Sodium formate-induced mitochondrial impairment and cytotoxicity in neuronal cells reveal crucial pathogenic mechanisms underlying diabetic neuropathy and retinopathy

Abstract

Formic acid, a toxic one-carbon metabolite formed from methanol and formaldehyde in the body, can cause neuronal dysfunctions. It was recently hypothesized that metabolic formation of toxic one-carbon metabolites (particularly formic acid) under hyperglycemic conditions may contribute to the pathogenesis of diabetic complications in humans. The present study aims to investigate the mechanism of formic acid-induced neurotoxicity using immortalized HT22 mouse hippocampal neurons as an in-vitro model. We found that treatment of cells with sodium formate (SF, a salt form of formic acid) causes a concentration-dependent loss of cell viability (based on MTT assay), whereas the cell number is reduced to a lesser degree when SF is present at lower concentrations. In addition, SF at the lower concentrations decreases cell proliferation by suppressing DNA synthesis, but at higher concentrations, SF induces cell death through apoptosis. SF can preferentially cause accumulation of mitochondrial ROS, disruption of mitochondrial structure, and suppression of mitochondrial functions (including ATP production). SF-induced mitochondrial ROS accumulation subsequently leads to the depletion of cellular glutathione, along with the buildup of cellular ROS and lipid-ROS. These changes jointly lead to increased permeability of both cytoplasmic and mitochondrial membranes, and ultimately the induction of apoptotic cell death. Analysis of the cellular transcriptomics revealed that the expression of genes for the relevant enzymes and proteins involved in mitochondrial function, energy metabolism and other cellular processes is altered in SF-treated cells. These findings highlight mitochondria as a crucial target in mediating SF-induced cytotoxicity, and also shed mechanistic lights on how formic acid accumulation may contribute to the pathogenesis of diabetic neuropathy and other diabetic complications.

Introduction

Formic acid, a clear and colorless liquid with a pungent odor, was first isolated from certain ants and thus was named after the Latin formica (meaning ants). Earlier studies have shown that formic acid is cytotoxic both in vitro and in vivo^1,2. In humans and animals, formic acid is a toxic one-carbon metabolite usually formed from methanol and formaldehyde³. In the literature, the toxicity of formic acid was mostly investigated in connection with methanol studies; in fact, there were very few studies specifically studying the toxic effects of formic acid itself^4,5,6,7. In methanol poisoning, formic acid accumulation in the body is largely responsible for the acute toxic effects such as metabolic acidosis, serious visual impairment (including blindness), damage to vital organs in the body (such as brain, kidney and heart), and even death^2,4.

Different from formaldehyde (a common metabolic precursor of formic acid), formic acid can readily accumulate in humans and non-human primates to far higher levels than formaldehyde following administration of methanol^4,5,6,7,8. For instance, an earlier study reported that when formate (the salt form of formic acid) was infused into monkeys, its accumulation in blood lasted 10 h, and after that, its blood levels were still at 10–30 mM^5,9. In another study, when the plasma pharmacokinetics of formic acid in monkeys, mice and rabbits were compared, it was found that formic acid accumulation in primates within the first 6 h was 43- and fivefold higher compared to its accumulation in mice and rabbits, respectively¹⁰.

It was recently hypothesized that the pathogenesis of some diabetic complications is attributable to the increased metabolic formation of the toxic one-carbon metabolites, namely, methanol, formaldehyde and formic acid, inside the human body, particularly in those tissues and cells that have elevated levels of cellular glucose, and these toxic one-carbon metabolites then drive the pathogenic processes associated with certain human diabetic complications. An important element of this hypothesis is that hyperglycemia-associated increase in cellular glucose levels in diabetic patients would lead to increased metabolic formation of endogenous methanol and/or formaldehyde, both of which are further metabolically converted to formic acid¹¹. Although the exact metabolic pathways as well as the enzymes involved are not fully clear at present, there is clear evidence showing that methanol, formaldehyde and formic acid are endogenously-formed compounds and present in humans at readily detectable levels, and that elevated levels of these toxic one-carbon metabolites are found in diabetic patients (reviewed in ref.¹¹).

It was previously reported that the mitochondrial cytochrome oxidase is a target of formic acid cytotoxicity^1,3,12. It was suggested that formic acid can inhibit the activity of cytochrome oxidase by binding at the sixth coordination position of its heme ferric iron¹³ resulting in inhibition of the cytochrome oxidase complex of the mitochondrial respiratory chain^1,3,13. It is expected that this inhibition by formic acid would reduce mitochondrial ATP synthesis^1,3,14, which may lead to increased anerobic glycolysis and potentially lactic acid accumulation. Additionally, inhibition of the mitochondrial respiratory chain by formic acid will also cause mitochondrial superoxide accumulation, which will result in lipid peroxidation, mitochondrial damage and even oxidative cell injury^15,16,17. Offering partial support for the role of ROS in methanol cytotoxicity, elevated lipid peroxidation products were found to be present in different tissues and urine of methanol-intoxicated animal models^18,19.

The purpose of this study is two-fold: (1) examine the cytotoxic effects of sodium formate (SF), a salt form of formic acid, using the cultured HT22 hippocampal neuronal cells as an in-vitro model, and (2) determine the mechanism by which SF elicits its cytotoxicity in this in-vitro model. We found that SF can cause neurotoxicity mainly through impairment of mitochondrial function and inhibition of cellular ATP synthesis. These observations may partially explain the pathogenic mechanisms by which accumulation of formic acid resulting from hyperglycemia in diabetic patients may contribute to the induction of diabetic neuropathy and retinopathy in humans.

Materials and methods

Chemicals and reagents

Sodium formate (SF; purity ≥ 99.0%, HPLC analysis) was purchased from Shanghai Aladdin Biochemical Technology (Shanghai, China). Dulbecco’s Modified Eagle’s Medium (DMEM, #12,800,017) and fetal bovine serum (FBS, # A5670701) were purchased from GIBCO (GIBCO, Paisley, UK). 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, #S6821), Ferrostatin-1 (Fer-1, #S7243), necrostatin-1 (Nec-1, #S8037), z-VAD(OMe)-FMK (z-VAD-FMK, #S7023) and propidium iodide (PI, ≥ 99.7%, HPLC analysis, #S6874) were purchased from Selleck Chemicals (Houston, TX, USA). Crystal violet staining solution (#C0121), 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA; #S0033M), BODIPY-581/591-C11 (#D3861), 3-amino,4-aminomethyl-2',7'-difluorescein diacetate (DAF-FM-DA; #S0019M), JC-1 working solution (#C2003S), ATP Assay Kit (#S0026), Mitochondrial Isolation Kit (#C3601), Calcein-AM/PI Cell Viability and Cytotoxicity Assay Kit (#C2015M) and Annexin V-FITC Apoptosis Detection Kit (#C1062M) were purchased from Beyotime Biotechnology (Shanghai, China). MitoSOX (#M36008), MitoTracker Deep Red-FM (#M22426), Click-iT EdU Alexa Fluor-488 Flow Cytometry Analysis Kit (#C10425) and MitoProbe Transition Pore Detection Kit (#34153) were purchased from ThermoFisher (Waltham, MA, USA). MitoTracker green (#9074s) was purchased from Cell Signaling Technology (Danvers, MA, USA). Succinate Dehydrogenase Activity Assay Kit (#BC0955), Reduced Glutathione Content Detection Kit (#BC1170) and CCK-8 Cell Proliferation and Cytotoxicity Assay Kit (#CA1210) were purchased from Beijing SolarBio Science and Technology (Beijing, China). Agilent Seahorse XFp Real-Time ATP Rate Assay Kit (#103591-100) and Agilent Seahorse XF Cell Mitochondrial Stress Test Kit (#103015-100) were purchased from Agilent Technologies (Santa Clara, CA, USA). Antibodies against succinate dehydrogenase (SDH) subunit A (#ab14715), PPARG coactivator-1α (PGC-1α, #ab191838), nuclear respiratory factor 1 (NRF1, #ab34682) and transcription factor A, mitochondrial (TFAM, #ab176558) were purchased from Abclonal (Wuhan, China). Antibodies against β-actin (#3700S) was purchased from Cell Signaling Technology (Danvers, MA, USA). The horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG and rabbit anti-mouse IgG were purchased from ProteinTech (Wuhan, China).

Cell culture and cell viability assay

The immortalized HT22 mouse hippocampal neuronal cells were originally subcloned from the HT-4 cell line²⁰. The HT22 cells were purchased from the Cell Bank of Shanghai Institute of Biological Science (Shanghai, China) and cultured in complete DMEM, supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. Cell culture dishes were kept at the 37 °C atmosphere containing 5% CO₂. Cells were passaged after reaching 60–80% confluence. The cells used in experiments were usually limited to 20 passages, and were authenticated by short tandem repeat profiling and routinely tested for mycoplasma contamination.

For shorter time (24 or 48 h) culture, HT22 cells were seeded in 96-well plates at a density of 2000 or 1500 cells/well, respectively, and then treated with chemicals as indicated. For longer time (72, 96 or 120 h) culture, HT22 cells were seeded in 24-well plates at a density of 5000 cells/well and treated with different chemicals. At the end of the drug treatment, MTT and CCK-8 assay were used to determine the change in gross cell viability. The 0.5 mg/mL MTT solution (100 μL for 96-well plates and 300 μL for 24-well plates) or CCK-8 solution (10 μL for 96-well plates and 30 μL for 24-well plates) was added to each well and the cells were incubated for 2–2.5 h at 37 °C under 5% CO₂. After incubation, the medium was removed and dimethylsulfoxide (DMSO, 100 μL for 96-well plates and 300 μL for 24-well plates) was added to each well to dissolve the MTT formazan. The absorbance was measured with a microplate reader (BioTek, Winooski, VT, USA) at 560 nm. The relative cell viability was compared with the vehicle-treated control cells.

