Identification of Compounds Protecting Pancreatic Islets against Oxidative Stress using a 3D Pseudoislet Screening Platform

Oxidative stress leads to a lower success rate of clinical islet transplantation. Here, FDA‐approved compounds are screened for their potential to decrease oxidative stress and to protect or enhance pancreatic islet viability and function. Studies are performed on in vitro “pseudoislet” spheroids, which are pre‐incubated with 1280 different compounds and subjected to oxidative stress. Cell viability and oxidative stress levels are determined using a high‐throughput fluorescence microscopy pipeline. Initial screening on cell viability results in 59 candidates. The top ten candidates are subsequently screened for their potential to decrease induced oxidative stress, and eight compounds efficient reduction of induced oxidative stress in both alpha and beta cells by 25–50%. After further characterization, the compound sulfisoxazole is found to be the most capable of reducing oxidative stress, also at short pre‐incubation times, which is validated in primary human islets, where low oxidative stress levels and islet function are maintained. This study shows an effective screening strategy with 3D cell aggregates based on cell viability and oxidative stress, which leads to the discovery of several compounds with antioxidant capacity. The top candidate, sulfisoxazole is effective after a 30 min pre‐incubation, maintains baseline islet function, and may help alleviate oxidative stress in pancreatic islets.


Introduction
While clinical islet transplantation has become more common as a treatment to type I diabetes, long-term graft survival is limited DOI: 10.1002/adbi.202300264due to diverse factors including host immune responses and oxidative stress.During the procedure, the islets of Langerhans experience stress both during the isolation process and after transplantation.During the harvesting and isolation of islets, it has been shown that disrupting their vasculature leads to high levels of oxidative stress, acute ischemia and cytokine production such as TNF further contributes to beta cell apoptosis, [1,2] decreased function, and viability. [3]Moreover, it has already been established that beta cells within the islets are especially sensitive to oxidative stress.Oxygen and nutrient deprivation, which naturally occurs during a transplantation procedure, results in significantly reduced beta cell viability. [4,5][8] Similarly, the clinical transplantation outcome, islet function, and insulin independence after transplantation can be predicted based on the islet oxygen consumption rate. [9,10]Evron et al. showed that enhanced oxygen supply improved the long-term viability of transplanted islets, once again showing the importance of oxygen during the transplantation. [11]Lastly, the initial lack of proper vascularization after transplanted islets engrafted in the liver may contribute to prolonged oxidative stress. [12]In all, these data argue that strategies aimed at reducing oxidative stress could present a promising way to improve islet viability and function, and as a result, improve transplantation success.Similar approaches could also have merit in biomaterials-based approaches for the treatment of type I diabetes. [13] direct and extensively studied strategy to reduce oxidative stress is the use of antioxidants.While many studies have found that antioxidants can have beneficial effects, [14][15][16][17] several have also shown controversial effects, as was reviewed previously. [18]or example, an excess of antioxidants may lead to disrupted signaling processes that are dependent on reactive oxygen species (ROS), which suggests that using antioxidants for such applications would require very precise treatment regimens. [18]Hence, other approaches are needed to provide a clinically applicable protective treatment for improving beta cell function when experiencing oxidative stress.
As an alternative to achieve better survival of islets, several studies tested compounds aiming to preserve beta cell function in patients. [19]One such study focused on non-targeted, wholeorganism screening for improving beta cell function. [20]In a zebrafish model, the authors were able to establish that 84 of the 4,640 small molecules they tested enhanced insulin expression under homeostatic conditions.Other screening studies have instead investigated beta cell proliferation.An elegant paper by Wang et al. showed that using a combination of small compounds could induce beta cell proliferation in both normal and type II diabetes conditions, [21] yet proliferation is not equivalent to beta cell function, as proliferating beta cells are mainly immature. [22]astly, viability has also been used as a read-out, and it has been shown that by improving viability, the islet function could be enhanced. [23]hile all these approaches have their merits, two points of attention remain.First, many experiments were done in diabetic conditions, e.g. using high glucose culture media, in which the hyper-stimulated beta cells will be prone to secrete more insulin prior to exhaustion and death.Moreover, the focus on beta cell function or proliferation does not necessarily mean that the underlying issues in terms of viability or oxidative stress are actually targeted by the compounds.As oxidative stress can reduce beta cell proliferation and function, as reviewed recently, [24] screening for modulators of oxidative stress, ideally with a concomitant improvement in viability and beta cell function, could present a more effective strategy.
To date, some papers have looked at cellular stress responses, such as endoplasmic reticulum (ER) stress.Fu et al. investigated the effect of small molecules on ER stress using a luciferase reporter system and identified azoramide as a compound that could alleviate such stress, and moreover, improve beta cell function and insulin sensitivity in murine obesity models. [25]Yet, screening-based approaches aimed at mediating oxidative stress, especially in combination with positive effects on viability and islet function, are lacking.
In this study, we screened 1280 FDA-approved compounds from the Prestwick library, which has previously been used for similar studies, [26][27][28] for their capacity to reduce oxidative stress and improve beta cell viability and function.A pre-selection was made for compounds that did not have a known antioxidant or functional effect in beta cells and were not expected to induce long-term toxicity.[31] This antioxidant activity could be beneficial for islet transplantation. [32,33]This screening was done in a 3D pseudoislet model, which is a more physiologically representative model system than monolayer cell culture.We found 59 compounds that successfully retained or improved viability by up to 80%, while reducing oxidative stress by up to 40%, compared to pseudoislets that were only challenged by menadione and H 2 O 2 treatment.We validated our top candidate, sulfisoxazole, in primary human islets.Together, the developed screening assay was successful in selecting a compound that maintains the function of primary human islets, which could be validated in vivo in the future and potentially impact the islet transplantation success rates.In addition, the screening methodology could be used to identify novel compounds for the treatment of diseases where oxidative stress plays a significant role.
To distinguish the cell types within the pseudoislet, the alpha cells were labeled with BFP2 (Addgene #113725) and the beta cells were labeled with mNeonGreen (Addgene #113727), as previously described. [34,35]Briefly, the stable cell lines were generated using lentiviral transduction, and positive cells were selected after 48 h using 1 μg mL −1 puromycin dihydrochloride (Sigma--Aldrich) added to the culture media for at least 7 days.Transduction efficiency was assessed by live cell imaging after a minimal culture period of 7 days.

