show response times in the range or 2.5-1200 s, 30-2700 s, and 2.0-300 s, respectively; [9]. Accuracy. To study the accuracy of the present transducer, p-nitrophenol was measured by the Kordel method over 10 min with 30 s intervals, and the release of p-nitrophenolate ions by the spectrophotometer method was compared with the present biosensor. The values obtained by both methods showed a good correlation, with r 2 = 0.9635 ( Figure 6A), which is superior to the PVC membrane biosensor and the CA membrane-bound enzyme biosensor (r 2 = 0.83 [23,57]) and comparable with the OD metric biosensor, EIS-CAP sensor, impedimetric/conductometric technique-based TG biosensor, and amperometric TG on pencil graphite electrode biosensor (r 2 = 0.9409 [60], r 2 = 0.9584 [58], r 2 =0.9801 [59], and r 2 = 0.99 [10], respectively).
While various types of triglyceride biosensors have been developed, a novel biosensor that uses an unreported polymer to immobilize CalB enzymes was implemented, and the immobilized enzymes were tested. The present biosensor offers an economical, affordable, and accessible option for monitoring TG levels.
In this context, trioctanoin is used in medicine and research as an alternative energy source for the body and was used as a standard in the determination of TGs in this work. The consumption of trioctanoin is an alternative to moderately increasing ketones without radically altering food intake or restricting carbohydrates. The diet with these TGs has been used for many years to treat refractory childhood epilepsy [61].
It is important to measure octanoic acid, since studies have identified it as a biomarker linked to the development of colorectal cancer [62].

Conclusions
An optical biosensor for TGs was established based on the immobilized CalB enzyme and a microfluidic device. All necessary parameters for constructing this biosensor have been determined, resulting in a detection limit and a response toward TG concentration.
This biosensor is excellent for detecting TGs with acceptable accuracy and precision. The experiments demonstrate the potential of this biosensor to detect TGs. Moreover, there are plans to further develop and enhance this specific TG biosensor to optimize its performance and customize it for use in natural blood serum. The objective is to improve its functionality and ensure its compatibility for practical applications in clinical settings.  Acknowledgments: This work would not have been possible without the financial support of Guadalajara University and the Division of Natural and Technological Sciences.

Conflicts of Interest:
The authors declare no conflicts of interest.

Abbreviations
The following abbreviations are used in this manuscript: