Biocatalytic membrane for membrane inlet mass spectrometry

Abstract

The study presents a method for the immobilization of enzymes on a polydimethylsiloxane membrane probe for membrane inlet mass spectrometry. The method uses an eggshell membrane as a gas-permeable layer containing amino functional groups of lysine residues, which are capable of covalent linkage with enzyme molecules through glutaraldehyde. The membrane probe with immobilized enzymes enables determination of analyte concentrations that do not themselves diffuse through the membrane, provided the analyte participates in an enzymatic reaction that generates or consumes a diffusible substrate or product. This study uses glucose oxidase as a model enzyme and employs it to determine glucose concentrations in complex samples. Wine, fruit juice, milk and bovine blood plasma, as well as possible interferents and substrate analogs, were analyzed using this method and compared with a glucose oxidase amperometric biosensor. The linear response range of the sensor was 50 to 1000 µmol·L^-1 glucose concentration, the response time was approximately 120 s, and the sensitivity was 67.8 µmol·L^-1 decrease in dioxygen concentration per 1 mmol·L^-1 of glucose. The effects of pH, temperature and buffer concentration, as well as stability during long-term storage, were tested. The eggshell membrane with immobilized enzymes was characterized by scanning electron microscopy.

Introduction

Direct protein immobilization onto the surface of a polydimethylsiloxane (PDMS) membrane is a complex and laborious process¹. Therefore, to provide an easy, fast and inexpensive method, the PDMS membrane probe of a membrane inlet mass spectrometer (MIMS) was wrapped in an eggshell membrane (ESM) serving as a biocompatible layer suitable for enzyme immobilization, an approach that has also been successfully used earlier in electrochemical biosensing methods^2–4. ESM is a thin biological membrane material permeable to gases and rich in protein (80–85%), with a lysine content of 3.5%^5,6. Due to the lysine residue content, enzymes can be immobilized using glutaraldehyde (GA).

Our paper presents an alternative method for determining glucose (Glc) concentrations using glucose oxidase (GOx) immobilized on a MIMS probe. This approach is not intended to compete with already established GOx biosensors; the objective is rather to present an innovative analytical framework that could employ enzymes for biosensing applications, particularly in scenarios where alternative analytical approaches may be advantageous. One such example is Candida rugosa lipase immobilized onto a PDMS membrane of a MIMS probe to determine the concentration of isopropyl benzoate and monoisopropyl phthalic acid¹. The lipase catalyzes the hydrolysis of the ester bond in these substrates, releasing isopropyl alcohol (IPA) as one of the products (Fig. 1). Being a small, volatile molecule, IPA readily pervaporates through PDMS membrane and is detected by the MIMS, while the parent esters themselves are non-volatile and cannot permeate the membrane directly.

Fig. 1.

Schematic representation of CRL catalysed reactions of isopropyl esters.

GOx was selected for this study as an inexpensive and stable protein⁷ commonly used for the analysis of Glc in biosensors, where it has the status of a model enzyme⁸. It also allows us to compare results from a MIMS biosensor with an electrochemical biosensor. GOx (EC 1.1.3.4, β-D-glucose:O₂ oxidoreductase) is a homodimeric protein with two non-covalently bound flavin adenine dinucleotide (FAD) coenzymes. GOx catalyzes the oxidation of β-D-glucose to D-gluconolactone using dioxygen (O₂), which is reduced to hydrogen peroxide (H₂O₂) (Fig. 2). Both products spontaneously decompose in aqueous solution. The cyclic form of D-gluconolactone is hydrolyzed, producing gluconic acid, and H₂O₂ undergoes disproportionation, releasing O₂ and H₂O⁹. GOx, usually extracted from Aspergillus niger, is a globular glycoprotein with a molecular mass of about 155 kDa and an average diameter 8 nm⁷; based on sequence data, its lysine residue content is estimated at 2.5%¹⁰.

Fig. 2.

Schematic representation of GOx catalysed oxidation of Glc.

Approaches to determining specific Glc concentration commonly include enzymatic and non-enzymatic sensors¹¹ that are mostly based on electrochemical¹² or optical methods¹³. In the past 25 years, nanomaterials have been utilized to further improve these techniques¹⁴. GOx biosensors have been a well-established analytical device since 1962¹⁵ and have found many applications, for instance, determining Glc concentrations in the blood^16,17, saliva¹⁸ and sweat¹⁹ of patients with diabetes, as well as in food industry quality control²⁰, product freshness assessment²¹ and in the monitoring of manufacturing processes²².

