Targeting prominin-2/BACH1/GLS pathway to inhibit oxidative stress-induced ferroptosis of bone mesenchymal stem cells

Abstract

Suppressing bone mesenchymal stem cell (BMSC) ferroptosis is expected to optimize BMSCs-based therapy for intervertebral disc degeneration (IVDD). Our previous study revealed that Prominin-2 could protect against ferroptosis by decreasing cellular Fe²⁺ content and inhibiting transcription regulator protein BACH1 (BACH1) expression. In this study we probed the molecular mechanisms underlying the Prominin-2/BACH1 pathway in BMSC ferroptosis. Using an array of in vitro and in vivo experiments we found that heat shock factor protein 1 (HSF1) activates PROM2 (encoding protein Prominin-2) transcription and elevated Prominin-2 expression. Furthermore, we showed that Prominin-2 attenuates ferroptosis induced by tert-butyl hydroperoxide (TBHP) through promoting BACH1 ubiquitination and degradation. Inhibition of BACH1 expression reversed TBHP-stimulated down expression of glutaminase kidney isoform, mitochondrial (GLS), which plays a crucial role in protecting BMSCs against ferroptosis. Targeting the Prominin-2/BACH1 axis has also been shown to improve BMSC survival post-transplantation and mitigate IVDD progression by inhibiting ferroptosis. Our results support a new mechanistic insight into the regulation of the Prominin-2/BACH1/GLS pathway in BMSC ferroptosis. These finding could lead to potential therapeutic targets to improve the survival of engrafted BMSCs under oxidative stress circumstances.

Background

It has been widely recognized that intervertebral disc degeneration (IVDD) predominantly causes chronic low back pain and disability [1]. Up to now, supplementation of transplanted bone mesenchymal stem cells (BMSCs) has manifested therapeutic efficiency on IVDD in basic and preclinical science research [2, 3]. Transplantation of BMSCs into degenerative intervertebral discs (IVDs) exerts therapeutic effects via a diversity of mechanisms, e.g., reducing functional cells’ death, differentiating into nucleus pulposus-like cells, lessening the expression of extracellular matrix, suppressing the release of pro-inflammatory cytokines [2]. Despite significant achievements in this area, unsolved weaknesses still hinder the therapeutic benefits of BMSCs. Among these restrictions, inferior BMSC retention in the oxidative stress (OS) microenvironment is a significant cause of BMSCs’ loss and inadequate therapeutic advantages after transplantation into degenerative IVDs [4, 5]. Therefore, it is urgently required to develop suitable strategies to enhance BMSC retention in the OS microenvironment of degenerative IVDs.

It is known that OS participates in the pathogenesis of IVDD, which is attributed to redox unbalance [6]. The process of IVDD increased reactive oxygen species (ROS) production and added to the harsh and complex microenvironment of degenerative IVDs. When withstanding OS exceeds their maximal adaptability range in IVDs, engrafted BMSCs will undergo an inevitable cell death procedure, instead of stably exerting therapeutic effects. Hence, improving the survival rate of BMSCs in OS circumstances is a crucial way to enhance their retention and fortify therapeutic benefits. We have previously elaborated on the molecular mechanisms of ferroptosis in mesenchymal stem cells (MSCs), emphasizing that it is an iron-reliant, OS-induced regulated cell death [4]. The persistent state of OS would overwhelm BMSCs’ antioxidant defense after being transplanted into degenerative IVDs, eventually bringing out oxidative disruptions of the intracellular microenvironment, thereby initiating ferroptosis. Our prior experimental results confirmed that ferroptosis is mainly responsible for BMSCs’ low survival in the early stages of in vitro ROS stress or after implantation into degenerative IVDs [5]. Thus, suppressing BMSC ferroptosis is expected to enhance their preservation in degenerative IVDs’ OS microenvironment and optimize BMSCs-based therapy.

In ferroptotic situations, ferrous iron (Fe²⁺) overload triggers ROS generation via the Fenton Reaction [7]. Excessive ROS then induces OS and initiates lipid peroxidation (known as OS biomarker), eventually driving ferroptosis [8]. Under cellular homeostasis, iron levels are coordinately regulated by ferritin (iron storage protein) and other iron transporters [9]. BMSCs exposed to OS showed differential expression of cellular iron storages and transporters [10, 11], which dysregulate iron homeostasis and make BMSCs more sensitive to ferroptosis. So, impelling iron efflux to reduce Fe²⁺ acceleration is vital to restrict ferroptosis. Prominin-2, a pentaspanin protein, was verified to prevent intracellular Fe²⁺ accumulation in answer to ferroptotic stress by transporting iron extracellular [5, 12]. We identified the differentially expressed PROM2 (encoding protein Prominin-2) associated with OS in BMSCs through bioinformatics analysis. The lack of PROM2 in BMSCs under OS sensitizes BMSCs to ferroptosis through exacerbating lipid ROS deposition [5]. In contrast, overexpression of Prominin-2 mitigated the sensitivity of tert-butyl hydroperoxide (TBHP)-induced ferroptosis in BMSCs [5]. Accordingly, Prominin-2 is supposed to be a novel regulatory factor for ferroptosis and a potential target for improving BMSCs’ transplantation efficiency. We previously found that Prominin-2 could also inhibit BTB and CNC homolog 1 (BACH1) protein expression, known as an OS-responsive regulator and could suppress antioxidant genes to disorder redox homeostasis [4, 5]. However, the molecular mechanisms that underlie the regulation of the Prominin-2/BACH1 pathway in BMSC ferroptosis remain unknown. In our study, we intended to probe the Prominin-2/BACH1 signaling pathway in regulating BMSC ferroptosis under OS circumstances. We would like to provide a new research basis for improving BMSCs’ survival against OS.

Materials and methods

BMSCs’ isolation and incubation

Based on our previous studies [5], we extracted primary BMSCs from Sprague–Dawley (SD) rats’ femur and tibia (4 weeks old, 150–180 g weight, male). We purchased the 4-week-old male SD rats (150–180 g weight, specific pathogen-free (SPF)) from the Animal Research Center of Shanghai (China). We fed them in the Southeast University Laboratory Animal Centre (SPF environment). SD rats were housed in standard conditions (22 to 25 °C, relative humidity of 45% to 55%, 12 h of light/dark cycle). Standard mouse chow and drinking water were offered ad libitum. The specific isolation and culture methods were previously described [5]. We expanded the primary BMSCs between passages 3 and 6. We identified the cell morphology and surface markers and assessed the multilineage differentiation capability of BMSCs. Under the light microscope (Olympus, IX73, Japan), BMSCs typically appear elongated, spindle-shaped, or irregular, with relatively small cell sizes (Fig. SA). The BMSCs also preserved their potential to differentiate into adipocytes, osteoblasts, and chondrocytes, demonstrating their capacity for multilineage differentiation (Fig. SB). The results illustrated that BMSCs showed positive Alizarin Red, Oil Red O, and Alcian Blue staining. Finally, we determined the surface markers of BMSCs. As illustrated in Fig. SC, these BMSCs expressed the stem cell markers CD73 and CD90, while they did not express CD45, the hematopoietic stem cell marker. Above all, those results verified that the extracted BMSCs retained their mesenchymal stem cell identity. Our experimental protocols were permitted by Southeast University’s Animal Experimental Ethical Inspection Committee (Number: 20230220017). The work has been reported in line with the ARRIVE guidelines 2.0.