Morphological analysis, manual cell counting and crystal violet staining

For morphological analysis, the HT22 cells were seeded and treated the same way as described above for MTT assay. Gross morphological change of the cells in different treatment groups was visualized with a Nikon Eclipse Ti-U inverted microscope (Nikon, Tokyo, Japan). For manual counting of cell numbers, the cells were seeded in 6-well plates at a density of 5 × 10⁴ cells/well and treated with different concentrations of SF for 24, 48 or 72 h, and after that, cell suspension was prepared for manual counting. For crystal violet staining, the cells were first fixed with 1% glutaraldehyde for 15 min and then stained with 50 μL of 0.5% (w/v) crystal violet solution (dissolved in 20% methanol and 80% deionized water) for 15 min at room temperature. Then images were visualized and captured with a Nikon Eclipse Ti-U inverted microscope. If quantitative testing was required, 200-proof pure ethanol (100 μL for 96-well plates and 300 μL for 24-well plates) was added to each well overnight to dissolve the crystal violet dye contained in cultured cells. The absorbance was measured using a microplate reader (BioTek, Winooski, VT, USA) at 405 and 570 nm. The relative number of cells was compared to the vehicle-treated control group.

Detection of cellular NO and ROS and mitochondrial ROS by fluorescence microscopy

The HT22 cells were plated in 24-well places at a density of 5 ⨯ 10⁴ per well and treated with different drugs for desired durations. For staining, cells were washed twice with HBSS and then incubated with DAF-FM-DA (5 μM), DCFH-DA (5 μM) or MitoSOX (5 μM) dissolved in 200 μL DMEM (free of serum and phenol red) for 20 min at 37 °C. Following three washes with HBSS, fluorescence images were captured with a Nikon Eclipse Ti-U inverted microscope, and the images were analyzed with the NIS-Elements software (Nikon).

Measurement of ROS, lipid-ROS and NO and mitochondrial density by flow cytometry

The HT22 cells were seeded in 6-well plates at a density of 1.5 × 10⁵ cells/well for 24 h before treatment with different drugs. Following drug treatment, cells were trypsinized, collected and suspended in phosphate-buffered saline (PBS). Cells were then centrifuged, and the resulting cell pellets were resuspended in DMEM (free of serum and phenol red) containing DCFH-DA (5 μM), BODIPY-581/591-C11 (5 μM), DAF-FM-DA (5 μM), MitoSOX (5 μM) or MitoTracker deep red (2 μM). After a 20-min incubation at 37 °C, the cells were washed three times with HBSS. Cellular and mitochondrial ROS, lipid-ROS and cellular NO levels and mitochondrial density were measured using a CytoFlex flow cytometer (Beckman Coulter, Brea, CA, USA), and the data were analyzed using the FlowJo software (v. 10.3; FlowJo, LLC, Ashland, USA).

Analysis of mitochondrial membrane potential (MMP)

The HT22 cells were seeded in 6-well plates at a density of 1.5 × 10⁵ cells/well for 24 h (the number of seeded cells varies depending on the duration of treatment). Following drug treatment, the cells were stained with the JC-1 working solution (10 μg/mL) and incubated for 20 min at 37 °C under 5% CO₂. The JC-1-treated cells were washed two times with the JC-1 binding buffer. Imaging analysis was performed with a LSM-900 confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany) at 490 nm (excitation) and 530 nm (emission) for JC-1 monomers and 525 nm (excitation) and 590 nm (emission) for JC-1 aggregates. The JC-1 fluorescence was also quantified using analytical flow cytometry, in which the red JC-1 aggregates were measured at the FL2 channel and the green JC-1 monomers were measured at the FL1 channel. A minimum of 10,000 cells were analyzed for each sample with the CytoExpert software (Beckman Coulter).

Apoptosis detection assay

The HT22 cells were usually seeded in 6-well plates at a density of 1.5 × 10⁵ cells/well for 24 h (the number of seeded cells varies depending on the required duration of drug treatment). Following treatment, the cells were collected by centrifugation at 1500 g for 5 min and washed three times with PBS, followed by detection of apoptotic cells using the Annexin V-FITC Apoptosis Detection Kit (Beyotime, #C1062M). Briefly, 195 μL annexin V-FITC binding solution was added to resuspend cells gently. Then 5 μL annexin V-FITC and 10 μL PI were added, followed by gentle mixing, incubation at room temperatures (20–25 °C) in the dark for 10–20 min. Then the samples were kept on ice for flow cytometry analysis (usually completed within 1 h). The gating strategy included: the annexin V-FITC single positive cells (Q3) are considered early apoptotic cells, and the annexin V-FITC and PI double positive cells (Q2) are late apoptotic/necrotic cells. The sum of Q3 and Q2 represent the percentage of total apoptotic cells. The flow cytometry assays were performed using a CytoFlex-S flow cytometer (Beckman Coulter) and the data were analyzed using the CytoExpert software (Beckman Coulter).

Measurement of mitochondrial permeability transition pore (mPTP) openness

The HT22 cells were seeded in 6-well plates at a density of 1.5 × 10⁵ cells/well for 24 h (the number of seeded cells varies depending on the duration of treatment). Following drug treatment, the cells were collected by centrifugation at 1500 g for 5 min and washed three times with PBS. Using the mPTP Assay Kit (ThermoFisher, USA), cells were stained with Calcein-AM and CoCl₂ for 40 min, and then were collected by centrifugation and resuspended in 300 μL HBSS. The fluorescence intensity of each sample was measured using a CytoFlex-S flow cytometer (Beckman Coulter) and the data were analyzed using the FlowJo software (v. 10.3; Ashland, USA).

Confocal microscopy

The HT22 cells were seeded on coverslips placed inside the 24-well plates (the number of seeded cells varies depending on the duration of treatment). Twenty-four h later, cells were treated with different drugs as indicated. Coverslips were washed in HBSS and incubated in HBSS containing BODIPY-581/591-C11 (5 μM), MitoSOX (5 μM), MitoTracker green (0.5 μM) or JC-1 working solution (10 μg/mL) for 20 min at 37 °C under 5% CO₂. Coverslips were then mounted on microscope slides for visualization. Slides were imaged using a LSM-900 confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany), and images were analyzed with the Zen software (Carl Zeiss).

Measurement of cellular and mitochondrial succinate dehydrogenase (SDH) activity

The HT22 cells were seeded in cell culture dishes, with the seeding density adjusted according to the intended treatment duration. After drug treatment for 24–72 h as indicated, the whole-cell lysates and mitochondrial-enriched lysates were separately prepared according to the manufacturer’s protocols. The lysates were then centrifuged at 11,000 × g for 10 min at 4 °C to collect the supernatants. The SDH activity in supernatants was analyzed using the SDH Activity Assay Kit (Beijing SolarBio Science and Technology, #BC0955). The absorbance at 600 nm was measured using a microplate reader (BioTek, Winooski, VT, USA). The activity was normalized and presented in the unit as "nmol of 2,6-dichlorophenol indophenol per mg protein per min". Liver lysates were prepared from mouse liver tissue (approximately 1 g wet weight) by following the instructions of the SDH Activity Assay Kit, and then the supernatants at a final protein concentration of 2.5 mg/mL were used for assaying the SDH activity in vitro with or without the presence of SF.

Measurement of cellular ATP levels

The cells were seeded in a cell culture dish (the number of seeded cells varies depending on the duration of treatment), and after 24‒96 h treatment, cells received drug treatments as indicated. The ATP Assay Kit (#S0026) from Beyotime Biotechnology (Shanghai, China) was used to measure cellular ATP levels. Specifically, after removal of cell culture medium, the cells were washed with PBS, and the lysis buffer (300 μL) was added to each well to lyse the cells. The cells were then centrifuged at 10,000 rpm at 4 °C for 5 min, and the supernatants were collected for measurement of ATP levels according to manufacturer’s instructions with a PerkinElmer EnVision (#2105-0010) multimode plate reader (Waltham, MA, United States). The ATP level (μmol/mg of protein) was normalized according to the total protein level in the corresponding cell lysate preparation.

Measurement of cellular reduced GSH levels

The reduced GSH levels in cells were determined using the Reduced GSH Assay Kit (S0052, Beyotime, China) following the manufacturer’s instructions. After SF treatment for 24–72 h, cell lysates were prepared and reacted with the assay solution for 2 min at 25 °C. The absorbance at 412 nm was measured on microplate reader (BioTek, Winooski, VT, USA). The cellular GSH levels were calculated and adjusted based on the protein content of the cell lysate.

Measurement of cell cycle and cell proliferation

Cell proliferation was assayed using the Click-iT EdU Imaging Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The HT22 cells were seeded in a cell culture dish, and then treated with SF for 24‒72 h. The attached cells were trypsinized, washed once with PBS, and then incubated with 10 μM EdU for 4 h at 37 °C. Following incubation, cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.5% Triton X-100 for 20 min. EdU incorporation was detected following a click chemistry reaction.

To measure the cell cycle, the cells received the same drug treatment as above, and then the attached cells were washed once with PBS, and digested with EDTA-free trypsin and collected by centrifugation at 2000 rpm for 5 min in an Eppendorf centrifuge (Germany). Cells were fixed with 70% ice-cold ethanol at 4 °C for 2 h, washed with PBS and stained with 10 μL PI (1 mg/mL) to assess the DNA content. Flow cytometry analysis was performed using a CytoFlex-S flow cytometer (Beckman Coulter). The cell proliferation data were analyzed using the CytoExpert software (Beckman Coulter), and the cell cycle distribution analysis was carried out using the FlowJo software (v.10.3 FlowJo, LLC, Ashland, USA).

Measurement of real-time ATP generation rate and cellular respiration

The Agilent Seahorse XFp Real-Time ATP Rate Assay Kit (#103591-100) and XF Cell Mitochondrial Stress Test Kit (#103015-100) were used to assess real-time ATP generation and mitochondrial respiratory capacity in HT22 cells. XF96 cell culture microplates were pre-coated with CellTak according to the manufacturer’s instructions. After 48-h treatment with SF (at different concentrations for ATP rate analysis and at 75 mM for respiratory capacity), cells were harvested, resuspended in XF assay medium containing glucose (5.5 mM or 10 mM), 2 mM L-glutamine, and 1 mM sodium pyruvate, and seeded at 5 × 10³ or 2 × 10³ cells per well, respectively. Following a 30-min calibration in a CO₂-free incubator at 37 °C, basal oxygen consumption rate (OCR) was recorded using the XF96 analyzer. For ATP rate analysis, 1.5 µM oligomycin (Port A) and 0.5 µM rotenone/antimycin A (Port B) were sequentially added. For respiratory capacity assessment, 1.5 µM oligomycin (Port A), 0.25 µM FCCP (Port B), and 0.5 µM rotenone/antimycin A (Port C) were used. Real-time ATP production (in units of pmol/min) and mitochondrial function were calculated based on the resulting OCR profiles.

Western blotting analysis

After treatment of HT22 cells with SF, cells were harvested with trypsin–EDTA and washed with PBS, and then suspended in 100 μL lysis buffer (containing 20 mM Tris–HCl, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and a protease inhibitor cocktail, pH 7.5). Protein levels were determined using the Bio-Rad protein assay (Bio-Rad, Hercules, CA). An equal amount of proteins was loaded in each lane, separated by 10% SDS–polyacrylamide gel electrophoresis (SDS-PAGE), and electrically transferred to a polyvinylidene difluoride membrane (Bio-Rad). After the membrane was blocked using 5% skim milk, target proteins were immunodetected using specific antibodies. Primary antibodies, i.e., the antibodies against SDH subunit A, PGC-1α, NRF1, TFAM or β-actin, were diluted at 1:3000 in TBST. Thereafter, the HRP-conjugated anti-rabbit or anti-mouse IgGs were applied as the secondary antibody (diluted at 1:5000 in TBST) and the membranes were developed using SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher, Waltham, MA, USA). The band intensity of the Western blot images was quantified with the ImageJ Software.

Transmission electron microscopy

Transmission electron microscopy (TEM) analysis of cells was carried out as described earlier²¹. The cell samples were immersion-fixed with 2.5% glutaraldehyde solution at 4 °C for 2 h. Thereafter, the ultrathin Sects. (60‒80 nm) were cut and stained with uranyl acetate and lead citrate. The images were obtained with a transmission electron microscope (Hitachi HT7700 120 kV; Sevier Bio, Wuhan, Hubei, China).

Transcriptomics analysis

The HT22 cells were plated at 2.5 × 10⁶ per 60-mm culture dish. After treatment with 12.5 and 25 mM SF for 48 h, cells were harvested with trypsin–EDTA and washed with PBS. The cell pellets were snap-frozen in liquid nitrogen and then stored at − 80 °C. Each treatment group had three replicates, and the same experiments (including all analyses) were repeated twice. Similar results were obtained (one representative data set from the two separate experiments were presented). The RNA-seq and transcriptomic data processing were completed by BGI (Shenzhen, China). Briefly, the total RNAs were isolated using TRIzol (Invitrogen, Waltham, MA, USA), and RNAs were purified using the Ribo-Zero rRNA removal kit (Illumina, San Diego, CA, USA) to remove rRNAs while keeping the other RNAs. The concentration of RNAs was measured using the Qubit RNA Assay Kit in a Qubit-3.0 fluorimeter (Life Technologies, Carlsbad, CA, USA). Only high-quality RNA samples were used to construct libraries for RNA-seq and utilized for further sequencing analyses. The KEGG database used in this study is covered by copyright permission, and authorization has been obtained for the use of the relevant pathways.

Statistical analysis

Most of the experiments described in this study were repeated three times or more to confirm the observations. Data presented in this study are the mean ± S.D. from multiple replicate measurements taken from one representative experiment. Statistical analysis was performed with GraphPad Prism 9.0 software (GraphPad, San Diego, CA, USA). For multiple comparison analysis, one-way ANOVA followed by Tukey’s multiple comparison tests was performed. Statistical significance was denoted by P < 0.05 (* or ^#) and P < 0.01 (** or ^##) for significant and very significant differences, respectively. In most cases, * and ** denote the comparison for statistical significance between the control group (cells treated with vehicle only) and the cells treated with SF, whereas ^# and ^## denote the comparison between the cells treated with SF and the cells jointly treated with SF and another compound.

Results

SF induces cytotoxicity in HT22 cells

The HT22 mouse hippocampal neuronal cells were treated with increasing concentrations (up to 100 mM) of SF for different durations. Based on MTT assay, SF elicited concentration- and time-dependent reductions in cell viability in HT22 cells (Fig. 1A‒F). Following 24-h exposure to SF, the cell viability was reduced in a concentration-dependent manner, and the apparent IC₅₀ was approximately 100 mM (Fig. 1A). When the treatment time was increased to 48, 72 and 96 h, SF produced a time-dependent cumulative reduction in cell viability (Fig. 1B‒D). In a separate experiment when the same concentrations of SF (such as 25 and 50 mM) were compared, the decrease in cell viability was cumulative when the exposure time was increased from 48 to 120 h (Fig. 1E, F). The results of the CCK-8 assay (Fig. 1G‒I) corroborated the MTT assay results.

Fig. 1

Concentration- and time-dependent cytotoxicity of SF in HT22 cells. (A–D) Concentration-dependent effect of SF (12.5–100 mM) on cell viability (MTT assay, n = 5). Cells were treated with 12.5–100 mM SF for different durations (A for 24 h, B for 48 h, C for 72 h, and D for 96 h). Culture medium was replenished every 24 h. (E, F) Time-dependent effect of 25 mM E and 50 mM SF F on cell viability (MTT assay, n = 5). Culture medium was replenished every 24 h. (G–I) Change in cell viability (CCK-8 assay, n = 5) following treatment with 12.5–100 mM SF for different durations (G for 24 h, H for 48 h, and I for 72 h). Culture medium was replenished every 24 h. (J–L) Change in relative cell density (based on crystal violet staining) following treatment with 12.5–100 mM SF for different durations (J for 24 h, K for 48 h, and L for 72 h; n = 5). Culture medium was replenished every 24 h. (M–O) Change in cell number (manual cell counting) following treatment with 12.5–100 mM SF for different durations (M for 24 h, N for 48 h, and O for 72 h; n = 3). Culture medium was replenished every 24 h. Each value is the mean ± S.D. * and ** denote for P < 0.05 and P < 0.01, respectively.

In this study, the crystal violet staining of cells was used to reflect the change in relative cell density. As shown in Fig. 1J‒L, SF decreased crystal violet staining in a concentration-dependent manner. It was noted that the reduction in apparent cell viability (based on MTT assay) was slightly bigger than the reduction in relative cell density. For instance, after 24-h treatment with 100 mM SF, the cell viability was decreased by 47% (Fig. 1A), but the relative cell density was decreased to a smaller degree, by 34% (Fig. 1J). Further manual counting of the cell numbers confirmed the observation that SF elicited a concentration-dependent reduction in cell density, and the decrease in cell viability (MTT assay) was slightly bigger than the reduction in cell density (Fig. 1M‒O). Given that the MTT assay mostly reflects the SDH activity in cultured cells^22,23, it is thus speculated that SF may preferentially disrupt mitochondrial function in addition to reducing cell density through inhibition of cell proliferation.

Using the Calcein-AM/PI staining method, next we assessed the esterase activity in cells exposed to SF for 24, 48 and 72 h. Calcein-AM is a substrate for cellular esterase, and is converted to a green fluorescence when inside live cells. We found that cells treated with higher concentrations of SF for longer durations had slightly higher levels of green fluorescence (reflecting higher esterase activity) compared to the control cells (Supplementary Fig. S1A, B, C for 24, 48 and 72 h exposure), suggesting that more Calcein-AM molecules (the substrate) get inside SF-treated cells. In addition, these cells also displayed slightly elevated PI staining (Supplementary Fig. S1A, B, C). Notably, very few dead cells were observed when the cells were exposed to lower concentrations of SF, but at higher concentrations, some dead cells were observed. These observations indicate that high concentrations of SF can mildly increase cytoplasmic membrane permeability; as a result, more Calcein-AM and PI molecules get inside the cells. As Calcein-AM is a substrate for the esterase, higher green fluorescence is thus seen in SF-treated cells.

In addition, it was observed that SF-treated cells appeared less healthy morphologically (Supplementary Fig. S2A, B, C). Notably, following exposure to higher concentrations of SF (50 and 75 mM) for 48 or 72 h, a significant fraction of the cells looked visibly larger in shape than the control cells (Supplementary Fig. S2B, C). When the cells were treated with lower concentrations of SF (≤ 25 mM), the changes in cell density and appearance were less apparent.

SF-induced changes in gene expression

To gain insights into the subcellular sites of SF action and the underlying mechanisms, comparative transcriptomics analysis of the HT22 cells treated with vehicle alone or 12.5 and 25 mM SF for 48 h was carried out using the transcriptome analysis. A total of 13,834 transcripts are identified based on analysis of all the samples. Principal component analysis (PCA) is performed to determine the differences between the transcriptomes of different treatment groups. The transcriptomics data of the control and SF-treated cells are broadly separated (Supplementary Fig. S3A). For cells treated with 12.5 mM SF for 48 h, 660 (4.8%) transcripts are significantly altered (Q_value ≤ 0.05) in SF-treated cells compared to the control cells (Supplementary Fig. S3B), with 386 genes upregulated (Q_value ≤ 0.5) and 274 genes down-regulated (Q_value ≤ 0.05). For cells treated with 25 mM SF for 48 h, 1493 (10.8%) transcripts are significantly altered (Q_value ≤ 0.05) in SF-treated cells compared to the control cells (Supplementary Fig. S3B), with 808 genes upregulated (Q_value ≤ 0.5) and 685 genes down-regulated (Q_value ≤ 0.05). In the volcano plots, an additional filter of |log₂ (SF/Con)|> 0.5 is applied to further refine the selection of genes. This filter helps to identify genes with significant differential expression and larger fold changes after 48-h exposure to different concentrations of SF (Supplementary Fig. S3C, D). The gene set enrichment analysis (GSEA) of the KEGG pathways^24,25,26 reveals that the expression of many genes which are altered in SF-treated cells are commonly associated with apoptosis, autophagy, neurodegeneration, Alzheimer disease, oxidative phosphorylation, and cell cycle (Fig. 2A).

