Slow-Release Hydrogen Sulfide Reduces Nerve Cell Death After Spinal Cord Blood Flow Injury by Blocking Iron-Related Cell Damage

All male Sprague–Dawley (SD) rats (188–200 g weight) in this study were obtained from the Animal Laboratory of Beijing Charles River Corporation (Beijing, China). The rats were housed with water and commercial diet available ad libitum in an air‐conditioned room with constant temperature and humidity and a 12‐h light/dark cycle and were acclimated to their surroundings for 1 week prior to experiments. All animal operations and care procedures were approved by the Qingdao University Laboratory Animal Welfare Ethics Committee (No. 20230820SD20231212042).

All rats were neurologically intact before the experiment. The SCIRI model was built as we previously described [26]. In brief, the rats were anesthetized and fixed supine, and a midline abdominal incision was used. The abdominal aorta was clamped under the right renal artery near the heart using a 50 g aneurysm clip for 60 min (which was not clamped in sham‐operated rats). After the operation, the rat was placed in a box at 28°C to recover from anesthesia and subsequently placed in a separate cage with food and water available at will.

3‐Methyladenine (3‐MA; 189,490) and ferrostatin‐1 (SML0583) were purchased from Sigma‐Aldrich. GYY4137 (HY‐107632), rapamycin (AY‐22989), and erastin (HY‐15763) were purchased from MedChemExpress. GYY4137, 3‐MA, erastin, ferrostatin‐1 (Fer‐1), and rapamycin (Rapa) were dissolved in dimethyl sulfoxide to yield a stock solution and further diluted in phosphate‐buffered saline (PBS) for the final dose before intraperitoneal injection.

The Sham group underwent the surgical procedure without aortic clipping. The SCIRI group was modeled with SCIRI, and an equivalent amount of vehicle solution was injected intraperitoneally immediately after reperfusion. The rats in the SCIRI+Fer‐1 group and SCIRI+3‐MA group also received the same procedure as the SCIRI group, but were treated with Fer‐1 (2 mg/kg) [13] or 3‐MA (2.5 mg/kg) [24] immediately after reperfusion, respectively. The rats in the SCIRI+GYY4137 group underwent the same surgical procedure as the SCIRI group, but were treated with GYY4137 (50 mg/kg) 30 min before the onset of spinal cord reperfusion [27]. The rats in the SCIRI + GYY4137 + Rapa group and SCIRI + GYY4137 + Erastin group also received the same procedure as the SCIRI + GYY4137 group, but were treated with rapamycin (0.5 mg/kg) [24] or erastin (20 mg/kg) [28], 30 min before GYY4137 administration, respectively.

The Basso, Beattie, and Bresnahan (BBB) open‐field locomotor scale ranges from 0 (complete paralysis) to 21 (normal locomotion) and was used to measure locomotor recovery after SCIRI [29]. The BBB scores at 1, 6, 12, and 24 h after reperfusion were recorded by two experienced investigators who were blinded to the experimental design.

According to the manufacturer's instructions, the concentration of the lipid peroxidation products malondialdehyde (MDA), glutathione (GSH), and iron was measured in the spinal cord lysates using the corresponding kits, respectively. MDA content assay kit (bc0025), GSH content assay kit (bc1175), and tissue iron content assay kit (bc4355) were purchased from Solarbio.

