Sodium thiosulfate attenuates liver injury in a rat model of sepsis by modulating ferroptosis and oxidative stress

Abstract

Sepsis is a dysregulated immune response to infection that can present with a pro-inflammatory endotype in which the ischemia/reperfusion injury leads to excessive production of reactive oxygen species (ROS), inflammatory mediators and the activation of cell death mechanisms. Sodium thiosulfate (STS) is a hydrogen sulfide producing agent that could reduce ischemia/reperfusion injuries induced ROS, inflammation and cell death. We used the peritoneal contamination and infection model in adult rats to induce sepsis. Animals from the treatment groups received either Trolox, an antioxidant medication or STS in doses of 500mg/kg b.w. (STS500 group) or 1000mg/kg b.w. (STS1000 group). Additionally all rats received fluid resuscitation therapy, antibiotic therapy and analgesic treatment. After 48 h the rats were sacrificed and blood and liver samples collected for analysis. Both STS500 and STS1000 reduced overall the ROS production and improved antioxidant defence mechanisms. They also reduced inflammatory cytokines IL-6 and TNF-α levels. In liver samples STS reduced inflammatory alterations and ferroptosis but had little effect on pyroptosis. STS could be a potential treatment option is sepsis pro-inflammatory endotypes by improving the redox balance and inhibiting ferroptosis and inflammation.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-29462-9.

Introduction

Sepsis is a dysregulated host immune response to infection and represents a global health burden with high mortality rates^1–3. During the early phases of sepsis the shock state and ischemia/reperfusion injuries cause increased production of pro-inflammatory cytokines that lead to excessive reactive oxygen species (ROS) release, organ dysfunction, cell damage and cell death^4,5. Inflammation-related cell death mechanisms involved in sepsis include necrosis, apoptosis, ferroptosis or pyroptosis⁶.

Ferroptosis is a non-apoptotic form of cell death in which iron imbalances enhance ROS production and leads to lipid-membranes peroxidation, and ultimately cause cell death⁷. Pyroptosis is a cell death mechanism in which pathogen factors such as endotoxins, pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) trigger inflammasome activation that mediate the formation of cellular membrane pores and the release of IL-1β and IL-18. This will trigger a cascade inflammatory reaction, where cytokines promote further cell death through pyroptosis activation^8,9. In experimental pro-inflammatory endotype sepsis models inhibition of these mechanisms has been proved to be beneficial in reducing organ damage and improving outcomes^10–12.

In sepsis, the liver plays a key role by producing pro- and anti-inflammatory cytokines, coagulation factors, and by clearing bacteria and bacterial debris¹³. It’s associated injury is both from the ischemia–reperfusion injury related with the haemodynamic alterations in sepsis, as well as from the pro-inflammatory reactions and the production of ROS in pro-inflammatory immune endotype models^14,15.

Hydrogen sulfide (H₂S) is a cysteine product with important intracellular antioxidant properties at low concentrations, but with cytotoxic properties at high concentrations^16,17. Sodium thiosulfate (STS) is a H₂S producing agent tested in clinical practice in cisplatin-chemotherapy induced hearing loss in paediatric patients, as an antagonist in cyanide poisoning and for calciphylaxis in peritoneal dialysis and haemodialysis patients^18–20.

By providing H₂S, STS has been proposed to reduce cell death associated with ischemia and reperfusion injuries. In the context of septic liver injury, it increases the antioxidant capacity of liver tissue and reduced the microcirculation damage in septic rats^21,22. Studies on STS in sepsis are scarce, particularly on ferroptosis and inflammatory cell death mechanisms. Because of its antioxidant capacity, STS can reduce the ischemia/reperfusion injury ROS formation and consequent lipid and protein peroxidation.

Based on this, STS might reduce ischemia/reperfusion injury and the shock induced inflammatory response linked with sepsis and sepsis-associated liver damage through inhibition of ROS. Therefore the present study aims to evaluate the effects of STS administration in a peritoneal contamination sepsis rat model. The action mechanism of STS was evaluated, specifically regarding the inflammatory cell death ferroptosis, and it’s effects were compared to Trolox, a known vitamin E analogue that reduced oxidative stress, inflammation and ferroptosis^23,24.

Materials and methods

Ethical statement

The experiment data are reported according to the ARRIVE 2.0 recommendations. All experiments were performed in line with the Romanian laws and regulation and following the 3R principles and the Minimum Quality Threshold in Pre-Clinical Sepsis Studies (Mqtipss) recommendations²⁵. The animal experiments were approved by the ethical board of the Iuliu Hatieganu University of Medicine and Pharmacy (Nr. AVZ71/10.05.2023).