Fig. 2

Copyright permission for the use of the KEGG database in this analysis has been obtained. (B) Effect of 12.5 and 25 mM SF (48-h treatment) on the expression of selected genes. Note that the RNA-seq and transcriptomic analysis were repeated twice, and one representative data set is shown here. Each quantitative value is the mean ± S.D (n = 3). Note that the mean intensity value for each control group is labeled in white numerical number in the black bar. * and ** denote for P < 0.05 and P < 0.01, respectively, compared to the respective control group.

Transcriptomic analysis of SF-treated HT22 cells. (A) The KEGG disease enrichment plot of genes ^24,25,26 showing SF-induced significant changes. The cells were treated with SF at 12.5 and 25 mM for 48 h, and then harvested for RNA-seq and transcriptomic analysis.

Genes associated with mitochondrial function and ATP synthesis

Several groups of relevant genes are briefly summarized below, which may help shed lights on the mechanism of SF-induced mitochondrial disruption in HT22 cells.

• (i) Ndufv3, Ndufb7 and Ndufb11, which are related to the mitochondrial respiratory chain complex I assembly, are down-regulated by approximately 20% in cells treated with 25 mM SF (Fig. 2B), which likely reflects the disruption of mitochondrial respiratory chain function and the coupling between electron transport and ATP synthesis.

• (ii) Several genes related to F-ATPase and V-ATPase in complex V, i.e., Atp5mf, Atp6v1a, Atp6v1g1, Atp6v0a2, Atp6v0d1 and Tcirg1, are slightly down-regulated (by 10–25%; Fig. 2B). Interestingly, the expression of Ppa1 in cells treated with 25 mM SF is upregulated by 24% (Q_value < 0.05), which may indicate that these cells are trying to enhance their cellular energy metabolism in response to mitochondrial oxidative stress.

• (iii) Changes in several other mitochondrial genes (Pdk4 and Camk2a) are also observed. For instance, Pdk4, which has ATP-binding activity and pyruvate dehydrogenase kinase activity²⁷, is located in mitochondrial inner membrane, is down-regulated by 50% (Q_value < 0.01; Fig. 2B).

• (iv) Notably, cells treated with 12.5 and 25 mM SF for 72 h only had a very small down-regulation of Sdhaf1 and Sdhaf4 (the genes for SDH subunits), and the expression of Sdhaf2 and Sdhaf3 was not changed (Supplementary Fig. S4).

• (v) Down-regulation of Idh1 (36%, Q_value < 0.01; Fig. 2B) and Idh2 (10%, Q_value < 0.05; Fig. 2B) at 25 mM SF suggests a decrease in the efficiency of the TCA cycle, which may lower ATP production. Additionally, a small upregulation of Dlat by 15% (Q_value < 0.05; Fig. 2B) may help maintain TCA cycle activity by supporting the conversion of α-ketoglutarate to succinyl-CoA, partially compensating for the reduced efficiency caused by lower Idh1 and Idh2 levels. In response, the upregulation of Pcx (40%, Q_value < 0.01; Fig. 2B) and Pck2 (35%, Q_value < 0.01; Fig. 2B) at 25 mM SF indicates an increase in gluconeogenesis, whereas the decrease in Acly (10%, Q_value < 0.05; Fig. 2B) suggests reduced fatty acid synthesis.

Genes related to cell cycle and growth regulation

The expression of several genes involved in cell cycle and growth regulation is significantly altered by treatment with 25 mM SF for 48 h (summarized in Fig. 2C). For instance, Cdkn2c, Cdkn2d and Cdkn1a (three genes encoding cyclin-dependent kinase inhibitors) are down-regulated by 30% (Q_value < 0.01), Tfdp2 (transcription factor DP2) is down-regulated by 45% (Q_value < 0.01), and Rbl1 (RB transcriptional corepressor like 1) is slightly down-regulated (by 15%, Q_value < 0.01). In comparison, Myc (involved in cell cycle regulation) is upregulated by 40% (Q_value < 0.01), and similarly, Stat3 (a transcription activator responsive to cytokines and growth factors) is upregulated by 50%. Together, changes in these genes likely suggest that the cells are under growth-suppressed and stressed situations. Notably, the expression of genes involved in cell cycle regulation (i.e., Mtbp, E2f2, Orc4, Cdk6, Anapc5, Orc2, Gadd45a and Zbtb17) are also altered, which are as expected.

Some other genes

In addition, treatment with 25 mM SF for 72 h causes an upregulation (by 80%) of Edn1 (Fig. 2D), a member of the endothelin family of peptides, and its increased expression will promote inflammatory responses²⁸. Mmp9, a member of the matrix metalloproteinase family (extracellular matrix-degrading enzymes), is upregulated by 70%. Mmp9 likely is involved in tissue remodeling, wound repair, progression of atherosclerosis and tumor invasion^29,30. Vegfd, Vegfa and Serpine1, three genes related to angiogenesis^31,32,33, are upregulated by 15 − 60%.

Notably, among the other genes which their expressions are altered by SF, many of them are known to be associated with the AGE-RAGE signaling pathway and diabetic cardiomyopathy pathway. As shown in Fig. 2D, SF upregulates eight genes (i.e., Cacna1i, Ctsd, Nfatc1, Tgfb3, Plcd1, Nfkb1, Col4a6 and Pik3r1), and down-regulates six other genes (i.e., Col1a1, Irs1, Col4a2, Vdac3, Tmem70 and Ppp1cb).

SF disrupts mitochondrial function and structure

Based on the insights gained from the KEGG analysis of SF-induced changes in transcriptomics, next we first sought to analyze effects of SF on mitochondrial structure and functions. The data are summarized below.

Changes in mitochondrial density and structure

Using analytical flow cytometry coupled with MitoTracker deep red staining, we first probed the change in relative mitochondrial density (Fig. 3A‒C). Our results indicate that the relative density of mitochondria in cells was increased by increasing SF dosage and exposure time. For instance, exposure to 25 mM SF for 24 h resulted in a small but statistically-insignificant increase in mitochondrial count (17%; P = 0.5791) (Fig. 3A), but prolonged exposure (48 and 72 h) led to a time-dependent cumulative increase (P < 0.05) (Fig. 3B, C).

Fig. 3

Effect of SF on mitochondrial density and structure in HT22 cells. (A–C) Effect of SF on the relative mitochondrial density (flow cytometry, stained with MitoTracker deep red) following treatment with 25, 50 and 75 mM SF for 24 h (A), 48 h (B) and 72 h (C). Each quantitative value is the mean ± S.D (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively, compared to the respective control group. (D–G) TEM images showing the structures of mitochondria in control cells. (H–K) Effect of SF on mitochondrial structures (TEM images) following treatment with 25 mM SF for 48 h. (L–O) Type II (L, M) and type III (N, O) mitochondrial autophagy. (P–S) Autophagosomes (P, Q) and autolysosomes (R, S).

In this study, SF-induced change in mitochondrial structure was analyzed by TEM. The mitochondria in untreated cells exhibited intact structures, with cristae clearly discernible at high magnification (Fig. 3D‒G). Following treatment with 25 mM SF for 48 h, alterations were observed in the cellular structures; the cell nuclei showed some degree of invagination, the number of mitochondria was increased, and the cristae in these mitochondria were lost. Additionally, some mitochondria appeared irregular in shape (Fig. 3H‒K). There were free membrane structures found near the inner surfaces of the damaged mitochondria in SF-exposed cells (marked with red arrows; Fig. 3L, M), which likely indicate the occurrences of type II mitochondrial autophagy^34,35,36. In addition, we also observed type III mitochondrial autophagy, in which damaged mitochondria form membrane structures that encapsulate parts of the organelle to generate mitochondria-derived vesicles (marked with red arrows; Fig. 3N, O). As the autophagy process progresses, these membrane structures gradually fuse to form a complete autophagosome that encapsulates the mitochondrion. SF exposure also induced the formation of autophagosomes (marked with red arrows; Fig. 3P, Q) and autolysosomes (marked with red arrows; Fig. 3R, S).

Changes in mitochondrial functions

The change in mitochondrial membrane potential (MMP) was commonly used to reflect alterations in mitochondrial function^37,38. Hence, analytical flow cytometry and confocal microscopy were jointly used to determine the change in MMP. When MMP is high, JC-1 accumulates in the mitochondrial matrix, forming J-aggregates and emitting red fluorescence; when MMP is low, JC-1 does not accumulate in the mitochondrial matrix and exists as monomers, emitting green fluorescence. It was observed that SF-induced MMP change was concentration- and time-dependent. For instance, after 48-h treatment with 75 mM SF, only 9.0% of cell clusters showed a decrease in MMP, but when SF exposure was extended to 72 h, approximately 35% of the cells exhibited decreased MMP (Fig. 4A). Confocal microscopic analysis revealed 28.7% (P < 0.01) and 39.7% (P < 0.01) reductions, respectively, in MMP after treatment with 75 mM SF for 48 and 72 h (Fig. 4B‒E).

Fig. 4

Effect of SF on mitochondrial membrane potential (MMP) and functional protein expression in HT22 cells. (A) Effect of SF on MMP (flow cytometry, stained with JC-1) following treatment with 25, 50 and 75 mM SF for 24, 48 and 72 h. The vertical axis represents the fluorescence intensity of JC-1 aggregates, and the horizontal axis represents the fluorescence intensity of JC-1 monomers. (B–E) Concentration- and time-dependent effects of SF on MMP (confocal microscopy, stained with JC-1, scale bar = 20 μm) following treatment with 25, 50 and 75 μM SF for 48 h (B) or 72 h (C). Panels D and E are the quantitative analysis of the images in B and C (using the ImageJ software) showing the red/green fluorescence ratios. Each value is the mean ± S.D. (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively.

