Evidence for a Cytokine-Sensitive Network of Iron-Associated Genes That Protects Pancreatic Islets Against Ferroptosis

TL;DR

This study shows that a network of iron-related genes protects pancreatic islets from cell death caused by inflammation and iron stress.

Contribution

The paper identifies a cytokine-sensitive iron-gene network that unexpectedly protects islets from ferroptosis.

Findings

Hamp, Steap4, and Lcn2 are the most cytokine-sensitive iron-associated genes in islets.

Pre-exposure to low-grade inflammation protects islets from ferroptosis by upregulating these genes.

Steap4 overexpression reduces islet dysfunction and upregulates Hamp and Lcn2.

Abstract

Background/Objectives: The micronutrient iron is closely connected to inflammation and is among the complex factors contributing to beta-cell failure in diabetes. High levels of dietary iron increase the risk of developing type 2 diabetes, and excessive iron uptake by beta-cells can cause oxidative stress and inhibit function. Elevated levels of proinflammatory cytokines in obese individuals, such as interleukin (IL)-1beta and IL-6, increase the risk of developing type 2 diabetes, and there is evidence that these low levels of circulating cytokines can lead to islet dysfunction. Methods: In this study, gene microarray and other data were analyzed for expression differences in islets treated for 48 h with 10 pg/mL IL-1beta + 20 pg/mL IL-6 as a model of low-grade inflammation versus untreated. Results: Three iron-associated genes were among the most cytokine-sensitive in the mouse genome: Hamp, Steap4, and Lcn2. These proteins are all involved with increasing/retaining cellular iron. We hypothesized that increased cellular iron would lead to increased susceptibility to ferroptosis. Surprisingly, 24 h pre-exposure to low-grade inflammation, which upregulates this iron-gene network, prevented subsequent erastin-induced ferroptosis. We also found that Steap4 overexpression reduced islet dysfunction caused by high-dose proinflammatory cytokines (10× low-dose), suggesting an overall protective effect. Steap4 overexpression also upregulated Hamp and Lcn2, suggesting Steap4 regulates these cytokine-sensitive iron genes.; in contrast, ferritin and ferroportin gene expression, which are not sensitive to cytokines, were unchanged. Conclusions: These data suggest an inflammation-induced network of genes involved in cellular iron uptake and retention plays a protective role in islets against oxidative stress and ferroptosis.

Figures

Full text

1. Introduction

Type 2 diabetes (T2D) is a metabolic disease characterized by hyperglycemia that increases the risk of developing complications such as cardiovascular disease, neuropathy, blindness, and nephropathy [1,2]. Pancreatic beta-cells are found in the islet of Langerhans, and under normal conditions, secrete insulin to lower blood glucose levels. However, this function is impaired in individuals with T2D. Following the onset of insulin resistance and resulting hyperglycemia, beta-cell function diminishes in T2D-prone individuals [3].

Obesity is the leading risk factor for developing T2D and contributes to beta-cell demise through oxidative stress resulting from chronic elevation of glucose, free fatty acids, and pro-inflammatory cytokines [3,4,5]. Under conditions of metabolic stress, the surplus of nutrients triggers a cellular immune response and leads to low-grade inflammation. Adipose and other tissues, as well as infiltrating macrophages, secrete pro-inflammatory cytokines, such as Interleukin (IL)-6, IL-8, tumor necrosis factor (TNF)-alpha, and monocyte chemoattractant protein (MCP)-1, that circulate in low concentrations in individuals with obesity [6,7]. Inflammasome activation further contributes to cytokine production via the secretion of IL-1beta [8]. Regardless of the source, concurrent elevation of circulating levels of IL-1beta and IL-6 has been linked to an increased risk of T2D [9], and there is evidence that increased serum levels of IL-1beta and IL-6 contribute to islet dysfunction in prediabetic mouse islets but not in healthy mouse islets [10]. Our lab has also reported that mouse islets exposed to cytokines for 24–48 h (short-term exposure) in concentrations mimicking low-grade inflammation cause disruptions in calcium handling that are associated with mildly impaired insulin secretion [10,11].

