Sterilization Effects on Nitric Oxide-Releasing Glucose Sensors

Abstract

Nitric oxide (NO) release from S-nitrosothiol-modified mesoporous silica nanoparticles imbedded in the diffusion limiting layer of a glucose sensor has been demonstrated as an effective strategy for mitigating the foreign body response common to sensor implantation, resulting in improved analytical performance. With respect to potential clinical translation of this approach, the effects of sterilization on NO-releasing biosensors require careful evaluation, as NO donor chemistry is sensitive to temperature and environment. Herein, we evaluated the influence of multiple sterilization methods on 1) sterilization success; 2) NO payload; and 3) sensor performance to establish the commercialization potential of NO-releasing glucose sensors. Sensors were treated with ethylene oxide gas, the most common sterilization method for intricate medical devices, which led to undesirable (i.e., premature) release of NO. To reduce NO loss, alternative sterilization methods that were studied included exposure to ultraviolet (UV) light and immersion in 70% ethanol (EtOH). Sterilization cycle times required to reach a 10⁻⁶ sterility assurance level were determined for both UV light and 70% EtOH against Gram-negative and -positive bacteria. The longest sterilization cycle times (258 s and 628 s for 70% EtOH and UV light, respectively) resulted in a negligible impact on benchtop sensor performance. However, sterilization with 70% ethanol resulted in a reduced NO-release duration. Ultraviolet light exposure for ~10 min proved successful at eliminating bacteria without compromising NO payloads or durations and presents as the most promising method for sterilization of these sensors. In addition, storage of NO-releasing sensor membranes at −20 and −80°C resulted in preservation of NO release for 6 and 12 months, respectively.

INTRODUCTION

Nitric oxide (NO), an endogenous free radical, plays multiple roles in human physiology, including wound healing, inflammation, and response to pathogens [1–3]. Recent work has shown that the use of exogenous NO may serve to treat diseases [4,5] and improve the biocompatibility of certain biomedical devices [6], including implantable sensors [7]. For example, implantable continuous glucose monitors (CGMs) are subject to the foreign body response (FBR), whereby proteins adsorb to the sensor surface upon implantation and initiate a cascading inflammatory response, resulting in the eventual formation of a collagen capsule that isolates the device from blood and glucose delivery and renders it inaccurate [8]. The release of NO from the sensor surface has been shown to mitigate the FBR and extend sensor accuracy to 28 days [7]. Of note, both Nichols et al. and Malone-Povolny et al. have reported that NO must be actively releasing in order to maintain FBR mitigation and improved sensor performance [7,9]. The most straightforward manner to deliver NO is through the use of chemical NO donors, including N-diazeniumdiolates (NONOates) and S-nitrosothiols (RSNOs), as NO is a highly reactive gas [10].While NONOates formed on secondary amines are often preferred in therapeutic applications for their ability to produce a large burst of NO spontaneously, RSNOs have proven more ideal for fabricating NO-releasing glucose sensors that release NO for extended durations (i.e., weeks versus days). Unlike NONOates, S-nitrosothiols release NO independent of water uptake or local pH [11], and the use of spatially confined locations such as those present in mesoporous silica facilitates the cage effect and prolonged release of NO [12]. Primary S-nitrosothiol modified mesoporous silica nanoparticles have been developed that, when doped into outer sensor membranes, facilitate NO release for approximately 28 days [11]. A NONOate-based system has not been synthesized that is capable of releasing NO over a similar duration. The NO from a RSNO is liberated through the homolytic cleavage of the S-N bond of the S-nitrosothiol resulting in the formation of thiyl and nitric oxide radicals [13]. Homolytic cleavage can be initiated by light or heat, at ultraviolet (330350 nm) and visible wavelengths (550–600 nm) and physiological temperatures [13]. This breakdown mechanism presents challenges in the fabrication and storage of NO-releasing biomedical devices. Nitric oxide payloads can be maintained by fabricating and storing the device in cold, dark conditions, although preservation of NO release has only been studied up to 96 h when stored at −20°C [11]. Of note, the primary RSNO donors used in this study are known to be less stable than their secondary or tertiary counterparts, due to less steric hinderance around the sulfur atom [14]. While the use of various sterilization methods have been reported previously to eradicate bacteria on NO-releasing medical devices, such studies were focused exclusively on a tertiary RSNO known as S-nitroso-N-acetylpenicillamine (SNAP) [15–19] and not other RSNOs or NO-releasing sensors that make use of less stable NO donors.