Pseudoislet Viability Screening
To find compounds that could be protective against cell death caused by oxidative stress, the pseudoislets were exposed to 10 μm of each of the 1280 compounds in the Prestwick chemical library (Prestwick Chemical Libraries) and incubated for 5 h.Additionally, 1 mm N-acetylcysteine (NAC) was tested as a known antioxidant control.The pseudoislets were then washed in PBS, after which 300 μm menadione (Sigma-Aldrich, M5625) was added for 30 min at 37 °C to induce oxidative stress.All experiments were performed in exposure medium comprising a 1:1 ratio of minimal essential DMEM (Gibco, 11880-028) supplemented with 15 mm HEPES, 0.1 mm non-essential amino acids, 1.5 g/L sodium bicarbonate, 3.0 g L −1 glucose, and SILAC medium (Gibco A2494201) supplemented with 1 mm sodium pyruvate, 10 mm HEPES, 3.0 g L −1 glucose and 2 mm l-glutamine.Cell viability was then assessed using the CellTiter-Glo 3D viability assay (Promega) by adding an equal volume of CellTiter-Glo 3D reagent to each well, mixing the contents for 5 min to induce lysis, and incubating the plate for 25 min to stabilize the luminescent signal.Luminescence was measured on an EnVision plate reader (PerkinElmer) and normalized to pseudoislets pre-treated with a vehicle-control and exposed thereafter to menadione.

