MLL1 promotes placental trophoblast ferroptosis and aggravates preeclampsia symptoms through epigenetic regulation of RBM15/TRIM72/ADAM9 axis

Abstract

This study explores the epigenetic mechanism of MLL1 regulating trophoblast ferroptosis in preeclampsia (PE). A murine model of PE was established, and HTR-8/SVneo cells were induced by Erastin to establish an in vitro cell model. GSH, MDA, Fe²⁺, and ROS levels were measured to assess ferroptosis. MLL1, RBM15, TRIM72, ADMAM9, ASCL4, GPX4, and FTH1 expressions were detected by qRT-PCR or Western blot. ChIP analyzed H3K4me3 enrichment and MLL1 enrichment on RBM15 promoter. The binding of YTHDF2 or m6A to TRIM72 mRNA was determined by RIP. TRIM72 mRNA stability was detected after actinomycin D treatment. The binding of TRIM72 to ADAM9 and the ADAM9 ubiquitination level were detected by Co-IP. MLL1 was highly expressed in placental tissues of PE mice. Inhibition of MLL1 improved PE symptoms in mice, repressed ferroptosis in placental tissues, and inhibited Erastin-induced ferroptosis in vitro. MLL1 elevated RBM15 expression by increasing H3K4me3 on RBM15 promoter. RBM15 promoted the binding of TRIM72 to YTHDF2 by enhancing m6A modification on TRIM72 mRNA, thereby repressing TRIM72 expression. TRIM72 bound to ADAM9 and ubiquitinated it for degradation. In conclusion, MLL1 promotes placental trophoblast ferroptosis and aggravates PE symptoms via epigenetic regulation of RBM15/TRIM72/ADAM9 axis.

Background

Preeclampsia (PE) is a gestational idiopathic disorder that clinically manifests as new-onset hypertension and proteinuria, which can lead to functional impairment of liver, kidney, and brain, or even maternal and fetal mortality if left untreated [1]. The pathomechanism of PE begins with placental dysfunction in early pregnancy, which is related to improper decidualization, vasculogenesis, angiogenesis, and spiral artery remodeling, leading to endothelial dysfunction [2]. Trophoblasts are important functional cells in the placenta essential for maintaining placental function [3]. Appropriate trophoblastic cellular death is important for healthy pregnancy, whereas inappropriate trophoblast death interferes cellular homeostasis, resulting in abnormal placentation [4]. Ferroptosis as a newly discovered modality of iron-dependent cell death is characterized by iron overload, lipid peroxidation accumulation, and redox imbalance [5]. Compelling studies have suggested the intricate association between trophoblast ferroptosis and PE pathogenesis [6,7,8,9]. Glutathione (GSH), glutathione peroxidase (GPx), and malondialdehyde (MDA) are important indicators to evaluate the degree of lipid peroxidation. Compared with normal maternal placenta, the levels of GSH and GPx4 in the trophoblasts of PE placenta are decreased, while the concentrations of Fe²⁺ and MDA are increased, indicating that ferroptosis occurs in PE [10, 11]. The decrease in GPx4 levels leads to an increase in lipid reactive oxygen species (ROS), which accumulates excessively on the membrane of trophoblasts and inhibits cellular biological functions such as proliferation, invasion, and migration, thus inducing poor invasion of trophoblasts and impaired remodeling of spiral arteries [12]. Therefore, exploring the molecular mechanism of PE from the perspective of ferroptosis is of substantial clinical value for timely treatment and early diagnosis.

Epigenetic changes including DNA methylation, histone modification, and non-coding RNAs, coordinate downstream effects that give rise to placenta dysfunction and subsequently the onset of PE [13]. Mixed lineage leukemia 1 (MLL1) acts as a histone methyltransferase responsible for catalyzing the trimethylation of H3K4 (H3K4me3), and H3K4me3 is considered as a mark of transcriptionally active promoters [14]. MLL1 plays a vital role in embryonic stem cell development [15], hematopoiesis [16], and neurogenesis [17] via its methyltransferase activity. In particular, a notable elevation of MLL1 expression has been observed in villus tissues of PE patients [18]. However, whether MLL1 participates in PE by regulating trophoblast ferroptosis remains unclear.

N6-methyladenosine (m6A) is the most prevalent internal co-transcriptional modification in eukaryotic RNAs, which is modified by m6A methyltransferases, or writers, such as METTL3/14, RNA binding motif protein 15 (RBM15), and WTAP, removed by demethylases, or erasers, including FTO and ALKBH5, and recognized by m6A-binding proteins YTHDF1/2/3, YTHDC1/2, and IGF2BP1/2/3, also termed “readers” [19]. m6A modification and RBM15 abundance are enhanced in PE patients, and overexpression of RBM15 retards the migration and invasion of trophoblasts [20]. As another well-characterized m6A methyltransferase, METTL3 augments ferroptosis but represses migration and invasion of trophoblasts, and exacerbates PE symptoms in vivo by catalyzing the m6A modification of ACSL4 mRNA [21]. At present, the role of RBM15 in ferroptosis of trophoblasts in PE has not been reported.

Tripartite motif protein 72 (TRIM72), a member of the tripartite motif family mainly found in striated muscle, possesses extensive functionalities by virtue of its classic membrane repair, anti-inflammation, and E3 ubiquitin ligase properties [22]. The therapeutic potential of TRIM72 has been extensively recognized in various diseases such as muscle injury [23], myocardial injury [24], and acute lung injury [25]. Of note, the E3 ubiquitin ligase characteristic of TRIM72 has been increasingly demonstrated to be associated with certain disease conditions. For example, TRM72 can directly interact with p53 to promote p53 ubiquitination and degradation, thereby repressing apoptosis and accelerating proliferation and migration of trophoblasts [26]. Epigenetic modifications of histones and m6A modifications on RNA both play critical roles in the PE process. However, it is unknown whether epigenetic modification enzymes are involved in the PE process through ferroptosis, especially the role of MLL1 and RBM15. Based on the preliminary findings of our study that MLL1 and RBM15 are highly expressed in PE, we speculate that MLL1 promotes RBM15 expression by increasing H3K4me3 on RBM15 promoter and then RBM15-mediated m6A modification mediates TRIM72 expression for the regulation of trophoblast ferroptosis. This study aims to investigate the epigenetic mechanism of MLL1 regulating RBM15-mediated ferroptosis of placental trophoblasts in PE, hoping to confer novel therapeutic targets for PE.

