All protocols involving human subjects were reviewed and approved by the Emory Institutional Review Board. All protocols involving mice were reviewed and approved by the Institutional Animal Care and Use Committee at Emory University.

Except as indicated, all chemicals were purchased from Sigma Chemical Corporation (Sigma, St. Louis, MO). Distilled, deionized water was used for analytical purposes. HPLC quality solvents were used for HPLC.

Human monocytic cells (U937, ATCC, Rockville, MD) were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (FBS, Atlanta Biologicals, Norcross, GA) and 10 U/ml penicillin and streptomycin sulfate. Cells were transferred to 0.5% FBS media 8–12 h prior to experimental manipulations.

To generate the desired range of extracellular redox potentials, the extracellular thiol/disulfide pool was altered by varying concentrations of Cys and CySS, added to cyst(e)ine-free RPMI, as previously described [11]. In these experiments, the total extracellular pool size of Cys+CySS was set at 200 µM, while concentrations of Cys and CySS were varied to obtain initial E_h values from −80 mV (physiological) to −46 mV (oxidized). The measured Cys/CySS redox range in human plasma is −120 to −20 mV [13]. E_h Cys/CySS is centered at −80 mV in healthy individuals and becomes oxidized with age, lifestyle factors such as smoking and alcohol abuse [14], [15], and pathologies such as atrial fibrillation [16], and age-related macular degeneration [9]. Therefore, the experimental range of −80 to −46 mV covers the physiological range and extends to oxidizing values that are reachable under pathological conditions.

Production of reactive oxygen species (ROS) was detected using 6-Carboxy-2′,7′-dichlorofluorescein diacetate (DCFH-DA, Molecular Probes, Eugene, OR, USA) [10]. U937 cells, plated into 96-well plate, were washed with KRH buffer and incubated with 100 µM DCFH-DA for 30 minutes (37°C, 5% CO₂). Cells were washed and exposed to physiological (−80 mV) and oxidized (−46 mV) Cys/CySS redox media at 37°C. Oxidation of DCFH-DA to fluorescent DCF was measured on a microplate reader (excitation, 485 nm; emission, 530 nm) [17].

Experiments were conducted using 10–14 week old, female C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME). Mice were housed in cages and maintained on a 12-h light-12-h dark cycle at the Division of Animal Resources at Emory University. All experiments were initiated during the light cycle. All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee.

Prior to the dietary intervention, all animals were fed pelleted rodent food (Test Diet 5015, Lab Diet Inc., Richmond, IN). Semi-purified diets were custom-prepared (Harlan-Teklad, Madison, WI, USA) in order to test the specific effects of sulfur amino acid (SAA) supplementation [18]. The SAA-adequate and SAA-supplemented diets were isocaloric and isonitrogenous and contained adequate and identical quantities of energy, nitrogen, carbohydrate, fat, fiber, micronutrients and essential amino acids. The SAA and nitrogen content was controlled at the desired experimental levels by varying the amount of cystine and methionine and the non-essential amino acids L-alanine, L-aspartic acid, glycine and L-serine. The SAA-supplemented diet contained 3-fold cystine, and 1.8-fold methionine compared to the SAA-adequate diet. Animals had free access to water at all experimental time points.

Escherichia coli O55:B5 LPS, dissolved in sterile PBS (100 µg/ml) was administered intraperitoneally to unanaesthetized animals at a dose of 1 mg LPS/kg body weight. Animals were sacrificed at 2 h post-LPS. The dose of LPS was used based on previous observations that it substantially increases plasma and lung IL-1β levels [19], and oxidizes plasma E_h Cys/CySS at 2 h [20].

Mice were anesthetized by isofluorane inhalation (Baxter Pharmaceuticals, Deerfield, IL). Opening of the blood brain barrier is reported at high concentrations of isoflurane [21]. The implications of this deleterious effect of isoflurane to measurements made in the current study are presently unknown. Subsequent to anesthesia, blood was collected by submandibular bleeding using a 4 mm mouse bleeding lancet (Medipoint, Inc. Mineola, NY). 0.18 ml of collected blood was immediately transferred to 0.02 ml of preservation solution containing γ-glutamyl-glutamate (γ-Glu-Glu) as an internal standard [22].