Chondrogenesis, adipogenesis, and osteogenesis evaluation

We plated BMSCs into the plate (24-well). When the BMSCs reached 80%, they were subjected to adipogenic or osteogenic differentiation induction. The BMSCs were cultured in a complete medium enriched with L-ascorbic acid (50 μM, Table 1), β-glycerophosphate (10 mM, Table 1), dexamethasone (10⁻³ μM, Table 1), and penicillin–streptomycin (Table 1) for osteogenic induction. The BMSCs were cultured in a complete medium enriched with dexamethasone (0.1 μM, Table 1), insulin (5 μg/mL, Table 1), indomethacin (50 μM, Table 1), penicillin–streptomycin (1%, Table 1) for adipogenic induction. Subsequently, following the manufacturer’s manuals, we stained the BMSCs using an Alizarin Red S Staining Kit (Table 1) to estimate osteogenic differentiation. Following the manufacturer’s manuals, we estimated the adipogenic differentiation by staining BMSCs with an Oil Red O dye solution (Table 1). Finally, we cultured BMSCs for 2 weeks in a medium enriched with insulin (500 ×, Table 1), ITS.T (1,000 ×, Table 1), and sodium selenite (10,000 ×) for chondrogenic induction.

Table 1 Reagent table

Flow cytometry

First, we harvested BMSCs, washed them twice with 1 × phosphate-buffered saline (PBS) by centrifugation at 350–500 × g for 5 min each time and discarded the supernatant. We then aliquoted BMSC samples to tubes or wells at a cell density of 1 × 10⁶ cells in 100 µL of 1 × Flow Cytometry Staining Buffer. We added the recommended amount of primary antibody and incubated for 20–40 min at 4 °C in the dark. We washed the BMSCs with 1 × PBS by centrifugation at 350–500 × g for 5 min and discarded the supernatant. We resuspended the BMSCs in 200–500 µL of 1 × Flow Cytometry Staining Buffer and analyzed them on a flow cytometer. BMSCs were identified with a FACScan flow cytometer (Becton Dickinson, United States, Accuri C6). We used the Flow cytometry to examine the BMSCs using fluorescein. Conjugate CoraLite® Plus 488-conjugated (anti-CD73, anti-CD90, and anti-CD45) antibodies (Table 1). Results were determined by flow cytometry with FlowJo_10.8.1 software for quantification.

Inhibitor studies

TBHP (50 μM), ferrostatin-1 (Fer-1, 3 μM), deferoxamine (DFO, 100 μM), Bis-2-(5-phenyl acetamido-1,2,4-thiadiazol-2-yl) ethyl sulfide (BPTES, 5 μM), Hemin (20 μM) were diluted with BMSC culture conditioned medium (Table 1). We cultivated BMSCs with TBHP in the presence or absence of Fer-1, DFO (both suppressing ferroptosis), or BPTES (suppressing Glutaminase kidney isoform, mitochondrial (GLS) expression) for 24 h.

After 12 h of cell culture in a complete medium containing TBHP, BMSCs received cycloheximide (CHX, 100 μg/mL) treatment for the indicated times and 20 μM (Carbobenzoxy-L-Leucyl-L-Leucyl-L-Leucinal, MG132) and chloroquine (CQ, 15 μM, Table 1) treatment for 6 h (Table 1). We collected the supernatant and washed BMSCs thrice with PBS after the intervention.

Quantitative reverse transcription-polymerase chain reaction

We assessed heat shock factor protein 1 (HSF1), PROM2, BACH1, and GLS transcript levels by real-time reverse transcription-polymerase chain reaction (RT-qPCR). We extracted BMSCs’ total RNA by utilizing the TRIzol reagent (Table 1). We synthesized complementary DNA from RNA by using an RT-qPCR assay kit (Table 1). We sequenced selected products by using an ABI PRISM 7500 Genetic Analyzer (Table 1). The 2^−ΔΔCT method has been used to quantify gene expression. Our results were normalized by glyceraldehyde-3-phosphate dehydrogenase. Primers for HSF1: frontward 5’-CCAGGAACTGGAAAGGCACTA-3’, reverse 5’-CAGTGGTTGGTCCCAGTCTT-3’. Primers for PROM2: frontward 5’-GCTCAGGAACCCAAACCTGT-3’, reverse 5’-GGCAGGCCATACATCCTTCT-3’. Primers for BACH1: frontward 5’-GCCCGTATGCTTGTGTGATT-3’, reverse 5’-CGTGAGAGCGAAATTATCCG-3’. Primers for GLS: frontward 5’-AGGGTCTGTTACCTAGCTTGG-3’, reverse 5’-ACGTTCGCAATCCTGTAGATTT-3’.

Western blot

We conducted immunoblotting as previously described (Table 1) [5]. We analyzed proteins with antibodies recognizing HSF1, Prominin-2, BACH1, Ubiquitin (NT) Rabbit Polyclonal Antibody, GLS, long-chain acyl-CoA synthetase 4 (ACSL4), prostaglandin-endoperoxide synthase 2 (PTGS2), Glutathione peroxidase 4 (GPX4), Ferritin, Histone-3, β-Actin, HRP conjugated goat anti-rabbit IgG (H + L), matrix metalloproteinase-9 (MMP-9), A disintegrin and metalloproteinase with thrombospondin motifs 5 (ADAMTS5), and matrix metalloproteinase-13 (MMP-13) (Table 2). We detected bands by using luminescent reagents (Table 1). We utilized ImageJ (version 1.8.0; National Institutes of Health) to perform quantification.

Table 2 Primary and secondary antibodies for Western Blot

Transcription factor-binding sites prediction

We inquired into the potential transcription factor-binding sites (TFBSs) of PROM2 based on the JASPAR database (http://jasper.genereg.net/). We first searched the promoter regions of PROM2 based on the web tool “Gene Expression Omnibus public database (https://www.ncbi.nlm.nih.gov/geo/)”. The sequence region within the range of 2000 bp upstream to 100 bp downstream of the starting gene point is generally thought to be the promoter region [13]. We then predicted the TFBSs of PROM2 by using the web-based tools “JASPAR” and “UCSC (http://genome.ucsc.edu)”. We set the relative profile score threshold at 85%.

Identification of differential gene expression

We downloaded BMSCs’ data from the Gene Expression Omnibus public database (GEO; www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35955). Due to BMSCs being vulnerable to excess OS in the aging microenvironment [14], we determined differentially expressed up and down-regulated genes among elderly group versus control group. Four aged BMSCs samples were collected (GSM878100, GSM878101, GSM878102, GSM878103) (Elderly group). Five control samples were collected (GSM878095, GSM878096, GSM878097, GSM878098, GSM878099) (Control group). In our study, the control group consisted of 5 individuals (4 Females and 1 Male) aged 57.6 ± 9.56 years, while the elderly group included 4 individuals (3 Females and 1 Male) aged 81.75 ± 4.86 years. We listed the information of the human subjects (including number of donors, age and gender) in Table 3.

Table 3 Human subjects’ characteristics

Lentiviral transfection

GenePharma (China) manufactured lentivirus packaging systems to carry HSF1 gene and PROM2 gene, respectively. Following the manufacturer’s manuals, we transfected BMSCs with lentiviral vectors carrying the HSF1 gene (LV-HSF1) and PROM2 gene (lentiviral vectors carrying PROM2 gene, LV-PROM2) utilizing the transduction enhancer polybrene (5 μg/ml). Twenty-four hours later, we replaced the medium and continued to culture BMSCs for two days.

Short hairpin RNA (ShRNA) lentiviruses against BACH1 were also manufactured by GenePharma (China). Sh-BACH1: 5’-TGTCTGTCTGTATGATGGGCATG-3’. Sh-negative control: 5’-GCATCGCATTGTCTGAGTAGGTG-3’. We transfected constructed lentivirus into BMSCs (Multiplicity of infection (MOI) = 90) when they extended to 70% confluence. We changed the culture medium after 2 days of transfection.