Materials and methods

Experimental animals

All animal experiment schemes were approved by the Animal Ethics Committee of Sichuan Provincial People’s Hospital and implemented based on the Guide for the Care and Use of Laboratory Animals [27].

CD-1 (ICR) mice were purchased from the Shanghai Laboratory Animal Center of the Chinese Academy of Sciences (Shanghai, China) and acclimated for at least 1 week before the beginning of the experiments. The animals were housed in a temperature and humidity regulated environment with a 12-h light cycle.

Establishment of a murine model of PE

In the timed pregnancy experiment, female mice (8–10 weeks old) were placed with a stud male for two nights. Embryonic day 0.5 (E0.5) was denoted as the first morning at which a vaginal plug was observed, and the female mice were removed from the cage. As described previously [28], pregnant ICR mice were injected with 6 × 10⁸ pfu of adenovirus expressing murine sFlt-1 via the tail vein on E8.5 to establish a well-characterized sFlt-1 overexpression model. The control mice were injected with an equal amount of adenovirus encoding mouse Fc protein (Ad-Fc). The blood, urine and tissues were collected at the end of the experiment. The mice were euthanized at E18.5, and the weights of the fetus and placenta were recorded. After weighing, the placenta was collected for further analysis. For gene intervention, adenovirus packaged sh-MLL1 or oe-RBM15 was injected into mice simultaneously with sFlt-1 adenovirus at E8.5. The above vectors and adenovirus packages were provided by Hanbio (Shanghai, China). The experimental mice were divided into 6 groups: Sham, PE, PE + sh NC, PE + sh-MLL1, PE + sh-MLL1 + oe-NC, and PE + sh-MLL1 + oe-RBM15 groups, with 12 mice in each group. Each mouse in each group underwent blood pressure and proteinuria testing. After euthanasia, 6 mice were randomly selected from each group for pathological staining, and the remaining 6 mice were subjected to tissue homogenization treatment.

Blood pressure measurement

The systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured from the tail of mice using CODA non-invasive plethysmography blood pressure transducer (Kent Scientific Corporation, Torrington, CT, USA). All mice underwent habituation measurements (10 measurements per day from E4.5-E6.5) before formal evaluation. Then, blood pressure levels throughout pregnancy were assessed at E7.5, E9.5, E11.5, E13.5, and E17.5.

Proteinuria detection

Urine was collected at E17.5 for proteinuria determination according to the protocol of the manufacturer of easy II protein quantitative kit (DQ111-01; Transgen Biotech, Beijing, China). The urine of each mouse was collected and the protein concentration was determined by bicinchoninic acid (BCA) assay. The absorbance at 595 nm was determined with a microplate reader, with bovine serum albumin (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) as the reference standard. According to the previously described method [29], the proteinuria detection was repeated at least three times.

Tissue staining

Placenta and kidney tissues were fixed with 4% paraformaldehyde for 48 h and processed according to routine procedures. Sections with a thickness of 3–5 μm were cut from the coated tissues, pasted on poly-L-lysine-coated slides, separated by xylene, dehydrated by alcohol, and then stained with hematoxylin and eosin (H&E) or periodic acid-schiff (PAS).

Cell culture

Human villous trophoblasts HTR-8/SVneo were purchased from ProCell (Wuhan, Hubei, China) and cultured in RPMI-1640 medium (Thermo Fisher Scientific, Waltham, MA, USA) containing 1% penicillin-1% streptomycin and 10% fetal bovine serum at 37℃ with 5% CO₂.

Cell treatment

MLL1, RBM15, TRIM72, A disintegrin and a metalloprotease 9 (ADAM9) overexpression vectors and their short hairpin RNAs (sh-MLL1, sh-TRIM72, sh-YTHDF2), empty vector (oe-NC), and shRNAnegative control (sh-NC) were provided by Hanbio and transfected into cells using Lipofactamine 2000 (Invitrogen, Carlsbad, CA, USA). Briefly, cells were seeded into 12-well plates (1 × 10⁵ cells/well), and the transfection reagents were bound with the indicated plasmids into each well, followed by 6 h of incubation at 37 °C with 5% CO₂. Afterward, the medium was changed to fresh intact medium, and cells were harvested after 48 h of transfection. The transfection efficiency was verified by qRT-PCR and Western blot. HTR-8/SVneo cells were treated with Erastin (MedChemExpress LLC, NJ, USA) at a concentration of 10 µmol/L for 24 h [30]. Proteasome inhibitor MG132 and protein synthesis inhibitor cycloheximide (CHX) were purchased from MCE. MG132 was added into the culture medium at a concentration of 10 μm after transfection, with an equal amount of DMSO as a control. CHX was added into the culture medium at a concentration of 25 µg/mL [31], and the expressions of target proteins in cells were detected after 0, 2, 4, and 8 h.

Cell counting kit-8 (CCK-8) assay

HTR-8/SVneo cells were incubated in 96-well plates (5 × 10³ cells/well) overnight and then treated with 10 µL CCK-8 reagent (Beyotime, Shanghai, China) at 37℃ for 2 h. The absorbance of HTR-8/SVneo cells at 450 nm was determined by a microplate reader (Thermo Fisher Scientific).