Animals and housing

Animals used in this study were adult Wistar albino female rats (mean weight = 210g), obtained from the Iuliu Hatieganu University of Medicine and Pharmacy animal facility and housed in the animal facility of the Physiopathology department of the Iuliu Hatieganu University of Medicine and Pharmacy. They were maintained under 12h:12h light/dark cycles with constant temperature and humidity. The animals were held for 1 week of acclimatization at the animal facility before the experiments.

Peritoneal contamination and infection sepsis model

The peritoneal contamination and infection model was used to induce sepsis in rats. This consisted in the intraperitoneal administration of a pre-characterized human stool suspension that was described previously²⁶. The faeces suspension was obtained by mixing the frozen rat faeces with a pre-determined volume of Ringer solution (based on the volume of faeces) and vortexed directly before use. Obtained solutions were immediately used and not refrozen.

Animals were allocated by a scientist blinded to the study protocol in groups (n = 8) and cages. All administered substances were pre-prepared and labelled with the cage number and administration date to keep the experimental groups blinded. Animals were divided into 5 groups: (1) negative control group—SHAM, (2) sepsis control group – SEPSIS, (3) sepsis and Trolox (50 mg/kg bodyweight (b.w.)/day) – TRX, (4) sepsis and sodium thiosulfate (500 mg/kg b.w./day) – STS500, (5) sepsis, and sodium thiosulfate (1000 mg/kg b.w./day)– STS1000. Animals from the sepsis groups received 1.5 mL of faecal slurry suspension intraperitoneally between 11 a.m. and 1 p.m. in order to account for circadian rhythm influences. Further, they received meropenem 25 mg/kg b.w. subcutaneous at 6 h after sepsis model induction and then every 12 h until the end of the experiment. Fluid resuscitation was performed with Ringer solution 10 mL/kg b.w. subcutaneous at 6 h after sepsis model induction and then every 12 h until the end of the experiment. Analgesia was provided with metamizole drops (500 mg/mL) every 6 h or more frequent if deemed necessarily based on clinical evaluation. Experimental treatments based on group allocation were performed at 6 h after sepsis induction and then every 24 h. Animals from the SHAM group received at every time-points injections with Ringer solution.

During the experiments animals were checked based on the Rat Grimace Scale (RGS) every 4 to 6 h based on the animals condition²⁷. Lethargic or RGS scores of 4 were considered humane end points and the animals were sacrificed.

At 48 h after the intraperitoneal injection animals were euthanized in a separate chamber under deep anaesthesia with ketamine (87 mg/kg) and xylazine (13 mg/kg) followed by cervical dislocation. Under deep anaesthesia blood samples were collected from the retroorbital plexus and the liver and spleen were harvested and stored at -80°C until analysis.

Oxidative stress analysis

For evaluating oxidative stress levels, plasma total oxidative stress (TOS), total antioxidant capacity (TAC), nitric oxide (NO), malondialdehyde (MDA) and total thiols (SH) were measured using Jasco V-350 UV–VIS spectrophotometer (Jasco International Co., Ltd., Tokyo, Japan). Oxidative stress index (OSI) was calculated using the formula OSI = TOS/TAC²⁸, TOS and TAC were determined with an automated colorimetric method^29,30 and MDA using thiobarbituric acid³¹.

The measurement from tissue lysates of malondialdehyde (MDA), a marker for membrane lipids peroxidation due to oxidative stress, was conducted using the MDA Colorimetric Assay Kit (Elab Science, Wuhan, China). The assessment of the antioxidant enzyme superoxide dismutase (SOD) activity was conducted from tissue lysates by using the Total Superoxide Dismutase (T-SOD) Activity Assay Kit from Elab Science, Wuhan, China. Readings were conducted at 450 nm with the correction wavelength set at 540 nm using the SpectraMax ID3 plate reader; results are presented as MDA—nMoles/mg protein, and SOD—units/g protein.