Chromatin immunoprecipitation-qPCR assay

We performed the chromatin immunoprecipitation-qPCR (CHIP-qPCR) analysis in light of the manufacturer’s manuals (Table 1). We first used control and HSF1 antibodies to precipitate DNA and then performed qPCR to detect the PROM2 promoter (Region 1-F GTGATGGGTCAGCGAGATGT, Region 1-R TTGGCACCCACACCTTTACC; Region 2-F CCAGATTGGCCTAGCAGAGG; Region 2-R TAAGACCAGACCGGCAAACC).

Immunofluorescence assay

We first used 4% paraformaldehyde to fix BMSCs for 15 min then 0.2% Triton X100 to permeabilize BMSCs. After blocking BMSCs with 10% goat serum, GLS antibody incubation was applied to BMSCs overnight at 4 degrees Celsius. Immunofluorescence with GLS antibody was then used for BMSCs detection (Table 2). We analyzed immunofluorescence under a confocal microscope (LEICA, STELLARIS STED, Germany). We utilized ImageJ (version 1.8.0; National Institutes of Health) to perform quantification.

Cell proliferation examination

We cultured BMSCs in a plate (96-well). We followed the manufacturer’s manuals to perform the cell counting kit-8 (CCK-8) examination (Table 1). We recorded the optical absorbance (450 nm) by the microtiter plate reader and converted it into cell survival rate. We also conducted cell proliferation detection with 5-ethynyl-2’-deoxyuridine (EdU) (Table 1). Briefly, we treated BMSCs with EdU staining buffer for two hours and then fixed them with 4% polyformaldehyde. We captured graphics under a fluorescence microscope (Olympus, IX73, Japan). ImageJ (version 1.8.0; National Institutes of Health) was utilized to perform quantification.

Lactate dehydrogenase release

The specific methods to measure lactate dehydrogenase (LDH) release were previously described [5] (Table 1). We recorded the optical absorbance (450 nm) by the microtiter plate reader and converted it into cell cytotoxicity rate.

ROS examination

According to our previous methods [5], we used 2’−7’-dichlorofluorescein diacetate (1:1000, 20 μmol/L) to measure ROS levels after TBHP treatment (Table 1). We captured graphics under a fluorescence microscope (LEICA, STELLARIS STED, Germany). We utilized ImageJ (version 1.8.0; National Institutes of Health) to perform quantification.

Iron Assay

According to the manufacturer’s instructions and our previous study [5], an Iron Assay kit was utilized to measure iron levels (Table 1). We recorded the optical absorbance (593 nm) by the microtiter plate reader. We calculated the cellular Fe²⁺ contents according to the standard curve.

Malondialdehyde levels

According to the manufacturer’s instructions and our previous study [5], an malondialdehyde (MDA) assay kit was utilized to measure the MDA levels (Table 1). We recorded the optical absorbance (532 nm) by the microtiter plate reader.

Transmission electron microscopy

After stimulation with TBHP, we fixed BMSCs at 2.5% ice‐cold glutaraldehyde overnight followed by osmium tetroxide post-fixation. We first dehydrated BMSCs with different alcohol concentrations and then rinsed them with propylene oxide and impregnated them with epoxy resin. We performed electron microscopy to monitor mitochondrial morphological changes and the occurrence of ferroptosis contrasted with uranyl acetate and lead citrate (HITACHI, H-7650 C, Japan).

Measurement of superoxide dismutase

We obtained the supernatant from the BMSCs and measured its absorbance at 560 nm based on the NBT reaction. The Total Superoxide Dismutase Assay Kit with NBT is a colorimetric assay based on the NBT reaction used to detect the activity of superoxide dismutase (SOD) in cells. By utilizing a Total Superoxide Dismutase Assay Kit with NBT (Table 1), we took all the steps in strict accordance with the manufacturer’s protocol.

Glutamate assay

We monitored glutamate uptake by using the Glutamate Assay Kit (Table 1). Briefly, BMSCs were either left untreated or treated for 24 h with TBHP and BPTES in a complete medium. According to the manufacturer’s manuals, we quantified the glutamate levels in the medium. We drew the standard curve by utilizing serial dilutions of glutamate.

Glutathione assay

According to the manufacturer’s instructions and our previous study [5], a glutathione (GSH) analysis kit was utilized to measure GSH levels (Table 1). We recorded the optical absorbance (412 nm) by the microtiter plate reader.

Establishment of rats’ lumbar IVDD model

According to our previous study [5], we executed this manipulation in SD rats (4 weeks old, 150–180 g weight, male). The rats were anesthetized with intraperitoneal injection of 2% (w/v) pentobarbital (50 mg/kg; C₁₁H₁₇N₂NaO₃). We punctured the exposed lumbar IVDs (5 mm). We rotated the needles 360° to achieve a full-thickness puncture. The needles were then pulled out five minutes later.

After one week, 5 μL PBS, 5 μL negative control lentivirus transfected BMSCs (LV-NC) (1 × 10⁴ cells/μL), and 5 μL BMSCs transfected by LV-PROM2 (1 × 10⁴ cells/μL), supplemented with or without BACH1 inhibitor (Hemin, 20 μM), were evenly administered into the IVDs via a 31-gauge needle. The injections were implemented every 2 weeks for 2 months. We also amputated the forelimbs and raised the food trough to create a lumbar IVD persistent degeneration model.

Evaluation of BMSCs’ engraftment

We transfected BMSCs with LV-PROM2 stably expressing the enhanced green fluorescent protein (EGFP). EGFP immunostaining and EGFP mRNA levels were utilized to evaluate engrafted BMSCs’ retention rate in IVDs (Anti-EGFP rabbit pAb; Table 2). We euthanized the rats 24 h after the BMSC injection, which was by institutional guidelines and approved protocols. The trained personnel performed the euthanasia by using CO₂ inhalation. The selected approach guaranteed swift and humane euthanasia, causing minimal suffering and discomfort to the rats. We assessed the samples under a confocal microscope (LEICA, STELLARIS STED, Germany). ImageJ (version 1.8.0; National Institutes of Health) was utilized to perform quantification.

Radiographic analysis

The IVDD progression was assessed post-surgery by X-ray radiography (4 weeks and 8 weeks). The rats were placed in the prone position. We utilized ImageJ (version 1.8.0; National Institutes of Health) to measure lumbar IVDs’ heights. Disc height index (DHI) (%) represented lumbar IVDD according to our previous study [5].

Histopathological analysis

According to our previous study [5], we performed Safranin O-fast green (S–O) (Table 1) and hematoxylin and eosin (H&E) (Table 1) staining. The specific methods were previously described [5]. We captured graphics under a light microscope (Olympus, IX73, Japan). Based on our previous study [5], we evaluated histological S–O staining and histological H&E staining.

Statistical analysis

All experiments conducted in our manuscript were repeated at least five times. We considered differences as statistically significant when p values < 0.05. We expressed all data by means ± standard deviation. GraphPad Prism 8 fulfilled statistical analyses. The differences between the two groups were evaluated via the unpaired student’s t-test. The comparison of diverse groups was executed by one-way analysis of variance (ANOVA) with Tukey’s multiple comparison post hoc test.