The majority of biomedical devices, including CGMs, are sterilized using ethylene oxide (EtO) gas as EtO does not usually interfere with device integrity or performance [20,21]. Quality of sterilization is dependent on concentration of gas, duration of exposure, temperature, and humidity [21]. Standard protocols for EtO sterilization require multiple phases of treatment and purging, as residual gas on or within the medical device may pose toxicity concerns [22]. With respect to NO-releasing biomedical devices, the duration of sterilization by EtO may result in the premature release of NO. An additional concern is the temperature employed spans 29 to 65°C, over which NO donor breakdown will happen [23]. Medical devices are also often sterilized with ionizing radiation (e.g., electron beam or gamma). Lin et al. reported that NO release was maintained following electron beam sterilization, but this work was also done with a tertiary RSNO, which are known to be more stable than the primary RSNOs evaluated herein [24]. To our knowledge, the effect of gamma radiation on NO release has not been reported and may provide a future avenue of study.

The objective of this study was to evaluate multiple sterilization protocols that meet current implantable glucose sensor sterilization standards (i.e., 10⁻⁶ sterility assurance level) for common pathogens Pseudomonas aeruginosa and Staphylococcus aureus while maintaining sensor performance and NO release [25]. Our hypothesis was that ultraviolet (UV) light and/or immersion in 70% ethanol would be optimal as both are readily implemented and generally safe (i.e., minimal impact of sterilization or degradation by-products on tissues). Short-wave UV light (UVC; 200–280 nm) damages DNA through dimer formation and is lethal to most micro-organisms [26]. Ultraviolet light is often used to sterilize surfaces exclusively as UV light will not penetrate into a material [27,28]. Indeed, UVC light has been utilized broadly to disinfect surfaces within hospitals in response to the COVID-19 pandemic [29,30]. Ethanol and water mixtures have been shown to denature proteins in cellular membranes and rupture/kill the bacteria [31]. However, ethanol is considered more of a disinfectant as it does not exert activity against the spores of many bacteria species [32]. Nevertheless, the use of UVC light or 70% ethanol, in combination with an aseptic manufacturing environment, may achieve sensor sterilization requirements and thus their efficacy and effect on sensor performance and NO release was the focus of this work. In addition to challenges with sterilization, storage of NO-releasing glucose sensors poses a hurdle for future commercialization. As such, we evaluated the effects of storage temperature (i.e., −80, −20, 4, and 25 ºC) and time (i.e., 1 day, 1 week, or 1, 3, 6, or 12 months) on three NO-release parameters for NO-releasing sensor membranes.

EXPERIMENTAL

Materials.

All solvents and reagents were of analytical grade and used as received unless noted otherwise. Cetyltrimethylammonium bromide (CTAB), sodium nitrite (NaNO₂), m-phenylenediamine (m-PD), glucose oxidase (GOx, type VII from Aspergillus niger), potassium iodide (KI), ethylenediaminetetraacetic acid (EDTA), copper (II) bromide (CuBr₂), and D-(+)- glucose were purchased from Sigma-Aldrich (St. Louis, MO). Ethanol (EtOH), methanol (MeOH), ammonium hydroxide (NH₄OH), hydrochloric acid (HCl), dimethylformamide (DMF) and tetrahydrofuran (THF) were purchased from Fisher Scientific (Waltham, MA). Tetraethylorthosilicate (TEOS), 3-mercaptopropyltrimethoxysilane (MPTMS), and trimethoxymethylsilane (MTMOS) were purchased from Gelest (Morrisville, PA) and stored under nitrogen atmosphere. Polyurethane HP93A-100 was purchased from Lubrizol (Wickliffe, OH). Tryptic soy agar (TSA) and tryptic soy broth (TSB) were purchased from Becton, Dickinson and Company (Franklin Lakes, NJ). Platinum-iridium wire and silver wire were purchased from A-M Systems (Sequim, WA). Nitrogen (N₂) and nitric oxide calibration gas (NO, 25.97 ppm in N₂) were purchased from Airgas National Welders (Radnor, PA). Water was purified using a Millipore Reference water purification system (Burlington, MA) to a resistivity of 18.2 MΩ cm and a total organic content < 6 ppb.

Synthesis of NO-Releasing Mesoporous Silica Nanoparticles.

Mesoporous S-nitrosothiol modified silica nanoparticles (RSNO-MSNs) were synthesized as previously described (Figure S1 A) [11]. Briefly, 175 mL of EtOH, 162 mL of H₂O, 11.8 mL of NH₄OH, and 280 mg of CTAB were stirred for 15 min to form a liquid crystal template. 1.395 mL of TEOS was added and the reaction was stirred for 2 hours to allow for condensation of TEOS to form silica nanoparticles. Particles were washed three times with ethanol, collected by centrifugation, and CTAB was removed from the pores with 10% HCl in EtOH. Particles were dried under vacuum before grafting with MPTMS to modify interior and exterior surfaces with thiol groups. Thiols were converted into S-nitrosothiol groups with nitrosation. Particles were suspended in 4 mL of MeOH and 4 mL of 5M HCl. An aqueous solution (2 mL) of sodium nitrite (400 mg) and EDTA (5 mg) was added dropwise to the particle solution at 0°C in the dark. The reaction was stirred for 1 hour and particles were collected by centrifugation and washed once with cold water and three times with cold MeOH. Nitrosated particles were dried under vacuum for 45 minutes before incorporation into polyurethane. Control particles were made following the same protocol but without nitrosation, denoted SH-MSNs. Particle morphology (Figure S2), surface area, pore volume, thiol content (Table S1), and NO-release parameters (Table S2) were characterized to ensure successful preparation of RSNO-MSNs.