ELISA analysis

For evaluating plasma inflammatory markers, blood samples were collected using EDTA tubes and centrifuged at 1000 × g for 10 min within 30 min of collection. The resulting plasma aliquots were stored at -80°C until analysis. Plasma levels of interleukin-6 (IL-6) (detection range: 12.5–800 pg/mL; sensitivity: 7.5 pg/mL; catalog number: E-EL_R0015) and tumor necrosis factor-alpha (TNF-α) (detection range: 15.63–1000 pg/mL; sensitivity: 9.38 pg/mL; catalog number: E-EL-R2856) were measured using commercially available ELISA kits (ElabScience, Houston, Texas, USA) following the manufacturer’s instructions. Absorbance readings were obtained with an 800 TS ELISA microplate reader (Agilent Technologies Inc., Santa Clara, CA, USA), and plate washing was performed using a Biotek Microplate 50 TS washer (Agilent Technologies Inc., Santa Clara, CA, USA).

Western-blot analysis

From liver samples we measured markers of ferroptosis (GPX4 and ACSL4), pyroptosis (cleaved caspase 1 and Gasdermin D), autophagy (Beclin 1 and LC3B), inflammation and oxidative stress (NRF2 and NOS2). The following primary antibodies were used: anti-ACSL4 (ACSL4-E-Ab-14661, made in rabbit), anti-GPX4 (GPX4-E-AB-67390, made in rabbit) both purchased from ElabScience (Wuhan, China), anti NRF2 (437C2a, sc-81342, made in mouse), anti-NOS2 (C11, sc-7271, made in mouse), anti-cleaved caspase 1(M211, made in rabbit), ELK Biotechnology, (Wuhan, China), Gasdermin D (GSDMD, made in rabbit), Wuhan Fine Biotech Co. Ltd. (Wuhan, China), Beclin 1 (E-8, sc-48341, made in mouse), GAPDH (G-9, sc-365062, made in mouse), purchased from Santa Cruz Biotechnology (Dallas, USA) and anti-LC3B (N-terminal, isoform B, PA1-46,286, made in rabbit), bought from Thermo Fisher Scientific (Rockford, USA). The secondary antibodies marked with HRP, anti-mouse were purchased from Promega (Madison, USA) and anti-rabbit from Cell Signalling Technology (Danvers, USA).

Liver samples were homogenized and lysed on ice for 1 h with agitation by using a lysis buffer made of Nonidet 0.1%, protease inhibitor cocktail in PBS, all purchased from Sigma Aldrich, (St Louis, USA) then supernatant was separated by centrifugation. Total protein concentration was measured by using the DC Assay Kit according to the manufacturer’s specifications (Biorad, Hercules, CA, USA), and bovine albumin as standard. For the Western Blot measurement, the proteins from the liver lysate samples (20 µg protein/lane) were separated by electrophoresis on SDS-PAGE gels and transferred to PVDF membranes, using the Trans Blot TurboTM Transfer System (BioRad, Hercules, USA). Blots were then blocked and further incubated with primary antibodies overnight. Afterwards, blots were washed and incubated with the corresponding secondary peroxidase-linked antibodies. Protein bands detection and quantification was done by using Bio-Rad Clarity Max ECL substrate, the Biorad ChemiDoc Imaging System and Image LabTM Version 6.0.0, build 25, Standard Edition, 2017, Bio-Rad Laboratories, Inc. analysis software. GAPDH was used as a protein loading control. Densitometry results were normalized to GAPDH and are presented as a ratio. All experiments were done in triplicate; gels were run in parallel in identical conditions. Full gel images are available in supplemental Fig. 4.

Fig. 4.

Liver inflammatory and cell death parameters results. Plots with min–max values/mean with images of the corresponding blots representing the experimental groups: TRX, Trolox administration group; STS500, sodium thiosulfate 500mg/kg; STS1000, sodium thiosulfate 1000mg/kg. Abbreviations: ACSL4, long-chain-fatty-acid—CoA ligase 4; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GPX4, glutathione peroxidase 4; GSDMD, gasdermin D; LC3B, microtubule-associated proteins 1A/1B light chain 3B; MDA, malondialdehyde; NRF2, nuclear factor erythroid 2-related factor 2; NOS2, nitric oxide synthase 2; SOD, superoxide dismutase. Only results from the multiple comparison using Tukey correction between the treatment groups and controls and between the control groups are shown on the graphs; *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant.

Histopathological analysis

For histopathological analysis, tissue samples were fixed in 10% formalin for 24 h and subsequently automatically processed using the Epredia™ Citadel 2000 Tissue Processor (Thermo Fisher Scientific, UK). The paraffin-embedded samples were sectioned at a thickness of 2 µm using the manual microtome HM 325 (Thermo Scientific) and stained using Hematoxylin and Eosin. Images were captured using the Olympus cellSens Entry 3.1 (Olympus Corporation).