In this study, the change in mitochondrial permeability transition pore (mPTP) opening was used as another parameter to reflect mitochondrial function. The effect of SF on mPTP opening was analyzed using flow cytometry coupled with Calcein-AM and CoCl₂ staining. Calcein-AM (a non-fluorescent substrate) is converted by cellular esterase to a fluorescent compound, while CoCl₂ acts as a fluorescence quencher to selectively label the mitochondria. We found that as the concentrations of SF and the duration of its exposure increased, there was a decrease in Calcein-AM fluorescence intensity, most notable in the 75-mM SF group, suggesting an increase in mPTP opening (Fig. 5A‒C). This change, most pronounced at 72 h, indicates that prolonged exposure and higher concentrations of SF have a stronger impact on mitochondrial function, potentially leading to mitochondrial dysfunction.

Fig. 5

Effect of SF on mitochondrial function in HT22 cells. (A–C) Effect of SF on mitochondrial permeability following treatment with 25, 50 and 75 mM SF for 24 h (B), 48 h (C) and 72 h (D). The mPTP opening was determined using flow cytometry following staining with Calcein-AM and CoCl₂. (D) Concentration-dependent effect of SF on cellular ATP levels (n = 6) following treatment with 25 and 50 mM SF for 48 h. (E) Time-dependent effect of SF on cellular ATP levels (n = 8) following treatment with 25 mM SF for 24, 48, 72 and 96 h. (F) Effect of SF on ATP generation (based on the Seahorse method; n = 3). Cells were treated with 25, 50, and 75 mM SF for 48 h, followed by the addition of 1.5 µM oligomycin, 0.5 µM rotenone, and 0.5 µM antimycin A. The rate of ATP production from both mitochondrial oxidative phosphorylation and glycolysis are then calculated. (G) Effect of SF on cellular respiratory capacity (based on the Seahorse method; n = 3). Cells were treated with 0 and 75 mM SF for 48 h, followed by the addition of 1.5 µM oligomycin, 0.25 µM FCCP and 0.5 µM rotenone/antimycin A. The OCR is then calculated.

The above results clearly show that the mitochondrial function is disrupted by SF exposure. Next, we sought to determine the change in cellular ATP levels following SF exposure for different durations. Treatment of cells with 25 and 50 mM SF for 48 h elicited a concentration-dependent reduction in cellular ATP levels (Fig. 5D). In addition, when the cells were treated with 25 mM SF for 24, 48, 72 and 96 h, there was a time-dependent cumulative reduction in cellular ATP levels with prolonged exposure (Fig. 5E).

By using the XF96 cellular energy metabolism analyzer, we also explored the impact of SF on real-time cellular ATP production. We found that SF primarily affects mitochondrial ATP synthesis. For instance, after 48-h treatment with 75 mM SF, the rate of mitochondrial ATP synthesis was reduced by 53%; in comparison, the rate of glycolytic ATP production was only modestly reduced (Fig. 5F). Additionally, using the XF96 analyzer, we found that 48-h SF exposure reduced the basal and maximal respirations by approximately 40% compared to the control group (Fig. 5G).

Changes in cellular SDH activity

It is known that MTT is reduced to purple protonated imidazole crystals (formazan) in live cells catalyzed mostly by the mitochondrial SDH. The crystal deposits in the mitochondria and the degree of deposition reflect mitochondrial metabolic activity of the cells. In this study, microscopic analysis of MTT-stained cells showed that SF exposure produced a small reduction in the deposition of the formazan crystals inside the cells, and this reduction was dependent on SF concentration and treatment duration (Fig. 6A, Supplementary Fig. S5).

Fig. 6

Effects of SF on mitochondrial SDH activity and mitochondrial regulatory proteins in HT22 cells. (A) Effect of SF on cellular MTT formazan deposition (microscopy images, scale bar = 25 μm). HT22 cells were treated with 12.5, 25, 50, 75 and 100 mM SF for 72 h, and then incubated with MTT (0.5 mg/mL) for 2 h to allow formazan formation in the cells. (B) Concentration- and time-dependent effects of SF on cellular SDH protein levels following treatment of HT22 cells with 25, 50 and 75 mM SF for 24, 48 and 72 h. The total cellular proteins (25 μg) were separated by 10% SDS-PAGE and immunoblotted with specific antibodies for SDHA and β-actin. The corresponding unprocessed Western blot images are shown in Supplementary Fig. S6, and the quantitative intensity values are shown in Supplementary Fig. S7. (C–E) Effect of SF on the SDH activity in the cell lysates prepared from HT22 cells treated with 25, 50 and 75 mM SF for 24 h (C), 48 h (D) and 72 h (E). (n = 3). (F–H) Effect of SF on the SDH activity in the mitochondrial fraction prepared from HT22 cells treated with 25, 50 and 75 mM SF for 24 h (F), 48 h (G) and 72 h (H). (n = 3). J. Direct inhibition by SF of the SDH enzyme activity in HT22 whole cell lysates in vitro. The cell lysates were prepared from untreated HT22 cells, and were used for the in-vitro assay of the SDH activity in the presence of 25, 50, 75 and 100 mM SF (n = 5). J. Direct inhibition by SF of the mouse liver SDH activity in vitro. The mouse liver samples were homogenized for preparation of tissue lysates, which were then used for the in-vitro assay of the SDH activity in the presence of 25, 50, 75 and 100 mM SF (n = 5). (K) Levels of PGC-1α, NRF-1 and TFAM proteins following treatment of HT22 cells with SF (25, 50 and 75 mM) for 24, 48 and 72 h. The total cellular proteins (25 μg) were separated by 10% SDS-PAGE under a reducing condition and immunoblotted with specific antibodies for PGC-1α, NRF-1, TFAM and β-actin. The corresponding unprocessed Western blot images are shown in Supplementary Fig. S8, and the quantitative intensity values are shown in Supplementary Fig. S9. Each quantitative value is the mean ± S.D. * and ** denote for P < 0.05 and P < 0.01, respectively.

Next, we compared SDH protein levels in HT22 cells treated with increasing concentrations of SF for 24, 48 and 72 h. We found that the total cellular SDH protein levels were not significantly affected by SF exposure for up to 72 h (Fig. 6B, Supplementary Fig. S6, S7). However, the SDH enzyme activity in their whole cell lysates was affected by SF in a concentration- and time-dependent manner (Fig. 6C‒E). For instance, after exposure to 25 mM SF for 48 h, the SDH activity was decreased by 22% (P = 0.117), and the activity in cells treated with 75 mM SF was decreased by ~ 50% (P < 0.01). Similarly, when the mitochondrial fraction from SF-treated HT22 cells was separately prepared and analyzed, similar concentration-dependent reductions in the SDH activity were observed (Fig. 6F–H).

These above observations suggest that SF may directly inhibit the mitochondrial SDH. This possibility was confirmed experimentally in this study. We found that incubation of HT22 whole cell lysates with SF in vitro produced a concentration-dependent inhibition of the SDH enzyme activity (Fig. 6I). Additionally, we also used the mouse liver lysates as an enriched source of mitochondrial SDH to help use more reliably determine whether SF can directly inhibit SDH activity. After incubation of the mouse liver lysates with increasing concentrations of SF at 37 °C for 30 min, we found that the SDH activity was also inhibited by SF in a concentration-dependent manner (Fig. 6J), thus confirming the observation.

Changes in mitochondrial regulatory proteins

In this study, we also determined the changes in three proteins TFAM, PGC1α and NRF1 as markers of mitochondrial dysfunction and oxidative stress. We found that the protein levels of TFAM, PGC1α and NRF1 were decreased with increasing concentrations of SF and prolonged treatment durations (Fig. 6G, Supplementary Fig. S8, S9). The reduction in these mitochondrial proteins would impact mitochondrial function, cellular energy metabolism and oxidative stress.

SF alters mitochondrial ROS levels

Fluorescence microscopic analysis showed that exposure of HT22 cells to increasing concentrations of SF caused a small but detectable increase in mitochondrial ROS (based on MitoSOX staining; Fig. 7A). This small increase in mitochondrial ROS appeared to be dependent on the concentrations of SF used (Fig. 7A). It is of note that based on flow cytometry analysis, mitochondrial ROS was significantly elevated following exposure to 25 mM SF for 24 h, (Fig. 7B); however, prolonged exposure to 25 mM SF did not cause a further increase in mitochondrial ROS levels (Fig. 7C‒F, Supplementary Fig. S10A, B). Similarly, when the cells were treated with a lower concentration (12.5 mM) of SF for 48 and 72 h, we did not observe a time-dependent further increase in mitochondrial ROS levels (Supplementary Fig. S10A, B).

Fig. 7

Effect of SF on Mitochondrial ROS levels in HT22 cells. (A) Effect of SF on mitochondrial ROS levels (fluorescence microscopy, scale bar = 25 μm) following treatment with SF (12.5, 25, 50 and 75 mM) for 48 h. (B–F) Effect of SF on mitochondrial ROS levels (flow cytometry) following treatment with 25 mM of SF for 24 h (B), 48 h (C), 72 h (D), 96 h (E), and 120 h (F). The left panels are the histograms, and the right panels are the corresponding quantitative intensity values (n = 3; the value for the control group is labeled on top of the bar). (G–J) Concentration- and time-dependent effects of SF on mitochondrial ROS levels (confocal microscopy, scale bar = 20 μm) following treatment with SF (25, 50 and 75 mM) for 24 h (G) or 48 h (H). Panels I and J are the quantitative analysis of the image results in G and H (using the ImageJ software) showing the red fluorescence intensity at 561 nm (excitation). Each value is the mean ± S.D. (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively.

Confocal microscopic analysis showed that the mitochondrial ROS (MitoSOX-staining) were mostly colocalized with the MitoTracker signal (Fig. 7G, H). It was confirmed that very small amount of mitochondrial ROS accumulation was detectable after 24-h exposure to 25 mM SF, and it increased with higher concentrations of SF (Fig. 7G–J).

SF alters cellular ROS, lipid-ROS, NO and GSH levels

It is speculated that increased buildup of mitochondrial ROS may lead to increased levels of cellular ROS, lipid-ROS and NO, ultimately affecting cellular GSH levels. Therefore, we have also examined the change in these reactive species in HT22 cells following exposure to SF for increasing durations.