Suspension of Nanoparticles in Polyurethane.

Polyurethane films were prepared by dissolving 50 mg mL⁻¹ HP93A polyurethane into 3:1 anhydrous THF:DMF via sonication (Figure S1 B). After full dissolution of the polymer, particles (RSNO-MSNs or SH-MSNs) were suspended in solution at a concentration of 25 mg mL⁻¹ and vortexed to ensure homogenous dispersal. Polyurethane membranes were deposited onto sensors in the dark using a solvent-casting method consisting of 10 layers of 8 μL each [11]. After fabrication, sensors were stored vacuum sealed at −20°C until use (i.e., sterilization). To evaluate the effect of storage conditions, each mock sensor was individually vacuum sealed in a centrifuge tube and stored at either −80°C, −20°C, 4°C, or 25°C.

Preparation and Characterization of Glucose Sensors.

First generation needle-type electrochemical glucose biosensors were fabricated and used for the evaluation of sterilization methods on sensor performance (Figure S3) [33]. Bare sensors were constructed by coiling a silver/silver chloride reference electrode around a polymer-insulated 90:10 Pt:Ir wire. A approximately 5 mm length of Pt:Ir wire was exposed by removing the polymer coating, which serves as the working electrode. A series of surface modification layers were then applied to the sensor to facilitate glucose detection [7]. First, a hydrogen peroxide selectivity layer of m-phenylenediamine (0.1 M in PBS) was deposited by cyclic voltammetry (0 → +1.0V, 20 cycles, 100 mV s⁻¹). Next, an enzyme containing sol-gel layer was deposited via dip-coating. Briefly, 50 μL of a solution of glucose oxidase (9 mg) and water (75 μL) was added to a solution of MTMOS (25 μL) and ethanol (100 μL) and vortexed to mix completely. Sensors were coated by dipping the working electrode in the sol-gel 15 times each (5 s dip, 10 s dry) and the gel was allowed to cure for 30 minutes. Finally, MSN polyurethanes were solvent cast onto the sensors following the procedure outlined above. Non-NO-releasing sensors were stored at 4°C and NO-releasing sensors were stored at −20°C when not in use for sterilization or calibration studies.

The analytical performance of the same sensors for each sterilization method was determined in PBS (37°C, pH 7.4) before and immediately after sterilization using a CH Instruments potentiostat (Austin, TX). Sensors were preconditioned by polarizing (+0.6V vs Ag|AgCl) until a stable background current was reached. Sensors were calibrated by incrementally increasing the buffer glucose concentration to 30 mM. Linear dynamic range was defined as the concentration range over which a linear fit trendline had a R² value ≥ 0.99. Sensitivity in buffer solution was defined as the slope of the trendline over the linear dynamic range.

Bactericidal Assays for Determination of Sterility Cycle Lengths.

Assays were adapted from previous publications [34,35]. To inoculate the sensors, bacterial cultures of Pseudomonas aeruginosa (PAO1) and Staphylococcus aureus (SF8300) were grown from frozen (−80°C) stocks on TSA. Colonies were isolated and suspended in TSB overnight at 37°C to reach a concentration of 1 × 10⁸ CFU mL^-1. Bacterial cultures were diluted to 1 × 10⁶ CFU mL⁻¹ in TSB and sensors were inoculated at 37°C and 100 rpm for 6 hours.

To sterilize sensors with ultraviolet light, inoculated sensors were washed with sterile phosphate buffered saline (PBS; 10 mM, pH 7.4) to remove non-adherent bacteria and placed parallel to the UV light (254 nm) in a biosafety cabinet at a distance of 63 cm from the source. Sensors inoculated with PAO1 were exposed for 10, 20, or 30 seconds and sensors inoculated with SF8300 were exposed for 30, 60, or 90 seconds. After initial UV irradiation, the sensors were flipped 180 degrees and illuminated again with UV light for the same time periods. Sensors were transferred to sterile PBS, sonicated for 15 minutes, and vortexed for 1 minute to remove any adherent bacteria. The resulting solution was serially diluted (1-, 10-, 100-, 1000-, 10,000-, 100,000-fold dilutions) and plated on TSA for overnight incubation at 37°C. Viability of bacteria following sterilization was determined using a Flash & Go colony counter (IUL; Farmingdale, NY) and enumerated as CFU mL⁻¹.