Data analysis and statistics

For the statistical analysis, we first assessed data normality using Q-Q plots and the Shapiro–Wilk normality test. Parametrical data was assessed with the ANOVA test and followed by post-hoc analysis with Tukey’s correction for multiple comparisons. We performed correlation analysis using Pearson’s test as the values did not show linear relationship (supplemental Fig. 3. All tests were considered statistically significant for a p-value below 0.05. For a better understanding of variables interaction we performed principal component analysis (PCA) and divided the results based on groups. ANOVA analysis and its graphical representation was performed using GraphPad Prism version 10 for MacOS (GraphPad Software, San Diego, California, USA, www.graphpad.com), correlation analysis and PCA were performed using R (R version 4.4.2) in RStudio (version 2024.12.0.467). Tables represent mean values with standard deviations, median and interquartile range based on groups. Graphical representation of the between group analysis results with significance levels and complete PCA and correlation analysis results are provided in the supplemental material section.

Fig. 3.

Histopathological features of the liver parenchyma in the studied groups. H&E stain. (A) Sham group, no notable pathological changes within the hepatic parenchyma, bar = 100 µm; (B) Sepsis group, hepatocytes exhibit a disorganised pattern, and multifocally the hepatocytes are shrunken with hypereosinophilic cytoplasm and pyknotic nuclei (necrosis)(black arrows), or they contain variably sized clear vacuoles (vacuolar degeneration)(blue arrow), bar = 100 µm; (C) Trolox group, multifocal to coalescing areas of vacuolar degeneration in the hepatocytes, bar = 100 µm; (D) and (E) Sodium Thiosulfate 500 mg group, rare vacuols are present within the cytoplasm of hepatocytes (blue arrows), bar = 100 µm and bar = 50 µm; (F) Sodium Thiosulfate 1000 mg group, diffuse, severe vacuolar degeneration of the liver parenchyma, bar = 100 µm.

The figures were created using RStudio (version 2024.12.0.467), GraphPad Prism (version 10) and Affinity Designer (version 2.6.2).

Data availability

The dataset analysed during the current study is available from the corresponding author on reasonable request.

Results

Blood analysis

Collected plasma was analysed for liver damage markers aspartate transaminase (AST) and alanine transaminase (ALT), oxidative stress parameters, antioxidants levels and inflammatory molecules IL-6 and TNF-α. For evaluating plasma oxidative stress, total oxidative status (TOS), malondialdehyde (MDA), nitric oxide (NO) and advanced oxidation protein products (AOPP) were measured. The antioxidant defence was quantified by determination of total antioxidant capacity (TAC) and total thiols (SH). Results are displayed in Figs. 1 and 2 with supplemental data available in supplemental table 1. ANOVA and Tukey-post-hoc test results are presented in supplementary tables 1 and 2.

Fig. 1.

Systemic oxidative stress analysis results. Boxplots in Tukey style representing the study groups: TRX, Trolox administration group; STS500, sodium thiosulfate 500mg/kg; STS1000, sodium thiosulfate 1000mg/kg. Only p-value results from the multiple comparison using Tukey correction between the treatment groups and controls and between the control groups are shown on the graphs; *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant. Abbreviations: AOPP, advanced oxidation protein products; MDA, malondialdehyde; NO, nitric oxide; TAC, total antioxidant capacity; TOS, total oxidative status.

Fig. 2.

Blood inflammatory markers analysis results. Boxplots in Tukey style representing the study groups: C-, sham group; C + , sepsis group; TRX, Trolox administration group; STS500, sodium thiosulfate 500mg/kg; STS1000, sodium thiosulfate 1000mg/kg. Only p-value results from the multiple comparison using Tukey correction between the treatment groups and controls and between the control groups are shown on the graphs; *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant. Abbreviations: IL, interleukin; TNF, tumoral necrosis factor.

TAC, TOS, and OSI index levels were significantly different between the groups. In the SEPSIS group compared with SHAM there was a significant pro-oxidative status due to both TOS increase (p < 0.001) and TAC reduction (p < 0.001) compared to SHAM. TOS and OSI were lowered by TRX (p < 0.05) and STS treatments (p < 0.01) while TAC was increased after TRX (p < 0.05) and STS administration (p < 0.01). There were no statistically significant differences between the 3 treatments.