Samples were centrifuged at 16000 g for 60 seconds to remove precipitated protein, and 0.1 ml of the supernatant was immediately transferred to an equal volume of ice-cold 10% (w/v) perchloric acid. Samples were immediately stored at −80°C.

For HPLC analysis (Gilson Medical Electronics, Middleton, WI), derivatized samples were centrifuged, and 50 µl of the aqueous layer was applied to the Supercosil LC-NH₂ column (25 cm×4.6 mm; Supelco, Bellefunk, PA). Derivatives were separated with a sodium acetate gradient in methanol/water and detected by fluorescence [23]. Concentrations of thiols and disulfides were determined by integration relative to the internal standard. Redox potential s (E_h) of the GSH/GSSG and Cys/CySS pools, given in millivolts (mV), were calculated from concentrations of GSH, GSSG and Cys, CySS in molar units with the following forms of the Nernst equation for pH 7.4: GSH/GSSG, E_h = −264+30 log ([GSSG]/[GSH]²), Cys/CySS, E_h = −250+30 log ([CySS]/[Cys]²)[24].

Lung samples were excised, snap frozen in liquid N₂ and stored at −80°C. Total RNA was extracted from tissue using an RNeasy Midi Kit (QIAGEN Inc., Valencia, CA) according to manufacturer's instructions. DNase treatment was performed to remove contaminating genomic DNA. RNA concentration was spectrophotometrically determined at 260 nm, and 0.5 µg of total RNA was used to synthesize 20 µl of cDNA (Invitrogen, Carlsbad, CA). Quantitative real-time PCR was performed on cDNA using gene-specific primers on an iCycler IQ Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA). Primers were designed using Beacon Designer Software 4.00 (PREMIER Biosoft International, Palo Alto, CA) (Table 1). Samples containing serial dilutions of known concentrations of cDNA, encoding the gene of interest, were amplified in parallel. Data were analyzed using the iCycler Software, and starting quantities of message levels of each gene were determined from constructed standard curves. Melt curves were examined to ensure amplification of a single PCR product. Expression level of IL-1β was normalized to β-actin in monocytes and 18S ribosomal RNA in mouse lung tissue.

Levels of IL-1β in the cell-supernatant were detected by ELISA (R&D Systems, Minneapolis, MN), and are expressed relative to total protein content in the supernatant. Western Blot analysis of the cell extract from U937 cells was performed to detect the precursor form of IL-1β. After washing with PBS, cells were lysed in homogenization buffer (50 mM NaCl, 50 mM NaF, 50 mM NaP₂0₇-10 H₂0, 5 mM EDTA, 5 mM EGTA, 2 mM Na₃ V0₄, 0.5 mM PMSF, 0.01% Triton X-100, 10 ug/ml leupeptin, 10 mM HEPES, pH 7.4) and the concentration of proteins was determined by using a Bradford reagent (Bio-Rad, Hercules, CA). Equal amounts of protein were loaded onto 10% acrylamide SDS/PAGE gradient gels (Bio-Rad).

Proteins were transferred onto nitrocellulose membranes using a semi-dry trans-blot apparatus set at 25 V for 1 h (Bio-Rad). Membranes were subsequently incubated overnight with an anti-human IL-1β polyclonal antibody (Cell Signaling Technology, Danvers MA). Rabbit polyclonal anti-β-actin (Abcam Cambridge, MA) was used as a loading control. After washing with tris buffered saline (TBS) (10 mM Tris-HCL, pH 8.0, 150 mM NaCl, 0.05% Tween-20), membranes were incubated for about 1 h at room temperature with goat anti-rabbit IgG conjugated to horseradish peroxidase-coupled secondary antibody. After further washing, immunoreactive signals were determined by chemiluminescence. Protein bands were quantified by densitometric scanning using a GS-800 Calibrated laser densitometer (Bio-Rad).