Prior to sterilization with ethanol, non-adherent bacteria were removed from the sensor with sterile PBS. Sensors were submerged in 70% EtOH for 20, 30, or 40 seconds when inoculated with PAO1 and 40, 60, or 80 seconds when inoculated with SF8300. Following exposure to 70% EtOH, sensors were allowed to dry and transferred to sterile PBS for removal of adherent bacteria, plated, and counted as described above.

Remaining CFU mL⁻¹ after sterilization with either UV or EtOH were plotted against sterilization time. The linear trendline equations were used to determine D-values (time to reduce bacterial burden by 1 log), which were used to establish sterilization cycle times in order to achieve at 10⁻⁶ SAL from an initial bacteria burden of 10⁶ CFU mL⁻¹.

Ethylene Oxide Sterilization.

NO-releasing glucose sensors were subjected to EtO sterilization with an Anprolene AN75 (Andersen Sterilizers; Haw River, NC). Samples were stored at −20°C before and after sterilization. The sterilization cycle ran for 24 h with a 3 h aeration time. Sterilization was performed at room temperature. The EtO sterilizer is housed in and maintained by the Department of Comparative Medicine at the University of North Carolina at Chapel Hill.

Nitric Oxide Release Measurements.

Nitric oxide release was measured using two methods. Short-term NO release was measured in real time using a Sievers 280i chemiluminescent nitric oxide analyzer (NOA; Fredrick, CO). The NOA was calibrated before use with a two-point calibration consisting of air passed through a zero NO filter and 25.87 ppm NO in N₂ as blank and standard values, respectively. While measuring NO release following sterilization, an unsterilized (control) or sterilized NO-releasing sensor was added to a sample flask with 30 mL of deoxygenated PBS at pH 7.4 and 37°C in the dark. Nitrogen gas was bubbled through the sample flask at 200 mL min⁻¹ to carry liberated NO from the solution to the instrument reaction cell. Measurements were concluded after 24 hours when evaluating short-term NO-release kinetics. For stability studies, NO-releasing sensor membranes were added to the NOA sample flask after a specific storage period and condition (i.e., temperature). Nitric oxide measurements were similar to the above protocol, except 200 μL of 2.5 M CuBr₂ (Cu (II)) was added to the sample flask after 24 h of measurement to more quickly liberate all NO. Of note, Cu (II) is useful for this purpose as it is soluble in water. When reacted with free thiols in the sample flask, Cu (II) is oxidized to Cu (I), which reacts with RSNOs and reforms Cu (II). Measurements were concluded when the measured NO dropped below the NOA limit of quantification (LOQ; 10 ppb).

Because of the low, extended NO flux of NO-releasing sensors that quickly drops below the LOQ of the NOA, a different NO measurement strategy must be used to capture the long-term NO release profile. Control and sterilized NO-releasing sensors were placed in 10 mL PBS (pH 7.4) and incubated at 37°C in the dark. At 1, 7, 14, 21, and 28 days, sensors were transferred to fresh PBS and a 1 mL aliquot of the soak solution was analyzed. Nitric oxide that was released into solution and exposed to oxygen was converted to nitrite. The soak solution aliquot was added to the reaction flask of the NOA, which contained 2 mM HCl and 5 mg potassium iodide, to convert nitrite to nitric oxide, which is then measured by the NOA. This method also accounts for dissolved NO in solution. Measurement was concluded when NO release dropped below the instrument LOQ.

Characterization of Polyurethane Morphology.

Polyurethane layer morphology and thickness were determined using a Hitachi S-4700 cold cathode field emission scanning electron microscope (Pleasanton, CA). Mock sensors in scanning electron micrographs were sputter-coated with 5.0 nm gold/palladium. Polyurethane layer thickness was measured using ImageJ software from three separate fabrications of mock sensors. Mock sensors were coated with polyurethane as described above, on steel wires of identical dimensions to the functional glucose sensors.

Statistical Analysis.

Data are reported as mean ± standard error of the mean, unless otherwise noted. Statistical comparisons were done using a student’s t-test, unless otherwise noted, with values of p < 0.05 considered statistically significant.

RESULTS AND DISCUSSION

Determination of Sterilization Cycle Length

Continuous glucose monitors must be sterilized using a method that can reach a sterility assurance level (SAL) of 10⁻⁶, or a one in a million probability that one microorganism survives the sterilization process and is found on the medical device [25,36,37]. To investigate the effects of sterilization on NO-releasing glucose sensors, two sterilization methods, 70% ethanol or UV light, were used to study killing activity against Pseudomonas aeruginosa (PAO1) and Staphylococcus aureus (SF8300) as representative Gram-negative and Gram-positive pathogens, respectively. As shown in Figure 1, viable colonies were quantified as a function of exposure time to each sterilization method.

Figure 1.