Pseudoislet Oxidative Stress Screening
To detect compounds that could be protective against oxidative stress, the pseudoislets were exposed to 10 μm of each the Prestwick chemical library compounds for 5 h at 37 °C.Oxidative stress was induced by either 750 μm H 2 O 2 (Sigma-Aldrich, H1009) or 300 μm of menadione for 30 min, depending on the experiment.To detect oxidative stress (both cytoplasmic and nuclear), the CellROX Deep Red Reagent (Thermo Fisher, C10422) was added to the wells at a final concentration of 10 μm for 30 min.For the validation of sulfisoxazole, hydroxyzine dihydrochloride, and tribenoside, the pseudoislets were additionally exposed to different concentrations (2.5, 10, 50, and 100 μm) and preincubation times (0.5, 5, or 16 h) of the top three compounds.Untreated pseudoislets without exposure to H 2 O 2 or menadione served as controls.
To optimize segmentation during analysis, wheat germ agglutinin (WGA) CF568 (Biotium, #29077-1) was added at a final concentration of 10 μg mL −1 for 15 min after the start of the Cell-ROX incubation.Afterward, pseudoislets were washed once in PBS and fixed in 4% formaldehyde for 15 min at room temperature.Pseudoislets were imaged within 24 h on an automated inverted widefield Ti-E microscope (Nikon), equipped with a Spectra X light source (Lumencor), Prime 95B sCMOS camera (Photometrics), a Z500-N TI z-stage (MCL NANO), and an X-Light V2 spinning disk unit (CrestOptics) with a pinhole size of 70 μm.
Automated acquisition was performed using the NIS Elements JOBS module.In the first phase of acquisition, each well was fully imaged at 10× magnification (CFI Plan Fluor DL 10×) to detect individual pseudoislets within the well based on the detection of TC1 cells (excitation 395/25 nm, emission 447/60 nm) in widefield mode.Online detection was then performed based on thresholding of the whole spheroid, with a sphericity of >0.7.Coordinate point sets were saved, and applied for phase two of acquisition, where randomly selected pseudoislets were imaged at 20× magnification (CFI Plan Apochromat K 20×, NA 0.75, WD 1 mm) in confocal mode.

Image Analysis
All images were analyzed using the NIS Elements GA3 module (Nikon, version 5.21).For pre-processing, channels for TC and INS1E were subjected to background subtraction (rolling ball), prior to 3D thresholding of TCand INS1E-positive signals to detect individual cells.To improve segmentation, the WGA signal was thresholded and subtracted from the detected TC1 and INS1E mask.Within the remaining 3D volume of the individual cell, the mean fluorescence intensity (MFI) of CellROX was measured and the cell volume (μm 3 ) was determined.Data were stored in individual.csvfiles, after which averages per well for MFI and cell volume were determined.

Primary Human Islet Culture
Human islets of Langerhans were isolated from two cadaveric organ procurements (Table S2, Supporting Information) at the Human Islet Isolation Laboratory of Leiden University Medical Center (Leiden, the Netherlands).The donor islets were used after being deemed unsuitable for clinical islet transplantation and with research consent present according to national laws and regulations.Isolated human donor islets were seeded on nonadherent polystyrene plates and cultured in CMRL-1066 (Cellgro) containing 5.5 mm d-glucose, supplemented with 10% FBS (Gibco), 100 U mL −1 streptomycin (Lonza), 2 mm l-glutamine, 10 mm HEPES (Gibco), and 1.2 mg mL −1 nicotinamide (Sigma-Aldrich).The medium was refreshed every 2 days.

Primary Human Islet Oxidative Stress Measurement
To assess the oxidative stress response, human islets (5 days after isolation to allow recovery from isolation-induced oxidative stress) were first switched to RPMI-1640 (Sigma-Aldrich, R1383) dissolved in 1 L sterilized water, and supplemented with 2 g L −1 sodium bicarbonate, 5 mm glucose, 10% FBS, 1% HEPES, 1 mm sodium pyruvate, 100 U mL −1 penicillin, 100 μg mL −1 streptomycin, and 10 μg mL −1 ciprofloxacin.Islets were washed in PBS and pre-treated with 10 μm sulfisoxazole for 5 h at 37 °C, after which they were washed again in PBS and exposed to 750 μm H 2 O 2 for 30 min at 37 °C in exposure medium consisting of SILAC RPMI 1640 Flex Media (Gibco), supplemented with 2 g L −1 sodium bicarbonate, 5 mm glucose, 1% HEPES, 1% sodium pyruvate, 2 mm l-glutamine, 100 U/mL penicillin, 100 μg mL −1 streptomycin, and 10 μg mL −1 ciprofloxacin.After exposure, Cell-ROX Deep Red Reagent (Thermo Fisher, C10422) was added to the wells at a final concentration of 10 μm for 30 min to determine oxidative stress.The islets were then washed twice in PBS prior to fixation in 4% formaldehyde for 10 min at room temperature.Unstimulated (no H 2 O 2 ) and untreated (vehicle control) primary islets served as controls.
Directly after staining, CellROX (excitation 640/30 nm, emission 692/40 nm) was imaged in the islets at 20× magnification (CFI Plan Apochromat K 20×, NA 0.75, WD 1 mm) in confocal mode.Brightfield images were also recorded to delineate individual islets.Each experiment was performed in duplicate and a minimum of 10 islets per well were imaged and analyzed by determining MFI using NIS Elements GA3 module (Nikon, version 5.21).