MDA levels significantly increased in SEPSIS group compared to SHAM (p < 0.001) while TRX (p < 0.01) and STS500 treatments (p < 0.05) significantly reduced MDA formation. In the STS1000 group MDA levels were not significantly reduced (p > 0.05) compared to SEPSIS.

NO was significantly elevated (p < 0.001) in SEPSIS group compared with SHAM, and they were not modified significantly by any treatments (p > 0.05).

AOPP level was higher in SEPSIS rats (p < 0.01) and TRX treatment diminished it without statistically significant difference (p > 0.05). Compared to SEPSIS, STS500 (p = 0.01) and STS1000 (p < 0.001) treatments significantly reduced AOPP levels in a dose-dependent manner.

SEPSIS reduced SH significantly (p < 0.001) but the treatments had no important effect on this antioxidant marker (p > 0.05).

IL-6 and TNF-α levels increased significantly in SEPSIS rats compared to SHAM and the treatments significantly reduced their levels: TRX (p < 0.001), STS500 (p < 0.01) and STS1000 (p < 0.001). For TNF-α there was no statistically significant difference between the treatments, while for IL-6 there was a significant reduction after TRX (p < 0.001) and STS1000 (p < 0.05) administration compared to STS500.

In line with findings in a mouse model, our model did not show high values of ALT or AST caused by liver cells injury (supplementary table 1)³².

Liver tissue analysis

Histopathological examination showed that in the SHAM group no significant alterations were observed in the liver parenchyma (Fig. 3A). In SEPSIS group, hepatocytes exhibited multifocal disorganization, with individual cell necrosis characterized by shrunken cells with hyper-eosinophilic cytoplasm and pyknotic nuclei. Additionally, vacuolar degeneration was noted in the hepatocytes (Fig. 3B). In TRX group, vacuolar degeneration was more pronounced compared to SEPSIS group (Fig. 3C).

In STS500 group, hepatocyte degeneration was less severe compared to TRX group, and only rare instances of vacuolar degeneration were observed (Fig. 3D and E). However, in STS1000 group, vacuolar degeneration was more pronounced compared to STS500 group (Fig. 3F).

For assessing liver cell death mechanisms, we determined the markers of ferroptosis: long-chain-fatty-acid—CoA ligase 4 (ACSL4) and glutathione peroxidase 4 (GPX4), pyroptosis: cleaved caspase-1 (c-casp1) and gasdermin D (GSDMD) and autophagy: beclin-1 and LC3B were assessed by western blot. For oxidative status and inflammation levels we measured the nuclear factor erythroid 2-related factor 2 (NRF2) and inducible nitric oxide synthase (NOS2) levels. Results are displayed in Fig. 4 with supplementary data in supplemental tables 3 and 4.

ACSL4 levels, as ferroptosis marker, diminished in the SEPSIS group compared with SHAM (p < 0.05) while the treatments did not significantly influence them (p > 0.05). TRX administration increased mean ACSL4 levels without statistical significance. GPX4 levels decreased significantly in SEPSIS group compared to SHAM (p < 0.001), TRX (p < 0.001) and STS500 (p < 0.001), while the treatments significantly improved them. Despite an increase of GPX4 in STS1000 compared to SEPSIS group, this was not statistically significant.

C-casp1 values increased in SEPSIS group, but not statistically significant compared to SHAM (p > 0.05). Also, treatments did not statistically modify c-casp1 levels. GSDMD was lower in the SEPSIS group compared to SHAM while the treatments increased its mean values, but without any statistical significance (p > 0.05).

Autophagy markers beclin1 and LC3B from the liver tissue were not significantly altered in the SEPSIS group compared to SHAM or following treatments (p > 0.05).

NRF2 levels were increased in the SEPSIS group, but the difference was not statistically significant compared to SHAM group (p > 0.05). Similarly, NOS2 diminished in the septic model groups. SEPSIS group exhibited significantly lower levels compared to SHAM (p < 0.001) while the treatments did not influence them (p > 0.05).

Superoxide dismutase (SOD) enzymatic activity was significantly reduced in the SEPSIS group compared to SHAM (p < 0.001). TRX and STS500 treatments had higher mean values of SOD (see supplemental table 4) but were not statistically significant (p > 0.05). Only the STS1000 treatment group showed a statistically significant increase in SOD levels compared to SEPSIS.

MDA levels in liver tissue were significantly increased in SEPSIS compared to SHAM and all treatments significantly reduced the levels. Of these the highest mean difference was for the STS1000 treatment group. MDA and SOD summary statistics and Tukey post-hoc test results are presented in supplemental tables 4 and 5.