Transwell

For the cell invasion experiment, cells (1 × 10⁵ cells/well) were seeded in the complete medium for 24 h of incubation after the invasion chambers pre-coated with extracellular Matrigel (356234, BD Biosciences, San Jose, CA, USA) were rehydrated. The movement of invading cells in the Transwell chambers was calculated using the Zen imaging software (Zeiss Inc, AG, Oberkochen, Germany).

For the cell migration experiment, except that there was no extracellular Matrigel pre-coated on the chamber, the other experimental procedures were consistent with those of the invasion experiment.

Detection of Glutathione (GSH), Malondialdehyde (MDA), Fe2+, and Reactive Oxygen Species (ROS)

Fresh tissues were rinsed with 5 mL phosphate-buffered saline (PBS) three times, then minced into a paste with ophthalmic scissors, added with 5 mL collagenase type I, digested at 37 °C for 40 min on a shaker, and placed into 5 mL complete Dulbecco’s modified Eagle’s medium (DMEM) to stop digestion. The cell suspension was collected and centrifuged to obtain cell microspheres. The cells were resuspended in 5 mL erythrocyte lysate, left at room temperature for 5 min, and centrifuged to obtain cell microspheres. After re-suspension and washing with 5 mL PBS, the cells were cultured in 2 mL DMEM.

BODIPY™581/591 C11 (D3861, Thermo Fisher Scientific) was used to analyze lipid ROS levels. Lipid ROS levels in cells were analyzed by C11-BODIPY staining. Briefly, cells were incubated with BODIPY™581/591 C11 working solution at room temperature for 30 min. DAPI staining was performed after cell washing. After 20 min of incubation, the images were captured for observation.

Fe²⁺ colorimetric method kit (E1042, Applygen Technologies, Beijing, China) was applied to detect Fe²⁺ levels of tissues or cells. All procedures were implemented according to the product instructions.

A commercial GSH detection kit (A006-1-1, JianCheng Bioengineering Institute, Nanjing, Jiangsu, China) was used for calculating GSH contents of tissues or cells, and the absorbance at 420 nm was captured.

Lipid peroxidation MDA assay kit was utilized to measure MDA levels of tissues or cells.

Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was extracted from tissues or cells using TRIzol reagent (15596026, Thermo Fisher Scientific), and the protein concentration was measured with a ultraviolet spectrophotometer (NanoDrop 2000, Thermo Fisher Scientific). Then, RNA was reverse transcribed into cDNA using reverse transcription kit (205311, Qiagen, Germany). PCR was performed on StepOnePlus instrument (7900HT, Applied Biosystems, Carlsbad, CA, USA) with SYBR Green PCR Master Mix (4344463, Applied Biosystems), and cDNA served as the template. The expression of target gene was calculated by the 2^−△△Ct method [32], with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the internal reference. PCR primers are shown in Table 1.

Table 1 PCR primers

Western blot

The extracted proteins were separated by SDS-PAGE (Bio-Rad, Hercules, CA, USA) and transferred onto polyvinylidene fluoride membranes (C3117, Millipore, Billerica, MA, USA). The membranes were blocked with 5% skim milk (A600669, Sangon Biotech, Shanghai, China) diluted in PBS with Tween 20 (PBS-T) (A100777, Sangon Biotech) at 4 °C overnight. Then, the membranes were detected with the primary antibodies anti-MLL1 (1:2000, NB600-256, Novus Biological Inc., Littleton, CO, USA), RBM15 (1:1000, ab315456, Abcam, Cambridge, MA, USA), YTHDF2 (1:1000, ab220163, Abcam), TRIM72 (1:1000, ab220163, Abcam), ADAM9 (1:10000, ab226459, Abcam), ASCL4 (1:10000, NBP2-92741, NOVUS), GPX4 (1:1000, ab125066, Abcam), and FTH1 (1:1000, 701934, Invitrogen) in tris-buffered saline with Tween 20 (TBS-T) containing 2% (w/v) bovine serum albumin (RPN4201, GE-Healthcare, USA) overnight at 4 °C. Afterward, the primary antibodies were detected with the secondary antibody IgG (1:1000, ab6721, Abcam) at room temperature for 1 h. The protein bands were imaged with the Chemiluminescent Imaging System (Tanon-5200, Tanon Science & Technology, Shanghai, China), with β-actin (1:1000, ab8227, Abcam) as the internal control.

Chromatin immunoprecipitation (ChIP)

Simple ChIP^® Enzymatic Chromatin IP Kit (9003, Cell Signaling Technology, Beverly, MA, USA) was applied for ChIP. Briefly, cell or tissue samples were fixed with formaldehyde and added with glycine (0.125 M) to stop fixation. Chromatin was digested into fragments and incubated with primary antibodies MLL1 (1:2000, nb600-256, NOVUS) and H3K4me3 (1:1000, ab213224, Abcam) at 4 °C overnight. The immune complexes were captured with protein G magnetic beads. After crosslinking reversal, purified DNA was subjected to standard PCR and qPCR analysis. All primers used in PCR experiments are listed in Table 1.

m6A quantitative analysis

m6A RNA methylation was detected using m6A RNA methylation detection kit (ab185912, Abcam) following the manufacturer’s protocol. Total RNA samples (400 ng) from each group were used to determine m6A percentage. The absorbance at 450 nm was measured with a microplate reader, and the percentage of m6A in total RNA was calculated.

RNA immunoprecipitation (RIP)

Magna RIP RNA binding protein immunoprecipitation kit (\17–700, Millipore) was used for RIP assay. Briefly, cells or tissue homogenates were lysed with RIP lysis solution. YTHDF2 (1:1000, ab220163, Abcam) or m6A (1:1000, MA5-33030, Invitrogen), IgG (1:50, ab172730, Abcam) were incubated with magnetic beads to obtain immunoprecipitated magnetic beads. The magnetic beads were incubated with cell lysate to immunoprecipitate the RNA binding protein-RNA complex, and then RNA was eluted from the complex for qPCR detection.