The antibodies used for immunofluorescence and immunohistochemical are as follows: anti‐NeuN (1:400, 94,403) and anti‐LC3B (1:200, 83,506) were purchased from Cell Signaling Technology. anti‐NeuN (1:200, 26975‐1‐AP) and anti‐GPx4 (1:200, 30388‐1‐AP) were purchased from Proteintech. anti‐ferritin (1:100, ab75973) and anti‐TFRC (1:1000, ab269513) were purchased from Abcam. Alexa Fluor 488‐labeled goat anti‐rabbit IgG (H + L) (1:300, A0423), Alexa Fluor 488‐labeled goat anti‐mouse IgG (H + L) (1:300, A0428), Alexa Fluor 555‐labeled donkey anti‐rabbit IgG (H + L) (1:300, A0453), and Alexa Fluor 555‐labeled donkey anti‐mouse IgG (H + L) (1:300, A0460) were purchased from Beyotime. The antibodies used for Western blotting are as follows: anti‐GPx4 (1:500, A11243) and anti‐SLC7A11/xCT (1:500, A13685) were purchased from ABclonal. anti‐ferritin (1:1000, ab75973), anti‐NCOA4 (1:1000, ab314553), anti‐SQSTM1/p62 (1:10000, ab109012), and anti‐LC3B (1:1000, ab192890) were purchased from Abcam. anti‐β‐actin (1:5000, BM0627), HRP Conjugated AffiniPure goat anti‐mouse IgG (H + L) (1:5000, BA1050), and HRP Conjugated AffiniPure goat anti‐rabbit IgG (H + L) (1:5000, BA1054) were purchased from Boster.

The protocol for immunofluorescence staining was based on our previously described method [24]. In brief, the spinal cord frozen sections were washed with PBS, permeabilized with 1% Triton X‐100, blocked with 5% bovine serum albumin (BSA), incubated with primary antibody, and then incubated with the corresponding secondary antibodies conjugated with Alexa Fluor‐labeled. Finally, the sections were mounted with ProLong Gold antifade reagent with DAPI to label the nuclei (Invitrogen, P36935↗). Traced sections were examined with a Zeiss microscope (Axioscope 5). Images and their colocalization or fluorescence intensity were quantified using ZEN Lite Viewer and ImageJ software.

The staining protocol was based on our previous report [30]. Briefly, paraffin sections were deparaffinized, rehydrated, and underwent antigen retrieval. Then the sections were incubated in hydrogen peroxide, blocked, incubated with primary antibody, and then incubated with the corresponding SABC‐HRP kit (Beyotime, P0612/P0615). Positive staining was visualized with DAB (Beyotime, P0202) and counterstained with hematoxylin. Traced sections were examined with a Nikon microscope (Ni‐U). Images and mean integrated optical density were quantified using NIS‐Elements Viewer and ImageJ.

The staining protocol was based on the instructions of the manufacturer of Nissl staining solution (Beyotime, C0117). In brief, paraffin sections were deparaffinized, rehydrated, and incubated in Nissl staining solution. Then, sections were dehydrated, transparent, and mounted. Traced sections were examined with a Nikon microscope (Ni‐U). Images and amounts were quantified using NIS‐Elements Viewer and ImageJ.

Dihydroethidium (DHE) staining was used to assess the level of oxidative stress as previously reported [31]. Briefly, the frozen spinal cord sections were incubated with the fluorescent dye DHE (Beyotime, S0063). Traced sections were examined with a Zeiss microscope. Images and fluorescence intensity were quantified using ZEN Lite Viewer and ImageJ software.

According to the manufacturer's instructions of the TUNEL kit (Roche, 12156792910) to detect apoptotic cells. In brief, the frozen sections were washed in PBS, permeabilized, blocked, and incubated with the TUNEL reaction mixture. Traced sections were examined with a Zeiss microscope. Immunofluorescence images and their colocalization were quantified using ZEN Lite Viewer and ImageJ software.

The protocol was based on previously described [27]. Briefly, the spinal cords were treated with ice‐cold RIPA (Beyotime, P0013C) containing a protease and phosphatase inhibitor cocktail (Beyotime, P1050), and the concentration of protein was measured with the BCA assay kit (Beyotime, P0012). Aliquots of protein were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride filter membrane (0.2 μm, Millipore, ISEQ00010). Subsequently, the membranes were blocked, incubated with the primary antibody, and incubated with the corresponding secondary antibody. Finally, the chemiluminescence results were recorded using an imaging system (FUSION SOLO S, VILBER). The bands relative intensities were calculated by ImageJ software.

All data were statistically analyzed using GraphPad Prism 9. The data are presented as the mean ± standard deviation (mean ± SD) of at least three independent experiments in this study. The normality of the data distribution was evaluated using the Shapiro–Wilk test. If the data were normally distributed, Student's t‐test or one‐way analysis of variance (ANOVA) was used to assess group differences. For data that were not normally distributed, a non‐parametric test (Kruskal‐Wallis test) was applied. Two‐way ANOVA and Tukey's post hoc test were employed to evaluate differences among multiple groups with repeated measures. For all statistical tests, a p‐value < 0.05 was considered statistically significant.