Principal components analysis results are displayed in Fig. 5 with additional data in supplemental Figs. 1 and 2.

Fig. 5.

Principal components analysis results. Legend: A-E blood analysis parameters, F-J liver analysis parameters. Abbreviations: AOPP, advanced oxidation protein products; ACSL4, long-chain-fatty-acid—CoA ligase 4; c-casp1, cleaved-Caspase 1; AST, aspartate transaminase; AST, alanine transaminase; GPX4, glutathione peroxidase 4; GSDMD, gasdemin D; IL, interleukin; MDA, malondialdehyde; NOx, nitrites and nitrates; NOS2, inducible nitric oxide synthase; NRF2, nuclear factor erythroid 2-related factor 2; TAC, total antioxidant capacity; TNF, tumoral necrosis factor; TOS, total oxidative status;

In the PCA analysis of blood parameters the main contributing factors to the principal components (PC) 1 and 2 for all groups analysis were TOS, TAC, OSI, NO, MDA, TNF-α and IL-6. In the TRX and STS500 groups AOPP also had a great contribution to the model. Most pro-oxidant values were positively correlated, while negatively correlated with SH and NO. TAC was negatively correlated with TOS and OSI indexes.

For the liver analysis parameters most notably contribution to the models for each group had Beclin1, Casp1, NOS, GSDMD and GPX4. Positive correlation can be seen between GPX4 and ACSL4 as well as between GPX4 and NRF2 and between GSDMD and Casp1.

Discussions

Current sepsis management involves mostly supportive treatment strategies associated with anti-infectious medication. Although several targets have been identified in pre-clinical sepsis studies, such as targeting oxidative stress, inflammation and cell death mechanisms, very few of them have showed efficiency in translating to clinical practice^1,33. STS acts as an antioxidant by releasing H₂S in a hypoxia-controlled manner that avoids it’s toxic effects³⁴. A previous study showed beneficial effects of STS in liver ischemia/reperfusion injury via ROS scavenging²¹. This type of injury is also seen in sepsis and therefore, STS could potentially reduce sepsis associated oxidative stress, inflammation and liver injury via its antioxidant properties. Since STS is approved for clinical use, it can represent an attractive therapeutic option as an adjuvant therapy in sepsis. In our study STS improved the oxidative status in favour of antioxidants and reduced the inflammatory process via modulating GPX4 of the ferroptosis pathway.

In the pro-inflammatory endotypes of sepsis the early stages are associated with shock and ischemia/reperfusion injuries³⁵. This involves the production of ROS and inflammation with the role to eliminate the source of infection by releasing acute-phase reactants and increasing ROS and inflammatory mediators^36,37. As part of the oxidative stress assessment, the markers which indicate the ROS presence and antioxidant defence include TOS and TAC and their relationship as the OSI index^38,39. More specific, macromolecules oxidation markers including MDA, AOPP and NO are also increased in experimental models of sepsis^40–42. The antioxidant defence is altered in sepsis states, particularly thiols homeostasis. Thiols reduce ROS by acting as free radicals scavengers.^43,44. In the current study, the SEPSIS group had higher TOS levels and lower TAC values and an increased OSI index compared to SHAM, which is suggestive of a pro-oxidant status. In the treatment groups TRX and STS significantly improved the oxidative status and decreased ROS presence in blood as seen with decreased TOS and increased TAC levels. The STS antioxidant effect included lipid and protein peroxidation reduction, with lower MDA and AOPP levels. The effects of STS on protein oxidation have not yet been studied, but are most probably related to STS acting as an antioxidant, which also reduced MDA levels and contributed to the overall ROS reduction^45,46. NO is a key mediator in sepsis associated with haemodynamic instability, which contributes to the multiple organ dysfunction⁴⁷. In a zebrafish model of hyperglycaemia, STS induced the upregulation of NO production, proving beneficial reduction of hyperglycaemia-induced renal damage⁴⁸. In the sepsis model we obtained similar results, with higher NO levels compared to SHAM. The treatments did not significantly influence the NO levels. SH levels are significantly reduced in septic patients and because they show both an increased production of ROS and a decreased antioxidant capacity, reduced SH levels have been associated with the risk of developing organ damage⁴⁹. In our model STS treatment increased mean SH concentrations without statistical significance. TRX administration showed the highest mean SH levels and reached statistical significance.