Iron homeostasis is closely connected to the inflammatory response and plays an important role in the pathogenesis of T2D [12,13,14]. Maintaining iron homeostasis is also crucial to maintaining normal cellular function. In pancreatic beta-cells, iron levels are tightly regulated [15]. Under normal conditions, transferrin-bound iron in the bloodstream is captured by the cell surface transferrin receptor and endocytosed. Within the low pH endosome, the Fe^3+^ is released and reduced to Fe^2+^ by a Steap-family metalloreductase and transported across the membrane by a divalent metal ion transporter. Once in the cytosol, Fe^2+^ can then be utilized for various metabolic purposes. However, if cytosolic Fe^2+^ levels are excessive, it is oxidized to Fe^3+^ by ferritin and stored as unreactive ferric oxyhydroxide within the hollow ferritin cage or exported out of the cell by the channel protein ferroportin. Proper beta-cell function is highly dependent on glucose metabolism, which in turn relies on proper mitochondrial function. In-turn, mitochondrial proteins involved in oxidative respiration rely heavily on iron, incorporated as iron-sulfur clusters, heme, and iron-ion cofactors [16]. While iron is an essential trace element, excess intracellular iron contributes to the production of toxic reactive oxygen species (ROS) via the Fenton reaction [17], which can also trigger ferroptosis [18]. Of note, high levels of dietary iron have been linked to an increased risk of developing T2D [19], and beta-cells exposed to IL-1beta have exhibited excessive iron uptake, subsequent oxidative stress, and inhibition of normal cellular function [20]. Together, these studies highlight the importance of cellular iron homeostasis to maintain redox reactions necessary for beta-cell function without contributing to oxidative stress.

Expression levels of several proteins involved in iron transport and homeostasis have demonstrated sensitivity to pro-inflammatory cytokines. Of particular interest, lipocalin-2 (Lcn2) and hepcidin (Hamp) are highly sensitive to inflammation and are known markers of inflammatory stress [21,22,23]. The Hamp protein is involved in the post-translational down-regulation of ferroportin (the only known iron efflux protein) [24,25], and Lcn2 secreted into the extracellular space sequesters iron and re-enters the cell complexed with Fe^3+^ [26]. Additionally, Steap4 is a metalloreductase protein involved in cellular iron uptake [27] and has been shown to be upregulated by numerous pro-inflammatory cytokines [28,29,30], including doses consistent with low-grade inflammation [31]. In fact, Steap4 has been suggested to protect cells from inflammatory stress; however, the exact mechanisms are unknown [32]. In this study, we provide evidence for a highly cytokine-sensitive network of iron-related genes in pancreatic islets involving Lcn2, Hamp, and Steap4 that appears to protect islets from ferroptosis. We propose a conceptual model for how these genes may play a protective role in beta-cells’ response to short-term exposure (24–48 h) to low-grade inflammation.

2. Materials and Methods

2.1. Mice

All data for this study were collected from isolated mouse islets (male) according to approved IACUC protocols from the University of Virginia (microarray data and RT-PCR) or Ohio University (cell death measurements). Four- or five-week-old male db/db mice and heterozygous controls were used to collect microarray data; adult CD-1 male mice Envigo (Indianapolis, IN, USA) at 8–20 weeks of age were used for all other studies.

2.2. Islet Isolation

Islets were isolated for all studies according to standard procedure using Collagenase P (Roche, Basel, Switzerland, #11249002001) injected into the common bile duct for enzymatic digestion for ~8 min and purification from acinar tissue using combined Histopaque 1.119 g/mL with 1.077 g/mL to produce a 1.100 g/mL gradient to enhance islet purity. (MilliporeSigma, Burlington, MA, USA). Further detailed procedures are published [33].

2.3. Microarray Data Collection and Analysis

Published microarray data were analyzed to focus on identifying a list of genes relevant to cellular iron homeostasis. Abbreviated details of the original study [31] are provided for clarity. Islets isolated from prediabetic 4- or 5-week-old male db/db mice and non-diabetic male heterozygous controls were exposed for 48 h to 10 pg/mL IL-1beta + 20 pg/mL IL-6 (low cytokines), 100 pg/mL IL-1beta + 200 pg/mL IL-6 (high cytokines), or left untreated for each strain using four pooled pancreata for each of two duplicates. Concentrations used for low cytokines were representative of circulating cytokine levels associated with low-grade inflammation, and high-dose cytokine treatment was ~10-fold greater [6]. The high-dose cytokine treatment was used as confirmation for the effect of cytokines on gene expression in the microarray [31]. Following the treatment, tissues were extracted for total RNA using a Qiagen RNeasy kit (Thermo Fisher Scientific, Waltham, MA, USA). Microarray analysis of Affymetrix chips was performed by the UVA Biomolecular Research Facility, essentially as previously described [34] using Affymetrix GeneChip Mouse Genome 430 2.0 oligonucleotide arrays. All microarray data have been deposited in GEO under accession GSE62868, available at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE62868 (last updated 11 February 2019; last accessed on 30 January 2026).

Probesets that did not produce an Affymetrix MAS5 present/marginal/absent call of either “present” or “marginal” for at least 2 of the 12 samples were regarded as “absent” and removed (17,350 probesets), leaving 27,751 probesets of 45,101 for further analysis [35]. Statistics on this reduced set of expressed probesets were calculated using moderated ANOVA across all six experimental groups, using the limma package, version 3.18.13, with intensity-trended, empirical Bayesian shrunken variance estimates, resulting in 8769 hits based on p < 0.10 in at least one of 16 different t-test comparisons among treatment groups. Additional details regarding the analysis to identify probe hits can be found in the original study [31].