RNA stability test

Cells were seeded into 6-well plates and treated with actinomycin D (5 µg/mL, Sigma-Aldrich) for 0, 3, and 6 h. Then total RNA was extracted for RT-qPCR.

Co-immunoprecipitation (Co-IP)

After pre-cooled PBS washing, cells were lysed in 500 mL Co-IP buffer (50 mM Tris-HCl pH 7.5, 5 mM EDTA, 150 mM NaCl, and protease inhibitor cocktail) and incubated with 10 µg primary antibodies anti-ADAM9 (1:100, ab226459, Abcam), TRIM72 (1:100, ab220163, Abcam), IgG (1:50, ab172730, Abcam), and 100 µL agarose A + G at 4℃ for 4 h. After washing, the binding protein was separated and detected by Western blotting. For ubiquitination detection, cells were re-suspended in lysis buffer containing protease inhibitors and lysed by sonication. After centrifugation at 15,000 g for 10 min, the supernatant was obtained and incubated with agarose beads conjugated with anti-ADAM9 (1:100, ab226459, Abcam) or anti-IgG (ab172730, Abcam) overnight at 4 °C. After extensive washing, they were boiled in 2 × loading buffer for 5 min, separated by SDS-PAGE, and then subjected to Western blot analysis with Ub (ab6721, Abcam).

Statistical analysis

Data analysis and map plotting were performed using the SPSS 21.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism 8.0 (GraphPad Software Inc., San Diego, CA, USA). The data were examined for normal distribution and homogeneity of variance. The t test was adopted for comparisons between two groups, and one-way or two-way analysis of variance (ANOVA) was employed for the comparisons among multiple groups, following Tukey’s multiple comparison test. A value of P < 0.05 indicated a significant difference.

Results

Inhibition of MLL1 ameliorates symptoms of PE mice and alleviates ferroptosis in placental tissues

A murine model of PE was established by injection of sFlt-1 adenovirus via tail vein at E8.5. PE mice presented with elevated systolic and diastolic blood pressures (P < 0.01, Fig. 1A), notably increased proteinuria (P < 0.01, Fig. 1B), and reduced weight of fetus and placenta (P < 0.01, Fig. 1C). In H&E and PAS staining, the placental tissues of PE mice exhibited underdeveloped villous blood vessels and disordered arrangement of trophoblast cells (Fig. 1E), formation of glomerular vascular branches in the kidney, and slight swelling of glomerular capillary endothelial cells (Fig. 1E). MLL1 was highly expressed in the placenta of PE mice (P < 0.01, Fig. 1D, J). We speculated that MLL1 affected PE symptoms in mice, so we inhibited MLL1 expression in the placenta by tail vein injection of sh-MLL1 adenovirus (P < 0.01, Fig. 1D, J). Inhibition of MLL1 led to a significant decrease in the blood pressure and proteinuria in PE mice (P < 0.01, Fig. 1A-B), an increase in the weight of fetus and placenta (P < 0.01, Fig. 1C), and an improvement in the pathological injury of placental and kidney tissues (Fig. 1E).

Fig. 1

Inhibition of MLL1 expression ameliorates symptoms of PE mice and alleviates ferroptosis in placental tissues. A murine model of PE was established by injection of sFlt-1 adenovirus via tail vein at E8.5, and the control mice were injected with adenovirus encoding Fc protein, followed by injection of sh-MLL1 adenovirus, with sh-NC adenovirus as the control. A: Measurement of blood pressure in mice, n = 12. B: Detection of proteinuria in urine, n = 12. C: Weight of placenta and fetus, n = 12. D: Detection of MLL1 expression in placenta by qRT-PCR, n = 6. E: Observation on the pathological structure of placenta and kidney by H&E staining and PAS staining, respectively. The black arrow indicates the placental villous vascular interstitium; ▲ indicates trophoblasts; the green arrows indicates pathological features of glomeruli in renal tissues, n = 6. F-I: Detection of ROS, MDA, Fe²⁺, and GSH levels in placenta, n = 6. J: Detection of MLL1, ASCL4, GPX4, and FTH1 expressions in placenta by Western blot, n = 6. Data in panels BCDFGHI were analyzed by one-way ANOVA, and data in panels AJ were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. **P < 0.01

Further detection of ferroptosis revealed that the placental tissues of PE mice had significantly elevated ROS, MDA, and Fe2 + levels (P < 0.01, Fig. 1F-H), decreased GSH level (P < 0.01, Fig. 1I), increased expression of ferroptosis related protein ASCL4, and decreased expressions of GPX4 and FTH1 (P < 0.01, Fig. 1J), while the above trends were reversed by inhibition of MLL1 (P < 0.01, Fig. 1F-J). The above results suggest that inhibition of MLL1 improves the symptoms of PE mice and alleviates ferroptosis in placental tissues.

Inhibition of MLL1 promotes the invasion and migration of HTR-8/SVneo cells and represses ferroptosis

To further explore the role of MLL1 in trophoblast ferroptosis, we cultured HTR-8/SVneo cells in vitro and established an in vitro cell model by adding ferroptosis inducer (Erastin). Meanwhile, we transfected sh-MLL1 into cells to inhibit MLL1 expression (P < 0.01, Fig. 2A, B, H). Erastin significantly weakened cell viability (P < 0.01, Fig. 2C), elevated intracellular ROS, MDA, and Fe2 + levels (P < 0.01, Fig. 2D, E, G), and diminished GSH level (P < 0.01, Fig. 2F), accompanied by increased ASCL4 expression and decreased GPx4 and FTH1 expressions (P < 0.01, Fig. 2H). The migration and invasion of HTR-8/SVneo cells were also abated significantly after Erastin treatment (P < 0.01, Fig. 2I). Inhibition of MLL1 enhanced cell viability (P < 0.01, Fig. 2C), diminished intracellular ROS, MDA, and Fe2 + levels (P < 0.01, Fig. 2D, E, G), increased GSH level (P < 0.01, Fig. 2F), decreased ASCL4 expression and elevated GPx4 and FTH1 expressions (P < 0.01, Fig. 2H). Moreover, the migration and invasion of HTR-8/SVneo cells were also enhanced (P < 0.01, Fig. 2I). In addition, we overexpressed MLL1 in HTR-8/SVneo cells (P < 0.01, Supplementary Fig. 1A-B), and the results showed that overexpression of MLL1 reduced cell viability (P < 0.01, Supplementary Fig. 1C) and inhibited the invasion and migration of HTR-8/SVneo cells (P < 0.01, Supplementary Fig. 1D). The above results indicate that inhibition of MLL1 suppresses ferroptosis but promotes migration and invasion of HTR-8/SVneo cells.