TNF-α and IL-6 are two of the main pro- inflammatory cytokines involved in sepsis, acting as biomarkers for the inflammatory response⁵⁰. In the current research both TNF-α and IL-6 were significantly increased in the SEPSIS group. STS therapy significantly decreased both cytokines’ levels in blood. The IL-6 decrease was more dependent on the STS dose than TNF-α, which was similarly reduced in all treated groups. Similar results were seen in a study on glial cells, where STS showed anti-inflammatory properties by reducing TNF-α and IL-6 secretions⁵¹. Also, in wound-healing models, tissue TNF-α secretion was reduced by topic STS-gels administration⁵². STS treatment can therefore reduce the general inflammatory response in sepsis.

The liver is one of the main organs involved in sepsis, by mediating the immune response and bacterial clearance. It’s main mechanisms of injury are related to the systemic inflammatory response and to the hypoxic insult associated with septic hypoperfusion and ischemia/reperfusion injury¹³. Generally liver dysfunction occurs early in the pathogenesis of murine sepsis models and liver injury biomarkers can be seen sooner than 24 h following the septic insult⁵³. In our model, the liver injury markers were not increased in the septic rats, but rather decreased, which can account for the success of fluid resuscitation and antibiotic therapy on liver damage. From a histological point of view the SEPSIS compared to SHAM group presented cell necrosis and vacuolar degeneration, phenomenon slightly ameliorated in the treated groups, but without statistically significant differences. NRF2 is a transcription factor that regulates the cellular production of ROS and inflammation^54,55. In sepsis models of liver injury it plays a protective role by regulating local defence mechanisms⁵⁶. In our experiment, NRF2 increased in SEPSIS group compared to control and STS administration significantly augmented NRF2 levels for STS500 group. This effect was similar to previous reports in murine models of acute liver failure, where STS treatment increased NRF2 levels and reduced liver injury⁵⁷. Inducible NOS (NOS2) decreased in the SEPSIS group compared to control, without being influenced by any of the applied therapies. This provides evidence that the increased NOS measured in the plasma did not originate in the liver-tissue and was not influenced by the treatments. Additionally, liver SOD enzymatic activity that reflects the liver antioxidant pool were not improved by any of the treatment groups. STS administration therefore directly reduced liver damage in the lower dose, but it did not increase the liver antioxidant capacity.

Cell death mechanisms are a key part in sepsis-associated organ damage⁵⁰. Ferroptosis has been extensively studied in liver models of sepsis and it’s inhibition was reported to provide beneficial effects^24,58–60. Glutathione peroxidase (GPX4) is a ROS scavenging enzyme, which prevents the formation of lipid peroxides and is considered the key-enzyme in regulating ferroptosis⁶¹. Long-chain-fatty-acid—CoA ligase 4 (ACSL4) is involved in converting fatty acids into fatty acid acyl-CoA esters that act as substrates for lipid peroxidation in ferroptosis^61,62. In the current study, liver ACSL4 levels decreased in septic rats, but they were not significantly affected by the treatments. TRX had a stronger effect on lipid peroxidation, yet only slightly increased ACSL4 levels, without statistical significance. GPX4 was significantly reduced in liver samples of SEPSIS group compared to SHAM and this effect was reversed by the TRX and STS treatments. The effect on GPX4 levels is likely because of STS H₂S release and antioxidant properties⁶³. MDA is a marker of lipid peroxidation, which in the serum of rats treated with STS was reduced to a statistical degree only by the lower dose treatment. In the liver samples, MDA levels were statistically significant reduced by both STS doses. STS treatment therefore did not influence the levels of ACSL4 and the formation of fatty acid acyl-CoA esters and lipid peroxidation substrates. However, STS acts as a free radical scavenger and it reduced the levels of lipid peroxidation itself, likely via the increase in GPX4²¹. All this considered, STS did not inhibit already initiated ferroptosis but was able to reduce lipid peroxidation and oxidative damage.