2.4. Treatment with Islet Stressors

Islets were treated with several different ‘stressors’ designed to evoke mild dysfunction via different pathways. The physiological effect of these conditions on islet function can be found in a previously published study [36]. Islets were exposed to each stressor in standard RPMI (RPMI 1640 medium (Invitrogen, Carlsbad, CA, USA)) containing 11 mM glucose and supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin for 48 h as follows: standard RPMI (untreated control), 28 G/glucotoxicity (28 mM glucose), free fatty acids (50 µM palmitate + 100 µM oleate + 50 µM linoleate), endoplasmic reticulum (ER) stress (100 nM thapsigargin), low-grade inflammation (10 pg/mL IL-1beta + 20 pg/mL IL-6), or mitochondrial stress (20 nM rotenone).

2.5. Quantitative PCR

Immediately following 48 h exposure to various stressors, total RNA from islets was isolated with the RNeasy mini kit (QIAGEN, Valencia, CA, USA). RNA purity and concentration were determined using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE, USA), and it was reverse transcribed into cDNA using Quantitect Reverse Transcription Kit (QIAGEN Inc.) according to the manufacturer’s protocol. Gene expression changes were made by PCR using SYBR Green-based technology with the Stratagene Mx3000P™ Real-Time PCR System (Agilent Technologies, Creek Blvd, Santa Clara, CA, USA) as described previously [10]. The relative gene expression was calculated using the 2^−ΔΔCT^ method [37] and normalized to the expression of beta-actin. Primer sequences are provided in Table 1.

2.6. Viral Transduction

Transduction-ready Lentiviral particles (Lentifect^TM^) were custom-made by Genecopoeia Inc. (Rockville, MD, USA) for overexpression of mouse Steap4. These viral particles were produced by packaging an expression construct for mouse Steap4, followed by purification and titer determination. The expression construct was generated by cloning the ORF for mouse Steap4 downstream of the CMV promoter in a lentiviral vector, pReceiver-Lv105 (Genecopoeia, West Pleasant View, CO, USA). This construct also had a puromycin resistance marker for the selection of transduced cells and the generation of stable cell lines. eGFP Lentifect^TM^ lentivirus particles were used as controls. Freshly isolated islets were placed in media containing 10 µL/mL of Steap4 lentivirus within 15 min of the final spin in the isolation process. Following 24-h exposure to the virus, islets were washed and given new media daily. For each transduction, a fraction of islets was tested by RT-PCR to confirm elevation of Steap4 message as compared to non-transduced and eGFP-transduced islets as controls. By 96 h post-virus exposure, Steap4 levels were elevated significantly.

2.7. Glucose-Stimulated Insulin Secretion

Islets were tested for glucose-stimulated insulin secretion as described previously [10]. Islets were preincubated at 37° and 5% CO_2_ for 1 h in a standard Krebs-Ringer-bicarbonate (KRB) solution, then washed and incubated in KRB supplemented with 3 mM glucose for 1 h, followed by a 1 h treatment with KRB containing 11 mM glucose. The supernatant was collected after each treatment, and insulin concentration in the supernatant was measured by an ELISA method (Mercodia, Uppsala, Sweden) according to the manufacturer’s instructions.

2.8. Intracellular Calcium

To determine changes in intracellular calcium, islets were loaded with 1 µM Fura-2 AM for 30 min in a modified KRB solution as previously described [38]. CellTracker Red was also included during Fura-2 loading for one of the two treatment groups in order to pair together and simultaneously image islets from both control and Steap4-overexpressing groups. Islets were then transferred to a recording chamber (Warner Instruments, Holliston, MA, USA. Cat: 64-0288) and perifused by passing solutions through an inline heater at ~500 µL/min to a temperature of 35 ± 3 degrees Celsius, which was mounted using a stage adapter (Warner Instruments, Cat: 64-0298). Observation of islets was performed using a Hamamatsu ORCA-Flash 4.0 digital camera (Hamamatsu Photonics K.K., Hamamatsu City, Japan, Model C11440-22CU) mounted on a BX51WIF fluorescence microscope with a 10X objective (Olympus, Tokyo, Japan). Excitation light was provided by a xenon burner supplied to the image field through a light pipe and filter wheel (Sutter, Novato, CA, USA, Model LB-LS/30) with a Lambda 10-3 Optical Controller (Sutter Instrument Co., Novato, CA, USA, Model LB10-3-1572). Images were taken sequentially every 5 s with 340 nm and 380 nm excitation to produce each ratio from emitted light at 510 nm. Data were analyzed using CellSens Dimension 1.13 imaging software (Olympus, Tokyo, Japan) [39].

