TNFα‐induced abnormal activation of TNFR/NF‐κB/FTH1 in endometrium is involved in the pathogenesis of early spontaneous abortion

Abstract Early spontaneous abortion (ESA) is one of the most common complications during pregnancy and the inflammation condition in uterine environment such as long‐term exposure to high TNFα plays an essential role in the aetiology. Ferritin heavy chain (FTH1) is considered to be closely associated with inflammation and very important in normal pregnancy, yet the underlying mechanism of how TNFα induced abortion and its relationship with FTH1 remain elusive. In this study, we found that TNFα and FTH1 were positively expressed in decidual stromal cells and increased significantly in the ESA group compared with the normal pregnancy group (NP group). Besides, TNFα expression was positively correlated with FTH1 expression. Furthermore, in vitro cell model demonstrated that high TNFα could induce the abnormal signals of TNFR/NF‐κB/FTH1 and activate apoptosis both in human endometrium stromal cells (hESCs) and in local decidual tissues. Taken together, the present findings suggest that the excessive apoptosis in response to TNFα‐induced upregulation of FTH1 may be responsible for the occurrence of ESA, and thus provide a possible therapeutic target for the treatment of ESA.

Researches have indicated that successful pregnancy requires the participation of inflammatory reaction, which is accompanied by the increase of cytokines and chemokines. [7][8][9] Tumour necrosis factor-alpha (TNFα) is thought to be one of the key cytokines and plays a vital role in the establishment and maintenance of pregnancy, whereas excessive TNFα has been linked to adverse pregnancy outcomes. Abnormally elevated levels of TNFα have been detected in both the local intrauterine and the serum of people and animals suffering abortion. 7,[10][11][12] Previous studies have proposed using the TNFα expression as a diagnostic indicator of miscarriage and applying to guide its clinical treatment. 12,13 Although TNFα blocking agents have been applied for some chronic immune-mediated inflammatory diseases during pregnancy, such as inflammatory bowel disease and psoriasis, and seem no harm for pregnancy and newborns, it is still controversial for its clinical safety. 14,15 Therefore, better understanding of the association between TNFα and ESA and its possible underlying mechanism will be helpful for avoiding the development of ESA.
Iron and iron-binding proteins such as ferritin are also closely related to inflammation regulation during pregnancy. Ferritin is a ubiquitous and highly conserved protein that can bind to numerous irons and convert them into a nontoxic and bioavailable form to store inside its protein shell. 16,17 Abnormal alteration of ferritin content is usually linked with a variety of diseases, such as anaemia, pregnancy complications, kidney disease and Parkinson's disease. [18][19][20][21][22][23] It has been reported that ferritin concentration is higher in women with ESA than women with healthy pregnancy. 24,25 Besides, increased serum ferritin is also associated with higher rate of preterm birth. 21,22 As far as we know that mammalian ferritins are consist of two subunits named ferritin heavy chain (FTH1) and ferritin light chain (FTL), of them, FTH1 was considered to play an essential role in the process of pregnancy because its expression continues to increase throughout the gestation in response to progesterone. 26 Besides, Ferreira, et al. delete the Fth gene in mice using homologous recombination and discover that the mice undergo early embryonic lethality at 9.5 gestation days. 16

| Subject sample collection
Eighteen women with normal pregnancy who chose induced abortion were enrolled as NP group, and seventeen women suffered from early spontaneous abortion were enrolled as ESA group. Their decidual tissues were collected via artificial abortion-vacuum aspiration. Ultrasound examination, menstrual cycle and blood hCG levels were used to confirm the viable intrauterine pregnancy and the gestational age of all subjects. All the subjects were enrolled according to the inclusion criteria as follows: i) The gestational age was ranged from 6 weeks to 12 weeks, ii) no malformation in reproductive tract and uterine, iii) no reproductive endocrine diseases or tract infections, and iv) no family history of heredity. This study was approved by the Ethical Review Board of West China Second University Hospital, Sichuan University. Informed consent was obtained from each subject in our study.

| Real-time PCR (RT-PCR)
Total RNA was isolated from human decidual tissues or hESCs by the TRIzol ® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and was converted to cDNAs using PrimeScript RT Reagent (Takara, Japan) in accordance with the manufacturer's instruction. SYBR Green Fast qPCR Mix (Bimake, China) was used to conduct RT-PCR on an iCycler RT-PCR Detection System (Bio-Rad Laboratories). Each assay was performed in triplicates for each sample, and GAPDH was applied as an internal control. The relative mRNA expression was quantified using the 2 −ΔΔCt method. 27 The primers for RT-PCR were listed in Table S1.