The PDMS-ESM-GA-GOx, a modified biocatalytic membrane probe for MIMS, specifically decreases the concentration of dissolved O₂ permeating through the ESM by a means of a GOx enzyme reaction in proportion to the Glc concentration in the solution. Gases, including O₂, then pervaporate via a semipermeable PDMS membrane which separates the vacuum inlet of the mass spectrometer from the analyzed solution. Pervaporation is an active diffusion process through a membrane, driven in MIMS by a pressure gradient. It is a three-step process that comprises the partitioning of analytes from the analytical matrix into a vapor phase. The first step involves selective adsorption of the analyte on the membrane surface, the second step concerns permeation through the membrane, often determining the rate of diffusion, and the third step is desorption from the membrane into the vacuum²³.

Material and methods

Chemicals and reagents

GOx from Aspergillus niger type X-S, in the form of lyophilized powder with a specific activity of 167,978 units·g^-1, was obtained from Merck (Kenilworth, NJ, USA). Bovine serum albumin (BSA) of 98% purity was also from Merck. D-Glucose (Glc) monohydrate (99.8%), K₂HPO₄ (99%) and KH₂PO₄ (99%) were acquired from Penta (Prague, CZ). The Glc stock solution (100 mmol·L⁻¹) was allowed to reach mutarotation equilibrium for at least 24 h before use. The KOH and H₃PO₄ (Penta) were 84.5% and 85%, respectively. GA was in the form of a 25% aqueous solution (Penta). NaBH₄ (Merck) was 98%. Sterile 10 × concentrated aqueous solution of phosphate buffered saline (PBS) pH 7.4 was obtained from Carl Roth (Karlsruhe, DE). Absolute ethanol (99.9%) (EtOH) was from VWR Chemicals (Radnor, PA, USA). Sodium citrate (99%) (Cit) dihydrate and 99.9% sucrose (Suc) were also from Penta. Sodium L-ascorbate (99%) (Asc) was also from Carl Roth. Both D-galactose (Gal) and D-mannose (Man) were 99.9% from Lachema (Brno, CZ). A gold target of 99.99% purity was used for sputtering samples for SEM. The water used throughout the process was Milli-Q grade, with its conductivity measured at ≤ 0.066 µS·cm^-1.

Cabernet Sauvignon dry red wine (13% (V/V) EtOH) from Bíza (Čejkovice, CZ) and Riesling semi-dry white wine (12.5% (V/V) EtOH) from Vladimír Tetur (Velké Bílovice, CZ) were donated by Masaryk University (Brno, CZ). Zweigeltrebe semi-dry rosé wine (13% (V/V) EtOH) from Zámecké vinařství Bzenec (Bzenec, CZ) was purchased at a local grocery store. Grape juice 100% natural Pinot Noir from Prifilom (Brno, CZ) and Mango Relax juice with a 25% fruit content from Maspex (Prague, CZ) were obtained from a local grocery store. Ultra-High Temperature (UHT) processed whole cow milk with a 3.5% fat content produced by Madeta (České Budějovice, CZ) was also from a local grocery store. Bovine blood plasma samples were provided by Mendel University (Brno, CZ).

Instruments

All measurements were taken using an HPR-40 DSA from Hiden Analytical (Warrington, GB). An MS GAS-100 from Photon Systems Instruments (Drásov, CZ) was used for comparison. Both quadrupole mass spectrometers were equipped with electron ionization and a Faraday detector. The scans consisted of 32 atomic mass unit (AMU) for dissolved O₂ monitoring and 40 AMU for dissolved argon as reference gas. The inlet interface of the HPR-40 DSA and MS GAS-100 was modified using a stainless-steel joint reducer from tubes available on the devices to a 1/16" 316L stainless-steel capillary probe. The capillary probe connection with a rubber O-ring finger-tight seal provided easy plugging of a 1/16" capillary PDMS membrane probe. For amperometric detection, a PalmSens4 potentiostat from PalmSense BV (Houten, NL) was used, with a single channel screen-printed electrode AC13.2.S with platinum (Pt) working electrode and a silver (Ag) reference electrode from BVT Technologies (Strážek, CZ). The temperature dependency experiment was carried out using a Haake SC100 water bath with a Haake S 5P module from Thermo Fisher Scientific (Waltham, MA, USA). The SEM analysis was carried out on a LYRA3 from Tescan (Brno, CZ). Prior to imaging, samples were coated by RF magnetron sputtering on a Leica ACE 600, Leica Microsystems (Wetzlar, DE) to avoid charging. The lyophilization was carried out using a Coolsafe™ from Scanvac.