2.9. Cell Death Detection

Apoptosis was measured using Annexin V (488 nm excitation/525 nm emission), which detects phosphatidylserine when it is exposed to the outer leaflet of the plasma membrane during apoptosis. A recent study showed that late-stage ferroptosis changes to the cell membrane (swelling and rupture) can be detected by AnnexinV [40]. Generalized cell death was detected using propidium iodide (535 nm excitation/620 nm emission), which is a cell exclusion dye. Islets were incubated for 15 min in a mixture of modified KRB solution (700 µL) + 7 µL of 30 mM stock of PI + 35 µL of AnnexinV conjugate (Invitrogen, Carlsbad, CA, USA). Regions of interest were drawn around individual islets to measure fluorescence intensity per islet. Additional details can be found in the following references [33,38]. Experiments were performed two or more times with N (ctr) = 50 islets (four repeats); N (cytokines+DMSO) = 21 islets (two repeats); N (cytokines+erastin) = 19 islets (two repeats); and N (erastin) = 30 islets (three repeats).

Lactate dehydrogenase (LDH) measurements were made with a commercially available lactate dehydrogenase kit—QUANT™ LDH Cytotoxicity Assay Kit (Invitrogen 20300). Islets were isolated following the standard collagenase protocol. Male CD-1 mice were used. The following treatments were performed: (1) untreated; (2) cytokines alone (10 pg/mL IL-1beta and 20 pg/mL IL-6, for total of 48 h, fresh cytokines were added at 24 h); (3) cytokines + vehicle (DMSO) (24 h of cytokines alone, followed by 24 h of cytokines + DMSO); (4) cytokines + erastin (20 µM in DMSO) (24 h of cytokines alone, followed by fresh media with cytokines + erastin for the last 24 h); (5) erastin alone (20 µM) (untreated for the first 24 h, followed by 24 h of erastin treatment). Additional islets were set up for the LDH test of maximum LDH release (as recommended by the manufacturer). Ten average-sized islets (150 µm diameter) were selected for each treatment. Treatments were performed in duplicate, and the experiments were repeated twice (the erastin-alone experiment was repeated three times). Islets were incubated in a humidity incubator in a 48-well plate with 500 µL complete RPMI media with appropriate treatments. In the last 24 h, the islets were transferred by pipetting into a fresh 48-well plate with 300 µL RPMI media with appropriate treatments. After 48 h of incubation, LDH activity was determined using the manufacturer’s protocol. The absorbance was measured with a plate reader (Synergy) at 460 and 680 nm. The cytotoxicity percentage was calculated by using absorbance values from untreated samples as spontaneous release, and absorbance values for each of the different treatments as the experimental, using the following formula:

$$[eqn]$$

2.10. Statistical Analysis

All other statistical comparisons were made by a two-tailed t-test for comparison of two conditions with significance at p < 0.05 or one-way ANOVA for multiple comparisons unless otherwise stated.

3. Results

3.1. Iron-Associated Genes in the Beta-Cell Are Highly Sensitive to Inflammation

We previously published results of a microarray study of low-grade inflammation in mouse islets that identified Steap4 as a highly cytokine-sensitive gene that was also associated with type 2 diabetes in humans [31]. In this study, we further analyzed these data to identify other genes involved with cellular iron homeostasis.

Islets isolated from diabetes prone mice (db/db) and heterozygous controls (db/+) were either left untreated or exposed to proinflammatory cytokines overnight (~20 h) at concentrations mimicking low-grade systemic inflammation (10 pg/mL IL-1beta + 20 pg/mL IL-6). It is important to note that such low-dose cytokine treatments do not induce cell death in normal healthy islets and have a limited impact on islet function as measured by intracellular calcium or glucose-stimulated insulin secretion [10,11]. Following treatment, mRNA was collected from these islets for an Affymetrix gene microarray as previously described [31].

Among 8769 cytokine-sensitive genes, we identified 45 genes involved in iron transport and homeostasis (Figure 1). Prominent genes related to iron homeostasis include Slc11a2, which codes for the divalent metal transporter 1 (DMT1) that transports iron across membranes [13,41], metallothioneins (MT1 and MT2) that protect against metal toxicity and oxidative stress [42], and the Steap family of metalloreductases, including Steap4, which has associations with obesity and T2D [31,43]. Of particular interest for further study, Hepcidin antimicrobial protein (Hamp), six transmembrane epithelial antigen of prostate-4 (Steap4), and lipocalin 2 (Lcn2) ranked within the top 10 most upregulated genes in response to cytokines of all 45,105 probes tested in this microarray for either db/db (Lcn2, 6th), db/+ (Hamp, 1st), or both (Steap4, top 5 for both).