Results

HSF1 overexpression activates the transcription of PROM2 in BMSCs under OS

Cellular adaptive response to OS could activate HSF1 expression, which functions as a regulator in cellular redox homeostasis and a defender against OS-induced ferroptosis [15]. However, persistent TBHP exposure decreased HSF1 mRNA and protein expressions (Fig. 1, A–E). Furthermore, ferroptosis inhibitors could abate TBHP-induced reductions in HSF1 protein expression, which verified that HSF1 downregulation could lead to TBHP-induced ferroptosis in BMSCs. HSF1 nuclear localization, which represented its activation [16], was also significantly weakened in BMSCs under OS but could be mitigated by DFO and Fer-1 (Fig. 1, B–E). Given that HSF1 acts as a crucial transcription factor for activating gene transcription following extracellular stress and HSF1 activates anti-ferroptosis gene transcription by binding to its promoter region [15, 17]. Therefore, we investigated whether HSF1 could achieve the anti-ferroptosis effect by regulating PROM2 transcription. We predicted that HSF1 binds specifically to the TFBSs of PROM2 promoter regions and also displayed the base sequence of possible HSF1-binding sites (Fig. 1, F–G). The significantly differentially down-expressed HSF1 and PROM2 were visualized by heatmaps (Fig. 1H). We then performed the CHIP-qPCR (Fig. 1, I-J). The enrichments of anti-HSF1-IP pull-down on the PROM2 promoter fragments (Region 1 and Region 2) in BMSCs were determined using the ChIP assay kit. This result becomes more pronounced in HSF1 overexpressed BMSCs (LV-HSF1). All these results indicated that HSF1 could bind to the PROM2 promoter.

Fig. 1

Tert-butyl hydroperoxide (TBHP) exposure-decreased heat shock factor protein 1 (HSF1) expression in bone mesenchymal stem cells (BMSCs) could be restored by ferroptosis inhibitors. A total of 1 × 10⁶ BMSCs were used in each group. A BMSCs were incubated with PBS or TBHP (50 μM) for 24 h. Then BMSCs were harvested and HSF1 mRNA levels were determined by RT-PCR. Data were normalized to the levels of GAPDH (n = 5). B and D BMSCs were incubated with TBHP (50 μM) in the presence of DFO (100 μM) or Fer-1 (3 μM) for 24 h. Then, BMSCs were harvested, and cytoplasmic HSF1 and nuclear HSF1 were evaluated by Western Blot (n = 5). Histone-3 was used as a loading control for nuclear fractions. C and E Quantitative analysis of the protein levels of cytoplasmic HSF1 and nuclear HSF1. F The base sequence of possible HSF1-binding sites of the PROM2 promoter region was shown based on the JASPAR database. The relative profile score threshold was set at 85%. G Detail of the results window: HSF1 binding sites were shown in blue using the web-based tools “JASPAR” and “UCSC”. H Heatmap showing differentially expressed genes among the control group (GSM878095, GSM878096, GSM878097, GSM878098, GSM878099) and BMSCs under the oxidative stress (OS) group (GSM878100, GSM878101, GSM878102, GSM878103). BMSCs’ data were downloaded from the GEO. I and J: IgG-normalized ChIP-qPCR of HSF1 binding to the PROM2 promoter. The control and HSF1 antibodies were used to precipitate DNA and then performed qPCR to detect the PROM2 promoter in BMSCs (n = 5). All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: unpaired student’s t-test (A, I, and J), ANOVA followed by a post hoc analysis (C and E); *P < 0.05, **P < 0.01, ***P < 0.001

Here we observed that PROM2 mRNA levels decreased in vitro ROS stress compared to the control group (Fig. 2A). And HSF1 positively correlated with PROM2 mRNA levels (Fig. 2B). Overexpression of HSF1 reversed the PROM2 mRNA levels, HSF1 and Prominin-2 expressions in BMSCs under OS (Fig. 2, C–F). In addition, HSF1 nuclear localization was increased in HSF1-overexpressed BMSCs, indicating HSF1 activity was elevated, which could upregulate HSF1 target genes (Fig. 2, G–H). These findings displayed that HSF1 might ameliorate TBHP-induced ferroptosis by activating the transcription of PROM2 and thus elevating Prominin-2 expression.

Fig. 2

Overexpression of heat shock factor protein 1 (HSF1) stimulates Prominin-2 expression by activating PROM2 (encoding protein Prominin-2) transcription. A total of 1 × 10⁶ bone mesenchymal stem cells (BMSCs) were used in each group. A BMSCs were incubated with PBS or TBHP (50 μM) for 24 h. Then BMSCs were harvested and PROM2 mRNA levels were determined by RT-PCR. Data were normalized to the levels of GAPDH (n = 5). B Correlation between HSF1 mRNA levels and PROM2 mRNA levels was performed using Pearson correlation analysis. Negative control lentivirus transfected BMSC (LV-NC) or lentiviral vectors carrying HSF1 gene transfected BMSC (LV-HSF1) incubated with PBS or TBHP (50 μM) for 24 h. C BMSCs were harvested and PROM2 mRNA levels were determined by RT-PCR. Data were normalized to the levels of GAPDH (n = 5). D and G BMSCs were harvested, and Prominin-2, cytoplasmic HSF1, and nuclear HSF1 were evaluated by Western Blot (n = 5). Histone-3 was used as a loading control for nuclear fractions. E and F Quantitative analysis of the protein levels of Prominin-2 and cytoplasmic HSF1. H Quantitative analysis of the protein levels of nuclear HSF1. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: unpaired student’s t-test (A), ANOVA followed by a post hoc analysis (C, E, F, and H); **P < 0.01, ***P < 0.001

Prominin-2 attenuates ferroptosis by facilitating BACH1 ubiquitination and degradation

In general, BACH1-mediated antioxidant genes downregulation leads to enhanced ROS generation upon cell exposure to OS [7, 18]. Meanwhile, increasing evidence has also illustrated that BACH1 is important in driving ferroptosis [8, 19]. Intriguingly, Prominin-2 physically interacted with BACH1 and overexpression of Prominin-2 mitigated BACH1 protein levels in BMSCs under OS circumstances based on our previous study [5]. Accordingly, we posited that Prominin-2-BACH1 interaction might prompt BACH1 degradation. We examined BACH1 protein levels in the control group and LV-PROM2 group treated by TBHP. As shown in Fig. 3, A–B, activation of PROM2 could reverse TBHP-stimulated BACH1 levels enhancement, implying that increased Prominin-2 expression inhibited BACH1 expression in BMSCs under OS. To expound on the potential mechanisms regulating BACH1 expression, we analyzed BACH1 mRNA expression levels in the control group, LV-PROM2 group, and LV-NC group (Fig. 3C). PROM2 did not alter BACH1 mRNA expression levels, manifesting that PROM2 restrains BACH1 expression in a transcription-independent mechanism. We next used CHX (inhibiting protein synthesis) to estimate this transcription-independent mechanism underlying Prominin-2-mediated inhibition of BACH1 protein expression. Prominin-2 overexpression could influence BACH1 protein stability and treatment of LV-PROM2 shortened the half-life of BACH1 after TBHP stimulation, thus hastening its degradation (Fig. 3, D–F). Additionally, using MG132 (inhibiting proteasomal degradation) obviously eliminated Prominin-2-impelled degradation of BACH1 (Fig. 3, G–H). We then found that Prominin-2 overexpression could accelerate BACH1 ubiquitination under OS, a post-translational modification that degrades a diverse variety of proteins and provides an elucidation for BACH1 degradation (Fig. 3I) [20]. The co-immunoprecipitation experiment showed that Prominin-2 overexpression increased the protein levels of ubiquitination in an OS microenvironment, which means the increased binding of BACH1 to Ubiquitin (Ub) (Fig. 3, J–K). These results signified the regulatory influence of Prominin-2 on BACH1 might rely on enhancing the ubiquitination and degradation of BACH1, resulting in a decrease in BACH1 levels. Previous studies also showed that BACH1 could be degraded by ubiquitination [21,22,23,24,25]. To further verify that the ubiquitin–proteasome pathway is the main pathway through which Prominin-2 promotes the degradation of BACH1, we also investigated the effect of the lysosomal proteolytic pathway on the degradation of BACH1. The ubiquitin–proteasome pathway and the lysosomal proteolytic pathway represent the two primary intracellular mechanisms for protein degradation, governing a variety of cellular activities, such as cell cycle regulation, cellular signaling, responses to stress, regulated cell death, autophagy, regulation of protein expression, and DNA transcription [23,24,25,26]. We thus individually added inhibitors for the ubiquitin–proteasome pathway (MG132) and the lysosomal proteolytic pathway (CQ). As anticipated, we observed that MG132 exerted a more pronounced cumulative effect on BACH1 in comparison to CQ (Fig. 3, L–M). This suggested that Prominin-2 primarily facilitated the degradation of BACH1 by enhancing its ubiquitin-dependent proteasomal degradation. Collectively, these results indicated that Prominin-2 attenuates TBHP-induced ferroptosis by propelling BACH1 ubiquitination and degradation.