ROS levels

Following exposure to SF (12.5, 25, 50 and 75 mM) for 24, 48 and 72 h, there were concentration- and time-dependent increases in cellular ROS levels (Fig. 8A, Supplementary Fig. S11A‒S11C). It is apparent that for the same treatment time, higher concentrations of SF caused a bigger increase in cellular ROS levels. In a separate experiment where the cells were exposed to 25 mM SF for up to 96 h, a time-dependent increase in cellular ROS buildup was further confirmed (Fig. 8B, Supplementary Fig. S11D).

Fig. 8

Effect of SF on cellular ROS levels in HT22 cells. (A) Effect of SF on cellular ROS levels (fluorescence microscopy, scale bar = 25 μm). Cells were treated with SF (12.5–100 mM) for 24, 48 and 72 h, and then stained with DCFH-DA for imaging analysis. Quantitative values for A are shown in Supplementary Fig. S11A‒S11C. (B) Time-dependent effect (24–96 h) of 25 mM SF on cellular ROS levels (fluorescence microscopy, scale bar = 25 μm). Quantitative values for B are shown in Supplementary Fig. S11D (C–G). Time-dependent effect of SF on cellular ROS levels (flow cytometry) following treatment with 25 mM SF for 24 h (C), 48 h (D), 72 h (E), 96 h (F), and 120 h (G). The left panels are the histograms, and the right panels are the corresponding quantitative intensity values (the value for the control group is labeled on top of the bar). Each quantitative value is the mean ± S.D (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively.

The concentration- and time-dependent increase in cellular ROS levels in SF-treated cells was also confirmed by flow cytometric analysis (Fig. 8C‒G, Supplementary Fig. S12A, S12B). For instance, while treatment of cells with 25 mM SF for 24 h caused a small, statistically-insignificant increase (20%; P = 0.1968) in cellular ROS levels (Fig. 8C), prolonged exposure to 48, 72, 96 and 120 h produced a time-dependent cumulative increase in ROS levels (P < 0.05) (Fig. 8D‒G), reaching a plateau at approximately 72 h. Similar concentration- and time-dependent change in cellular ROS levels was observed in another experiment when the cells were treated with 12.5 and 25 mM SF for 48 and 72 h (Supplementary Fig. S12A, S12B). While 12.5 mM SF produced a very small increase (statistically insignificant) in cellular ROS levels at both 48 and 72 h (P = 0.6651 and 0.5093, respectively), 25 mM SF consistently produced a significant increase in cellular ROS levels at these two time points (P < 0.05; Supplementary Fig. S12A, S12B).

Lipid-ROS levels

Confocal microscopy was used to characterize the accumulation and subcellular localization of lipid-ROS in SF-treated cells. We found that SF elicited a time- and concentration-dependent increase in cellular lipid-ROS (Fig. 9A, B, Supplementary Fig. S13A, B). For instance, after exposure of cells to 12.5 mM SF for 48 h, the green fluorescence appeared and intensified with increasing concentrations of SF. Similarly, when cells were treated with the same concentrations of SF for a longer duration, higher accumulation of lipid-ROS was observed.

Fig. 9

Effect of SF on lipid-ROS levels in HT22 cells. (A, B) Concentration- and time-dependent effects of SF on lipid-ROS levels (confocal microscopy, scale bar = 20 μm) following treatment with 12.5, 25 and 50 mM SF for 48 h (A) and 72 h (B). Quantitative values for A, B are shown in Supplementary Fig. S13A, S13B. (C–G) Flow cytometry analysis of lipid-ROS levels (flow cytometry) following treatment with 25 mM SF for 24 h (C), 48 h (D), 72 h (E), 96 h (F), and 120 h (G). The left panels are the histograms, and the right panels are the corresponding quantitative intensity values (the value for the control group is labeled on top of the bar). Each quantitative value is the mean ± S.D (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively.

Analytical flow cytometry was also used to quantify the change in cellular lipid-ROS levels (Fig. 9C‒G, Supplementary Fig. S14A, B). We found that SF elicited a time- and concentration-dependent increase in cellular lipid-ROS levels. For instance, while treatment of cells with 25 mM SF for 24 h caused a small but statistically-insignificant increase (34%; P = 0.4959) in cellular lipid-ROS levels (Fig. 9C), prolonged exposure to 48, 72 and 96 h produced time-dependent cumulative increase (P < 0.05) (Fig. 9D‒F). Similar observations were made in another experiment when the cells were treated with a lower concentration (12.5 mM) of SF for 48 and 72 h (Supplementary Fig. S14A, B).

NO levels

NO accumulation in cells treated with different concentrations of SF (12.5 − 75 mM) for different durations was jointly determined by jointly using fluorescence microscopy and analytical flow cytometry. We found that SF induces NO accumulation in a concentration- and time-dependent manner (Fig. 10A‒D, Supplementary Fig. S15A, B). For instance, when the cells were treated with 25 mM SF, the NO signal was detected at 48 h after exposure, but when the cells were treated with an even lower concentration (12.5 mM) of SF for a longer duration (72 h), a similar increase in NO signal was detected, indicating that extending the length of SF exposure can lead to a cumulative increase in cellular NO levels.

Fig. 10

Effect of SF on NO and GSH levels in HT22 cells. (A) Effect of SF on cellular NO levels (fluorescence microscopy, scale bar = 25 μm) following treatment with SF (12.5–100 mM) for 48 and 72 h. Quantitative values for (A) are shown in Supplementary Fig. S15A, S15B. (B–D) Effect of SF on cellular NO levels (flow cytometry) following treatment with SF (25–75 mM) for 24 h (B), 48 h (C), and 72 h (D). (E–G) Effect of SF on cellular GSH levels following treatment with SF (25, 50 and 75 mM) for 24 h (E), 48 h (F), and 72 h (G). Each quantitative value is the mean ± S.D (n = 3). * and ** denote for P < 0.05 and P < 0.01, respectively.

GSH levels

Reduced GSH is a crucial cellular antioxidant, and its cellular levels reflect the antioxidant capacity of the cells. In this study, we determined the change in cellular reduced GSH levels in HT22 cells treated with different concentrations of SF (25, 50 and 75 mM) for different durations (24‒72 h). We found that in cells treated with SF, there was a concentration- and time-dependent reduction in reduced GSH levels (Fig. 10E‒G). This observation is in accord with the observations described above that chronic SF exposure can induce cumulative oxidative stress in exposed cells, which will then lead to increased consumption of reduced GSH to counteract SF-induced oxidative stress and cellular injuries.

SF-induced changes in cell cycle, cell proliferation and apoptosis

Results from our transcriptomic analysis clearly indicate that SF also alters the genes involved in regulating cell cycle, proliferation and cell death. Based on these observations, we also investigated the effects of SF on cell cycle, proliferation and cell death.

Cell cycle

Flow cytometry analysis of the cell cycle showed that cells exposed to SF exhibited a concentration-dependent increase in the population of S-phase cells (Fig. 11A, Supplementary Fig. S16). For instance, after treatment with 75 mM SF for 24 h, the population of S-phase cells increased from approximately 23.0% in the control group to 41.5% (P < 0.05). This trend appeared to be more pronounced when the cells were treated with lower concentrations of SF (25 and 50 mM) for longer durations.

Fig. 11

Effect of SF on cell cycle and apoptotic cell death in HT22 cells. (A) Effect of SF on cell cycle (flow cytometry, stained with PI, n = 3) following treatment with 25, 50 and 75 mM SF for 24, 48 and 72 h. Each value is the mean ± S.D. (B) Effect of SF on cell proliferation (flow cytometry, stained with Click-iT EdU, n = 3) following treatment with 25, 50 and 75 mM SF for 24, 48 and 72 h. Each value is the mean ± S.D. (C) Concentration- and time-dependent effects of SF on apoptosis induction (flow cytometry, labeled with annexin V-FITC and PI) following treatment with 25, 50 and 75 mM SF for 72 h. (D–F) Effects of z-VAD-FMK (D), Nec-1 (E), and Fer-1 (F) on SF-induced cytotoxicity (MTT assay, n = 5) following treatment with 75 mM SF ± z-VAD-FMK or Nec-1 or Fer-1 for 72 h. The cell culture medium was replaced every 24 h.  (G) Protective effect of z-VAD-FMK against SF-induced apoptosis (flow cytometry, labeled with annexin V-FITC and PI, n = 3) following treatment with 75 mM SF ± 5 μM z-VAD-FMK for 72 h. Quantitative values (mean ± S.D.) for different quadrants are shown in Supplementary Fig. S19. Each quantitative value is the mean ± S.D. * or ^# denotes for P < 0.05 and ** or ^## denotes for P < 0.01, respectively.

Cell proliferation

During the S phase of the cell cycle, 5-ethynyl-2'-deoxyuridine (EdU), a thymidine analog, can be incorporated into the newly-synthesized DNA³⁹. Flow cytometry analysis of EdU-incorporated cells showed that SF-treated cells exhibited concentration- and time-dependent reductions in cell proliferation (Fig. 11B, Supplementary Fig. S17). For instance, after treatment with 25 mM SF for 24, 48 and 72 h, the rate of cell proliferation was reduced in a time-dependent manner (Fig. 11B, Supplementary Fig. S17).

Apoptosis

Using annexin V-FITC/PI double staining (coupled with flow cytometry analysis), we explored whether SF exposure induces apoptotic cell death. We found that the number of PI-negative but annexin V-positive cells was increased with increasing SF concentration and treatment time (Fig. 11C for 72-h exposure; Supplementary Fig. S18 for 24 and 48 h exposure). For instance, cells treated with 75 mM SF for 72 h had 16.1% of cells singly positive for annexin V, but the population of late apoptotic cells (i.e., PI and annexin V double positive cells) was very small, only approximately 3.2% (Fig. 11C). This result is in accord with the results of Calcein-AM/PI staining (Supplementary Fig. S1A, B, C) and cell cycle analysis of the sub-G₁ population (Fig. 11A, Supplementary Fig. S16).