Primary Human Islet Glucose-Stimulated Insulin Secretion Tests
Static glucose-stimulated insulin secretion (GSIS) tests were performed using triplicate samples of 30 human islets handpicked 5 days after isolation.The islets were washed in PBS and pretreated with 10 μm of the compounds for 5 h at 37 °C, washed again in PBS, and exposed to 750 μm H 2 O 2 for 30 min at 37 °C in exposure medium.The islets were then transferred to the top of 12 mm polycarbonate Millicell cell culture inserts (Merck; 12 μm pores) that were equilibrated for 1.5 h in a 24-well plate with Krebs buffer containing 0.2% BSA.Equilibrated samples were sequentially incubated with Krebs buffer containing low glucose (1.7 mmol L −1 ), and high glucose (16.7 mmol L −1 ) concentrations, each for 1.5 h at 37 °C.The islets were washed in PBS followed by cold acid ethanol overnight at 4 °C for cell lysis and extraction of total insulin content.After each incubation step, the supernatant was collected and stored.Insulin concentration was measured using the human ELISA kit (Mercodia) according to the manufacturer's instructions.Results are presented as low and high glucose stimulation profiles, normalized to the total insulin content of the treated islets.The stimulation index was calculated as the ratio of the insulin values after high glucose stimulation divided by those after low glucose stimulation.

Statistical Analysis
The compound screening for viability was performed once (N = 1) with 169 pseudoislets being measured (n = 169).The compound screen for oxidative stress was performed in independent triplicates (N = 3); in every experiment, duplicates of all conditions were included, and 10-20 pseudoislets were imaged in each well (n = 10-20).To investigate the effect of each compound, two-sided independent t-tests were performed, comparing compounds to their treatment control after normalization to the negative control.Furthermore, to examine the overall effect of the compounds, independent of oxidative stress induction, a two-way two-sided ANOVA with Dunnett's post-hoc test was performed.To study the effect of different concentrations and incubation times, two-way ANOVA was also employed with Dunnett's post-hoc test.Data from intensity measurements and GSIS of human islets were analyzed using a one-way ANOVA with Bonferroni post-hoc test to compare individual groups.All statistical analyses were performed using Graphpad 9.1.Data were represented as mean ± SEM, and statistical significance is indicated as *p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001, unless indicated otherwise.

Pseudoislet Viability upon High Oxidative Stress was Protected by 59 Compounds
To determine which compounds could be of interest for promoting pancreatic islet survival, all 1280 Prestwick library compounds were first screened for only their effect on viability because it is a more amendable read-out for screening with the high-throughput format (Figure 1; N = 1, n = 169).To validate the screening setup, we first determined the baseline viability and ROS production in pseudoislets upon stimulation with H 2 O 2 or menadione (Figure S1A,B, Supporting Information) in independent experiments.For the full compound screen, pseudoislets were then formed and incubated with 10 μm of each compound for 5 h prior to oxidative stress induction using 300 μm menadione, and viability was measured using luminescence.In parallel to the compound screen, independent control plates were run (Figure S1C,D, Supporting Information).From the 1280 compounds, 59 compounds met our criterion of a z-score ≥4, representing at least 75% viability compared to vehicle controls (Table S1 and Figure S1E, Supporting Information).From these 59 hits, ten compounds (Table 1) were selected for further validation based on a ranking of compounds based on three criteria: i) not having a known antioxidant effect, ii) not having a known effect on islet function, and iii) having low expectancy of long-term toxic effects.