Fig. 3

Prominin-2 attenuates ferroptosis by facilitating BTB and CNC homolog 1 (BACH1) ubiquitination and degradation. A total of 1 × 10⁶ bone mesenchymal stem cells (BMSCs) were used in each group. A Negative control group BMSC or lentiviral vectors carrying PROM2 gene transfected BMSC (LV-PROM2) incubated with PBS or TBHP (50 μM) for 24 h. Then BMSCs were harvested and BACH1 were evaluated by Western Blot (n = 5). B Quantitative analysis of the protein levels of BACH1. C The mRNA levels of BACH1 were determined in the control group, negative control lentivirus transfected BMSC (LV-NC) group, and LV-PROM2 BMSC group (n = 5). Data were normalized to GAPDH. D: After 12 h of cell culture in a complete medium containing TBHP (50 μM), control group or LV-PROM2 BMSC group received CHX (100 μg/mL) treatment for the indicated times. Then, BMSCs were harvested, and Prominin-2 and BACH1 were evaluated by Western Blot (n = 5). E and F Quantitative analysis of the protein levels of Prominin-2 and BACH1. G After 12 h of cell culture in a complete medium containing TBHP (50 μM), control group or LV-PROM2 BMSC group received 20 μM (MG132) treatment for 6 h. Then, BMSCs were harvested, and Prominin-2 and BACH1 were evaluated by Western Blot (n = 5). H Quantitative analysis of the protein levels of Prominin-2 and BACH1. I–J After 12 h of cell culture in a complete medium containing TBHP (50 μM), control group or LV-PROM2 BMSC group were harvested and the ubiquitination of BACH1 was evaluated by Western Blot (n = 5). K Quantitative analysis of the protein levels of ubiquitination. L After 12 h of cell culture in a complete medium containing TBHP (50 μM), LV-PROM2 BMSC group received 20 μM (MG132) or chloroquine (CQ, 15 μM) treatment for 6 h. Then, BMSCs were harvested, and BACH1 was evaluated by Western Blot (n = 5). M Quantitative analysis of the protein levels of BACH1. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: unpaired student’s t-test (K), ANOVA followed by a post hoc analysis (B, C, E, F, H and M); *P < 0.5, **P < 0.01, ***P < 0.001

Depletion of BACH1 remits TBHP-evoked GLS inhibition

GPX4 opposes ferroptosis by decreasing lipid peroxides utilizing GSH and BACH1 facilitates ferroptosis by repressing gene expression participated in GSH synthesis [4, 7]. GLS, mediating GSH synthesis, is also implicated in regulating ferroptosis [9]. We then speculated that BACH1 might expedite ferroptosis by restricting GLS expression. Heatmap visualized the differentially down-expressed GLS in BMSCs under OS. In an attempt to validate our speculated mechanism of BACH1 in TBHP-induced injury, we monitored GLS mRNA levels by RT-qPCR and GLS protein levels by Western Blot and immunofluorescence. Data from Fig. 4, A–E illustrated that ablation of BACH1 reversed TBHP-stimulated GLS attenuation, implying that knocking down BACH1 enhanced GLS expression in BMSCs under OS.

Fig. 4

Depletion of BTB and CNC homolog 1 (BACH1) remitted tert-butyl hydroperoxide (TBHP)-evoked glutaminase kidney isoform, mitochondrial (GLS) inhibition. A total of 1 × 10⁶ bone mesenchymal stem cells (BMSCs) were used in each group. The control BMSC group, nontargeting negative control BMSC group (Sh-NC), and a short hairpin RNA (shRNA) targeting BACH1 transfected BMSC group (Sh-BACH1) were incubated with PBS or TBHP (50 μM) for 24 h. A BMSCs were harvested, and GLS mRNA levels were determined using RT-PCR. Data were normalized to the levels of GAPDH (n = 5). B BMSCs were harvested, and GLS were evaluated by Western Blot (n = 5). C Quantitative analysis of the protein levels of GLS. D BMSCs were harvested and GLS expression levels in BMSCs were measured by immunofluorescence assay with DAPI labelling of nuclei (n = 5). White Bars = 50 μm. E Relative fluorescence signals associated with GLS (red staining) were statistically analyzed. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: ANOVA followed by a post hoc analysis (A, C, and E); **P < 0.01, ***P < 0.001

BPTES treatment aggravates TBHP-induced ferroptosis of BMSCs

To verify whether GLS expression could delay ferroptosis progression, we blocked glutamine metabolism with the GLS inhibitor BPTES. To elucidate the cytotoxic effect of BPTES on TBHP-induced BMSC injury, we assessed cell viability, LDH levels, and lipid ROS levels. The cell viability detected by CCK-8 and EdU examinations was decreased in the TBHP-induced group and BPTES-treated group, while the combination treatment with TBHP and BPTES aggravated inhibition of BMSCs’ proliferation compared with TBHP and BPTES alone (Fig. 5, A–C). The LDH and ROS levels were increased in the TBHP-induced group and BPTES-treated group compared with the control group, we found that treatment with BPTES enhanced the LDH and ROS levels in TBHP-induced BMSCs (Fig. 5, D–F). These results suggested that BPTES aggravates the cytotoxic effect of TBHP on BMSCs.

Fig. 5

Bis-2-(5-phenyl acetamido-1,2,4-thiadiazol-2-yl) ethyl sulfide (BPTES) enhances the cytotoxic effect of tert-butyl hydroperoxide (TBHP) on bone mesenchymal stem cells (BMSCs). A total of 1 × 10⁶ BMSCs were used in each group. BMSCs were incubated with PBS, TBHP (50 μM), BPTES (5 μM), or TBHP (50 μM) in the presence of BPTES (5 μM) for 24 h. A and B Cell proliferation was detected by CCK-8 (n = 5) and 5-ethynyl-2’-deoxyuridine (EdU) assay (n = 5). Representative EdU-positive cells (red) are shown. White Bars = 100 μm. C EdU-positive cells (red) were counted to calculate the percentage. D Lactate dehydrogenase (LDH) activity was measured by LDH kit (n = 5). E The ROS levels were measured by 2’−7’-dichlorofluorescein diacetate (1:1000, 20 μmol/L) (n = 5). White Bars = 100 μm. F Relative reactive oxygen species (ROS) intensity (green staining) was statistically analyzed. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: ANOVA followed by a post hoc analysis (A, C, D, and F); **P < 0.01, ***P < 0.001

We then investigated whether TBHP combined with BPTES aggravated ferroptosis, five hallmarks of ferroptosis, i.e., Fe²⁺ content, MDA levels, mitochondrial morphological changes, SOD activity, and ferroptosis-related marker protein levels, were assessed in BMSCs. Compatible with our anticipation, following co-treatment with TBHP and BPTES, Fe²⁺ content was remarkably increased (Fig. 6, A). In addition, we observed remarkably increased MDA levels after co-treatment with TBHP and BPTES in BMSCs (Fig. 6, B). Treatment with BPTES not only changed mitochondrial morphology in BMSCs (including appearing smaller volume and performing increased vestigial cristae and condensed membrane density) but also aggravated the TBHP-induced typical mitochondrial morphological characteristics of ferroptosis (Fig. 6, C). As exhibited in Fig. 6, D, the reduced SOD (the cellular enzymatic antioxidant) was noted in the TBHP-induced group and BPTES-treated group. Co-treatment with TBHP and BPTES further reduced SOD activity. Ferroptosis-related marker protein (ACSL4, PTGS2, GPX4, Ferritin) changes also indicated that ferroptosis may be exacerbated after co-treatment with TBHP and BPTES in BMSCs (Fig. 6, E–G). Collectively, we revealed the crucial role of GLS in TBHP-induced ferroptosis.