To confirm that SF induces apoptotic cell death, we analyzed the effects of z-VAD-FMK (a pan-caspase inhibitor⁴⁰), necrostatin-1 (Nec-1, a necroptosis inhibitor⁴¹), and ferostatin-1 (Fer-1, a ferroptosis inhibitor⁴²) on SF-induced cytotoxicity. Based on MTT assay, z-VAD-FMK exhibited a concentration-dependent partial protection against SF-induced cytotoxicity (Fig. 11D). By contrast, joint treatment of cells with Nec-1 and Fer-1 did not display a meaningful protection against SF-induced loss of MTT activity (Fig. 11E, F). Here, it is of note that when the apoptotic cells were selectively measured with annexin V-FITC/PI double staining, we found that the presence of z-VAD-FMK prevented ~ 90% of SF-induced apoptotic cell death (Fig. 11G). The apparent discrepancy that z-VAD-FMK prevents SF-induced apoptosis but only weakly restores overall MTT activity suggests that SF-induced reduction in cell proliferation is a major contributor to the overall reduction in MTT activity.

Discussion

Previous studies have established that formic acid is cytotoxic both in vitro and in vivo^5,18. In the present study, we found that treating HT22 mouse hippocampal neuronal cells with SF (a salt form of formic acid) causes mitochondrial ROS accumulation, disrupts mitochondrial structure, impairs mitochondrial functions (e.g., MMP, mPTP opening and ATP production), and ultimately, resulting in reduced cell proliferation and accelerated cell death. Provided below is a discussion of the effects of SF on mitochondrial and cellular functions and the underlying mechanisms.

SF disrupts mitochondrial structure and functions

Mitochondria are complex cellular organelles responsible for the production of energy for normal cellular function and survival⁴³. As key regulators of bioenergetic status and cellular viability, mitochondria are also involved in various forms of cell death, including classical apoptosis and necroptosis^44,45, as well as ferroptosis⁴⁶ and autophagic cell death⁴⁷. During apoptosis, the mitochondria release cytochrome C into the cytoplasm, leading to the formation of apoptosomes and activation of the caspase cascade, which orchestrate the intrinsic and extrinsic apoptotic pathways. Altered functions of the double-membrane structure of mitochondria play a pivotal role in the execution of apoptosis⁴⁸.

The results of our present study show that SF exposure significantly impairs mitochondrial functions. Previous literature has indicated that the mitochondrial cytochrome oxidase is a primary target of formic acid toxicity^1,3,12,49, which may disrupt the flow of electrons through the electron transport chain (ETC) and lead to mitochondrial ROS production^1,3,12,14,50. We hypothesized that SF-induced accumulation of mitochondrial ROS may serve as an initial event leading to the disruption of mitochondrial membrane integrity, which leads to a cascade of cellular stress responses that increase cytosolic levels of ROS, lipid-ROS and NO. These oxidative stress responses may further exacerbate the loss of mitochondrial membrane integrity, increase membrane permeability, and initiate mitochondria-mediated cell death pathways⁵¹. As briefly summarized below, a number of experimental evidence are obtained in this study which offer support for this hypothesis.

First, our results indicate that treatment of cells with even relatively low concentrations of SF (25 mM) can lead to significant mitochondrial ROS accumulation, which is accompanied by the opening of mPTP and loss of MMP. This phenomenon becomes more pronounced with prolonged SF exposure. MMP is a key indicator of mitochondrial function, and the observation that SF reduces MMP in a concentration-dependent manner confirms that SF can induce mitochondrial dysfunction. As suggested above, the observed mitochondrial ROS accumulation most likely stems from SF’s direct inhibition of the mitochondrial cytochrome oxidase. This suggestion is in line with an earlier observation that SF at a rather low concentration (10 mM) could directly stimulate ROS production in the isolated liver mitochondrial preparations⁴⁹. The accumulation of mitochondrial ROS can damage mitochondrial membranes, which then leads to mPTP opening and loss of MMP, along with increased permeability of the inner and outer mitochondrial membranes. mPTP pore opening can further lead to dissipation of the proton gradient, thereby prohibiting ATP synthesis; these changes are often accompanied by mitochondrial swelling and additional ROS accumulation, and ultimately, irreversible mitochondrial damage.

Second, in line with the above explanation, we found that SF induces a concentration- and time-dependent decrease in cellular ATP levels. Following prolonged SF exposure, the decrease in cellular ATP becomes more pronounced. It is believed that mitochondrial impairment is the underlying cause of the energy deficit observed in SF-treated cells. This suggestion is further supported by the observation that ATP production through glycolysis is only mildly affected in SF-exposed cells, whereas mitochondrial ATP production is significantly reduced. Notably, similar observations were also made by others in cultured cells^52,53,54, and the ATP synthase, along with the mitochondrial precursor of the ATP synthase α, was reduced by formic acid exposure^52,55.

Third, we find that the activity of mitochondrial SDH (Complex II) is reduced in HT22 cells by SF in a concentration-dependent manner. However, the total SDH protein levels are only mildly reduced in SF-exposed cells. Similarly, the expression of the Sdhaf1, Sdhaf2, Sdhaf3 and Sdhaf4 genes is not significantly affected by SF exposure. Together, these results suggest that the reduced mitochondrial SDH activity in SF-exposed cells is mostly due to functional inhibition of SDH by SF. While the precise mechanism for the functional SDH inhibition by SF is not fully clear at present, the following factors may contribute to the observed inhibition: i. Direct inhibition of cytochrome oxidase (Complex IV) by SF may lead to ubiquinol accumulation and create an overly-reduced coenzyme Q pool, which would diminish the thermodynamic driving force for SDH-catalyzed oxidation of succinate⁵⁶. ii. Complexes I, III and IV are known to form supercomplexes to streamline electron transfer^57,58. Dysfunction of Complex IV in SF-exposed cells can destabilize the supercomplex assemblies and thus may indirectly impair electron entry from SDH. Additionally, when Complex IV is compromised, the upstream ROS production from Complex II will increase, and the elevated ROS may oxidize the Fe–S clusters in SDH, further impairing its catalytic function⁵⁹. iii. Decreased MMP in SF-exposed cells will promote oxaloacetate accumulation, which then inhibits SDH; in comparison, high MMP enhances reverses electron flow, maintains NADH reduction, lowers oxaloacetate levels, and thus relieves SDH inhibition⁶⁰.

Here, it is also of note that MTT is reduced to purple protonated imidazole crystals (formazan) in live cells by SDH. The MTT crystals deposit inside the mitochondria, and the degree of deposition reflects the mitochondrial metabolic activity of a cell. The ability of SF to directly inhibit the mitochondrial SDH is in accord with the observations that there is a bigger reduction in the MTT activity in SF-treated cells than the reduction in cell density.

Fourth, based on electron microscopy analysis, 24-h SF exposure causes severe mitochondrial impairment, including mitochondrial swelling, loss of cristae, and disruption of the mitochondrial membranes. Increased mitochondrial autophagy (mitophagy) is also observed in this study, likely indicating that SF-exposed cells are trying to remove the damaged mitochondria. In agreement with our observations, it was reported earlier that SF-exposed cells had an elevated autophagy activity along with LC3II upregulation⁵¹.

Lastly, it is observed in this study that there is a marked increase in the apparent mitochondrial density in SF-treated cells, and the increase is dependent on SF concentrations. It is believed that this increase in mitochondrial density is a compensatory feedback response, due to deficiency in the normally-functioning mitochondria in SF-treated cells. In support of this speculation, we found that SF-treated cells exhibit a marked reduction in three mitochondrion-related proteins, namely, TFAM, PGC-1α and NRF1. PGC-1α is a transcriptional coactivator, which regulates mitochondrial functions, including oxidative phosphorylation and ATP production⁵⁴. A reduction in PGC-1α level would suggest a reduced mitochondrial function. Similar observations were also made in vivo showing that SF administration reduced PGC-1α and ATP levels in the intestinal tissue⁵⁵. TFAM, which located in the mitochondrial matrix, is essential for maintaining and transcribing mitochondrial DNA (mtDNA), and a reduction in its levels also similarly reflects a reduced mitochondrial function.

In summary, SF disrupts mitochondrial function through multiple mechanisms, ultimately leading to severe impairment of mitochondrial energy metabolism. SF-induced opening of mPTP results in the loss of MMP and dissipation of the proton gradient, which prevents ATP synthase through the proton gradient. Additionally, as SDH is a bridge between the TCA cycle and ETC, inhibition of SDH by SF impairs the function of the TCA cycle and thus adds to cellular energy crisis. Moreover, the loss of membrane potential may lead to leakage of electrons from ETC, which, in turn, reacts with oxygen to produce excess ROS, further damaging the mitochondria and exacerbating cellular dysfunction.

SF alters cell viability, proliferation and functions

The results of this study showed that SF at relatively lower concentrations exhibits relatively mild cytotoxicity in cultured HT22 hippocampal neurons, but its cytotoxicity clearly follows the concentration- and time-dependent patterns. Importantly, the relatively mild cytotoxicity of SF can accumulate with prolonged exposure, suggesting that even low concentrations of SF may cause cumulative cytotoxicity over time.

Cell cycle analysis showed that SF at relatively low concentrations can inhibit cell proliferation by prolonging the S phase of the cell cycle. It was confirmed by the EdU incorporation experiments that the DNA synthesis in SF-exposed cells is inhibited. Under normal conditions, cells would increase their mitochondrial numbers and function during the G₁ and S phases to meet the high energy demands of DNA replication during cell division. Therefore, it is speculated that the energy shortage in SF-exposed cells is a major factor contributing to the observed inhibition of cell proliferation. Additionally, the accumulation of ROS in SF-exposed cells may damage genetic materials and thus slows down DNA replication, which is another factor contributing to reduced cell proliferation. Accordingly, the extended cell cycle may represent an adaptive response to SF-induced energy crisis, which would help the cells to survive under unfavorable conditions.

