Hypoxia causes trophoblast cell ferroptosis to induce miscarriage through lnc-HZ06/HIF1α-SUMO/NCOA4 axis.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, [email protected] (H. Zhang).

Peng Tian: Writing – original draft, Validation, Investigation, Conceptualization. Zhongyan Xu: Writing – original draft, Validation, Investigation, Conceptualization. Jiarong Guo: Validation, Investigation. Jingsong Zhao: Methodology. Weina Chen: Methodology. Wenxin Huang: Methodology. Manli Wang: Methodology, Investigation. Chenyang Mi: Methodology. Ying Zhang: Resources, Methodology. Yang Yang: Resources, Methodology. Huidong Zhang: Writing – review & editing, Funding acquisition, Data curation, Conceptualization.

In order to compare differences and explore the potential pathogenesis of unexplained recurrent miscarriage (RM), we collected villous tissue samples from unexplained RM patients with gestation in 10-11 weeks and their matched healthy control (HC) groups (each n = 30). The known causes, such as chromosomal abnormalities, hormonal abnormalities and uterine deformation, have been excluded [ [46] , [47] , [54] ]. During 10-11 weeks of healthy gestation, placenta has reached to a normoxic condition (about 8-11 % O 2 ) [ 23 , 24 ]. However, the continuous hypoxia in placenta might be associated with miscarriage [ 59 ]. Therefore, in this study, we focused on their association and potential mechanisms. The protein levels of HOXA10, a miscarriage indicator that is generally lowly expressed in RM [ 60 ], were significantly lower in unexplained RM vs HC villous tissues ( Fig. 1 A and B). The protein levels of HIF1α were higher in RM vs HC tissues ( Fig. 1 A, C, D), showing that the levels of hypoxia were higher in RM vs HC villous tissues. However, the mRNA levels of HIF1α were almost unchanged in both groups ( Fig. S1A ). Moreover, Pearson correlation analysis showed that the relative protein levels of HIF1α were negatively correlated with those of HOXA10 in RM group ( Fig. 1 E). Therefore, these data showed that the levels of hypoxia were higher in unexplained RM vs HC villous tissues and the hypoxia levels in villous tissues in 10-11-week gestation may be associated with the occurrence of miscarriage. Fig. 1 Higher levels of hypoxia and ferroptosis in unexplained RM vs HC villous tissues. (A) Western blot analysis of the protein levels of HOXA10, HIF1α, GPX4, and Ferritin in unexplained RM and HC villous tissues (n = 30), with Actin as internal standard. (B, D, F, G) The relative expression levels of each protein band were quantified and plotted in both RM and HC groups. (C) IHC analysis and relative quantification of HIF1α protein levels (brown color) in HC and unexplained RM villous tissues (n = 5). (E) Person correlation analysis of the protein levels of HIF1α and HOXA10 in unexplained RM and HC villous tissues (n = 12). (H-K) Analysis of MDA levels (H), GPx activities (I), free Fe 2+ ion levels (J), and COX-2 mRNA levels (K) in unexplained RM and HC villous tissues (n = 30). (L and M) Person correlation analysis of HOXA10 protein levels with MDA levels or free Fe 2+ levels in unexplained RM and HC villous tissues (n = 12). (N and O) Person correlation analysis of HIF1α protein levels with MDA levels or free Fe 2+ levels in unexplained RM and HC villous tissues (n = 12). HC-health control group; RM-unexplained recurrent miscarriage group. Student's t -test analysis for (B, D, F-K). Person correlation analysis for (E, L-O). P  < 0.05 was considered as significant compared with HC. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 1 Higher levels of hypoxia and ferroptosis in unexplained RM vs HC villous tissues. (A) Western blot analysis of the protein levels of HOXA10, HIF1α, GPX4, and Ferritin in unexplained RM and HC villous tissues (n = 30), with Actin as internal standard. (B, D, F, G) The relative expression levels of each protein band were quantified and plotted in both RM and HC groups. (C) IHC analysis and relative quantification of HIF1α protein levels (brown color) in HC and unexplained RM villous tissues (n = 5). (E) Person correlation analysis of the protein levels of HIF1α and HOXA10 in unexplained RM and HC villous tissues (n = 12). (H-K) Analysis of MDA levels (H), GPx activities (I), free Fe 2+ ion levels (J), and COX-2 mRNA levels (K) in unexplained RM and HC villous tissues (n = 30). (L and M) Person correlation analysis of HOXA10 protein levels with MDA levels or free Fe 2+ levels in unexplained RM and HC villous tissues (n = 12). (N and O) Person correlation analysis of HIF1α protein levels with MDA levels or free Fe 2+ levels in unexplained RM and HC villous tissues (n = 12). HC-health control group; RM-unexplained recurrent miscarriage group. Student's t -test analysis for (B, D, F-K). Person correlation analysis for (E, L-O). P  < 0.05 was considered as significant compared with HC. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) To explore whether ferroptosis might occur in unexplained RM villous tissues, several typical ferroptosis biomarkers were detected in both RM and HC tissues. The markers of ferroptosis include peroxides (free Fe 2+ , ROS, COX-2), oxidation products (MDA), and antioxidant enzymes (GPx, GPX4) [ 61 , 62 ]. The levels of MDA (malondialdehyde, the final product of lipid peroxidation), free Fe 2+ ion, and COX-2 (prostaglandin-endoperoxide synthase 2) were all higher, whereas the activity of GPx (glutathione peroxidase) and the protein levels of GPX4 (glutathione peroxidase 4) and ferritin were all lower, in RM vs HC villous tissues ( Fig. 1 A, F-K), indicating that ferroptosis levels were significantly higher in unexplained RM vs HC tissues. Moreover, Pearson correlation analysis showed that the relative levels of MDA and free Fe 2+ ion were negatively correlated with the protein levels of HOXA10 in RM group ( Fig. 1 L and M). These data implied that ferroptosis levels were higher in RM vs HC villous tissues and the ferroptosis levels in villous tissues might be associated with the occurrence of RM. Meanwhile, the relative levels of MDA and free Fe 2+ ion were also positively correlated with the protein levels of HIF1α in RM group ( Fig. 1 N and O), implying that hypoxia might be associated with ferroptosis in unexplained RM villous tissues. Taken together, the levels of hypoxia and ferroptosis were significantly higher in RM vs HC villous tissues and both of them might be associated with the occurrence of unexplained RM. To explore whether hypoxia might induce ferroptosis, we constructed hypoxic trophoblast cell model by treating Swan 71 and HTR-8/Svneo cells with 1 % O 2 or 400 μM CoCl 2 , as described previously [ 48 , 52 ]. After treatments, the levels of hypoxia indicator HIF1α were all increased ( Fig. 2 A, S2A ), indicating that the hypoxic trophoblast cells were successful constructed. In hypoxic trophoblast cells, the levels of LDH (lactate dehydrogenase), MDA, free Fe 2+ , COX-2 mRNA, and ROS were all increased ( Fig. 2 B-F), whereas cell viability and GPx activity were both decreased ( Fig. 2 G and H). Treatment with Erastin (an agonist of ferroptosis) further enhanced these changes ( Fig. 2 B-H). However, treatment with Fer-1 (an inhibitor of ferroptosis) diminished or abolished these changes ( Fig. 2 B-H). Treatment of trophoblast cells with CoCl 2 gave the similar results: the levels of LDH, MDA, Fe 2+ ion, COX-2 mRNA, and ROS were all increased, whereas cell viability and GPx activity were both decreased ( Figs. S2B and C ). Moreover, treatment with Fer-1 diminished these changes ( Figs. S2B and C ). Collectively, these data indicated that hypoxia indeed induced ferroptosis in human trophoblast cells. Fig. 2 Hypoxia induced ferroptosis by promoting HIF1α/NCOA4 pathway in hypoxic trophoblast cells. (A) Western blot analysis and relative quantification of HIF1α protein levels in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 3, 6, 12, or 24 h, with Tubulin as internal standard. (B-H) Analysis of cytotoxicity (B), MDA levels (C), free Fe 2+ levels (D), COX-2 mRNA levels (E), ROS levels (F), cell viability (G), and GPx activity (H) in Swan 71 or HTR-8/SVneo cells treated with DMSO or ferrostatin-1 (Fer-1, 10 μM) for 2 h and then treated