As shown in the inset of Figure 1, Hamp, Steap4, and Lcn2 were among multiple iron genes upregulated in both db/db (no fill) and db/+ (filled bars) when comparing the low-dose cytokine treatment and untreated islets. Furthermore, treatment with IL-1beta and IL-6 in concentrations representative of low-grade inflammation upregulates Hamp, Steap4, and Lcn2 to a similar extent as treatment with high doses of cytokines, confirming the high sensitivity of these genes to cytokines (Figure 1, inset). For all three genes, expression was slightly lower in db/db compared to db/+ for the untreated condition and the low-dose cytokine condition. Hamp, in particular, had significantly reduced expression in the low-dose condition compared to db/+ controls.

3.2. Upregulation of Iron-Regulating Genes Is Specific to Cytokine Action

As a follow-up to the gene microarray, expression levels of Steap4, Lcn2, Hamp, and other iron genes were evaluated by qPCR in response to various conditions hypothesized to induce mild beta-cell stress (Figure 2) [36] using normal healthy islets from the outbred CD1 strain. RT-PCR was performed using mRNA collected from islets following 48 h exposure to 20 nM rotenone as a model of mitochondrial stress, 100 nM thapsigargin as a model of ER stress, 10 pg/mL IL-1beta + 20 pg/mL IL-6 as a model of low-grade inflammation (same as was performed for microarray discussed above), 28 mM glucose as a model of glucotoxicity, or a combination of free fatty acids (50 µM palmitate + 100 µM oleate + 50 µM linoleate) as a model of lipotoxicity. The concentrations of each stressor were carefully chosen to cause similar effects on islet function, specifically glucose-stimulated calcium response, without causing detectable increases in cell death [36].

Cytokine treatment upregulated Steap4 expression by 90-fold (p < 0.001), and upregulation to a smaller extent was observed for rotenone and thapsigargin treatment (Figure 2A). Cytokine exposure induced a 35-fold upregulation of Lcn2, whereas thapsigargin induced an 8-fold (p < 0.05) change. Rotenone, 28 mM glucose, and free fatty acid treatment induced no significant expression changes in Lcn2 (Figure 2B). Exposure to rotenone, 28 mM glucose, and free fatty acids for 48 h induced no significant changes in Hamp1 expression (Figure 2C), whereas 48 h treatment with cytokines induced Hamp1 upregulation by a fold change of 210 (p < 0.01). These data confirm the cytokine sensitivity of Steap4, Hamp1, and Lcn2, and demonstrate that upregulation is specific to cytokine-induced stress.

Hamp2 expression (Figure 2D) was not significantly upregulated under any of the stress conditions. Of note, two distinct forms of the hepcidin protein exist in mice [44], while only one isoform is present in humans. It has been previously demonstrated that IL-6 induces upregulation of Hamp1, whereas Hamp2 expression is unresponsive [45]. Prior to this study, the presence or lack of Hamp1 and Hamp2 in mouse islets remained unknown; therefore, both isoforms were included. Hamp2 was only marginally detectable in mouse islets (threshold cycle (CT) of 34.8 ± 6.4) compared to Hamp1 (28.5 ± 2.8) (Figure 2C,D).

To compare the cytokine-specific response of Hamp1, Steap4, and Lcn2 to genes responsible for iron storage or export, islet expression levels of ferritin and ferroportin were measured using RT-PCR following exposure to stress-inducing conditions identical to those described above (Figure 2E,F). In contrast to Hamp1, Steap4, and Lcn2, ferritin and ferroportin were not significantly up- or down-regulated by cytokines.

3.3. Overexpression of Steap4 Results in Upregulation of Lcn2 and Hamp1

To investigate potential crosstalk between iron-regulating genes, we examined the effect of overexpressing Steap4 on the expression levels of other iron genes. Given the previous identification of a SNP in human Steap4 (7q21.12, rs2538898) that increases the risk of T2D [31], Steap4 overexpression was of particular interest. Lentiviral transduction of the Steap4 construct increased Steap4 expression by ~25 fold; however, transduction of an inert green fluorescent protein (GFP) construct did not significantly upregulate Steap4 (Figure 3A) compared to untreated islets. Following transduction of Steap4 in mouse islets, Hamp (Hamp1) was upregulated ~10 fold (p = 0.08) when compared to non-transduced islets, and Lcn2 saw significant upregulation (p < 0.01) (Figure 3B). When compared to non-transduced islets, expression levels of ferritin and ferroportin genes were not changed upon Steap4 transduction (Figure 3B).

3.4. Steap4 Overexpression Attenuates Cytokine-Mediated Islet Dysfunction

We next examined whether overexpressing Steap4 had effects on islet function by examining glucose-stimulated insulin secretion. Overnight treatment with low levels of cytokines (10 pg/mL IL-1beta + 20 pg/mL IL-6) had no significant impact on glucose-stimulated insulin secretion in either control or Steap4-transduced islets (Figure 4). This is not surprising since minimal effects of low-dose cytokines on insulin secretion have been reported previously in normal healthy islets [10,36], and cytokines may even have protective effects [46].