Sterilization of glucose sensors with UV light (triangles/dotted line; exposure time per side) or 70% ethanol (squares/solid line). (A) P. aeruginosa (B) S. aureus. Sensors inoculated with bacteria were exposed to each sterilization method for various lengths of time and bacteria viability was quantified. Data points represent the average number of surviving bacteria from 3 biological replicates, each with 3 technical replicates.

The reciprocal slope of the linear fit trendline represents the D-value, or the time required to achieve a 90% (1 log) reduction in viable bacteria [35,38]. To determine the sterilization cycle length, the total log reduction required is multiplied by the D-value. In this work, a starting bacterial burden of 10⁶ CFU mL⁻¹ to a 10⁻⁶ SAL is an overall 12 log reduction, and sterilization cycle lengths were 12 times the D-value [39]. As bacterial burden in a clean manufacturing environment is unlikely to reach 10⁶, the time required for a 12 log reduction is excessive and should be more than sufficient. D-values and sterilization cycle lengths for each pathogen with both sterilization methods are summarized in Table 1.

Table 1.

Calculated D-values and sterilization cycle lengths for each method of sterilization.

Pathogen	70% Ethanol		UV Exposure
	D-value (s)a	Sterilization Cycle (s)b	D-value (s)a	Sterilization Cycle (s)b
P. aeruginosa	11.4	137	15.4	185
S. aureus	21.5	258	52.3	628

^aDetermined from negative reciprocal slope of linear fit trendline.

^bCycle length to reach a 12 log reduction.

D-values for Gram-negative P. aeruginosa using 70% EtOH or UV light were 11.4 s and 15.4 s (7.7 s per side), respectively. A similar trend was observed for Gram-positive S. aureus albeit for longer periods, 21.5 s and 52.3 s (26.2 s per side) for 70% EtOH and UV light, respectively. Indeed, P. aeruginosa proved more sensitive than S. aureus to both methods of sterilization. Gram-negative bacteria (e.g., P. aeruginosa) are surrounded by an outer membrane over a thin peptidoglycan layer, whereas Gram-positive bacteria (e.g., S. aureus ) do not have an outer membrane and are enclosed by a thicker and harder to disrupt peptidoglycan layer [40]. The thin outer membranes of Gram-negative bacteria appear to be more susceptible to sterilization. Both pathogens were also more susceptible to sterilization by 70% EtOH compared to UV light. This enhanced susceptibility may be due to the direct chemical interaction of ethanol with proteins in the bacterial membrane rather than the delayed response by DNA modification via UV light. For further experiments evaluating the effects of sterilization on sensor performance and NO release, sterilization cycle times for S. aureus (258 s and 628 s for 70% ethanol and UV exposure, respectively) were followed, as they represent the longest possible sterilization cycle. The full cycle time for UV sterilization corresponded to a UVC dose greater than 100 mJ cm⁻² (Figure S4), which is on par with commercial UV sterilization devices.

Glucose Sensor Performance After Sterilization

To evaluate the impact of sterilization of the glucose sensors on their analytical performance, linear dynamic range (LDR) and sensitivity were established via amperometry before and after sterilization. Sensors were fabricated and calibrated, then sterilized with either 70% EtOH, UV light, or EtO gas. The glucose sensors were calibrated again and directly compared to the identical group of sensors from before sterilization. Normally, LDR should cover 0 – 24 mM glucose to account for severe hypoglycemia (2.5 mM) and severe hyperglycemia (22 mM) [25]. Sensitivities greater than 3 nA mM⁻¹ on these lab-fabricated sensors have been associated with an in vivo performance that maintains implantable glucose sensor accuracy standards [41,42]. However, chemical or physical alterations to the polyurethane diffusion limiting layer on these glucose sensors often lead to changes in LDR and/or sensitivity. While degradation of polyurethane could be of concern after exposure to UV light, such effects are more likely to be apparent with weeks, not minutes, of exposure [43]. Damage to glucose oxidase might also negatively impact sensor performance. House et al. previously reported that UV light decreased the biological activity of immobilized glucose oxidase [44]. For our sensor design, however, any effect of UV light on the glucose oxidase should be minimal as the enzyme is beneath a polyurethane layer (i.e., sensor membrane) that prevents deep penetration by UVC radiation.

Indeed, sensors sterilized with UV light maintained acceptable linear dynamic ranges (≥ 24 mM) before and after sterilization with negligible changes in sensitivity (Figure 2, Table S3, Figure S5).

Figure 2.

Glucose sensor performance before (solid) and after (checks) sterilization. Error represents standard error of the mean for n ≥ 6 sensors. (A) Linear dynamic range. Determined as the concentration range, beginning at 0 mM, over which a linear trendline yielded an R² values ≥ 0.99. (B) Sensitivity. Determined as the slope of the trendline fit over the linear dynamic range.