PDMS membrane probe

PDMS hollow-fiber membrane tubing was Silastic acquired from Dow Corning (Midland, MI, USA), with respective outer and inner diameters of 1.96 mm and 1.47 mm. As the tubing was of medical grade, it was used directly without further cleaning. The PDMS membrane tubing was cut to a length of 10.5 mm and secured on one end with a cylindrical stainless-steel pin, with a 2 mm diameter and 3.5 mm length. To support the shape of the PDMS membrane under vacuum conditions, the other end of the tube was fitted with a stainless-steel spring with the following parameters: 0.2 mm wire diameter, 1.5 mm spring diameter, 3.2 mm spring length, and 7.5 threads. A 7.2 mm length of a 1/16" stainless-steel capillary was then inserted into the PDMS hollow fiber membrane from the open end. The resulting probe, with the securing pin, the supporting spring and a quarter of the stainless-steel capillary length covered with the PDMS membrane, had a total length of 13.9 mm. The sampling area of the probe for the MIMS method is in the region of the supporting spring, with a surface area of approximately 20 mm², the working pressure inside the vacuum chamber of the mass spectrometer being approximately 1.2 × 10^–6 mbar when immersed in a solution.

Preparation of the GOx biocatalytic membrane probes

Ordinary hen eggs were purchased at a local grocery store, purchasing the freshest available. The egg was cracked, the white and yolk were removed, and the inside of the shell was rinsed with water. The ESM was then peeled off and placed on a Petri dish, where it was rinsed multiple times with water. The cleaned ESM was placed onto aluminum foil and cut into a square shape of approximately 6 × 6 mm. The square piece of ESM was wrapped one and a half times around the PDMS membrane fitted onto the probe, following the natural curvature of the ESM so that the outer surface, which would receive GOx immobilization, faced outward. The wrapped ESM was secured onto the membrane probe with cotton sewing thread around the stainless-steel capillary and stainless-steel securing pin. The membrane probe, wrapped with the ESM, was then submerged in a freshly prepared 0.5 mL 3% solution of GA in PBS for half an hour in 1 mL Eppendorf tube at room temperature (24 °C). Subsequently, the membrane probe was rinsed with water and submerged in 100 µl 1 mg·mL^-1 freshly prepared solution of GOx in PBS for one hour at 24 °C in 200 µl Eppendorf tube, rinsed with water again, and the Schiff base secondary amines were reduced using a freshly prepared 0.5 mL 4 mg·mL^-1 solution of NaBH₄ in PBS for 15 min at 24 °C. These steps were conducted in static conditions. The resulting reduced PDMS-ESM-GA-GOx probe was rinsed again with water. Probes were stored in the fridge at 4 °C in an Eppendorf tube inside glass tubes sealed with parafilm and containing a water reservoir to prevent the ESM, which tends to crack, from drying. A schematic illustration of the ESM-wrapped MIMS probe is shown in Fig. 3.

Fig. 3.

Schematic of MIMS PDMS probe with ESM. 1: stainless-steel hollow capillary connected to mass spectrometer, 2: stainless-steel spring supporting PDMS membrane, 3: ESM wrapped around the probe, 4: stainless-steel sealing pin, 5: PDMS membrane.

Preparation of electrodes with immobilized GOx

The GOx enzyme was immobilized onto the working Pt electrode surface via a cross-linked protein aggregate. BSA at 50 mg·mL^-1 and GOx at 25 mg·mL^-1 were prepared in PBS. GA was diluted from a 25% stock solution to a final 3% concentration in water. The protein immobilizing mixture was prepared from 10 µL of GOx solution, 9 µL of BSA solution, 50 µL of PBS and 2 µL of GA solution. A 1 µL droplet of the protein-GA mixture was deposited onto the Pt electrode surface. The electrode was placed in a Petri dish, sealed with parafilm, and stored overnight at 4 °C in humid conditions maintained by water-soaked tissue placed inside the dish. After incubation, the electrode was rinsed with water and stored dry at 4 °C until use.