Fig. 2

Inhibition of MLL1 promotes the invasion and migration of HTR-8/SVneo cells and represses ferroptosis. HTR-8/SVneo cells were transfected with sh-MLL1, with sh-NC as the control. A: Detection of transfection efficiency by qRT-PCR; then HTR-8/SVneo cells were treated with 10 µmol/L Erastin for 24 h to establish a cell model. B: Detection of MLL1 expression in cells by qRT-PCR. C: Detection of cell viability by CCK-8 assay. D-F: Detection of MDA, Fe²⁺, and GSH levels in cells by kits. G: Detection of ROS levels by fluorescence. H: Detection of MLL1, ASCL4, GPX4, and FTH1 expressions in cells by Western blot. I: Detection of cell invasion and migration by Transwell. The cell experiments were repeated three times independently. Data in panel A were analyzed by t test. Data in panels BCDEFGI were analyzed by one-way ANOVA, and data in panel H were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. **P < 0.01

MLL1 promotes RBM15 expression by increasing H3K4me3 modification; RBM15 enhances m6A modification to promote YTHDF2 binding to TRIM72 mRNA and degrade TRIM72, thereby repressing TRIM72 expression

MLL1 promotes the expression of downstream gene by increasing H3K4me3 modification, and RBM15 is highly expressed in PE [20]. We found an enrichment of MLL1 and H3K4me3 at the RBM15 promoter (H3K4me3 site is shown in Supplementary Fig. 2), and such an enrichment was weakened after MLL1 inhibition (P < 0.01, Fig. 3A). RBM15 expression was significantly elevated in placental tissues of PE mice and Erastin-treated HTR-8/SVneo cells (P < 0.01, Fig. 3B-C), but decreased after MLL1 inhibition (P < 0.01, Fig. 3B-C). m6A modification was enhanced in placental tissues of PE mice and Erastin-treated HTR-8/SVneo cells, but attenuated after inhibiting MLL1 expression (P < 0.01, Fig. 3D). It has been reported that upregulation of TRIM72 contributes to trophoblast migration and proliferation [26]. YTHDF2, as a “reader” of m6A modification, can bind to m6A modification and promote mRNA degradation. We observed m6A modification and YTHDF2 enrichment on TRIM72 mRNA (Supplementary Fig. 3), and the enrichment was weakened after inhibition of MLL1 (P < 0.01, Fig. 3E). Also, the TRIM72 mRNA stability was decreased in HTR-8/SVneo cells treated with Erastin, but increased after inhibition of MLL1 (P < 0.01, Fig. 3F). Further, we inhibited YTHDF2 expression in HTR-8/SVneo cells (P < 0.01, Fig. 3G-H), which led to a decrease in the enrichment of YTHDF2 on TRIM72 mRNA (P < 0.01, Fig. 3E), an enhancement in TRIM72 mRNA stability (P < 0.01, Fig. 3F), and an increase in TRIM72 protein expression (P < 0.01, Fig. 3C). Corresponding to the mRNA stability of TRIM72, the expression of TRIM72 was declined in PE but elevated after inhibition of MLL1 (P < 0.01, Fig. 3B-C). The above results show hat MLL1 promotes the expression of RBM15 by increasing H3K4me3 modification and then increases m6A modification to promote YTHDF2 binding to TRIM72 mRNA and facilitate its degradation, thereby repressing the expression of TRIM72.

Fig. 3

MLL1 promotes RBM15 expression by increasing H3K4me3 modification; RBM15 enhances m6A modification to promote YTHDF2 binding to TRIM72 mRNA and degrade TRIM72, thereby inhibiting TRIM72 expression. A: Analysis of the enrichment of MLL1 and H3K4me3 on RBM15 promoter in tissues (n = 6) and cells (n = 3) by ChIP. B-C: Detection of RBM15 and TRIM72 expressions in tissues (n = 6) and cells (n = 3) by qRT-PCR and Western blot. D: Quantitative analysis of m6A modification in tissues (n = 6) and cells (n = 3). E: Analysis of the m6A modification on TRIM72 mRNA and the enrichment of YTHDF2 in tissues (n = 6) and cells (n = 3) by RIP. F: Detection of TRIM72 mRNA stability in cells (n = 3) by actinomycin D; HTR-8/SVneo cells were transfected with sh-YTHDF2, with sh-NC as the control. G-H: Detection of YTHDF2 expression in cells (n = 3) by qRT-PCR and Western blot. Data in panels AEF were analyzed by two-way ANOVA, and data in panels BCD were analyzed by one-way ANOVA, followed by Tukey’s multiple comparisons test. Data in panels GH were analyzed by t test. **P < 0.01

Overexpression of RBM15 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells

We transfected RBM15 overexpression vector into HTR-8/SVneo cells (P < 0.01, Fig. 4A-B) and then performed a combined experiment with sh-MLL1 to verify the above mechanism. Compared with sh-MLL1 transfection alone, the combined treatment significantly diminished TRIM72 expression (P < 0.01, Fig. 4B), weakened cell viability (P < 0.05, Fig. 4C), augmented ferroptosis (P < 0.01, Fig. 4B, D-G), and abated cell migration and invasion (P < 0.05, Fig. 4H). These results suggest that overexpression of RBM15 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells.