Fig. 6

Bis-2-(5-phenyl acetamido-1,2,4-thiadiazol-2-yl) ethyl sulfide (BPTES) treatment enhances tert-butyl hydroperoxide (TBHP)-evoked ferroptosis in bone mesenchymal stem cells (BMSCs). A total of 1 × 10⁶ BMSCs were used in each group. BMSCs were incubated with PBS, TBHP (50 μM), BPTES (5 μM), or TBHP (50 μM) in the presence of BPTES (5 μM) for 24 h. A The cellular iron contents were measured by an Iron Assay kit in BMSCs (n = 5). B Malondialdehyde (MDA) assay was used to quantify lipid peroxidation in BMSCs (n = 5). C Observation of BMSCs morphologic change by a transmission electron microscope. Black Bars = 5.0 μm. Red arrows: The typical mitochondrial morphological characteristics of ferroptosis (including appearing smaller volume and performing increased vestigial cristae and condensed membrane density). D The superoxide dismutase (SOD) activity was measured by a Total Superoxide Dismutase Assay Kit with NBT (n = 5). E BMSCs were harvested, ACSL4, PTGS2, GPX4 and Ferritin were evaluated by Western Blot (n = 5). F and G Quantitative analysis of the expression levels of ferroptosis-related marker proteins. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: ANOVA followed by a post hoc analysis (A, B, D, F, and G); *P < 0.5, **P < 0.01, ***P < 0.001

Silencing BACH1 alleviates GLS levels to rescue TBHP-induced ferroptosis in BMSCs

To further explore whether BACH1 might act as an upstream of GLS under OS circumstances, we silenced BACH1 via lentiviral vectors. Our study found that following treatment with BACH1 ablation increased cell proliferation and viability under the TBHP-stimulated microenvironment, as shown by the EdU assay (Fig. 7, A-B). Compared with the TBHP-induced group, the ROS and MDA levels were mitigated in the TBHP-treated group transfected by Sh-BACH1 (Fig. 7, C–E). Transfection of a ShRNA targeting BACH1 decreased ferroptosis sensitivity of BMSCs induced by TBHP. In contrast, BPTES reversed the beneficial effects of BACH1 ablation in BMSC ferroptosis induced by TBHP. To further confirm the BACH1/GLS axis on ferroptosis of BMSCs treated with TBHP, we detected the cellular glutamate and GSH content levels. As expected, knockdown of BACH1 raised intracellular glutamate levels and postponed GSH depletion (Fig. 7, F–G). Then, we used Western Blot to analyze the intracellular BACH1 and GLS protein levels, and the results exhibited that knockdown of BACH1 in BMSCs induced by TBHP diminished its expression and elevated its downstream target GLS (Fig. 7, H–J). The results of cell proliferation and viability, ROS levels, MDA levels, glutamate levels, and GSH content levels, were consistent with BACH1 and GLS protein levels. These results indicated that BACH1 participates in the regulation of GLS under OS circumstances and that BPTES could promote BMSC ferroptosis via GLS suppression.

Fig. 7

Silencing BTB and CNC homolog 1 (BACH1) alleviates glutaminase kidney isoform, mitochondrial (GLS) levels to rescue tert-butyl hydroperoxide (TBHP)-induced ferroptosis in bone mesenchymal stem cells (BMSCs). A total of 1 × 10⁶ BMSCs were used in each group. Nontargeting negative control BMSCs (Sh-NC) and a short hairpin RNA (shRNA) targeting BACH1 BMSCs (Sh-BACH1) were incubated with PBS, TBHP (50 μM), BPTES (5 μM), or TBHP (50 μM) in the presence of BPTES (5 μM) for 24 h. A Cell proliferation was detected by 5-ethynyl-2’-deoxyuridine (EdU) assay (n = 5). Representative EdU-positive cells (red) are shown. White Bars = 100 μm. B EdU-positive cells (red) were counted to calculate the percentage. C The ROS levels were measured by 2’−7’-dichlorofluorescein diacetate (1:1000, 20 μmol/L) (n = 5). White Bars = 100 μm. D Relative reactive oxygen species (ROS) intensity (green staining) was statistically analyzed. E Malondialdehyde (MDA) assay was used to quantify lipid peroxidation in BMSCs (n = 5). F Quantitative analysis of glutamate content levels was measured by using the Glutamate Assay Kit (n = 5). G Quantitative analysis of glutathione (GSH) levels was measured by a GSH analysis kit ((n = 5). H BMSCs were harvested, BACH1 and GLS were evaluated by Western Blot (n = 5). I and J Quantitative analysis of the protein levels of BACH1 and GLS. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: ANOVA followed by a post hoc analysis (B, D, E, F, G, I and J); *P < 0.05, **P < 0.01, ***P < 0.001

Targeting the prominin-2/BACH1 axis to ameliorate rats’ lumbar IVDD process

By puncturing SD rats’ lumbar IVDs, amputating the forelimbs, and elevating the food trough, we built the lumbar IVD persistent degeneration model. We randomized the SD rats into the following four groups: non-punctured group (Control), LV-NC injection with puncture group (IVDD + LV-NC), LV-PROM2 injection with puncture group (IVDD + LV-PROM2), LV-PROM2 + BACH1 inhibitor (Hemin) injection with puncture group (IVDD + LV-PROM2 + Hemin). We mainly explored whether ferroptosis resistance caused by BACH1 inhibition affects BMSC retention and therapeutic effects after they were engrafted into degenerative IVDs. In comparison to the BMSCs transfected with LV-PROM2 (IVDD + LV-PROM2), co-treated with LV-PROM2 transfection and Hemin pretreatment significantly elevated BMSCs’ retention in the degenerative IVDs at 24 h after transplantation, as evidenced by increased the EGFP fluorescent intensity and EGFP mRNA expression (Fig. 8, A–C). Correspondingly, the therapeutic effects of BMSCs on degenerative IVDs were ascended by BACH1 inhibition. At 4 weeks and 8 weeks after the operation, we performed the X-ray examinations, which exhibited markedly lower DHI in the IVDD + LV-NC and IVDD + LV-PROM2 than in the IVDD + LV-PROM2 + Hemin group (Fig. 8, D–H). The degeneration scores (reflected by DHI) of the lumbar IVDs that received BMSCs treatment were improved when BACH1 expression was inhibited. Suppressing BMSC ferroptosis via co-treatment of Prominin-2 overexpression and BACH1 inhibition also significantly improved the histological structure of lumbar IVDs, which was analyzed by H&E and S–O. Notably, the IVDD + LV-PROM2 + Hemin group demonstrated a more integral IVD structure compared to other groups (Fig. 8, I and J). We evaluated the degenerative scores based on H&E and S–O at 4 weeks and 8 weeks after transplantation (Fig. 8, K–N). To further validate the BMSC transplantation efficiency based on ferroptosis resistance, we collected samples of IVD tissues at 8 weeks after surgery and detected the MMP-9, ADAMTS5, and MMP-13 protein levels. In line with the available evidences [2, 27], BMSC engraftment exerts observable protective effects on puncture-induced IVDD, as evidenced by the decreased expressions of MMP-9, ADAMTS5, and MMP-13 extracted from degenerated IVDs, all of which are the major enzymes participated in the IVDD (Fig. 8, O–R) [28, 29]. A further declining trend was seen in the LV-PROM2 + Hemin group. Western Blot results confirmed that treatment with Prominin-2 overexpressed BMSCs pretreated by Hemin effectively ameliorates IVDD in SD rats and exerts better long-term effects for IVDD. Together, these results corroborated that ferroptosis resistance targeted by the Prominin-2/BACH1 axis significantly enhances BMSC retention and therapeutic efficacy in degenerative IVDs.