Scanning electron microscopy (SEM)

Samples were prepared in the same way as described in 2.4: one sample with the GOx enzyme and one blank sample without the enzyme, serving as a reference. Samples were then frozen at -80°C, lyophilized at -70°C under vacuum overnight, and stored in a desiccator with silica gel until use. Prior to imaging, lyophilized samples were coated with a 10 nm-thick gold film to avoid charging. The SEM analysis was carried out in secondary emission mode with a 10 kV acceleration voltage and a working distance of about 9 mm. For SEM characterization, multiple 10 µm field-of-view images of both the blank and GOx-immobilized ESM samples were acquired. The pixel size on these images was 7.8 × 7.8 nm and the tool for measuring distances on images was: https://tools-py.e-cyanobacterium.org/ “Cell size – filamentous strains”. The GOx molecules (approximately 8 nm in diameter) were sputter-coated with a 10 nm gold layer. Structures approximately four pixels in size, clearly distinguishable from the native ESM surface, were sought in the images.

Measurements

The determination of Glc concentrations using the GOx-modified MIMS probe was carried out in a glass spectrophotometric cuvette at room temperature (24 ± 1 °C) with 2 mL of a 100 mmol·L^-1 potassium phosphate buffer solution (pH 7.0) which was stirred with a cuvetteMIXdrive 1, a magnetic stirrer from 2mag AG (Munich, DE), at the maximum rotation speed of 1,200 rpm. The magnetic stirrer bar was 5 mm in length and 3 mm in diameter. For pH-effect experiments, 200 mmol·L^-1 buffers prepared from a corresponding mixture of K₂HPO₄ and KH₂PO₄ were employed; the pH was then measured with a pH electrode, and its value was adjusted using 1 mol·L^-1 KOH or 1 mol·L^-1 H₃PO₄. Glc samples were introduced into the cuvette via a micropipette and the subsequent decrease in O₂ concentration was monitored by MIMS. Dissolved O₂ concentration was quantified by single-point calibration against air-equilibrated buffer following²⁴ (analytical precision 1%). The equilibrium O₂ concentration at 24 °C and atmospheric pressure (101.325 kPa) was calculated as 252 µmol·L⁻¹ using Henry’s Law²⁵. Changes in dissolved O₂ concentration following sample addition were then determined by linear scaling from the baseline signal. Chronoamperometry using GOx immobilized on a Pt electrode was conducted under the same conditions as MIMS measurements, but with a buffer volume of 3 mL; the production of H₂O₂ was monitored. The working Pt electrode potential was set to + 600 mV vs. Ag reference electrode. Uncertainties in all figures, and the table are expressed as standard deviations of three replicates.

Results and discussion

Response behavior of MIMS probe modified with GOx

The capacity of MIMS to accurately measure the concentration of gases dissolved in a solution was harnessed to determine the decrease in O₂ concentration in the presence of Glc using the membrane probe with the immobilized GOx enzyme. All experiments were conducted with three repetitions using the same probe. The responses to Glc of PDMS, PDMS-ESM-GA and PDMS-ESM-GA-GOx membrane probes are compared in Fig. 4. A slight baseline drift is common in MIMS signals. In our case, normalization of the O₂ signal with the Ar signal to address this drift was not an option due to the noise being multiplied from both signals. Therefore, the signal values were read out 150 s after adding a sample, or, more precisely, 120 s after the signal deviation from the baseline became clearly visible. Reported response time with advanced GOx electrochemical biosensors is typically 3–10 s^26,27. The GOx-modified probe was calibrated over a broad range of Glc concentrations and exhibits a limit of detection of 50 µmol·L^-1 (9 ppm), linear response range 50–1000 µmol·L^-1 Glc, with a sensitivity of 67.8 µmol·L^-1 O₂ concentration decrease per 1 mmol·L^-1 of Glc (Fig. 5). For comparison, the typical linear range of electrochemical GOx biosensors is 0.1–30 mmol·L^-1 with an indicated sensitivity of 0.2–221 µA·mmol·L^-1·cm^-226,27. The cross-linked BSA-GOx amperometric biosensor used for comparison had a response time of 50 s and a sensitivity of 5.08 µA·mmol·L^-1·cm^-2.