Fig. 4

Overexpression of RBM15 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells. HTR-8/SVneo cells were transfected with oe-RBM15, with oe-NC as the control. A: Detection of transfection efficiency by qRT-PCR; then HTR-8/SVneo cells were treated with sh-MLL1 for a combined experiment. B: Detection of RBM15, TRIM72, ASCL4, GPX4, and FTH1 in cells by Western blot. C: Detection of cell viability by CCK-8 assay. D-F: Detection of MDA, Fe²⁺, and GSH levels in cells by kits. G: Detection of ROS levels in cells by fluorescence. H: Detection of cell invasion and migration by Transwell. The cell experiments were repeated three times independently. Data in panel A were analyzed by t test. Data in panels CDEFGH were analyzed by one-way ANOVA, and data in panel B were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01

Inhibition of TRIM72 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells

Similarly, we transfected sh-TRIM72 into cells (P < 0.01, Fig. 5A-B) and then performed a combined experiment with sh-MLL1. Compared with sh-MLL1 transfection alone, the combined treatment resulted in reduced cell viability (P < 0.01, Fig. 5C), enhanced ferroptosis (P < 0.01, Fig. 5B, D-G), and weakened cell migration and invasion (P < 0.05, Fig. 5H). These results suggest that inhibition of TRIM72 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells.

Fig. 5

Inhibition of TRIM72 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells. HTR-8/SVneo cells were transfected with sh-TRIM72, with sh-NC as the control. A: Detection of transfection efficiency by qRT-PCR; then HTR-8/SVneo cells were treated with sh-MLL1 for a combined experiment. B: Detection of TRIM72, ASCL4, GPX4, and FTH1 expressions in cells by Western blot. C: Detection of cell viability by CCK-8 assay. D-F: Detection of MDA, Fe²⁺, and GSH levels in cells by kits. G: Detection of ROS levels in cells by fluorescence. H: Detection of cell invasion and migration by Transwell. The cell experiments were repeated three times independently. Data in panel A were analyzed by t test. Data in panels CDEFGH were analyzed by one-way ANOVA, and data in panel B were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01

TRIM72 ubiquitylates ADAM9 to inhibit its protein expression

The ubiquitin ligase TRIM72 could bind to ADAM9 in HTR-8/SVneo cells (P < 0.01, Fig. 6A). ADAM9 is reported to be highly expressed in PE [33]. We speculated that TRIM72 could bind to ADAM9 and ubiquitinate it for degradation. The ADAM9 ubiquitination level was low in placental tissues of PE mice and Erastin-treated HTR-8/SVneo cells (Fig. 6B), while the ADAM9 ubiquitination level was increased after inhibition of MLL1 but decreased after overexpression of RBM15 or inhibition of TRIM72 (Fig. 6B). Further, we used CHX to treat HTR-8/SVneo cells and harvested ells at the indicated time (0, 2, 4, 8 h) for Western blot. The results unveiled that inhibition of TRIM72 significantly prolonged the half-life of ADAM9 (P < 0.01, Fig. 6C) and decreased the ubiquitination of ADAM9 (Fig. 6B). Overexpression of TRIM72 promoted the ubiquitination of ADAM9 (Fig. 6B), while the ubiquitination of ADAM9 was significantly abated after adding MG132 (Fig. 6B). The change of TRIM72 expression (P < 0.01, Fig. 6D) did not affect the mRNA expression of ADAM9 (P > 0.05, Fig. 6E), but affected the protein expression of ADAM9 (P < 0.01, Fig. 6F). The above results indicate that TRIM72 could bind to ADAM9 to promote its ubiquitination degradation, thus repressing the protein expression of ADAM9.

Fig. 6

TRIM72 binds to ADAM9 and ubiquitinates it for degradation, thereby inhibiting the protein expression of ADAM9. A: Detection of the binding between TRIM72 and ADAM9 in cells by Co-IP. B: Detection of ADAM9 ubiquitination level in tissues (n = 6) and cells (n = 3). C: Cycloheximide (CHX) was used to treat sh-TRIM72-transfected HTR-8/SVneo cells, and cells were harvested at the indicated times (0, 2, 4, 8 h) for Western blot analysis. D: Detection of transfection efficiency of sh-TRIM72 and oe-TRIM72 by Western blot. E-F: Detection of ADAM9 expression in tissues (n = 6) and cells (n = 3) by qRT-PCR and Western blot. The cell experiments were repeated three times independently. Data in panels DEF were analyzed by one-way ANOVA, and data in panel C were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. **P < 0.01

Overexpression of ADAM9 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells

We overexpressed ADAM9 in HTR-8/SVneo cells (P < 0.01, Fig. 7A-B) and then performed a combined experiment with sh-MLL1. Compared with sh-MLL1 transfection alone, the combined treatment resulted in reduced cell viability (P < 0.05, Fig. 7C), enhanced ferroptosis (P < 0.01, Fig. 7B, D-G), and weakened cell migration and invasion (P < 0.05, Fig. 7H). These results suggest that overexpression of ADAM9 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells.