To better assess the inflammatory status in vivo, levels of IL-1β, and tumor necrosis factor (TNF)-α were measured. IL-1β and TNF-α in human plasma, mouse plasma, and mouse lung homogenates were detected by immunofluorescence using a multiplex panel assay (R&D Systems, Minneapolis, MN). Briefly, the antibody coupled beads were incubated with the sample followed by incubation with a detection antibody. Fluorescence intensity was read on a Bioplex suspension array system (Bio-Rad, Hercules, CA).

To evaluate IL-1β gene transcription, U937 monocytes were electroporated with a pIL-1 (4.0 kb) luciferase promoter construct, as previously described [25]. Cells were plated in 6-well plate and were incubated for 8 h in specific redox media. Cells were lysed in reporter lysis buffer (Promega, Madison, WI) and assayed for luciferase activity. Measurements were done using a Labsystems Luminoskan Ascent Plate Luminometer. Results were recorded as luciferase units and were adjusted for total protein content.

This study was reviewed and approved by the Emory Institutional Review Board. A total of 16 healthy volunteers were recruited by posting fliers in public locations in the Atlanta/Emory University community. Following written informed consent, participants were admitted to the outpatient unit of the Emory University Hospital General Clinical Research Center (GCRC), where potential subjects were screened using a medical history, physical examination, urinalysis, standard chemistry profile and a complete blood count. Eligibility was established based on the following criterion: absence of acute or chronic illness (other than a medical history of well-controlled hypertension), BMI<30, non-smokers, and compliance in discontinuing nutritional supplements, if consumed, 2 weeks prior to study entry. Eligible participants were then scheduled for a 24 h inpatient visit within 2 weeks of screening in the GCRC. Characteristics of study subjects are presented in Table 2.

Participants were instructed not to eat after 10 pm the night prior to the inpatient visit in order to standardize baseline levels of metabolites. Participants were admitted to the GCRC at 7.00 on Day 1 and a heparin-lock catheter was placed in a forearm vein for blood sampling at 8.00. After a 30-minute supine resting period, 3 ml blood samples were drawn every hour for 24 consecutive hours. Participants were given breakfast at 9:30, lunch at 13:30, dinner at 17:30, an evening snack was provided at 21:30 immediately following the timed blood draw for that hour. The composition of the meals and snack was standardized for all subjects, as previously reported [26]. Water was provided ad libitum throughout the study period.

Data are presented as means+SEM. Statistical analysis was done using SAS v 9.1 (SAS Institute Inc., Cary, NC, US). Data from in vitro experiments were analyzed using an unpaired t-test. Data from in vivo murine experiments were analyzed using a one-way ANOVA with treatment specified as the main effect. To analyze data from human plasma, a linear mixed model was used. The association of Cys, CySS, and E_h Cys/CySS with IL-1β, and TNF-α was determined after controlling for time of day, BMI, age, and gender. Because none of the potential confounders had statistically significant regression coefficients, parameters for BMI, age, and gender were excluded to arrive at the most parsimonious model. The residuals for thiol/disulfide redox were normally distributed; therefore thiol/disulfide redox was specified as the response variable in all constructed models to improve the stability of the computed regression coefficients. The human study by its very design cannot establish a cause-effect relationship; causality is therefore not invoked by specification of thiol/disulfide redox as dependent and cytokine as the independent variable. The best model from all subsets was determined using the Akaike Information Criterion (AIC). Significance was set at a P value <0.05 for all tests.

Measurements of Cys and CySS in human plasma show that E_h Cys/CySS is centered at approximately −80 mV in healthy young individuals [24] and is oxidized in association with inflammatory disease states [9]. In vitro, oxidized extracellular E_h Cys/CySS activates nuclear factor (NF)-ΚB, a key pro-inflammatory transcription factor [10], [11]. Together, these data indicate that oxidized E_h Cys/CySS is likely to modulate pro-inflammatory cytokine production. Because peripheral blood monocytes are constantly exposed to the extracellular redox environment, we determined whether levels of IL-1β, a major monocyte-derived cytokine, are increased by oxidized extracellular E_h Cys/CySS.