Eight Compounds Reduced Induced Oxidative Stress in Beta Cells
To evaluate the potential antioxidant effect of these ten compounds, we developed a 3D screening platform for measuring oxidative stress by integrating an automated imaging pipeline and automated analysis with low-attachment microwell plates for high throughput.For each experiment, we measured baseline oxidative stress, H 2 O 2 -induced and menadione-induced oxidative stress, and the effect of a 5 h preincubation with 10 μm of each of the individual compounds.In addition, a known antioxidant, 1 mm N-acetylcysteine (NAC), was included as a control (Figure 2).

Seven Compounds Reduced Induced Oxidative Stress in Alpha Cells
One of the benefits of our screening approach is that we could study the effect of these compounds on both the beta cells and the alpha cells within the same system.Therefore, BFP2-labeled  Likewise, when oxidative stress was induced by H 2 O 2 (Figure 4B), we also observed a significant reduction of H 2 O 2oxidative stress from seven of ten compounds compared to pseudoislets only stimulated with H 2 O 2 (without compounds).Sulfisoxazole showed the highest reduction (−43.3%, p = 0.0013), followed by hydroxyzine dihydrochloride (−30.2%,p = 0.0060), similar to the reduction of menadione-induced oxidative stress.Moreover, isradipine (−29.3%, p = 0.0088), irsogladine maleate (−29.3%, p = 0.0088), tribenoside (−25.0%,p = 0.0240), alclometasone dipropionate (−25.3%, p = 0.0229), and astemizole (−24.2%,p = 0.0292) also reduced oxidative stress levels significantly, although under these conditions, 1 mm NAC was surprisingly not able to significantly reduce oxidative stress levels.Similar to our results from INS1E cells, sulfisoxazole was also able to reduce baseline oxidative stress levels in  cells (−26.3%, p = 0.0178; Figure S2B, Supporting Information).
While all seven compounds seemed viable candidates, we additionally investigated whether they could have potentially unwanted effects that could make them less suitable candidates for clinical application.We therefore examined the morphology of the pseudoislets and the cell volume within them, as oxidative stress can induce cytoskeleton changes. [37]We eliminated four compounds after this analysis: astemizole, isradipine, and chlormezanone treatment resulted in pseudoislets with a tendency to be less spherical compared to untreated controls (Figure S3, Supporting Information), and irsogladine maleate resulted in TC1 cells showing a significant increase in cell volume (Figure S4, Supporting Information).Based on these factors, we continued with sulfisoxazole, hydroxyzine dihydrochloride,  and tribenoside to determine the optimal concentration and preincubation time, as they exhibited the highest overall oxidative stress reduction without affecting morphology or cell volume.

Optimal Concentration and Pre-Incubation Times for Sulfisoxazole, Hydroxyzine Dihydrochloride, and Tribenoside
In this set of three compounds, we set out to determine the optimal concentration for the oxidative stress-reducing effects at the previously used pre-incubation time of 5 h prior to oxidative stress induction (N = 3, n = 10-20).We evaluated 5, 10, 50, and 100 μm concentrations, and determined the reduction in oxidative stress per cell type (Figure 5).Following a challenge with menadione (Figure 5A), sulfisoxazole was most effective at the 10 μm concentration, reducing oxidative stress by 40% compared to the treatment control for INS1E (p = 0.0198) and TC1 (p = 0.0239); hydroxyzine dihydrochloride at 10 μm showed a trend toward oxidative stress reduction (p = 0.0539 and p = 0.0548 for INS1E and TC1, respectively), whereas tribenoside did not show significant oxidative stress reduction (p = 0.2160 and p = 0.3152 for INS1E and TC1, respectively) at 10 μm.Similarly, following a challenge with H 2 O 2 (Figure 5B), all three compounds showed the greatest oxidative stress-reducing effect at 10 μm.
Having concluded that 10 μm indeed was an optimal concentration for lowering oxidative stress for all three compounds, we also established whether shorter or longer pre-incubation times could be used (N = 3, n = 10-20).We chose 30 min as a shorter pre-incubation time and 18 h (1080 min) for the longer pre-incubation time (Figure 6).For sulfisoxazole, a 30 min preincubation significantly reduced menadione-induced oxidative stress in both INS1E (−38.0%,p = 0.0213) and TC1 (−43.5%,p = 0.0137), and was equally effective as a pre-incubation of 5 h, which resulted in a 37.5% (p = 0.0098) and 39.8% (p = 0.0122) reduction for INS1E and TC1, respectively.For hydroxyzine dihydrochloride, a pre-incubation of 30 min showed a 30% reduction in menadione-induced oxidative stress for TC1 (p = 0.0333) but did not significantly reduce oxidative stress in INS1E (p = 0.0703).Interestingly, in TC1, hydroxyzine dihydrochloride was also more effective after a 30 min pre-incubation than after a 5 h pre-incubation (−34.4%,p = 0.0333).For tribenoside, the 5 h pre-incubation was most effective but did not significantly reduce oxidative stress in these experiments.A longer pre-incubation of 18 h did not reduce oxidative stress with any of the three compounds, meaning either the compounds lost their direct oxidative stress-reducing effect after such a long time, or compensatory mechanisms in the cell reduced the effectiveness of the compounds after long pre-incubation.