We next used intracellular calcium, the proximal step to insulin secretion, as a more robust and sensitive marker of islet viability and function [47]. Measurements of intracellular calcium can serve as a proxy indicator of glucose stimulation [48,49] and islet health [47], although it should be noted that certain features, such as the amplifying pathway, are not detected by intracellular calcium measurements [50,51]. As shown in Figure 5A,B, there was no apparent effect of Steap4 overexpression on calcium in low glucose (3 mM) or during the phase 1 (peak) or phase 2 (plateau) response to 11 mM glucose stimulation under control conditions.

We next looked at whether Steap4 overexpression affected cytokine-induced islet dysfunction. Following overnight exposure to 10 pg/mL IL-1beta + 20 pg/mL IL-6 (low dose), islets showed the expected attenuation of glucose-stimulated calcium response, but no impact of STEAP4 overexpression on islet function was observed (Figure 5C,D). Islets exposed to a higher dose of cytokines (100 pg/mL IL-1beta + 200 pg/mL IL-6) showed an elevation in basal calcium in low glucose and a substantial reduction in the phase 1 and phase 2 response to glucose stimulation as previously described in detail [10,11]. Steap4 overexpression reduced the severity of these effects (Figure 5E,F). Indeed, the impact of Steap4 overexpression was highly significant for phase 1 (*** p < 0.001) and phase 2 (** p < 0.01), indicating a protective effect against high-dose cytokines. Note that we did not attempt to measure insulin secretion with the higher concentrations because of the increased likelihood of dead or dying cells, which can produce false insulin secretion results from lysed cells releasing insulin content upon death.

3.5. Cytokine Action and Ferroptosis

Ferroptosis is defined as iron-mediated lipid peroxidation leading to cell death, and cytokines like IL-1beta and IL-6 are thought to be triggers of ferroptosis [52]. Because our previously published synchrotron X-ray fluorescence data suggest redistribution of iron into iron puncta after exposure to low-level cytokines [53], we hypothesized that low-level stimulation of these cytokine pathways promotes cellular sequestration of iron, which should be anti-ferroptotic because the capacity for lipid peroxidation would be lowered by reducing iron availability. Note that we apply the term ‘sequestration’ with the assumption that iron is being stored in a non-labile form by ferritin cages or other means. We have not definitively identified the valences of sequestered iron detected by synchrotron X-ray fluorescence. To test this hypothesis, islets were first treated with proinflammatory cytokines at low doses for 24 h, and subsequent treatment for 12 h with erastin, a potent activator of lipid peroxidation and the ferroptotic pathway, was unable to efficiently kill beta-cells and islets, as measured by propidium iodide (Figure 6A, a cell-membrane impermeable dye that can enter cells undergoing ferroptosis) [40], AnnexinV (Figure 6B, a fluorescent probe that binds to phosphatidylserine when it translocates to the outer-membrane during late-stage ferroptosis) [40], and LDH (Figure 6C, lactate dehydrogenase, a protein measured extracellularly that is only released by dead or dying cells with degraded cell membranes). All three measures of cell death showed consistent results; pre-exposure to cytokines reduced or prevented erastin-induced cell death, rather than augmenting it. These data are consistent with the view that low-grade inflammation triggers pathways associated with iron homeostasis to protect islets from ferroptosis.

4. Discussion

Our study provides insight into the complex relationship among several iron-regulating genes and their role in the beta-cell’s inflammatory response. Of note, the cytokine exposure tested in this study is relatively short-term in duration and with a low concentration of cytokines. Hepcidin [21,25,54], Steap4 [27,31,55,56,57], and Lipocalin 2 [22,58], have each been shown to be independently involved in iron homeostasis, T2D, and the inflammatory response. We show that Steap4 can induce expression of Lcn2 and Hamp1 to activate a protective mechanism in beta-cells in response to inflammation. In the following sections, we discuss how these iron-regulating genes could contribute to or protect against inflammation and ferroptosis under short-term inflammatory challenge.

4.1. A Network of Iron Genes: A Pathway to Destruction or Protection

Proinflammatory cytokines are known agents of destruction that are utilized as chemical messengers by the immune system and other cells under stress. The most direct link between cytokine-induced cell death and iron is through ferroptosis. It is thought that IL-6 activation of JAK/STAT increases HAMP-mediated degradation of ferroportin to prevent iron export, thus triggering iron-induced ferroptosis [52]. IL-1beta, via activation of the NF-kB pathway, is thought to reduce key antioxidant enzymes while also increasing LCN-2-mediated iron entry, thus triggering ferroptosis [52]. We also observed increased Hamp and Lcn2 in response to low-dose IL-1beta plus IL-6. In previously published work, we provided evidence of increased iron retention in conditions of low-grade inflammation in beta-cells [53].