| Western blotting
The proteins of human decidual tissues or the cultured cells were extracted and separated using 12.5% SDS-polyacrylamide gels.
Whereafter, the proteins were blotted into a polyvinylidene difluoride (PVDF) membrane (Millipore) and blocked with 5% skim milk for 1 h at room temperature. Subsequently, the membrane was incubated with primary antibodies, including anti-GAPDH, anti-TNFα and anti-FTH1, for 4℃ overnight, and then incubated with corresponding secondary antibodies (goat anti-mouse IgG or goat antirabbit IgG, Abcam, US) for 1 h at room temperature. ECL Western blotting kit (Millipore) was applied to develop the blots.

| Enzyme-linked immunosorbent assay (ELISA)
The proteins were isolated from human decidual tissues, and the concentration of TNFα and FTH1 of the protein lysis buffers was detected by using Human TNF alpha Uncoated ELISA (Invitrogen, USA) and Human

| Immunofluorescence staining (IF)
For the immunofluorescence staining of human decidual tissues, the deparaffinized and rehydrated slides were boiled in 10 mM citrate buffer (pH 6.0) to enable antigen retrieval. Then, the slides were blocked with 10% normal donkey serum for 1 h at room temperature and incubated with primary antibodies which included anti-TNFα For the immunofluorescent staining of cultured cells, the slides were fixed with 4% paraformaldehyde and punched with 0.3% Triton X-100 in PBS for 15 min. After that, the slides were blocked with 5% BSA and incubated with primary antibodies including anti-NF-κB (Proteintech, China). The following steps were consistent with the immunofluorescent staining of decidual tissues. The slides were observed and imaged by a laser scanning confocal microscope (Olympus, Japan).

| Cell counting kit-8 analysis
The proliferation and cell viability of hESCs were evaluated by cell counting kit-8 (CCK-8, Bimake, China) according to the manufacturer's instructions. 5000 cells were seeded in each well of a 96well plate and cultured overnight. After being treated with drugs, 10 μL CCK-8 was added into each well, and the optical densities which reflected the cell numbers were determined at the absorbance of 450 nm after the incubation for 2 hours.

| Apoptosis assays
After hESCs were treated with or without 10 ng/ml TNFα for 72 or 96 h, the cells were collected to stain with FITC-conjugated Annexin V and propidium iodide (PI) using Annexin V-FITC Apoptosis Detection Kit (BD Biosciences) in accordance with the recommendation of the manufacturer. The percentage of apoptotic cells was analysed by flow cytometry (BD Biosciences).

| Cell cycle analysis
Detection of the cell cycle of hESCs treated with or without 10 ng/ml TNFα for 72 or 96 h follows the instructions of Cell Cycle Analysis Kit (4A Biotech). After being treated with or without TNFα, the cells were collected and fixed with 75% ethanol overnight. Then, PI was applied to stain the cells at 37℃ for 30 min.
The determination of the cell cycle was conducted by Flow cytometry (BD Biosciences).

| Statistical analysis
Statistical analysis was performed using SPSS 25.0; all the data were presented as means ± standard deviation (SD). For the analysis of two groups, a two-tailed unpaired Student's t-test was used; for the analysis of multiple comparations, one-way ANOVA with Tukey's

| Increased expression of TNFα and FTH1 were found in the ESA group
The clinical characteristics of patients from the NP group and the ESA group were investigated and no significant differences in age and BMI between the two groups were found (Table 1)

| TNFα induced the apoptosis of human endometrium stromal cells
To further understand the effect of abnormal expression of TNFα in decidual stromal cells, 10 ng/ml TNFα was used to stimulate hESCs to imitate the intrauterine environment of patients with ESA. After being treated with TNFα for 72 h, the cell proliferation was inhibited and the inhibition was more obvious for 96 h (Figure 2A). Previous studies suggested that the inhibition of cell proliferation may be due to the increase of cell death and the arrest of cell cycles. 28 Hence, the alteration of cell death and cell cycle of hESCs treated with or without TNFα for 72 or 96 h were analysed, and the percentage of apoptotic cells were found to be increased ( Figure 2B), besides, it could be partially rescued by apoptosis inhibitor ZVAD-FMK, but not ferroptosis inhibitor ferrostatin-1 (Fer-1) or necroptosis inhibitor necrostatin-1 (Nec-1) ( Figure 2C; Figure S2A). While there was no effect on the cell cycle after being treated with TNFα ( Figure

| The expression of FTH1 was upregulated by TNFα in hESCs
Given that the expression of FTH1 and FTL were significantly positively correlated with the expression of TNFα in the human decidual tissues, we therefore aim to examine whether TNFα could regulate FTL and FTH1 expression. As expected, the expression level of FTH1 was changed throughout processing time and rose to a peak after being stimulated with 10 ng/ml TNFα for 48 h (Figure S3A,B).
However, 48 h treatment of TNFα did not induce the upregulation of FTL ( Figure S3C). Subsequently, different concentrations of TNFα Nanobody (TNB) were used to attenuate the stimulation of TNFα, of which 2 μg/ml TNB showed complete neutralization ( Figure S3D).
Therefore, 2 μg/ml TNB was used to further verify the action of