Fig. 8

Targeting the Prominin-2/BTB and CNC homolog 1 (BACH1) axis to ameliorate rats’ lumbar intervertebral disc degeneration (IVDD) process. Six male Sprague–Dawley (SD) rats aged 4 weeks were used in each group. Establishment of rats’ lumbar IVDD model via puncture surgery. A 24 h after puncture surgery, the rats’ spinal tissue was observed via the EGFP immunostaining under a confocal microscope (n = 6). White Bars = 100 μm. B The intensity of EGFP fluorescence signals was represented (green staining). C 24 h after puncture surgery, the rats’ spinal tissue was harvested and EGFP mRNA levels were determined by RT-PCR (n = 6). D and E and F At pre-surgery, four weeks post-surgery, and eight weeks post-surgery, representative radiographic images of rat lumbar were recorded by X-ray (n = 6). Black arrows: The lumbar intervertebral disc heights. G and H The disc height index (DHI %) was calculated at four weeks and eight weeks post-surgery. I At four weeks post-surgery and eight weeks post-surgery, representative safranin O-fast green (S–O) staining images were shown by performing S–O staining (n = 6). White Bars = 500 μm. J At four weeks post-surgery and eight weeks post-surgery, hematoxylin and eosin (H&E) staining images were shown by performing H&E staining (n = 6). White Bars = 500 μm. K and M Quantitative analysis of S–O staining analysis. L and N Quantitative analysis of H&E staining analysis. O At eight weeks post-surgery, degenerated IVDs were collected, and the proteins were extracted. MMP-9, ADAMTS5, and MMP-13 were evaluated by Western Blot (n = 5). P-R: Quantitative analysis of the protein levels of MMP-9, ADAMTS5, and MMP-13. All data are representative of at least five separate experiments. All these experiments were repeated by at least five times. Analysis: ANOVA followed by a post hoc analysis (B, C, G, H, K, L, M, N, P, Q and R); *P < 0.05, **P < 0.01, ***P < 0.001

Discussion

BMSCs implanted into degenerative IVDs exert therapeutic effects via various mechanisms and thus are supposed to be a boon for treating IVDD [2, 3]. However, inadequate BMSC retention leads to rapid exhaustion of transplanted BMSCs and thus weakens their therapeutic efficiency owing to their intolerance of the OS microenvironment of the degenerative IVDs. As the key physiological mechanism of cells, redox regulation could resist OS to promote BMSCs’ survival. Considerable evidence validated that the redox environment imbalance inhibits cell proliferation and survival and leads to subsequent occurrence of cell death [9, 30, 31]. Our previous review illustrated that ferroptosis participates in BMSCs’ survival after being transplanted, which in turn reduces clinical efficacy [4]. In our in vitro experiments, we also verified the changed expressions of ferroptosis-related marker proteins and enhanced OS levels in BMSCs in vitro ROS stress. The long-term OS microenvironment disorders redox status in BMSCs, which is also the characterization of ferroptosis, disrupting ROS/antioxidant ratio and accelerating BMSC death. Therefore, inhibiting this OS-induced regulated cell death might be a promising strategy to optimize BMSCs’ retention and therapeutic effectiveness.

Our previous studies identified Prominin-2 as a critical protein regulating BMSCs’ ferroptosis susceptibility, especially under OS circumstances [5]. Using lentiviral gene transfection technology, we validated that Prominin-2 expression is responsive to ROS stress and ferroptotic irritation, and its silence renders BMSCs vulnerable to Fe²⁺ overload or TBHP-induced ferroptosis [5]. Interestingly, we confirmed the interaction between Prominin-2 and BACH1 by using co-immunoprecipitation and colocalization assays in our further experiments [5]. Our previous review revealed the harmful effects of BACH1 in MSC transplantation therapy [4]. We discussed the probability that BACH1-mediated promotion of ferroptosis under OS circumstances through suppression of BACH1 target genes, including solute carrier family 7 member 11 (SLC7 A11), glutamate–cysteine ligase regulatory subunit (GCLM), and glutamate–cysteine ligase catalytic subunit (GCLC), all of which are involved in the synthesis of GPX4 (the most effective enzyme in reducing lipid ROS and a critical anti-ferroptosis protein) [4]. As another crucial mechanism of ferroptosis, the disturbance of cellular antioxidant defense is mainly due to the depletion of GPX4 [30]. The system Xc⁻ (composed of SLC7 A11 and solute carrier family 3 member 2 (SLC3 A2)) is responsible for the uptake of cystine [31]. Once transported into cells via transporters, glutamine is transformed into glutamate by GLS (The opposite process depends on glutamine synthetase (GSS)) [19]. The GCLM and GCLC function as rate-limiting enzymes to catalyze cystine and glutamate to synthesize GSH, which is then used for GPX4 synthesis [32] (Fig. 9). Hence, exploring the specific mechanisms taking part in the Prominin-2/BACH1 pathway is vital for long-term BMSC post-transplant survival rates.

Fig. 9

Our study supports a new mechanistic insight into the regulation of the Prominin-2/BACH1/GLS pathway in BMSC ferroptosis. Heat shock factor 1 (HSF1) activates Prominin-2 expression, which protects Bone Mesenchymal Stem Cells (BMSCs) from ferroptosis induced by Tert-butyl hydroperoxide (TBHP). This protective effect occurs through two mechanisms: extracellular iron transport and promoting the ubiquitination and degradation of BTB and CNC homolog 1 (BACH1). Inhibition of BACH1 expression reversed TBHP-stimulated down regulation of Glutaminase (GLS) and therefore protects BMSCs from ferroptosis. Fe²⁺: Ferrous iron; Fe³⁺: Ferric iron; TBHP: Tert-butyl hydroperoxide; HSF1: Heat shock factor protein 1; PROM2: encoding protein Prominin-2; ROS: Reactive oxygen species; BACH1: BTB and CNC homolog 1; GLS: Glutaminase kidney isoform, mitochondrial; GSS: Glutamine synthetase; GCLC: Glutamate–cysteine ligase catalytic subunit; GCLM: Glutamate–cysteine ligase regulatory subunit; GSH: Glutathione; GPX4: Glutathione peroxidase 4; Gln: glutamine; SLC3 A2: Solute carrier family 3 member 2; SLC7 A11: Solute carrier family 7 member 11; BMSCs: Bone mesenchymal stem cells

A series of significant and original research results have been achieved in our present study to solve the problem mentioned above. Reportedly, HSF1 might act as a key mediator for Prominin-2 [33]. Correspondingly, HSF1 was predicted to interact with the promoter regions of PROM2 based on the JASPAR database, highlighting its important role in inducing Prominin-2 expression. Meanwhile, the heatmap corroborated the differentially down-expressed HSF1 and PROM2 in BMSCs under OS compared with control group. Thus, we explored whether HSF1 could regulate the transcription of PROM2 in BMSCs. Consistent with these researches, our data revealed that HSF1 could bind to the PROM2 promoter and HSF1 activation reversed the PROM2 mRNA levels under OS circumstances. Recently, HSF1 was reported to play a meaningful role in ferroptosis [15] and it could protect against OS-induced cellular injury [34]. A new study illustrated that cancer cells with elevated HSF1 expression resist ferroptosis [35]. Another study discovered that HSF1 activation alleviated palmitic acid-induced lipid ROS and ferroptosis in cardiomyocytes [36]. In our study, we corroborated that HSF1 could activate Prominin-2 expression to cope with TBHP-induced ferroptosis in BMSCs.