Similarly, sensors sterilized with EtO gas had non-significant changes in LDR and sensitivity after sterilization. While EtO gas has been used successfully as a sterilization method for commercial glucose sensors, a slight increase in average sensitivity was noted, indicating that it may impact glucose oxidase and/or the polyurethane sensor membrane. Ethanol sterilization has the potential to alter the polyurethane diffusion limiting layer as well, because HP93A polyurethane is soluble in many organic solvents. Sensors immersed in 70% ethanol, however, maintained an acceptable linear dynamic range (0 – 24 mM) with only a slight, albeit not significant increase in average sensitivity (8.1 to 12.6 nA mM⁻¹). The slight increase in sensitivity might indicate that the outermost layers of polyurethane in the solvent cast layer became compromised, which in turn may increase glucose diffusion. Yet, scanning electron microscopy indicated no changes in the physical nature or thickness of the polyurethane (Figure 3, Table S4) after sterilization, regardless of sterilization mode. It is possible that the polyurethane became slightly more permeable upon immersion in 70% EtOH, which in turn may have enhanced glucose diffusion. Still, the average sensitivity falls within normal range for these glucose sensors and is not expected to impact in vivo sensor performance.

Figure 3.

Scanning electron microscopy images of cross sections of sterilized mock sensors. (A) Unsterilized. (B) Sterilized with 70% ethanol. (C) Sterilized with UV light. (D) Sterilized with EtO gas.

Effect of Sterilization on Nitric Oxide Release

Although the three sterilization techniques do not affect analytical sensor performance, the preservation of NO release remains the key parameter in future commercialization of NO-releasing glucose sensors. The NO donor system used herein, S-nitrosothiol modified mesoporous silica nanoparticles doped into polyurethane, releases NO under physiological conditions for approximately 28 days [11]. Such release is critical in both mitigating the FBR and maintaining the analytical performance of the glucose sensors. As such, the NO-release payloads and kinetics were evaluated as a function of sterilization method. Specifically, maximum achievable instantaneous NO concentration ([NO]_max) and total NO liberated after 24 hours ([NO]_{24 h total}) were quantified for all sterilization methods (Table 2).

Table 2.

Short-term nitric oxide release of sterilized and unsterilized glucose sensors.^a

Sterilization Method	[NO]max (pmol cm−2)b	[NO]24 h total (μmol cm−2)c
None	53.6 ± 5.1	1.48 ± 0.08
70% Ethanol	69.2 ± 8.0	1.88 ± 0.13
UV Light	57.4 ± 8.4	1.74 ± 0.18
Ethylene Oxide Gas	24.7 ± 1.2	0.78 ± 0.03

^aError represents standard error of the mean for n = 3 substrates.

^bMaximum instantaneous NO concentration.

^cTotal NO release after 24 hours.

Sensors sterilized with EtO gas displayed a significant change in short-term NO release (Table 2, Figure S6). The [NO]_max was reduced by over 50%, from 53.6 to 24.7 pmol cm⁻² upon sterilization with ethylene oxide. This decrease in [NO]_max indicates a significant premature release of NO during the sterilization process and a loss of the initial large burst of NO common to unsterilized sensors. The [NO]_{24 h total} were also indicative of a significant loss of NO. For example, the [NO]_{24 h total} for NO donor-modified sensors sterilized with ethylene oxide gas was only 0.78 μmol cm⁻, representing a 47% loss compared to the unsterilized sensors (1.48 μmol cm⁻²). Such diminished NO payloads would likely be detrimental to FBR mitigation and in vivo sensor performance. Of note, the model of ethylene oxide sterilizer that was used in this study completes the cycle at room temperature, which is not conventional for EtO sterilization (i.e., typically done from 29–65°C) [23]. It is highly likely that significantly more NO would be lost in a sterilizer run at higher temperatures.

Although the maximum absorbance for S-nitrosothiols is 330–350 nm, UVC light (254 nm) common for sterilization with light also has the potential to degrade the NO donor, as it is higher energy and would cleave S-N bonds. However, the short exposure time (628 s) and potential for absorption of UV light in the range of 300–390 nm by the polyurethane may lessen the potential for premature NO donor breakdown [45]. Indeed, the [NO]_max was not significantly different than the control (57.4 and 53.6 pmol cm⁻², respectively) following sterilization with UV light. The [NO]_{24 h total} after UV sterilization (1.74 μmol cm⁻²) was also similar to the control (1.48 μmol cm⁻²).

Analogous effects were observed after 70% ethanol sterilization, as [NO]_max (69.2 pmol cm⁻²) was not significantly different than that of the control. While [NO]_{24 h total} (1.88 μmol cm⁻²) was not technically significantly different from the control (p = 0.06), the [NO]_{24 h total} increased with respect to the unsterilized sensors. An increase in total NO released may be indicative of premature liberation and diffusion of NO from S-nitrosothiols through the polyurethane during the sterilization process, lowering the overall duration of NO release. To evaluate the effects of sterilization on extended NO release, measurement of NO liberated from sensors sterilized with 70% ethanol or UV light was carried out for 28 days.