U937 monocytes were exposed to E_h of −80 mV (physiological) or −46 mV (oxidized) for 8 h and levels of IL-1β were determined in the cell extract by western blot using an antibody that detects both the precursor and mature forms of IL-1β. Figure 1A shows an increase in cellular levels of the IL-1β precursor at −46 mV compared to −80 mV (lane 1 versus lane 3; P<0.05). IL-1β levels were also determined in cells treated with LPS (100 µg/ml) under different redox conditions, to assess whether IL-1β is increased in oxidized extracellular E_h under inflammatory states. The data show increased cellular levels of the IL-1β precursor at −46 mV compared to −80 mV with LPS treatment (lane 2 versus lane 4; P<0.05).

Because an increase in the precursor form of IL-1β was observed in the cell extract under oxidized conditions, we next attempted to quantify secreted IL-1β levels in the cell-supernatant by ELISA. In monocytes and macrophages, the release of mature IL-1β is dependent on stimulation of caspase-1 by extracellular ATP [27]. Because ATP was not added to the media, the form of IL-1β measured by ELISA is likely the precursor form. Figure 1B shows a 1.7-fold increase in precursor form of IL-1β in cells treated with E_h of −46 mV compared to −80 mV (P<0.01).

To determine whether increased pro-IL-1β in response to oxidized E_h occurred due to increase in mRNA abundance, IL-1β mRNA was quantified by real-time PCR. As shown in Figure 1C, IL-1β mRNA was more abundant in response to E_h of −46 mV, 4 h after treatment (P<0.05). Experiments with U937 cells expressing the human IL-1β promoter fused to a luciferase reporter gene revealed significant increase in luciferase activity at −46 mV, suggesting transcriptional induction of IL-1β by oxidized E_h (Figure 1D; P<0.05).

Previous studies have shown that induction of IL-1β by extracellular stimuli, such as ATP, is associated with generation of ROS and up-regulation of genes involved in GSH synthesis [27]. Therefore, we determined whether increase in IL-1β in response to oxidized extracellular E_h Cys/CySS is associated with intracellular redox changes. Measurement of cellular GSH (Figure 2A) and E_h GSH/GSSG (Figure 2B) revealed no significant differences between the −80 and −46 mV treatment groups at 2 h and 8 h.

Next, we examined whether oxidized extracellular E_h Cys/CySS stimulated cellular ROS production. Cells were pre-incubated with a ROS-sensitive dye, DCFH-DA, prior to treatment with −80 mV and −46 mV Cys/CySS redox media. Oxidation of DCFH-DA to fluorescent DCF was measured on a microplate reader. Cells treated with glucose oxidase served as positive controls. Within 5 minutes, cells exposed to oxidized E_h of −46 mV showed a 5-fold increase in DCF fluorescence compared to cells exposed to a physiological redox potential of −80 mV (P<0.001) (Figure 2C). When measured 2–6 h after treatment, the magnitude of DCF fluorescence increased in both −80 and −46 mV conditions, with at least a 2-fold higher fluorescence at −46 mV (P<0.001, data not shown). This indicates that oxidized E_h-induced ROS production was sustained over-time, and occurred in the absence of oxidation to cellular E_h GSH/GSSG.

We subsequently determined whether increase in ROS, in response to oxidized extracellular E_h Cys/CySS, was sensitive to the oxidation of redox-sensitive membrane-bound thiols. To this end, monocytes were pre-treated with 0.25 mM 4-acetamide-4′-amleimidylstilbene-2,2′-disulfonic acid (AMS), a non-permeant alkylating agent, prior to exposing the cells to given E_h Cys/CySS. Pre-treatment with AMS decreased (P<0.001), but did not completely inhibit, cellular ROS levels in response to oxidized E_h indicating that increase in cellular ROS occurs in part due to oxidation of membrane-bound thiols [10].