Ferroptosis has been linked to beta-cell dysfunction in a type 2 diabetes mouse model [59], and cellular iron accumulation is a contributor to the ferroptotic cell death pathway [60]. In fact, cytokine-induced iron accumulation, mediated at least in part through DMT1, has demonstrated damaging effects on beta-cells by increasing ROS [20]. Of note, DMT1 was upregulated in our microarray study (Slc11a2), suggesting yet another cytokine-sensitive gene involved in iron homeostasis. These collective observations suggest that increased iron is linked to increased susceptibility to ferroptosis.

However, we did not observe increased ferroptosis; we observed the opposite. When first exposing beta-cells to proinflammatory cytokines at low levels, we observed a sharp decrease in erastin-mediated cell death. Increased expression of ferritin, an iron-sequestering protein that forms iron ferritin cages, has been shown to inhibit ferroptosis, suggesting a protective effect of iron sequestration under conditions of cellular stress. In this study, we have not observed increased ferritin transcripts after exposure to inflammation; however, it is known that beta-cells have very high endogenous expression of ferritin protein [61] and that ferritin can be regulated translationally [62]; thus, to have a clearer mechanistic understanding of how ferritin contributes to the proposed cytokine-sensitive network, we would need to quantify ferritin protein levels in future experiments. Furthermore, ferritin lysosomal degradation, called ferritinophagy, has also been linked to ferroptosis and is excellently discussed in several reviews [63,64,65]. Inhibiting ferritinophagy in cultured beta-cells (MIN6) has been shown to decrease arsenic-induced ferroptosis [66]. In other words, there is a need to further investigate ferritin’s contribution to the observed cytokine-induced protection against ferroptosis in beta-cells. Additionally, Hamp and Lcn2 have been linked to ferroptosis inhibition [67,68], in contrast to what was noted above [52], and we have observed that Steap4 overexpression partially attenuates cytokine-induced islet dysfunction at higher concentrations of cytokines. Collectively, this suggests that upregulation of these genes may be involved in a protective mechanism that makes labile iron unavailable for ferroptotic pathways.

Additionally, iron retention in beta-cells may point to antibacterial efforts, as LCN2 and HAMP are key proteins involved in the innate immune system’s strategy of limiting extracellular iron available to pathogens [26,69]. Further sequestration of iron could provide necessary protection against ROS production that would otherwise accompany excess intracellular iron. Iron safely stored in its ferric state, such as in ferritin cages, is redox-inert and thus incapable of progressing the Fenton reaction and hydroxyl radical production [70]. However, excess iron in its reactive form can be harmful to beta-cells. Further investigation of labile iron pools would greatly contribute to the mechanistic understanding of how iron is redistributed during the inflammatory response and requires future experimentation.

It should be noted that this cellular protection is based on observations at a single time point involving islets in isolation from other factors and using cytokines at much lower concentrations than in most published studies. The dose, duration, and milieu of cytokines are key considerations in conducting experiments and interpreting data [6]. For example, cytokine-induced DNA damage can be reversed by reducing damage from nitric oxide or removing cytokines; however, this damage becomes irreversible after 24 to 36 h [71,72]. Evidence also suggests that IL-1beta at low enough concentrations may stimulate insulin and beta-cell protective pathways [46,73,74], whereas higher concentrations inhibit insulin and promote beta-cell death [75,76,77], a concept known as hormesis. A recent study mirroring our findings showed that first treating beta-cells with low concentrations of IL-1beta triggered an adaptive response that protected the viability and function of these beta-cells from subsequent cytokine exposure [78].

4.2. A Model of Cytokine-Mediated Iron Redistribution

Our findings show that elevated levels of pro-inflammatory cytokines IL-6 and IL-1beta, consistent with obesity-induced low-grade inflammation, substantially upregulate several iron-regulating genes. As shown in Figure 7, we hypothesize these proteins work together to collect and retain non-labile iron to reduce the potential for lipid peroxidation and ferroptosis. In this model (Figure 7), increased amounts of lipocalin 2 protein are secreted by the cell to bind with iron-laden siderophores, and LCN2 re-enters the cell through receptor-mediated endocytosis [79]. Fe^2+^ is then released into the cytosol, facilitated by metalloreductases like the Steap family proteins and transporters like DMT1. Upon hepcidin binding, ferroportin is degraded, preventing cellular iron export [24]. Thus, upregulation of hepcidin may also lead to increased iron accumulation within beta-cells. This cytokine-triggered network thus acquires and retains iron as a protective mechanism. It should be noted that there are other cytokine actions beyond iron regulation; however, iron seems to be a prominent feature of cytokine action at the levels involved with low-grade inflammation. Additionally, further investigation of labile iron pools and quantification of protein levels of proteins involved would be needed to further complete and refine the model proposed in this study.