Fig. 7

Overexpression of ADAM9 partially reverses the inhibitory effect of MLL1 inhibition on ferroptosis of HTR-8/SVneo cells. HTR-8/SVneo cells were transfected with oe-ADAM9, with oe-NC as the control. A: Detection of transfection efficiency by qRT-PCR; then HTR-8/SVneo cells were treated with sh-MLL1 for a combined experiment. B: Detection of ADAM9, ASCL4, GPX4, and FTH1 expressions in cells by Western blot. C: Detection of cell viability by CCK-8 assay. D-F: Detection of MDA, Fe²⁺, and GSH levels in cells by kits. G: Detection of ROS levels in cells by fluorescence. H: Detection of cell invasion and migration by Transwell. The cell experiments were repeated three times independently. Data in panel A were analyzed by t test. Data in panels CDEFGH were analyzed by one-way ANOVA, and data in panel B were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01

Overexpression of RBM15 partially reversed the ameliorative effect of MLL1 inhibition on symptoms in PE mice

Finally, we overexpressed RBM15 in PE mice (P < 0.01, Fig. 8A, J) and performed a combined experiment with sh-MLL1 to verify the mechanism. Compared with inhibition of MLL1 alone, the combined treatment led to the aggravation of PE symptoms in mice, which was manifested by the elevation of blood pressure (P < 0.01, Fig. 8B), the increase of proteinuria (P < 0.01, Fig. 8C), and the decrease of placental and fetal weight (P < 0.01, Fig. 8D), accompanied by the aggravation of placental and renal tissue pathology (Fig. 8E) and the enhancement of ferroptosis in placental tissues (P < 0.05, Fig. 8F-J). Moreover, TRIM72 expression in placental tissues of mice in the combined treatment group was decreased, while ADAM9 expression was increased (P < 0.01, Fig. 8J). According to the above results, MLL1 in PE can elevate RBM15 expression by promoting H3K4me3 modification and then enhance m6A modification, thereby promoting the degradation of TRIM72 mRNA by YTHDF2 and repressing the expression of TRIM72, which suppresses the ubiquitination and degradation of ADAM9 by TRIM72, stabilizes the expression of ADAM9, and finally facilitates trophoblast ferroptosis and aggravates PE symptoms.

Fig. 8

Overexpression of RBM15 partially reversed the ameliorative effect of MLL1 inhibition on symptoms in PE mice. A murine model of PE was established by injection of sFlt-1 adenovirus via tail vein at E8.5, and the control mice were injected with adenovirus encoding Fc protein, followed by injection of oe-RBM15 adenovirus for a combined experiment with sh-MLL1, with oe-NC adenovirus as the control. A: Detection of RBM15 expression in placenta by qRT-PCR. B: Measurement of blood pressure in mice, n = 12. C: Detection of proteinuria in urine, n = 12. D: Weight of placenta and fetus, n = 12. E: Observation on the pathological structure of placenta and kidney by H&E staining and PAS staining, respectively. The black arrow indicates the placental villous vascular interstitium; ▲ indicates trophoblasts; the green arrows indicates pathological features of glomeruli in renal tissues, n = 6. F-I: Detection of ROS, MDA, Fe²⁺, and GSH levels in placenta, n = 6. J: Detection of RBM15, TRIM72, ADAM9, ASCL4, GPX4, and FTH1 expressions in placenta by Western blot, n = 6. Data in panels ACDFGHI were analyzed by one-way ANOVA, and data in panels BJ were analyzed by two-way ANOVA, followed by Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01

Discussion

PE is a pregnancy-specific syndrome associated with placental pathology, which predisposes women to cardiovascular diseases in later life [1]. Ferroptosis as a newly discovered modality of regulated cell death has been demonstrated to mediate PE pathogenesis [6]. This study found for the first time that MLL1 promoted RBM15 expression through H3K4me3 modification, which elevated the m6A modification level to enhance the degradation of TRIM72 mRNA by YTHDF2 and then repress the TRIM72 expression, thereby inhibiting the ubiquitination of ADAM9 by TRIM72, stabilizing the expression of ADAM9, and finally promoting trophoblast ferroptosis and aggravating PE symptoms (Fig. 9).

Fig. 9

MLL1 promotes RBM15 expression by increasing H3K4me3 on the RBM15 promoter; RBM15 upregulates TRIM72 mRNA m6A modification, promotes the binding between YTHDF2 and TRIM72, and then degrades TRIM72 mRNA and inhibits TRIM72 expression, thereby reducing the ubiquitination degradation of ADAM9, stabilizing the expression of ADAM9, and eventually promoting ferroptosis of trophoblasts and aggravating PE

PE is usually considered a two-stage disease. In early pregnancy, insufficient remodeling of the uterine spiral artery can lead to vascular stenosis, and later, the placenta may suffer from ischemia, hypoxia, and oxidative stress damage, thus releasing toxic circulating factors such as sFlt-1 and sEng into the maternal circulation and leading to vascular and glomerular endothelial injury [34]. Treating pregnant mice with sFlt-1-Exo or recombinant mouse sFlt-1 induces a preeclampsia-like phenotype, characterized by elevated blood pressure, proteinuria, increased plasma sFlt-1 and adverse pregnancy outcomes [35]. Taking into account the operability and funding of the experiment, we chose injection of sFlt-1 adenovirus to establish a murine model of PE. Dysregulation of cell biological processes within the villous trophoblasta of the placenta results in complex processes that end up in a cell death spectrum [36]. Trophoblasts are prone to ferroptosis, and trophoblast ferroptosis is a critical mechanism for the pathogenesis of PE [37]. Previous data have revealed that iron-dependent lipid peroxidation is augmented in PE placenta, and ferroptosis arises in the placenta of PE patients [38,39,40]. MLL1 is a histone methyltransferase responsible for catalyzing H3K4me3 mark at the gene promoter and modulating target gene expression [14]. A previous study has unveiled a novel epigenetic role of MLL1 regulating trophoblast syncytialization, a pivotal process closely related to several pregnancy-specific disorders, including PE [18]. MLL1 expression is notably upregulated in villus tissues of PE patients [18]. Consistently, our results demonstrated an upregulation of MLL1 expression in placenta of PE mice, and inhibition of MLL1 ameliorates symptoms of PE mice and alleviates ferroptosis in placental tissues. Moreover, we cultured HTR-8/SVneo cells in vitro and established an in vitro cell model by adding ferroptosis inducer Erastin, and found that inhibiting MLL1 accelerated HTR-8/SVneo cell invasion and migration and repressed ferroptosis.