Taken together the in vitro data show that oxidized extracellular E_h Cys/CySS induces a rapid and sustained increase in ROS in monocytes compared to physiological E_h, which could be involved in oxidized E_h mediated increase in IL-1β. Based on these in vitro findings, we wanted to determine whether increase in IL-1β during inflammation in vivo can be attenuated by preservation of E_h Cys/CySS from oxidation.

Based on previous research showing that supplementation with sulfur amino acids (SAA) increases plasma Cys and shifts the E_h to a more reduced potential [18], we applied this approach to a mouse model of inflammation in which LPS/endotoxin is administered to increase IL-1β [19]. In this design, mice were fed either a SAA-adequate diet or an isonitrogenous SAA-supplemented diet for 7 days prior to endotoxin challenge. Mice were sacrificed 2 h post-endotoxin, a time point that coincides with maximal IL-1β production [19]. Plasma was collected prior to sacrifice for analysis of Cys, CySS, GSH and GSSG by HPLC. Thiol/disulfide redox parameters in untreated controls, fed pelleted rodent chow, are shown for comparison. Redox potential values for GSH and Cys are approximately 10–15 mV more reduced in mice plasma compared to human plasma.

SAA-supplementation protected against endotoxin-induced decrease in plasma Cys (Figure 3A; Cys (µM) - SAA supplemented, 19.6±1.9; SAA-adequate, 13.5±1.2; P<0.05). The decrease in plasma Cys in response to endotoxin is likely due to clearance of Cys, and oxidation of Cys to CySS. Ex vivo studies in hepatic membrane vesicles show that LPS stimulates hepatic influx of Cys to support gluconeogenesis and synthesis of acute phase proteins [28]. LPS also increases peroxide levels [29], and this could decrease plasma Cys via oxidation to CySS, an interpretation supported by the increase in plasma CySS levels in mice fed SAA-adequate diet (Figure 3B). However, because the magnitude of increase in CySS exceeds the decrease in Cys, mechanisms related to altered transport or recycling of Cys and/or CySS may also contribute to alterations in Cys/CySS homeostasis during endotoxemia.

The difference in plasma Cys between SAA-supplemented and SAA-adequate group is consistent with the predicted increase in Cys due to excess SAA intake, based on previous work done in our laboratory [18]. However, plasma CySS levels were not affected by excess SAA intake [18]. Thus, the lack of increase in plasma CySS in the SAA-supplemented group, 2 h post endotoxin, suggests additional effects on ROS homeostasis due to SAA supplementation (CySS (µM) - SAA-adequate, 141±9.7; SAA supplemented, 85.2±1.2; P<0.01). As a consequence of the higher Cys and decreased CySS, E_h Cys/CySS was on average 15 mV more reduced in response to endotoxin in the SAA-supplemented group compared to SAA-adequate group (Figure 3C; SAA-adequate, −76.4±2.6; SAA supplemented, −91.5±3.7; P<0.01).

Measurements of plasma GSH, GSSG (data not shown) revealed that plasma E_h GSH/GSSG (Figure 3D) was not oxidized after endotoxin treatment, and was comparable across treatment groups. These results confirm that the in vivo model is adequate to test whether attenuating oxidation of plasma E_h Cys/CySS decreases plasma IL-1β levels in inflammation.

Next, we determined whether the more reducing plasma E_h Cys/CySS, in SAA-supplemented animals is associated with a decrease in plasma IL-1β, in response to endotoxin. Plasma IL-1β levels in untreated controls were un-detectable (not shown). As shown in Figure 4A, plasma IL-1β decreased by 1.6-fold in SAA-supplemented animals (IL-1β (pg/ml) - SAA adequate, 190±22; SAA supplemented, 116±12; P<0.01). Induction of IL-1β in the lung is an early response to endotoxin [20], so we determined whether a protective effect on IL-1β was also observed in the lung, in response to a reduced extracellular E_h Cys/CySS. Lung IL-1β levels were decreased by 2-fold in the lung homogenate from the SAA-supplemented group (Figure 4B; P<0.001). Measurements of IL-1β message levels by quantitative real-time PCR revealed a 3-fold decrease in lung IL-1β mRNA levels (Figure 4C; P<0.05). Thus, preserving extracellular E_h Cys/CySS during endotoxemia decreased tissue and circulating levels of IL-1β.