Second, based on our previous findings, Prominin-2 is associated with BACH1 protein expression. However, there was no correlation between PROM2 and BACH1 in mRNA levels. Since BACH1 stability is regulated by ubiquitin-mediated proteolysis after its nuclear export [21, 37], we assumed Prominin-2 might influence its ubiquitination degradation. Indeed, our data showed that Prominin-2 facilitates BACH1 degradation in a ubiquitin-dependent behavior. Considering that Prominin-2 cannot direct ubiquitination regulation, this function might be attributed to Prominin-2 promoting BACH1 nuclear exporting. Emerged evidence suggested that the ubiquitination of BACH1 exerts a crucial role in ferroptosis [21, 38, 39]. Irikura et al. [18] pointed out that BACH1 re-expression could promote ferroptosis by repressing GSH synthesis and disordering intracellular iron metabolism under OS circumstances. It is widely recognized that BACH1 protein expression is closely correlated with ubiquitination degradation [21, 38]. Taken together, we verified that Prominin-2 could restrict ferroptosis by impelling BACH1 ubiquitination in the present study, which might enhance the transplanted BMSC survival rate under OS circumstances. In addition, our future research will focus on this specific ubiquitination regulation. Given that BACH1 inhibition markedly alleviated Fe²⁺ accumulation, ROS generation, and endogenous MDA levels (lipid peroxidation product) in BMSCs. We pretreated BMSCs to restrain BACH1 protein expression. Pretreatment with hemin (specifically degrading BACH1) has been shown to enhance the cardioprotective properties of MSCs on myocardial infarction in mouse models [40]. We also identified that transplantation of Prominin-2-overexpressed BMSCs pretreated with hemin (20 μM) significantly improved BMSC short-term survival under OS circumstances and attenuated IVDD pathology in rat IVDD models. In this paper, targeting the Prominin-2/BACH1 axis has been shown to improve BMSC survival post-transplantation and attenuate IVDD progression by inhibiting ferroptosis.

Third, we found that BACH1 could target the inhibition of GLS and drive ferroptosis in BMSCs induced by TBHP. Nishizawa et al. [7] reported that BACH1 accelerates ferroptosis by directly suppressing genes that participated in the synthesis of GSH, including GCLM and GCLC. We validated that GSH content levels were reversed upon BACH1 down-expression in BMSCs induced by TBHP. Notably, GLS functions as a rate-limiting enzyme that is also involved in GSH generation [41]. Blocking GLS downregulates GLS-mediated glutamine metabolism, reduces GPX4 protein levels, accelerates ROS accumulation, and ultimately induces ferroptosis [9]. Li et al. [42] revealed that overexpression of GLS rescued the cell growth inhibition and avoided ferroptosis in gallbladder cancer cells treated with lithocholic acid. Our findings also displayed that GLS inhibition aggravated cell proliferation suppression and enhanced TBHP-induced ferroptosis in BMSCs. Therefore, we hypothesized that BACH1 might subdue the crucial role of GLS in protecting BMSCs against ferroptosis under OS circumstances. Indeed, downexpression of BACH1 in BMSCs treated with TBHP redeemed cell viability and alleviated ferroptosis by reducing ROS and MDA levels and could promote glutamine metabolism, thus raising glutamate contents. Our findings disclosed that BACH1 functions as an inhibitor for GLS in BMSCs under OS by regulating downwards GLS-mediated glutamine metabolism and consequently activating ferroptosis.

Increasing evidence indicated that ferroptosis could also promote the occurrence and progression of inflammation [43,44,45]. Tsurusaki et al. [43] revealed that ferroptosis inhibition effectively shields hepatocytes from regulated cell death and suppresses subsequent immune cell infiltration and inflammatory responses. Results suggested hepatic ferroptosis plays a vital role as the trigger for initiating inflammation in steatohepatitis and could serve as a therapeutic target to avert the progression of the condition. Similarly, ferroptosis is pivotal in the pathogenesis of bacterial keratitis [45]. Ferroptosis inhibition showed potential for alleviating inflammation, decreasing corneal scarring, and ultimately improving the prognosis of bacterial keratitis. Numerous research studies have demonstrated that inflammation is intricately linked to the pathological processes of IVDD, including the ongoing loss of extracellular matrix, cell death, and senescence, highlighting the vital importance of inflammation in the underlying mechanisms of IVDD [46, 47]. The inflammatory microenvironment within degenerated IVDs is associated with high levels of pro-inflammatory factors, which can be toxic to transplanted BMSCs, affect their differentiation potential, and interfere with communication between the BMSCs and host cells [48, 49]. This suggested that targeting ferroptosis could improve BMSCs’ survival rate while improving their anti-inflammatory capacity in the degenerative IVD microenvironment, ultimately improving the efficiency of BMSC transplantation for repairing degenerated IVDs. In addition, the occurrence of ferroptosis also involves mitophagy, which is crucial for sustaining mitochondrial homeostasis [50,51,52]. OS stimulation could activate the cyclic GMP-AMP synthase-stimulator of interferon genes protein pathway, which induces excessive mitophagy and exacerbates ferroptosis in tendon stem cells [52]. However, another study showed that promoting mitophagy could prevent ferroptosis in stem cells [51]. This might be attributed to the double-edged sword effects of mitophagy, where normal levels of autophagy exert a protective effect, while excessive autophagy can harm the stem cells [53, 54]. These evidences suggested that targeting BMSC ferroptosis may involve other essential mechanisms. Future validation of the roles of these mechanisms (inflammation and mitophagy) is expected to further enhance the therapeutic efficiency of BMSCs in treating degenerated IVDs.

Mechanistically, we demonstrated that the Prominin-2/BACH1/GLS pathway confers protection against TBHP-induced ferroptosis in BMSCs. Based on our previous study, Prominin-2/BACH1 pathway disruption could drive ferroptosis [5]. In this study, Prominin-2 functions as a regulator in controlling oxidative homeostasis by facilitating BACH1 ubiquitination degradation. We also unveiled HSF1 and GLS as important defenders in TBHP-induced BMSC ferroptosis by regulating Prominin-2 expression and mediating GSH synthesis, respectively. We discovered that treatment with BPTES markedly enhanced the cytotoxicity of TBHP on BMSCs and stimulated TBHP-induced ferroptosis by repressing GLS expression. However, there are some deficiencies in our present study. First, we only focused on the Prominin-2/BACH1/GLS axis in BMSC ferroptosis. We did not explore the entire HSF1/Prominin-2/BACH1/GLS pathway in the survival of BMSCs under the OS microenvironment. Second, we only evaluated BMSC retention or the long-term efficacy of transplanted BMSCs for degenerative IVDs. We did not assess the direct impact of the Prominin-2/BACH1/GLS pathway intervention on the occurrence of BMSC ferroptosis in vivo. Third, we did not quantitatively demonstrate the ubiquitination levels of BACH1, which should be prioritized in future studies. Hence, we should pay more attention to these deficiencies in future studies.

Conclusions

Our results supported new mechanistic insight connected the regulation of the Prominin-2/BACH1/GLS pathway with BMSC ferroptosis and represented potential therapeutic targets for improving the survival of engrafted BMSCs under OS circumstances. Prominin-2 activated by HSF1 increases the antioxidant capacity of BMSCs by interacting with BACH1 and facilitating BACH1 ubiquitination and degradation. BACH1 inhibition ameliorates TBHP-induced ferroptosis by propelling glutamine metabolism by relieving GLS inhibition. Our data suggested that Prominin-2/BACH1/GLS pathway has the potential to be applied in preclinical research for IVDD diseases. These findings will be helpful in developing attractive treatment strategies for BMSC survival and stem cell-based therapy.