As S-nitrosothiol modified-mesoporous silica nanoparticles have a low, extended NO flux, generally the NO release falls below the limit of quantification (10 ppb) of the NOA in approximately 33 hours, well before 28 days. To best quantify long-term NO release, NO-releasing sensors were placed in pH 7.4 PBS and stored in the dark at 37°C to account for NO donor breakdown by heat only. At each timepoint (day 1, 7, 14, 21, and 28) sensors were transferred to fresh PBS. The nitrite concentration and dissolved NO in the remaining solution was then quantified to determine NO release per total surface area of the sensors (Figure 4, Table S5). Of note, long-term NO release was not determined for EtO sterilization because of the dramatic loss of NO in the first 24 hours of measurement, which is not favorable for sustained release or FBR mitigation. Unsterilized controls and sensors treated by EtOH or UV sterilization released similar amounts of NO on day 1, and days 7 to 28. The total amount of NO released during days 1 to 7 for 70% EtOH sterilized sensors (93.9 nmol cm⁻²) was significantly lower than both unsterilized (598.7 nmol cm⁻²) and UV sterilized (566.4 nmol cm⁻²) sensors, indicating a greater than anticipated loss of the most labile NO during the ethanol sterilization process. Total NO released is statistically similar between the unsterilized, 70% ethanol sterilized, and UV light sterilized sensors for days 7 to 28, as longer-term release kinetics do not seem to be affected by either sterilization method. However, the reduced NO payload during days 1 to 7 from sensors sterilized using 70% ethanol may weaken mitigation of the foreign body response. Coupled with slight changes to sensor performance and short-term NO release following sterilization with EtOH, there is reason to believe there are small physical changes in the outermost polyurethane layer of the sensors, increasing both diffusion of glucose and premature release of NO, making immersion into 70% EtOH solutions less than ideal for sterilizing NO-releasing glucose sensors.

Figure 4.

Long-term nitric oxide release from unsterilized sensors (solid), sensors sterilized with 70% ethanol (dots), and sensors sterilized with UV light (stripes). Inset provides a smaller y-axis range to allow for visualization of the later timepoints. * p<0.05 relative to unsterilized control.

Preservation of Nitric Oxide Release via Storage Temperature

To investigate the effect of storage temperature and duration on NO release, three different NO-release parameters were evaluated. The first parameter was the NO]_max, which represents the initial burst of NO. Two different NO totals were also evaluated: a 24 h total ([NO]_{24 h total}) and a copper total ([NO]_{Cu total}). The [NO]_{24 h total} was the total NO released in 24 h in the dark at 37°C, indicative of release under physiological conditions. After 24 h, Cu (II) was added to catalytically release remaining NO and is reported as [NO]_{Cu total}, which includes cumulative NO release, in order to evaluate potential to maintain long-term release.

In order to preserve translational potential, NO release from glucose sensors must be maintained through shipping and storage. As such, [NO]_max, [NO]_{24 h total}, and [NO]_{Cu total} of NO-releasing mock sensors were evaluated after storage for 12 months at −80°C, −20°C, 4°C, or 25°C (Figure 5). After storage for up to 12 months at −80°C, there were no significant differences in any of the NO-release parameters between day 1 and 1 week, 1 month, 3 months, 6 months, or 12 months. As [NO]_{Cu total} was maintained, it is likely that long-term NO release would also be maintained after storage for 12 months at −80°C.

Figure 5.

(A) Maximum instantaneous NO concentration reached, (B) total NO released after 24 hours, and (C) total NO released after copper bolus of NO-releasing mock sensors stored at −80°C, −20°C, 4°C, or 25°C for 1 day (black), 1 week (dark gray), 1 month (light gray), 3 months (white), 6 months (checks), or 12 months (stripes). Data represented as mean ± standard error of the mean for n ≥ 3 separate mock sensors. * denotes p-value < 0.05 with respect to day 1 at the same storage temperature determined via one-way ANOVA. # denotes p-value < 0.01 with respect to day 1 at the same storage temperature determined via one-way ANOVA. ^ denotes p-value < 0.05 with respect to day 1 at −80°C determined via one-way ANOVA.

Nitric oxide-releasing mock sensors also proved stable when stored at −20°C for up to 6 months. However, at month 12, there was a decrease in [NO]_max and [NO]_{24 h total} (25 pmol cm⁻² and 0.68 μmol cm⁻², respectively) compared to day 1 (42 pmol cm⁻² and 0.90 μmol cm⁻² for [NO]_max and [NO]_{24 h total}, respectively). These data, combined with the observed decrease in [NO]_{Cu total}, are indicative of the beginning of loss of long-term NO payloads. Analogous NO-releasing sensors might not be suitable for FBR mitigation. Of note, commercially available CGMs report sensor expiration dates of 6 to 18 months following manufacturing. In this respect, the shelf-life of NO-releasing sensor membranes should match those of commercial CGMs to maximize translational potential.