| Apoptosis was mediated by TNFα/TNFR/NF-κB/FTH1 axis in hESCs
As literature reported, engagement of TNFα with its cognate receptors TNFR1 and TNFR2 results in the recruitment of many important adaptor proteins like TRADD, TRAF2, RIP and FADD, thereby activating and initiating downstream events leading to apoptosis, as well as NF-κB activation, and so on. 30,31 To further understand the mechanism of TNFα-induced apoptosis and the function of FTH1 in it, the downstream molecules were explored in the hESCs stimulated by TNFα. RT-PCR and Western blotting showed that the expression of TNFR1 and TNFR2 was increased both at the transcription level and at the protein level ( Figure 4A,B). In addition, NF-κB was confirmed to translocate from the cytoplasm to the nucleus by immunofluorescence staining (Figure 4C), suggesting the activation of NF-κB by TNFα. Furthermore, pro-caspase3 was also induced by TNFα ( Figure 4C), indicating the occurrence of apoptosis. The above results had demonstrated that FTH1 was upregulated by TNFα; thus, we speculated that FTH1 was the downstream target of NF-κB through transcriptional regulation. JASPAR database (http://jaspar.gener eg.net) was used to predict the transcription factor-binding sites in the FTH1 promoter sequences, 32 and NF-κB was found to be likely to bind with the promoter of FTH1 (Table S2). To validate the binding, NF-κB inhibitor maslinic acid (MA) was applied to repress the activation of NF-κB. After being treated with MA, the nucleus translocation of NF-κB was suppressed ( Figure 4C); meanwhile, the increased expression of FTH1 was inhibited ( Figure 4D,E). Together, the results support the notion that TNFα induces apoptosis through mediating of TNFR/NF-κB/FTH1 axis in hESCs.

| Validation of abnormal TNFR/NF-κB/FTH1/ apoptosis signals in human decidual tissues
To elaborate that the occurrence of ESA was closely associated with the function of TNFα/TNFR/NF-κB/FTH1/apoptosis signalling, the induction of this pathway was further testified in human decidual tissues. The immunofluorescence staining showed a significant enhancement of TNFR1 and TNFR2 in human decidual tissues from the ESA group ( Figure 5A). Additionally, NF-κB was nucleus translocated ( Figure 5B), suggesting the activation of its transcriptional activity in the ESA group. Besides, decidual tissues from the ESA group exhibited a strongly positive staining of caspase3 ( Figure 5C) and an obvious increase of TUNEL positive cells ( Figure 5D), indicating the exacerbated apoptosis. These results were consistent with those data in hESCs and further verified the findings that the abnormal activation of TNFα/TNFR/NF-κB/FTH1/apoptosis signalling may account for the pathogenesis of ESA, at least in some cases ( Figure 6).

| DISCUSS ION
Previous evidences have emphasized the important role of ferritin in pregnancy; the imbalance of ferritin levels is often associated with adverse pregnancy outcomes. 21,22,24,25 Unfortunately, little researches focus on the functions of its subunits FTH1 during pregnancy. So far, just one literature has reported that FTH1 deletion would contribute to early embryonic lethality in mice, 16    Decidua (a specialized endometrium) is an abundant source of cytokines, including TNFα, interleukin-1α (IL-1α), interleukin-1β (IL-1β), interleukin-6 (IL-6), interferonγ (IFNγ) and so on. 44 The disorders of these cytokines would lead to adverse outcomes, containing miscarriage, preeclampsia, premature labour and so on. 45,46 However, the content of cytokines other than TNFα was not detected in our study, especially interleukin-1β (IL-1β), interleukin-6 (IL-6) and interferonγ (IFNγ), which are also considered risk stimuli of apoptosis and pregnancy loss. 7,47 Moreover, it has been reported that the FTH1 could be triggered by IL-1α, IL-1β, IL-6 and IFNγ as well. 17,[48][49][50] Therefore, it was difficult to conclude that the occurrence of ESA was just accounted for TNFα-induced upregulation of FTH1. Since the literatures have proposed that FTH1 could be regarded as a marker to reflect iron stores of the body, and plays an important role in balancing iron homeostasis, 51 it was suspectable that the increase of apoptosis may be related to the disorder of iron metabolism induced by dysregulation of FTH1. More studies are needed to understand the detailed mechanism underlying the association among inflammation, FTH1 and iron metabolism during pregnancy.
In conclusion, our study has found that the level of TNFα is significantly increased in the decidual tissues from the ESA group, and long-term stimulation of high TNFα level is proved to induce the aberrant activation of TNFR/NF-κB/FTH1 signals, which further led to the excessive apoptosis of the decidua, resulting in unstable implantation and ESA. This is the first study linking the TNFα-upregulated FTH1 to the development of ESA, providing a possible target of drug therapy for ESA.

ACK N OWLED G EM ENTS
The authors thank all the patients participate in this study.