To verify the existence of neuron damage in a rat model of SCIRI (Figure 1a), we performed Nissl staining to evaluate changes in neuronal viability, a TUNEL assay to evaluate the viability of cells around the anterior horn, and the motor function was assessed by the BBB score.

The results of Nissl staining revealed that the number of Nissl bodies in the anterior horn of the spinal cord was significantly decreased in the SCIRI group compared to the Sham group (Figure 1b,c). In addition, the numbers of TUNEL‐positive cells were obviously increased in the SCIRI group compared to the Sham group (Figure 1d,e). Moreover, BBB scores demonstrated that motor function was largely decreased after reperfusion (Figure 1f). In summary, we generated a SCIRI rat model that was characterized by neurological dysfunction, which was consistent with increased cell death and loss of motor neurons.

Substantial evidence revealed the crucial role of ferroptotic machinery in driving cell death [32, 33]. Our previous study showed that SCIRI could induce lipid peroxidation, which is one of the key events of ferroptosis. Herein, we assumed that ferroptosis might contribute to neuron death in SCIRI. To verify this assumption, we assessed the concentrations of GSH, MDA, and Fe²⁺, and these are typical indicators for ferroptosis. As expected, the results demonstrated that the higher concentrations of MDA and Fe²⁺, and the lower GSH in the SCIRI group compared to the Sham group. This means that ferroptosis events, depletion of GSH, peroxidation of lipid, and accumulation of Fe²⁺, were indeed triggered by SCIRI (Figure 2a). The dihydroethidium (DHE) staining results also proved this (Figure 2d,e). Moreover, we conducted Western blotting for the expression of GPx4 and solute carrier family 7 member 11 (Slc7a11), both considered critical for anti‐ferroptosis (Figure 2b). The results showed that the expression of Slc7a11 and GPx4 was obviously decreased in the SCIRI group (Figure 2c). Together, these findings suggested that ferroptosis was activated by SCIRI.

To further clarify whether ferroptosis was required for neuron death, we inhibited ferroptosis by Ferrostatin‐1 (Fer‐1, a lipid ROS scavenger), which is an inhibitor of ferroptosis. The results showed that ferroptosis events were reversed by Fer‐1 (Figure 2f). Moreover, the expression of Slc7a11 and GPx4 was obviously increased in SCIRI rats with Fer‐1 treatment compared to the SCIRI group (Figure 2g,h). The GPx4 immunohistochemistry staining results also proved this (Figure 2i,j). Besides, the number of Nissl bodies and BBB scores were significantly increased with Fer‐1 treatment compared to the SCIRI group (Figure 2k–m), and these were consistent with the results of TUNEL staining (Figure S1). Taken together, our findings suggested that SCIRI initiated ferroptosis, which might be crucial for inducing neuronal death.

Our previous study reported that SCIRI could activate autophagy, which has been identified to play a key role in driving cells toward ferroptotic death [30, 34, 35]. We conducted Western blotting for the expression of LC3 and p62, which are two typical markers of autophagy (Figure 3a). The results showed that the SCIRI group decreased the expression of p62 and upregulated the protein ratio of LC3‐II/I compared to the sham group (Figure 3b), and these were consistent with the results of immunofluorescence staining of LC3 (Figure 3c,d). Together, these findings suggested that autophagy was activated in SCIRI rats.

To elucidate whether autophagy was necessary for neuron death, the autophagy inhibitor 3‐methyladenine (3‐MA) was further used. Western blotting analysis showed that treatment with 3‐MA reduced the LC3‐II/I ratio and increased the p62 protein level compared to the SCIRI group (Figure 3e,f). We also performed immunofluorescence colocalization of LC3 with NeuN to investigate LC3 expression and localization. The results indicated that autophagy occurred in neurons following SCIRI and was inhibited by 3‐MA treatment (Figure 3g,h). Additionally, the number of TUNEL‐positive cells significantly decreased in the SCIRI rats with 3‐MA treatment (Figure 3i,j). Altogether, these results suggested that SCIRI activated autophagy, which was critical for inducing neuronal death.