Fig. 4.

Response comparison of non-enzymatic and enzyme modified membrane probes. Addition of 5 µL 100 mmol·L^-1 glucose into 2 mL of buffer solution to a final glucose concentration of 250 µmol·L^-1 indicated by the black arrow.

Fig. 5.

Calibration plot for measured decrease of dissolved O₂ concentration in response to glucose concentration in the solution.

Two different mass spectrometers, HPR-40 DSA and MS-GAS 100, were tested with the same probe for Glc response behavior. The sequence was first the MS-GAS 100, and then the same probe was connected to the HPR-40 DSA (Fig. 6). The difference between those two instruments in response is due to HPR-40 DSA having a new ion source recently installed and being tuned to maximum peak heights, while GAS MS-100 was an old prototype instrument tuned to optimum peak width, and its tuning was not permitted to be changed. One probe was measured over a period exceeding 10 months, demonstrating exceptional durability, as shown in Fig. 7, which confirms that ESM is a biocompatible material capable of preserving enzyme activity over an extended period.

Fig. 6.

Comparison of MS GAS-100 and HPR-40 DSA responses using the same probe.

Fig. 7.

Sensitivity dependency on long term storage. Glucose was added as 5 µL of 100 mmol·L^-1 solution into 2 mL of 100 mmol·L^-1 buffer solution to a final glucose concentration of 250 µmol·L^-1.

The effect of buffer concentration, pH and temperature on probe sensitivity

Because previous studies have shown that the sensitivity of electrochemical GOx biosensors employing an ESM depends on the potassium phosphate buffer concentration², we evaluated whether this effect occurs in the GOx-modified membrane probe using MIMS detection. Probe sensitivity showed no significant dependence on buffer concentration (Fig. 8); therefore, a 100 mmol·L⁻¹ potassium phosphate buffer (pH 7.0) was considered sufficient. The differences most likely arise from the fundamentally different nature of the two methods: in amperometric biosensors, higher ionic strength increases electrical conductivity and thus improves signal quality, whereas this factor is not relevant when using a membrane inlet in MIMS. For experiments examining pH dependency, the phosphate buffer concentration was increased to 200 mmol·L^-1 because the lower buffer strength was insufficient at some pH values. The pH dependency of the GOx-modified membrane probe was tested in the range of pH 3 to 10 and indicated a broad range of suitable pH values, spanning from pH 4.0 to pH 8.0 (Fig. 9). It is well established that covalent immobilization shifts the pH optimum of GOx toward more neutral values²⁸; furthermore, a broadened pH activity range for GOx immobilized specifically on ESM has been consistently reported^2–4, suggesting that the ESM matrix provides a favourable microenvironment for enzyme activity at physiological pH. The temperature dependence of the GOx probe is shown in Fig. 10, with the highest sensitivity observed at 40 °C and a gradual decrease at higher temperatures, consistent with the thermal denaturation of the enzyme.

Fig. 8.

Potassium phosphate buffer pH 7 concentration effect on GOx enzyme modified membrane probe sensitivity. Glucose was added as 5 µL of 100 mmol·L^-1 solution into 2 mL of buffer solution to a final glucose concentration of 250 µmol·L^-1.

Fig. 9.

Sensitivity dependency on pH of potassium phosphate buffer. Glucose was added as 5 µL of 100 mmol·L^-1 solution into 2 mL of 200 mmol·L^-1 buffer solution to a final glucose concentration of 250 µmol·L^-1.

Fig. 10.

Sensitivity dependency on temperature. Glucose was added as 5 µL of 100 mmol·L^-1 solution into 2 mL of 100 mmol·L^-1 buffer solution to a final glucose concentration of 250 µmol·L^-1.