Pyroptosis is an inflammatory cell death mechanism. Pyroptosis canonical activation involves the inflammasome-mediated cleavage of caspase-1 to c-casp-1 and the formation of cellular pores by GSDMD molecules with the release of pro-inflammatory cytokines IL-1β and IL-18⁶⁴. In the current study, GSDMD decreased slightly and c-casp-1 increased, but not statistically significant, suggesting that the pyroptosis-mediated cell death was involved to a lesser extent in our experimental model. Autophagy is a cell death mechanism amplified in sepsis, with Beclin1 and LC3 as key regulators^65–67. Their modulation leads to improved organ outcomes and reduced damage, suggesting a role of autophagy inhibition in sepsis^68,69. In our model the septic insult led to more pronounced activation of ferroptosis and pyroptosis in the liver compared to autophagy. This reduced cell-death involvement could be explained mostly by the experimental model. We have used a model that included supportive treatment with antibiotics and early fluid resuscitation, that reduced the overall organ damage by reducing the initial ischemia/reperfusion injury seen in models without fluid resuscitation or antibiotics. It is likely that if kept for more time, the animals would have developed liver injury and increased cell death markers.

There were minimal differences between the two doses of STS used in this study (500mg/kg and 1000mg/kg). Significant differences were seen in the plasma IL-6 and liver GPX4 levels. The higher STS dose significantly reduced IL-6 levels in plasma but decreased GPX4 levels. The histological analysis of the liver samples show less damage in the lower STS dose group. It is possible that while the higher dose of STS is more beneficial in reducing general oxidative stress and inflammation, it could cause tissue damage. These data are contrary to the current literature, where STS has been reported as increasing the liver antioxidant capacity²¹ and the microcirculatory oxygenation at a similar dose as used in our experiment (1000mg/kg)²². Current treatment strategies in humans use varying STS doses from 200 mg/kg to 12.5 g or higher doses to treat or prevent calciphylaxis, in cyanide poisoning and platinum-chemotherapeutics induced toxicity^70–72. Our doses would be an equivalent of 80 mg/kg and 161 mg/kg, so fairly low doses compared to the ones used in practice⁷³. It is likely that the septic ischemia/reperfusion injury was involved in reducing the tolerated STS levels, therefore we would advise when translating these results to human patients to begin with lower doses than those currently in use.

Inaccurate animal models are probably of the reasons why in the last years, although a great number of studies showed the benefit of several treatments in sepsis, none of them showed translational benefits. To address this problem our model, compared to most models in the literature, included both early fluid resuscitation and antibiotics administration. This likely provided a source of higher heterogeneity between subjects, as the initial septic insult was diminished, but not in a homogenous fashion, despite the rats having similar age, weight, gender and climatization conditions.

The current study has some limitations. Several parameters presented large standard deviations and results which influenced the analysis. While ferroptosis inhibition is beneficial in the first phases of sepsis, it is not clear if continuous inhibition does not lead to immunological dysfunction and immunosuppression in later phases. Also, since the patients that show this specific type of increased inflammatory response in sepsis would probably benefit more of the anti-inflammatory and antioxidant medication, future studies should be designed to stratify the results based on the baseline inflammatory response of the experimental animals to better mimic the clinical scenario.

Conclusion

In our experimental animal model, the bacterial peritonitis induced a septic reaction with increased reactive oxygen species production and consequently macromolecules oxidation and liver inflammation and ferroptosis. Sodium thiosulfate treatment improved the redox imbalance and provided an antioxidant effect, by reducing both systemic and liver oxidation, increased GPX4 levels and reduced ferroptosis, increased liver NRF2 levels and to a lesser extent reduced inflammation. The treatment did not significantly influence other mechanisms such as the formation of lipid peroxidation substrates, pyroptosis or autophagy. The results suggest that STS administration acted as an antioxidant at a systemic level and in liver tissue, without modulating already initiated ferroptosis. Future studies should focus to find the optimal dose and treatment duration of sodium thiosulfate that elicits beneficial effects and consider the treatment effects on survival and long-term outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

A.O.M. Conceptualization, Methodology, Investigation, Data curation, Formal analysis, Visualization, Writing—original draft, Writing—review&editing; I.B. Formal analysis, Supervision, Visualization, Writing—review&editing; G.A.F. Conceptualization, Project administration, Supervision, Writing—review& editing. R.M.P. Formal analysis, Visualization, Writing—original draft. M.A.N. Methodology, Supervision, Writing-original draft; B.M. Validation, Writing—Original draft, Writing- Reviewing and Editing; A.N. Formal analysis, Writing—review&editing; A.T.P. Conceptualization, Funding acquisition, Writing—review&editing; A.E.P. Conceptualization, Methodology, Resources, Project administration, Supervision, Writing-review&editing.

Funding

Adrian Press acknowledged the German Research Foundation (DFG), Grant ID: 852813223 for funding.

Data availability

The dataset analysed during the current study is available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.