Sulfisoxazole Protected Primary Human Islets from Oxidative Stress and Maintained Islet Function
Collectively, sulfisoxazole showed the greatest protective effect in our screening experiments and was promising in terms of potency after both the short (30 min, = ) as well as the long (5 h) pre-incubations.To establish its potential for clinical use, we tested sulfisoxazole in primary human islets with a similar procedure.We pre-incubated the primary islets for 5 h with the compound prior to induction of oxidative stress by 750 μm H 2 O 2 , as menadione treatment induced cell death.Overall CellROX intensity was imaged per islet, and quantified (N = 2, n = 7-38).We observed that, as expected, oxidative stress increased after stimulation with H 2 O 2 (Figure 7A).Pre-treatment with sulfisoxazole protected against H 2 O 2 -induced oxidative stress compared to the treatment control (−30.1%,p < 0.0001) (Figure 7B).Unlike in the pseudoislets, we did not detect an effect on baseline oxidative stress (i.e., without H 2 O 2 ).
In addition to measuring oxidative stress levels, we also investigated the effect of sulfisoxazole on islet function by performing a glucose-stimulated insulin secretion test on the treated islets.Prior to glucose stimulation, insulin secretion was similar in all islets (Figure 7C), though we observed a tendency toward increased insulin secretion in donor 1 prior to glucose stimulation (p = 0.0686, Figure S5, Supporting Information), probably due to excessive cellular stress after H 2 O 2 . [38]Sulfisoxazole Figure 6.The effect of various pre-incubation times of compounds on oxidative stress levels after A) menadione-induced oxidative stress and B) H 2 O 2induced oxidative stress compared to baseline stress levels (no oxidative stress induction, solid line).In menadione-induced oxidative stress, the effect was more pronounced, and reduction of oxidative stress was already present after a 30 min pre-incubation with most compounds, where sulfisoxazole significantly reduced oxidative stress in INS1E (*, p = 0.0213) and TC1 (#, p = 0.0137), and hydroxyzine dihydrochloride significantly reduced oxidative stress in TC1 (#, p = 0.0333).A 5 h pre-treatment with the compounds also showed a similar reduction in oxidative stress, which again was significant for sulfisoxazole (INS1E: *, p = 0.0098; TC1: #, p = 0.0122) and hydroxyzine dihydrochloride (INS1E: *, p = 0.0484; TC1: #, p = 0.0490).Longer preincubation times did not reduce oxidative stress levels.For H 2 O 2 -induced oxidative stress, levels were lowest after 5 h pre-incubation with the various compounds, while oxidative stress levels tended to increase after 18 h pre-incubation.Data are presented as mean ± SEM (N = 3, n = 10-20).
pre-treatment resulted in a twofold increase in insulin secretion after oxidative stress induction by H 2 O 2 compared to islets challenged with H 2 O 2 alone (Figure 7C, p = 0.0179).When calculating the stimulation index (SI), we could clearly observe that sulfisoxazole pre-treatment preserved a SI comparable to islets that were not challenged with H 2 O 2 (p = 0.0316, Figure 7D).