Alternative substrates and possible interferents

Responses to alternative substrates (Gal, Man and Suc) and possible interferents (Cit, Asc, EtOH) were measured with both MIMS and amperometric biosensors and compared to Glc (Tab. 1). Asc showed a strong interference effect in amperometric measurements, with its response exceeding even that of Glc, due to its direct oxidation on the electrode surface²⁹. In contrast, the MIMS response to Asc was substantially lower than that of Glc, though still measurable, likely because Asc spontaneously reduces dissolved O₂ in solution, consuming it before it can reach the membrane and be detected³⁰. While EtOH appeared to have no interference effect at 0.5% (V/V) in either MIMS or amperometric measurements, subsequent measurements revealed a cumulative interaction with the PDMS membrane. Although EtOH is not among the strongly swelling solvents for PDMS³¹, repeated introductions drive progressive partitioning into the membrane, causing incremental changes in permeability and gas transport properties. This manifested as a slight increase in MIMS signal drift and a baseline offset upon repeated EtOH sample introductions. The effect was mitigated by extending the buffer equilibration time between samples, though it remains a practical consideration for long-term MIMS deployment.

Table 1.

Responses of MIMS and amperometric biosensor to substrate analogs and possible interferents, and glucose concentrations determined in real samples.

Alternative		MIMS	Amperometry
substrates and interferents	concentration (mmol·L-1)	O2 concentration decrease (µmol·L-1)	Current, I (nA)
Glucose	0.25	15.54 ± 1.40	75.45 ± 8.45
Galactose	3.2	1.55 ± 0.46	16.28 ± 0.72
Mannose	3.2	0.51 ± 0.04	8.12 ± 0.35
Sucrose	3.2	0.36 ± 0.07	4.84 ± 0.30
Citrate	3.2	0.30 ± 0.14	0.29 ± 0.31
Ascorbate	3.2 a / 0.32 b	5.12 ± 0.19	236.24 ± 12.56
Ethanol	85.6 (0.5% V/V)	– c	0.00
Analyzed samples		Glucose concentration (mmol·L-1)
Cabernet Sauvignon red		5.24 ± 0.31	4.42 ± 0.42
Riesling white		6.20 ± 0.31	5.73 ± 0.56
Zweigeltrebe rosé		13.60 ± 0.55	13.30 ± 1.00
100% natural grape juice		369.78 ± 5.79	371.66 ± 9.41
Relax Mango		52.32 ± 0.66	47.49 ± 3.74
UHT whole cow milk (3.5% fat)		0.49 ± 0.01	0.49 ± 0.03
Bovine blood plasma 1		4.54 ± 0.15	–
Bovine blood plasma 2		4.51 ± 0.14	–
Bovine blood plasma 3		4.27 ± 0.13	–
Bovine blood plasma 4		4.69 ± 0.12	–

^aMIMS, ^bamperometry, ^cPDMS membrane swelling observed; see Section "Alternative substrates and possible interferents".

Wines, fruit juices, milk and bovine blood plasma samples

Samples were measured with both MIMS and amperometric techniques. Wine sample analysis was conducted with additional precautions to mitigate EtOH-induced swelling of the PDMS membrane: the buffer volume was increased to 3 mL to reduce the final EtOH concentration after sample addition, and the buffer equilibration time was extended. When tested with bovine blood plasma, the amperometric electrodes failed irreversibly, showing unstable signals after sample introduction and becoming unusable for further Glc determinations; therefore, no amperometric results are reported for these samples. The MIMS method, in contrast, successfully determined Glc concentrations in blood plasma within the physiological range. For wine, fruit juice, and milk samples, both methods yielded comparable results (Table 1).