Storage of NO-releasing mock sensors at 4°C or 25°C led to dramatic losses of NO. At 4°C, after 1 month of storage, [NO]_max and [NO]_{24 h total} (13 pmol cm⁻² and 0.37 μmol cm⁻², respectively) were significantly lower compared to day 1 (52 pmol cm⁻² and 1.13 μmol cm⁻² for [NO]_max and [NO]_{24 h total}, respectively). While not significant, the [NO]_{Cu total} also decreased after 1 week and 1 month of storage (2.62, 1.75, and 0.97 μmol cm⁻² for 1 day, 1 week, and 1 month, respectively). After 6 months of storage, only 0.003 μmol cm⁻² NO is released in the first 24 hours with a correspondingly lower [NO]_{Cu total} compared to day 1 (p < 0.01). After 12 months of storage at 4°C, the sensors are no longer capable of releasing NO. Indeed, all NO-release parameters for were roughly zero. Similar trends were seen for storage of NO-releasing mock sensors at 25°C. The [NO]_max and [NO]_{24 h total} (17 pmol cm⁻² and 0.42 μmol cm⁻², respectively) after only 1 day of storage at 25°C were significantly lower than the same parameters after 1 day of storage at −80°C (51 pmol cm⁻² and 1.03 μmol cm⁻² for [NO]_max and [NO]_{24 h total}, respectively), indicating that room temperature is not suitable for storage of NO-releasing glucose sensors for any length of time.

CONCLUSIONS

The unique properties of S-nitrosothiol NO donors (i.e., sensitive to light and heat) make it challenging to find a suitable sensor production methods that do not compromise the NO release required to mitigate the FBR. Sterilization of glucose sensors with ethylene oxide gas, the most common method for medical devices, resulted in compromised NO release over the first 24 hours. Both 70% ethanol and UVC light exposure proved to be promising strategies for sensor sterilization by determining sterilization cycle times that achieve a 10⁻⁶ SAL for representative Gram-negative and -positive pathogens. Neither sterilization method greatly impacted sensor performance or short-term NO release. However, 70% EtOH lowered NO payloads over extended periods (28 days). Sterilization with UV light proved to be superior, as sensor performance and short- and long-term NO release were most like unsterilized controls. Incorporation of UV irradiation, as well as aseptic processing, into sensor manufacturing represents a simple method for sterilizing glucose sensor surfaces while maintaining sensor performance and NO release. Sterilization with UV light should also be considered with other NO-releasing medical devices, such as vascular catheters, which are also commonly sterilized with ethylene oxide gas. While preservation of NO release following sterilization is one challenge to the translational potential of NO-releasing glucose sensors, storage conditions are another important consideration. Indeed, NO-releasing sensor membranes should be stored at cold temperature to circumvent undesirable loss in NO payloads. The evaluation of sterilization and shelf-life is critical for conveying the potential translation of new FBR mitigation strategies, particularly when reporting on future implantable glucose sensor designs.

Highlights.

• Ethylene oxide gas sterilization negatively impacts nitric oxide payloads

• Nitric oxide payloads maintained after UVC exposure

• UVC light exposure for only 10 min results in a 12-log reduction in bacteria counts

• Nitric oxide-releasing glucose biosensors can be stored for at least 12 months at −80ºC

Biographies

Taron M. Bradshaw received her BS degree in Chemistry in 2018 from Clemson University. She is currently a PhD candidate in Chemistry at the University of North Carolina at Chapel Hill with research interests spanning chemical sensors, electrospun fibers, and strategies to mitigate the foreign body response.

Courtney R. Johnson received her BS degree in Chemistry & Biochemistry in 2020 from Kansas State University. She is currently a PhD candidate in Chemistry at the University of North Carolina at Chapel Hill with research interests in on-demand release of nitric oxide via light for wound healing and anti-biofouling applications.

Christopher A. Broberg received his PhD in Molecular Microbiology in 2011 from the University of Texas Southwestern Medical Center, Dallas. His research interests are at the interface of chemistry and microbiology.

Darci E. Anderson received her BS degree in Biomedical Engineering in 2023 from the University of North Carolina at Chapel Hill.

Mark H. Schoenfisch received his PhD degree in Chemistry in 1997 from the University of Arizona. He is currently the Peter A. Ornstein Distinguished Professor at the University of North Carolina at Chapel Hill in the Department of Chemistry and Division of Pharmacoengineering and Molecular Pharmaceutics in the Eshelman School of Pharmacy. His research interests include analytical sensors, biomaterials, and the development of nitric oxide release scaffolds for biomedical and therapeutic applications.