Discussion
In this study, we set out to identify compounds with the potential to protect pancreatic islets against oxidative stress, e.g., during clinical islet transplantation.We employed a screening strategy using pseudoislets based on maintaining or improving viability, in which we identified ten promising compounds that we then tested for their effect on induced oxidative stress in vitro.We found that many of these compounds could reduce oxidative stress in both alpha and beta cells, with sulfisoxazole, hydroxyzine dihydrochloride, and tribenoside showing the highest reduction in oxidative stress without affecting cell volume or morphology.This finding suggests that cell survival, our first screening parameter, and oxidative stress status are linked.
Oxidative stress in beta cells can be the result of an oversupply of nutrients, such as glucose, that overload the mitochondrial respiration and create an overdemand for insulin, leading to ER stress, an unfolded protein response, and hypoxia.[41][42][43] These low endogenous antioxidant levels could contribute to JNK-mediated loss of beta cell differentiation and finally lead to beta cell dysfunction, expression of stress genes, and apoptosis. [39]Antioxidants can reduce oxidative stress and thereby increase insulin secretion. [44]nterestingly, even though not known for their antioxidant properties, many of our compounds, including our three finally selected compounds, outperformed both known antioxidants theobromine and NAC (Figures 3 and 4), which were included as a reference.This shows the benefit of a screening approach as was performed here for unveiling novel compounds that can modulate oxidative stress responses.
Our screening of (FDA-approved) drugs/compounds was done in a physiologically relevant model system, one that reflects the complexity of the native tissue.The pseudoislet contained the alpha and beta cells, which together comprise 90% of an adult pancreatic islet, and was recently shown to exhibit a completely different response upon induction of ROS compared to using single cell type models, or monolayer culture. [34]While this still does not fully recapitulate the full complexity of human islets of Langerhans, and lacks several cell types as well as vasculature, it is a useful alternative for pre-screening compounds instead of needing to test these on sparsely available primary human islets.Therefore, we believe that this model better reflects the physiologic state than 2D systems or 3D spheroids comprising only beta cells, which was supported by our sulfisoxazole results in the primary human islets.This validation in primary tissue, however, is crucial for translation of the results.
Sulfisoxazole effectively protected against induced oxidative stress in primary human islets and maintained beta cell function as determined by glucose-stimulated insulin secretion (Figure 7).[47] Sulfisoxazole is generally known as an antibiotic.
Antioxidant effects and antibacterial effects are not directly related, [48] although some antioxidants also show an antibacterial effect and some antibiotics can increase the endogenous level of antioxidants. [49,50]For example, many polyphenols have known antibacterial activities.These antibacterial activities depend mostly on the surface characteristics of the polyphenol structure and differ per bacterial strain, lipophilic character, and electric charge. [51]ore insight into the antioxidant effect of sulfisoxazole may be derived from its temporal effect, as it is most potent within the first 5 h of incubation, while after 18 h the antioxidant effect was much lower.Decreased antioxidant capacity has also been seen for compounds like edaravone used in the treatment of cardiovascular disease, after stimulation with an antioxidant for longer times. [52]Timing is essential to increase specific endogenous antioxidant systems. [53]Different antioxidants could induce different mechanisms that increase the overall antioxidant capacity of the cell.Compounds could directly work in the short term as an antioxidant scavenger of some damaging ROS, and as such reduce baseline oxidative stress, as we observed for sulfisoxazole, while the Keap1-Nrf2 pathway could be activated after a few hours, leading to an increase in the production of endogenous antioxidants. [54]An overactivation of the Keap1-Nrf2 pathway could, however, also lead to increased mitochondrial ROS production and a decrease in the antioxidant effect. [55,56]n future studies, it could also be of interest to study sulfisoxazole and its derivatives to help elucidate the underlying molecular mechanisms, for example by investigating whether sulfisoxazole could increase NRF2 translocation and upregulate endogenous antioxidants as GSH in beta cells, and to elucidate if sulfisoxazole derivatives could also be effective in ROS reduction in (pseudo)islets.
The level of oxidative stress induced by menadione after pretreatment with sulfisoxazole was much lower than the level of oxidative stress induced by H 2 O 2 after pretreatment with sulfisoxazole, indicating the mechanism of the antioxidant effect of sulfisoxazole.It is important to note that menadione induces oxidative stress in a different way than H 2 O 2. [57] While H 2 O 2 is a ROS itself and induces oxidative stress directly, menadione induces oxidative stress via redox cycling both in the mitochondria and the cytoplasm. [58]Some studies indicate that this process is accompanied by cytochrome-c release by the mitochondria, while other studies underline the importance of mitochondrial permeability.It takes up to 30 min after exposure for menadione to induce oxidative stress because it must be first taken up by the cell. [59]owever, once taken up, a lower concentration of menadione can induce more oxidative stress compared to H 2 O 2 , which we observed clearly in our study and has also been seen in bacterial cells. [57,60]This could have potential physiological consequences, as menadione has been shown to induce murine pancreatic acinar cell apoptosis. [61]In addition, ROS (superoxide) generated by menadione also decreased endothelial nitric oxide synthase activity and less cyclic guanosine monophosphate generation leading to increased endothelial dysfunction. [62]Endogenous antioxidants such as thioredoxin reductase-1 even exaggerate the generation of ROS via redox cycling. [63]6] Nevertheless, our findings demonstrate that sulfisoxazole is effective at decreasing oxidative stress, an important factor during islet transplantation.During transplantation, the instant bloodmediated inflammatory reaction activates a local inflammatory cascade following the secretion of inflammatory mediators and leads to the production of superoxide anion and H 2 O 2 , among others, by tissue-resident macrophages. [43,67]From this moment, upward of 60% of the transplanted islets will be lost.Interestingly, we observed that sulfisoxazole has a better ROS-decreasing function in case of menadione treatment, supporting its use as a pre-transplant ROS-reducing therapy.Sulfisoxazole was also able to maintain baseline oxidative stress levels in primary human islets upon induction of oxidative stress, and simultaneously maintain normal insulin secretion.In all, alleviating oxidative stress could enhance islet viability and function, [40,[68][69][70] and potentially increase the success rate of clinical islet transplantation.This could be achieved either ex vivo, protecting the islets prior to the transplantation, or incorporated in novel therapies, e.g. in stem cell-based therapies in combination with encapsulation devices, as some materials used for encapsulation are known to induce oxidative stress. [13,71,72]