Thereafter, we shifted to investigating the mechanism of MLL1 mediating trophoblast ferroptosis. m6A is recognized as the most universal posttranscriptional modification in eukaryotes and functionally mediates fundamental biological processes of PE [41]. A plenty of m6A regulatory proteins have been demonstrated to be differentially expressed in PE, among which G3BP1, YTHDF2, and RBM15 expressions are dramatically increased in the PE group [20]. m6A methyltransferase RBM15 is mainly responsible for installing m6A modification, and YTHDF2 as the m6A reader protein can bind to m6A modification and promote mRNA degradation. We found an enrichment of MLL1 and H3K4me3 on the RBM15 promoter, and MLL1 inhibition weakened such an enrichment. RBM15 expression and m6A modification in Erastin-treated HTR-8/SVneo cells and PE mouse placental tissues were significantly enhanced, but were attenuated after MLL1 inhibition.

TRIM72, a member of the tripartite motif protein family, possesses diverse functionalities including membrane repair, anti-inflammation, and E3 ubiquitin ligase property [42]. Particularly, the E3 ubiquitin ligase property of TRIM72 impacts the pathogenesis of certain diseases under the conditions of inflammation, hypoxia, and oxidative stress [25, 43]. m6A modification and YTHDF2 enrichment were observed on TRIM72 mRNA, and the mRNA stability and expression of TRIM72 in Erastin-treated HTR-8/SVneo cells were decreased. Inhibition of MLL1 reduced YTHDF2 enrichment and improved mRNA stability and expression of TRIM72. Further, inhibition of YTHDF2 in HTR-8/SVneo cells resulted in decreased YTHDF2 enrichment on TRIM72 mRNA, enhanced TRIM72 mRNA stability, and increased TRIM72 protein expression. These results suggested that MLL1 promoted RBM15 expression by increasing H3K4me3 modification and then enhanced m6A modification to facilitate YTHDF2 binding to TRIM72 mRNA, thus degrading TRIM72 mRNA and repressing TRIM72 expression. RBM15 modulates the functions of trophoblasts by enhancing the binding between YTHDF2 and CD82 3’UTR to diminish CD82 expression [20]. TRM72 alleviates apoptosis but elevates proliferation and migration of trophoblasts by directly promoting P53 ubiquitination degradation [26]. Similarly, our functional rescue experiments showed that overexpression of RBM15 or inhibition of TRIM72 partially reversed the inhibitory effect of MLL1 inhibition on ferroptosis in HTR-8/SVneo cells.

ADAM9 is a membrane-anchored protein that mediates various physiological processes through the disintegrin domain for adhesion and the metalloproteinase domain for shedding of cell surface proteins [44]. ADAM9 exhibits aberrant expression patterns in PE, and PE patients are about three times more likely to have mutations in ADAM9 gene than normal puerperae [45]. ADAM9 has been reported to be highly expressed in PE and is associated with the proliferation, invasion, migration, and angiogenesis ability of trophoblast [33, 46]. We found that TRIM72 could bind to ADAM9 in HTR-8/SVneo cells. The ubiquitination level of ADAM9 in placental tissues of PE mice and Erastin-treated HTR-8/SVneo cells was low. Inhibition of TRIM72 prolonged the half-life of ADAM9 and declined the ubiquitination level of ADAM9, while overexpression of TRIM72 promoted the ubiquitination of ADAM9. The ubiquitination level of ADAM9 was notably reduced after adding MG132. The change of TRIM72 expression did not affect the mRNA expression of ADAM9, but affected the protein expression of ADAM9. The above results indicated that TRIM72 could bind to ADAM9 for ubiquitination degradation to depress the protein expression of ADAM9. ADAM9 is highly expressed in PE, and functionally, ADAM9 is deubiquitinated by USP22 to suppress the proliferation, migration, and invasion of trophoblasts [33]. Silence of ADAM9 abolishes the inhibition of trophoblast cell progression triggered by miR-223-3p inhibitor [46]. We found that overexpression of ADAM9 partially reversed the inhibitory effect of MLL1 inhibition on ferroptosis and the promotion effect on cell invasion and migration in HTR-8/SVneo cells.

Conclusion

To conclude, MLL1 promotes placental trophoblast ferroptosis and aggravates PE symptoms via epigenetic regulation of RBM15/TRIM72/ADAM9 axis. Our findings may provide a theoretical basis for the clinical treatment of PE and the development of novel therapeutic targets. This study also has certain limitations. First of all, the rodent model of PE can not fully simulate human PE, so our conclusions need to be further supported by clinical data. The detection of human placental tissues can enhance the persuasiveness of our findings, but the collection of clinical samples requires ethical approval and informed consent from every participant involved in the study, which may consume a considerable amount of time. At present, our study is still in the preclinical stage and we have not yet carried out any related work on clinical sample collection. Secondly, gene sequencing enables the acquisition of more relevant genes, thereby enhancing the convenience and validity of the study. However, our current experimental conditions and funding do not support us in conducting this test. Thirdly, although HTR8 has been widely used in the study of trophoblast functions, HTR8 cells have evolved from extravillous trophoblasts and may not be the optimal choice for studying PE during late gestation. Fourthly, H3K4me3 modification may involve complex regulatory processes in PE, and different histone modification enzymes participate in regulating histone modifications of different genes. Similarly, m6A modification is no exception. Our manuscript currently only focuses on MLL1 regulating RBM15 through H3K4me3 modification and RBM15 regulating TRIM72 through m6A modification. Other regulated genes downstream of MLL1 or other target genes downstream of RBM15 have not been tested yet. In the future research, we will collect clinical data to verify our conclusions and also identify more downstream target genes of MLL1 and RBM15 in placental trophoblast ferroptosis of PE. In addition to trophoblast ferroptosis, we will also determine whether MLL1/RBM15/TRIM72/ADAM9 is also involved in other processes of PE, such as placental angiogenesis.