4.3. A Note on STEAP4 and Metabolic Function

We showed that STEAP4 overexpression had protective effects on islets against cytokine damage. We were successful at confirming over-expression at the transcriptional level, but we acknowledge that protein expression was very difficult to confirm. Although we did not show the opposing situation of knocking down STEAP4 expression, other published studies of STEAP4 deficiency appear to support our findings. The first study of whole-animal STEAP4 (also called STAMP2) deficiency [55] showed that STEAP4 knockout mice developed obesity, glucose intolerance, dyslipidemia, and fatty liver disease [55]. STEAP4 deficiency also resulted in exaggerated inflammatory responses in adipose tissue [55]. Although the authors did not look at islet function, these data suggest substantial metabolic dysfunction in the absence of STEAP4. Another study showed reduced STEAP4 expression in association with visceral adipose tissue dysfunction [80], and we showed similarly reduced STEAP4 expression in human islets associated with increased HbA1c in T2D donors [43]. Of note, several other studies showed that overexpressing STEAP4 provided metabolic benefits, including improvement in diabetic renal function [81], attenuation of adipocyte angiogenesis and insulin resistance [82], reduction in atherosclerosis, and stabilization of plaques [83]. Together, these data suggest a protective role of STEAP4 with regard to metabolic function.

4.4. Other Considerations

As discussed earlier, it is important to consider the dose, duration, and milieu of cytokines when comparing effects [6]. The cytokine milieu used in this study (20 pg IL-1beta + 10 pg IL-6) is consistent with circulating cytokine levels found in pre-diabetic mice and individuals experiencing metabolic stress [9,10]. The duration of cytokine exposure in our studies is sufficient to produce physiological effects [84], but would still be considered an early adaptive response to inflammation rather than chronic low-grade inflammation associated with years of obesity-induced increases in circulating cytokines. A long-term study of low-dose cytokine exposure would require experiments in vivo, which would be difficult to conduct and interpret.

Cytokines in the concentrations and durations used in this study have been shown to alter glucose-stimulated calcium responses in islets isolated from pre-diabetic mice, whereas this toxic effect is less pronounced in normal murine and human islets [10]. In fact, IL-1beta can have stimulatory effects on islets with lower concentrations and/or shorter doses [74,85]. We observed no significant changes in insulin secretion due to overexpression of Steap4 or exposure to low-dose cytokines. However, overexpression of Steap4 appeared to attenuate the damaging effects of higher doses of cytokines on intracellular calcium, suggesting that Steap4 provides some protection against pro-inflammatory cytokines. Of note, the expression levels for Hamp, Lcn2, and Steap4 are lower for islets from db/db mice compared to het controls, especially for Hamp in low-grade inflammation. If the actions of these genes are indeed protective against iron-associated cell death, reduced expression in islets from db/db mice suggests that this iron-sequestering mechanism may be impaired in the prediabetic condition.

Similar considerations should be made with regard to iron. Excess labile iron would contribute to lipid peroxidation and ferroptosis, so it is crucial to have mechanisms in place to store iron in its non-labile form. However, increasing cellular iron in response to inflammation may be protective, provided that there is sufficient capacity to prevent the accumulation of ROS at high levels. We previously reported the identification of small iron-dense structures in beta-cells by synchrotron X-ray fluorescence microscopy that we termed ‘iron puncta’ [53]. Following cytokine exposure, the density of iron increased in these puncta, suggesting that iron may be sequestered in these structures in response to low-grade inflammation. The identity of the puncta and the valence of iron contained in these structures are not known, but puncta could provide a depot for iron storage under conditions of cell stress (see Figure 7). Quantifying a time course of labile iron would additionally contribute to understanding how iron could be moving from one cellular depot to another and would provide a better mechanistic understanding of the proposed model.

5. Conclusions

In summary, we show that a key feature of cytokine stimulation at low levels is the upregulation of several genes/proteins involved with iron homeostasis, including Steap4, lipocalin2, and hepcidin. These proteins appear to act in concert to acquire and/or retain iron, which looks to be protective based on two additional observations: (1) overexpressing Steap4 protects against islet dysfunction caused by high-dose cytokines, and (2) pre-exposing islets or beta-cells to low-dose cytokines (which we have shown to upregulate iron-associated genes and increase iron levels) prevents cell death caused by erastin, a key ferroptosis-inducing agent that triggers lipid peroxidation. Due to their inherently low antioxidant defenses, this mechanism of iron regulation may provide a unique approach to regulating oxidative stress in beta-cells.