Tumor necrosis factor (TNF)-α is another prototypical pro-inflammatory cytokine that is induced in concert with IL-1β in response to infection, injury, and immunological challenge [30]. To better assess the inflammatory status in vivo, we determined whether attenuation of TNF-α also occurred in SAA-supplemented mice. Results showed a greater than 2-fold decrease in plasma TNF-α levels with SAA-supplementation (Figure 4D; P<0.05). TNF-α in the lung was also significantly decreased (Figure 4E; P<0.05). Thus, the in vivo observations show that preservation of plasma E_h Cys/CySS during endotoxemia is associated with a decrease in IL-1β levels. These data extend the in vitro observations, and the combined findings support a mechanistic role for Cys redox potential in determining IL-1β levels.

To investigate whether Cys redox potential could represent a determinant of pro-inflammatory cytokine levels in humans, we examined IL-1β, TNF-α, Cys, and CySS in plasma samples from 16 healthy adults. The characteristics of the study participants are shown in Table 2.

Because plasma Cys and E_h Cys/CySS, and cytokines exhibit well documented circadian rhythms [26], [31], the study was designed to examine whether diurnal variation in E_h Cys/CySS was associated with variation in IL-1β and TNF-α. For this purpose, hourly samples were collected for an entire 24 h period from 16 individuals. Diurnal variation in plasma Cys and E_h Cys/CySS was closely related to meal intake, as reported previously (data not shown) [26]. However, due to considerable inter-individual variation in plasma IL-1β and TNF-α, we did not detect significant time-dependent variations in cytokine levels. Therefore, we used a secondary analysis to determine whether variation in E_h Cys/CySS correlated with variation in IL-1β, independent of time of day.

Because repeated measures were obtained from the same individual, we used a linear mixed procedure to model variation in redox parameters with cytokine levels controlling for time of day, BMI, age, and gender. As none of the potential confounders had statistically significant regression coefficients, parameters for BMI, age, and gender were excluded to arrive at the most parsimonious model. Cys, CySS and E_h Cys/CySS were specified as response variables in the analyses because the residuals for these biomarkers were normally distributed. Regression coefficients for the mixed model are presented in Table 3.

Examination of redox parameters for Cys redox potential revealed a strong positive association between E_h Cys/CySS and IL-1β (P<0.001). As seen in Figure 5A, oxidized values of E_h Cys/CySS correlate with higher levels of plasma IL-1β. The regression parameter for IL-1β indicates that a 1 unit increase in plasma IL-1β is associated with a 1.1 mV oxidation of plasma E_h Cys/CySS. Analysis of the association between plasma Cys and IL-1β revealed a significant negative association indicating that low levels of plasma Cys correlate with high IL-1β levels (Figure 5B; P<0.05). No significant correlation was observed between plasma CySS and IL-1β. While an association does not establish causality, together with our in vitro and in vivo data the present findings strongly suggest that Cys redox potential is an important determinant of IL-1β. Plasma TNF-α was also significantly associated with E_h Cys/CySS (Figure 5C) and CySS levels (Figure 5D) (P<0.05) suggesting that oxidized Cys redox potential is also a determinant of TNF-α. It must be noted that a caveat that may limit extrapolation of the results to the general population is the age (62 years) and the BMI (24 kg/m²) of the population studied. Because old age and overweight are risk factors for inflammation [32], [33], the association between E_h Cys/CySS and IL-1β and TNF-α may be less-strong in a younger population with an optimum BMI range.