Conclusion
Overall, this study showed that screening compound libraries has the potential to uncover novel functional effects on cells and tissues, e.g., their effects on oxidative stress in 3D cell aggregates.This study yielded single-cell information that helped uncover novel antioxidant effects of compounds for use in regenerative medicine approaches.Specifically, we discovered ten promising compounds, of which one, sulfisoxazole, may present a promising compound to protect islets of Langerhans during clinical islet transplantation.

Figure 3 .
Figure 3.The percentage of oxidative stress after pre-incubation with compounds and oxidative stress induction by menadione A) and H 2 O 2 B) in alpha cells (INS1E) (N = 3, n = 10-20).Data are compared to their treatment control and are presented as mean ± SEM; *p < 0.05, **p < 0.005.

Figure 5 .
Figure 5.Effect of various concentrations of compounds on menadione-induced oxidative stress A) and H 2 O 2 -induced oxidative stress B) compared to pseudoislets without induced oxidative stress or added compound (solid line).Menadione induced a strong response.Sulfisoxazole was most effective at a 10 μm concentration, with a 50% reduction in oxidative stress for INS1E (*, p = 0.0198) and TC1 (#, p = 0.0239), respectively.Hydroxyzine dihydrochloride showed a trend toward reduced menadione-induced oxidative stress in both INS1E and TC1 (p = 0.0539 and p = 0.0548 for INS1E and TC1, respectively), whereas tribenoside did not show a statistically significant reduction in oxidative stress at a 10 μm concentration.H 2 O 2 only induced minor oxidative stress, and we could not observe a statistically significant reduction in oxidative stress by the compounds.Data are presented as mean ± SEM (N = 3, n = 10-20).

Table 1 .
Compounds selected for screening for their potential oxidative stress-lowering effects.