The MIMS method yielded slightly higher values for wine samples, likely due to increased membrane permeability caused by EtOH-induced swelling of the PDMS membrane. Beyond EtOH, other potential interferents were evaluated based on available literature. Cit is commonly present at high concentrations in fruit juices, particularly those of citrus origin, reaching up to 67.4 mmol·L^-1 in grapefruit juice³². Grape juices, however, contain considerably lower Cit concentrations³³, with the highest reported value being 181 mg·L^-1 (0.94 mmol·L^-1). The Relax Mango juice has a declared composition of: “drinking water, mango concentrate (10%), sugar, apple concentrate (5%), orange concentrate (3%), and lemon concentrate (2%), also citric acid as a pH regulator of unspecified concentration.” Due to their high Glc content, juice samples required substantial dilution prior to measurement. For grape juice, 2 µL was added to 2 mL of buffer for MIMS, and 5 µL to 3 mL for amperometry. For Relax Mango juice, 15 µL was added to 2 mL of buffer for MIMS, and 20 µL to 3 mL for the amperometry method. To assess the worst-case Cit interference from citrus juices, a hypothetical scenario was considered: grapefruit juice at 67.4 mmol·L⁻¹ Cit, diluted identically to the Relax Mango sample, would result in signal interference of only 0.3% for MIMS and 0.05% for amperometry. For grape juice, interference is negligible in both methods. For wine, the maximum acceptable Cit concentration is 1 g·L^-1 (5.2 mmol·L^-1)³⁴; at the applied dilution of 100 µL to 3 mL buffer for both methods, the resulting interference is negligible. UHT whole cow milk contains Cit in the range of 7–11 mmol·L^-135. Taking the upper end of this range, the applied dilution of 500 µL of milk to 2 mL buffer for MIMS yields 1.7% signal interference, while for amperometry, 200 µL of milk to 3 mL buffer results in only 0.08%. For bovine blood plasma, the endogenous Cit concentration is relatively low, with a reported mean of 110.3 ± 47.2 µmol·L^-136, an amount negligible for samples used without anticoagulation pretreatment as in our case; however, in citrate-anticoagulated plasma samples, Cit is present exogenously at a final concentration of 10.9 mmol·L^-137, representing the dominant source of Cit in such samples. At a dilution of 100 µL of plasma to 2 mL of buffer, this would result in 3.3% signal interference for MIMS.

As Asc is a strong interferent in amperometry, appropriate precautions were taken during sample treatment. Asc in grapes is 7.8 mg ·L^-1 (44.3 µmol·L^-1)³⁸ and in mango 160 mg ·L^-1 (908 µmol·L^-1)³⁹; in the case of the Relax Mango product, L-ascorbic acid is added as an antioxidant in an unspecified amount stated in the product composition. To reduce the Asc concentration, the samples were transferred into test tubes with half the volume left for headspace, sealed, gently shaken, and stored in the fridge (4 °C) for 48 h to allow Asc to be naturally oxidized. Wine is sometimes treated with exogenous Asc, especially white wines, to prevent colour changes up to 250 mg·L^-1 (1.42 mmol·L^-1), which represents an extremely strong interference; however, analysis of real samples reveals rather non-detectable concentrations in red wines and 23–110 mg·L^-1 (131–623 µmol·L^-1) in white wines⁴⁰, still within the range of high interference. Our wine samples originated from traditional manufacturing without additional preservatives and additives — stated in the product description and composition as only 100% grape juice and sulfur dioxide. To reduce any endogenous Asc remaining after fermentation, the wine samples were transferred to test tubes with half liquid and half headspace, sealed, slightly shaken, and stored in the fridge (4 °C) for over 48 h. For UHT whole cow milk, a concentration of 2258–3342 µg·L^-1 (12.8–19.0 µmol·L^-1) of L-ascorbic acid is reported⁴¹, and blood plasma in fattening cattle contains Asc concentrations of 1.49–3.33 mg·L^-1 (7.88–17.6 µmol·L^-1)⁴². In our case, for the amperometric biosensor where Asc is considered an interferent, 200 µL of milk sample was added to 3 mL of buffer solution, resulting in a signal interference on the order of 0.4%.

SEM characterization

For SEM characterization of the GOx-immobilized ESM sample, 46 images were acquired, of which 33% were considered to contain structures suitable for analysis. Among these 15 images, 26 analyzed structures matched the size of the GOx enzyme. For the blank sample, 26 images were acquired; 8% (2 images) contained structures for analysis and 1 structure was positively identified as a match to GOx size. A comparison of SEM images for the blank and GOx-immobilized ESM samples is given on Fig. 11. SEM characterization clearly demonstrates that GOx-immobilized ESM contains spherical structures, consistent in size with gold-coated enzyme molecules, at a significantly higher frequency than the blank.

Fig. 11.

SEM images of ESM. Left image blank sample, right image immobilized GOx. Black circles show matching structures for GOx.