However, at higher concentrations (≥ 50 mM), SF consistently induces cell death and reduces cell density. Notably, earlier studies have also reported that exposure of retinal photoreceptor cells⁵¹ or primary dissociated mouse cerebrocortical cells⁵⁸ to SF induced time- and concentration-dependent toxicity. Interestingly, SF at 20–60 mM primarily affects large polygonal neurons, but at higher concentrations (≥ 120 mM), it can induce nonspecific cytotoxicity¹⁴. The results of this study clearly show that SF at higher concentrations (50 and 100 mM) can induce apoptotic cell death, which can be effectively rescued by z-VAD-FMK, a pan-caspase inhibitor. This observation is in line with an earlier observation reporting that while Bcl-2 expression was reduced, Bax expression was increased in SF-exposed retinal photoreceptor cells, along with increased cleavage of caspases⁵¹.

Previous studies have shown that SF can increase ROS production^61,62. In this study, we have carefully examined the effect of SF on cellular ROS, lipid-ROS and NO levels, focusing on the effects of relatively lower concentrations of SF (12.5 and 25 mM) since cell numbers are not significantly affected under these conditions. Our results show that after 48-h SF exposure, cellular total ROS/lipid-ROS levels are significantly elevated, but this increase is not observed in the first 24 h. In comparison, SF-induced mitochondrial ROS accumulation occurs earlier compared to that of cellular ROS/lipid-ROS. As the mitochondria are the primary source of intracellular ROS^63,64, we speculate that the inhibition of the mitochondrial cytochrome oxidase by SF is the initial cause of ETC dysfunction, which then increases electron leakage and mitochondrial ROS production. With mPTP opening, mitochondrial ROS can further leak into the cytoplasmic compartment, increasing total cellular ROS levels.

GSH is one of the primary antioxidants in the cells, responsible for neutralizing ROS generated during mitochondrial respiration and preventing oxidative stress⁶⁵. It is observed that 24-h SF exposure causes significant reduction in cellular GSH levels, likely due to its consumption in neutralizing the mitochondrial ROS being produced. Cellular GSH inadequacy together with elevation of cellular ROS/lipid-ROS and NO levels in SF-exposed cells will create a state of chronic oxidative stress, contributing to the observed impairment of membrane integrity. Somewhat in line with this notion, it was reported earlier that significant leakage of LDH (which reflects cellular membrane integrity) only occurs when higher concentrations (≥ 60 mM) of SF are present⁵². Given the crucial role of GSH in maintaining mitochondrial function and regulating apoptosis, the reduction of GSH cellular levels may be a factor promoting mitochondria-mediated apoptosis, and it is possible that this reduction may also facilitate ferroptosis/necroptosis under certain conditions. Additionally, the reduced GSH levels may further impair a cell’s ability to counteract oxidative stress, thus creating the potentials for a protracted vicious cycle.

In summary, the cytotoxic effects of SF in HT22 mouse hippocampal neurons involve multiple mechanisms. SF primarily induces mitochondrial dysfunctions (e.g., SDH inhibition, reduced ATP production, increased mPTP opening and decreased MMP) and mitochondrial structural damage. These mitochondrial impairments coupled with cellular energy deficit will disrupt DNA synthesis and cell proliferation. In the meantime, mitochondrial damage will lead to elevated cellular ROS/lipid-ROS and NO levels, which will amplify oxidative stress and further damage mitochondrial and cytoplasmic membranes, exacerbating mitochondrial dysfunction and cellular injury and apoptosis.

Insights gained from transcriptomic analysis

The results from transcriptomic analysis not only guides us in our search for SF’s cellular targets, but the data also shed lights on our understanding of the underlying mechanisms of SF’s cytotoxicity. The observed changes in mitochondrial function and ATP synthesis align closely with the results of transcriptomic analysis of SF-treated cells. We found that many genes in the pathways related to oxidative phosphorylation, the TCA cycle, and diabetes are altered, which are predominantly associated with mitochondrial complex deficiencies.

First, at the transcriptional level, several key genes involved in mitochondrial function are down-regulated, particularly those related to mitochondrial respiratory chain complexes. For instance, genes responsible for the assembly of complex I, such as Ndufv3, Ndufb7 and Ndufb11, are down-regulated by ~ 20% in 25-mM SF group. This down-regulation likely disrupts complex I assembly, impairing the coupling between electron transport and ATP synthesis. Since complex I is crucial for NADH oxidation and proton translocation, its dysfunction reduces NADH dehydrogenase activity, which, in turn, diminishes proton pumping along the electron transport chain, directly impairing oxidative phosphorylation and ATP synthesis efficiency.

Additionally, the genes associated with complex V (F-ATPase and V-ATPase), such as Atp5mf, Atp6v1a, Atp6v1g1, Atp6v0a2, Tcirg1 and Atp6v0d1, are down-regulated by 10 − 25%. Complex V is essential for ATP synthesis by utilizing the proton gradient generated by the electron transport chain⁶⁶. The down-regulation of these genes reduces the proton-translocating ATPase activity and binding efficiency, further compromising mitochondrial ATP production. Given that complex V is the final step in ATP synthesis, its impaired function would severely limit a cell’s ability to generate ATP, thus exacerbating the energy deficit.

The down-regulation of Camk2a (calcium/calmodulin-dependent protein kinase IIα) in SF-exposed cells may further compound the energy deficit. Camk2a is involved in calcium signaling and in the regulation of glucose and lipid metabolism, and its reduced expression may impair energy balance, affecting ATP production and increasing susceptibility to metabolic stress. This impairment of metabolic regulation, combined with mitochondrial dysfunction, may exacerbate the energy crisis in SF-exposed cells.

The gene product of Pdk4 regulates the conversion of pyruvate to acetyl-CoA, a critical step in fueling the TCA cycle and oxidative phosphorylation⁶⁷. The observed down-regulation of Pdk4 may reduce the efficiency of this metabolic pathway, contributing to the energy deficit.

Second, Vdac3 is involved in the regulation of calcium ion transport across the mitochondrial membrane ⁶⁸. SF-induced upregulation of Vdac3 (a gene often induced by mitochondrial ROS) may indicate an attempt to mitigate calcium dysregulation (such as calcium overload). In addition, mitochondrial calcium overload can trigger mPTP opening, contributing to mitochondrial damage and ROS accumulation⁶⁹.

Notably, the combined effects of down-regulation of mitochondrial metabolic regulators like Camk2a and Pdk4 and the upregulation of Vdac3 suggest that SF exposure disrupts both mitochondrial ATP synthesis and other metabolic processes, leading to a compounded energy deficit, calcium dysregulation, and increased cellular stress.

Third, in diabetic patients, the combined effects of oxidative stress and inflammation promote pathological vascular proliferation, characterized by abnormal endothelial cell proliferation and neovascularization. This process is closely associated with the dysregulation of several key genes identified in our analysis. For example, the upregulation of Myc following SF exposure indicates that cells are under proliferative stress. Myc is a well-known oncogene, and its upregulation is commonly associated with increased cellular proliferation pressure and metabolic changes.

At the same time, the down-regulation of cell cycle regulatory genes Cdkn2c, Cdkn2d, and Cdkn1a, along with the upregulation of Stat3 (its product is a nuclear transcription activator responsive to cytokines and growth factors). As Stat3 upregulation is commonly associated with cell survival and proliferation, it likely indicates that cells are adapting to unfavorable environmental conditions by sustaining their proliferative capacity through compensatory mechanisms.

In diabetic vascular proliferation, inflammation is often a precursor to abnormal angiogenesis, with inflammatory responses playing a critical role in initiating and sustaining the process. This study highlights key gene expression changes that align with this mechanism, particularly the upregulation of Mmp9 (matrix metallopeptidase 9), which is associated with tissue remodeling, wound repair, and even tumor invasion. Mmp9 encodes a proenzyme that, once activated, becomes a zinc-dependent endopeptidase capable of degrading collagens such as types IV, V, and XI⁷⁰. This degradation reshapes the extracellular matrix (ECM), facilitating angiogenesis by creating space for new blood vessels.

Further supporting the link between inflammation and angiogenesis, the upregulation of Col1a1 (collagen type Iα1) and Col4a2 (collagen type IVα2), coupled with the down-regulation of Col4a6 (collagen type IVα6), suggests a shift in collagen production from type IV to type I. This remodeling of the ECM is characteristic of the microenvironments which promote new blood vessel formation, such as in wound healing and diabetic complications. The upregulation of Vegfa and Vegfd, two important factors in angiogenesis, would further enhance this response, ensuring adequate blood supply for biological processes such as tissue repair or regeneration.

Moreover, the upregulation of inflammatory mediators such as Serpine1 and Edn1 in SF-treated cells highlights their roles in coordinating inflammatory reactions that not only contribute to tissue repair but also promote angiogenesis. These factors are extensively studied in the context of injury repair and cancer, where inflammation drives both the degradation of existing tissue and the formation of new vasculature to support growing or regenerating tissues. In diabetic patients, a similar mechanism may be at play, where inflammation triggers vascular remodeling and excessive angiogenesis, leading to pathological changes in blood vessels⁷¹.

Conclusions

The results of our present study reveal that SF can disrupt mitochondrial function through multiple mechanisms, leading to severe impairment of cellular energy metabolism. Under SF-induced oxidative stress, the opening of mPTP results in the loss of MMP and dissipation of the proton gradient, thereby halting ATP synthesis and potentially triggering mitochondria-mediated cell death. The weakening or loss of MMP prevents ATP synthase from utilizing the proton gradient to generate ATP, compromising the cell’s energy supply. Moreover, the loss of membrane potential may also lead to the leakage of electrons from the ETC, which, in turn, reacts with oxygen to produce excessive ROS. These ROS can further damage the mitochondria, creating a potentially vicious cycle that exacerbates cellular dysfunction and may ultimately lead to apoptotic cell death.

Collectively, the results of this study highlight the multi-faceted toxicity of SF in neuronal cells and underscores mitochondria as the crucial cellular organelles targeted by SF which ultimately disrupts mitochondrial energy metabolism and cellular survival. This study also offers an explanation for the hypothesis that cellular accumulation of formic acid resulting from hyperglycemia may contribute importantly to the induction of certain diabetic complications (such as diabetic neuropathy and retinopathy) in humans.