It has been demonstrated that NCOA4 selectively autophagically degrades ferritin, the major protein for intracellular iron storage, thereby accumulating labile iron and promoting ferroptosis [13, 36]. So, we focus on NCOA4‐mediated ferritinophagy in SCIRI. we conducted Western blotting for the expression of ferritin and NCOA4 (Figure 4a), which are vital proteins for NCOA4‐mediated ferritinophagy. The results showed that the SCIRI group had decreased the expression of ferritin and had increased the expression of NCOA4 compared to the sham group (Figure 4b). In addition, the ferritin immunofluorescence staining was consistent with the above results (Figure 4c,d).

To further verify that ferritinophagy was involved in SCIRI‐induced cell death, we inhibited autophagy with 3‐MA. Western blotting results showed that 3‐MA treatment increased the expression of ferritin and decreased the expression of NCOA4 compared to the SCIRI group (Figure 4e,f). Furthermore, the immunofluorescence colocalization of NeuN and ferritin supported the above findings (Figure 4g). It showed that with 3‐MA treatment, the ferritin expression was increased in neurons compared to the SCIRI group (Figure 4h). Collectively, these results imply that NCOA4‐mediated autophagic‐dependent ferroptosis might be the pivotal process for SCIRI‐induced neuron death.

We previously reported that H₂S could inhibit autophagy and reduce oxidative stress in SCIRI rats [24], and these are key events of ferritinophagy and ferroptosis [17, 34]. In addition, we found that GYY4137, a sustained release H₂S donor, not only regulated autophagy but also ferroptosis (Figures S2, S3). Consequently, it is reasonable to hypothesize that the therapeutic effect of H₂S in SCIRI might be associated with inhibited ferritinophagy and ferroptosis. To test our hypothesis (Figure 5a), we first assessed the concentrations of GSH, MDA, and Fe²⁺. The results showed that depletion of GSH, peroxidation of lipids, and accumulation of Fe²⁺ were reversed with GYY4137 treatment compared to SCIRI, whereas these effects were abolished by the combination of GYY4137 with rapamycin (autophagy activator) or erastin (ferroptosis inducer) (Figure 5b). The DHE staining results also supported these findings (Figure 5c and Figure S4a). Moreover, we performed Western blotting to assess the degree of ferroptosis. Compared to the SCIRI group, GYY4137 treatment significantly reduced the expression of GPx4 and slc7a11, whereas GYY4137 combined with rapamycin or erastin had no such effect (Figure S3). These were consistent with the results of immunohistochemistry staining GPx4 and TFRC (Figure 5d and Figure S4b).

Then, the expression of Ferritin and NCOA4 proteins was reversed in the GYY4137 treatment group compared to the SCIRI group, but these were not different in GYY4137 combined with rapamycin or erastin compared to the SCIRI group. (Figure 5e,f). Additionally, the LC3 and Ferritin immunofluorescence colocalization staining was used to detect ferritinophagy (Figure 5g). As increased LC3 expression occurred, Ferritin expression decreased in the SCIRI group compared to the sham group, and these changes could be reversed with GYY4137 treatment (Figure S4c). Rapamycin or erastin could abolish the above effect of GYY4137 (Figure S4c).

Furthermore, we performed the TUNEL and NeuN immunofluorescence colocalization staining (Figure 6a). The results showed that TUNEL‐positive neurons were significantly reduced after GYY4137 treatment compared to the SCIRI rats, and there was no difference in GYY4137 combined with rapamycin or erastin compared to the SCIRI rats (Figure 6b). Meanwhile, we further tested Nissl staining and BBB scores (Figure 6c,e). As expected, GYY4137 significantly restored lower limb motor function and ameliorated the loss of neurons, but these effects disappeared when it was combined with rapamycin or erastin (Figure 6d,e). More generally, our results suggested that hydrogen sulfide attenuated neuronal death by inhibiting SCIRI‐induced ferritinophagy and ferroptosis.