Conclusion

The concept of an ESM-wrapped MIMS membrane probe modified with immobilized enzymes was demonstrated to be functional with GOx as a model enzyme. Moreover, the immobilization method presented here, applying ESM onto a PDMS membrane surface as a biocompatible protein layer rich in lysine residues for covalent coupling of protein molecules via GA, is efficient and fast to perform. The method has limitations in response time, concentration range, and sensitivity compared to standard GOx biosensors. However, enzyme-immobilized layers offer significant promise, being less restrictive to permeation processes. These may include, for example, functional nanolayers capable of biomolecule immobilization through amino functional groups, prepared via polymerization driven by plasma deposition⁴³. Another plasma-based approach may involve using organic polymer MIMS membranes enriched with reactive radicals that allow direct covalent binding of protein molecules from solution without additional linkers⁴⁴. Employing plasma-based methods and comparing their parameters with the ESM immobilization presented here will be the focus of future work.

Moreover, the use of MIMS for biosensing offers several additional benefits. Most notably, a mass spectrometer can detect a broad range of analytes; in the case of MIMS, this includes small non-polar gaseous or volatile species such as alcohols, hydrocarbons, and atmospheric gases. Coupling MIMS with immobilized enzymes extends the technique beyond its conventional detection range, enabling indirect quantification of analytes that are not directly membrane permeable. This approach broadens the applicability of MIMS as an analytical platform wherever target compounds participate in reactions yielding detectable volatile products. One such example involves measuring isopropyl alcohol, released by the action of Candida rugosa lipase, using MIMS to assess environmental pollution caused by organic esters with detection limit 1.5 µmol·L^-1 (230 ppb) for monoisopropyl phthalic acid¹.

In our future work, we plan to use an ethylene-forming enzyme (EFE)^45,46 with a MIMS membrane probe capable of detecting its O₂ substrate and C₂H₄ and CO₂ products, to measure the concentrations of its substrates 2-oxoglutarate and L-arginine, as well as its cofactor Fe²⁺ (Fig. 12). Using an EFE biosensor based on the MIMS method could provide a tool for screening organisms, cultivation conditions and limiting factors to enhance the biotechnological production rate of ethylene^47,48. Such improvements are crucial for achieving sustainable and competitive ethylene production⁴⁹. Limit of detection for ethylene by MIMS is reported 606 µmol·L^-1 (17 ppm) using Faraday cup detector⁴⁸.

Fig. 12.

Schematic representation of EFE reactions. Top reaction showing L-Arginine hydroxylation. Bottom reaction representing oxidative fragmentation of 2-Oxoglutarate.

Another advantage of MIMS over electrochemical or optical methods is that the analysis does not require the use of electric potential or electromagnetic irradiation, therefore requiring no special precautions for sensitive species like Fe²⁺, which serves as the EFE reaction cofactor. On the other hand, mass spectrometers are large and heavy devices, far exceeding the compact dimensions of current-generation biosensors. However, miniaturization is being studied even for quadrupole mass filters, using microelectromechanical systems manufacturing to produce chip-size quadrupoles⁵⁰. The use of MIMS with enzyme-modified membrane probes remains relatively uncommon, yet it represents a functional approach with promising future applications owing to the immobilization method presented, which can be easily implemented in any laboratory equipped with a MIMS instrument.

Abbreviations

AMU

Atomic mass unit

Asc

Ascorbate

BSA

Bovine serum albumin

Cit

Citrate

CRL

Candida rugosa Lipase

EFE

Ethylene-forming enzyme

ESM

Eggshell membrane

EtOH

Ethanol

FAD

Flavin adenine dinucleotide

IPA

Isopropyl alcohol

GA

Glutaraldehyde

Gal

Galactose

Glc

Glucose

GOx

Glucose oxidase

Man

Mannose

MIMS

Membrane inlet mass spectrometry

PDMS

Polydimethylsiloxane

PBS

Phosphate buffered saline

SEM

Scanning electron microscopy

Suc

Sucrose

UHT

Ultra-high temperature

Author contributions

TZ, JC and TK originated the project idea. TZ was responsible for the experimental design and performing all experimental work except scanning electron microscopy, which was carried out by BK. The work was under the supervision of JC and TK. TZ wrote the draft of the manuscript, with critical feedback from JC, BK and TK.

Funding

This work was supported by the Technology Agency of the Czech Republic (TAČR), National Centres of Competence Program TN01000048.

Data availability

The experimental data are available on the Harvard Dataverse repository: 10.7910/DVN/4LOPH6.

Declarations

Competing interests

The authors declare no competing interests.