with normoxia, hypoxia, or Erastin (5 μM) for 12 h. (I-K) Analysis of MDA levels (I), intracellular free Fe 2+ levels (J), and Ferritin protein levels (K) in Swan 71 or HTR-8/SVneo cells transfected with NC or si1-NCOA4 for 12 h and then incubated under hypoxia for 12 h. (L and M) The mRNA levels of NCOA4 and protein levels of NCOA4 and HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 3, 6, 12, or 24 h. (N-Q) The mRNA levels of NCOA4 and protein levels of NCOA4 and HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown or overexpression of HIF1α, with Tubulin as internal standard. (R) ChIP assay analysis of the levels of NCOA4 promoter region enriched by HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells, with IgG as negative control; and agarose gel image of the PCR products of NCOA4 promoter region enriched by HIF1α. (S) The relative luciferase activity of HIF1α binding with wild-type (WT) or mutant (MT) HRE of NCOA4 ( Table S1 ) in hypoxic Swan 71 or HTR-8/SVneo cells. Data in (A-S) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, I-S), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (B-H). P  < 0.05 was considered as significant difference. Fig. 2 Hypoxia induced ferroptosis by promoting HIF1α/NCOA4 pathway in hypoxic trophoblast cells. (A) Western blot analysis and relative quantification of HIF1α protein levels in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 3, 6, 12, or 24 h, with Tubulin as internal standard. (B-H) Analysis of cytotoxicity (B), MDA levels (C), free Fe 2+ levels (D), COX-2 mRNA levels (E), ROS levels (F), cell viability (G), and GPx activity (H) in Swan 71 or HTR-8/SVneo cells treated with DMSO or ferrostatin-1 (Fer-1, 10 μM) for 2 h and then treated with normoxia, hypoxia, or Erastin (5 μM) for 12 h. (I-K) Analysis of MDA levels (I), intracellular free Fe 2+ levels (J), and Ferritin protein levels (K) in Swan 71 or HTR-8/SVneo cells transfected with NC or si1-NCOA4 for 12 h and then incubated under hypoxia for 12 h. (L and M) The mRNA levels of NCOA4 and protein levels of NCOA4 and HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 3, 6, 12, or 24 h. (N-Q) The mRNA levels of NCOA4 and protein levels of NCOA4 and HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown or overexpression of HIF1α, with Tubulin as internal standard. (R) ChIP assay analysis of the levels of NCOA4 promoter region enriched by HIF1α in hypoxic Swan 71 or HTR-8/SVneo cells, with IgG as negative control; and agarose gel image of the PCR products of NCOA4 promoter region enriched by HIF1α. (S) The relative luciferase activity of HIF1α binding with wild-type (WT) or mutant (MT) HRE of NCOA4 ( Table S1 ) in hypoxic Swan 71 or HTR-8/SVneo cells. Data in (A-S) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, I-S), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (B-H). P  < 0.05 was considered as significant difference. Recently, it has been reported that NCOA4 (nuclear receptor coactivator 4) [ 63 ] could promote ferritin degradation, increase intracellular Fe 2+ levels, and ultimately induce ferroptosis in glioblastoma cells [ 35 ]. However, whether NCOA4 might induce ferroptosis in hypoxic trophoblast cells was completely unclear. After NCOA4 was silenced by transfecting with its two distinguished siRNAs in hypoxic Swan 71 or HTR-8/Svneo cells ( Figs. S2D and E ), the levels of MDA, Fe 2+ ion, LDH, and ROS were all decreased; whereas the levels of ferritin, GPx activity, and cell viability were all increased ( Fig. 2 I-K, S2F-I). Overexpression of NCOA4 gave the opposite results ( Figs. S2J and K ). These data showed that NCOA4 promoted ferroptosis in hypoxic human trophoblast cells. By aligning sequence, there was an HRE (Hypoxia-Responsive Element, 5’-ACGTG-3’) in the promoter region of NCOA4 (sequence in Table S1 ). This inspired us to hypothesize that HIF1α might act as a transcription factor of NCOA4 to induce ferroptosis through HIF1α/NCOA4 pathway in hypoxic trophoblast cells. To verify this, we found that the mRNA and protein levels of NCOA4 were higher in hypoxic trophoblast cells ( Fig. 2 L and M). Moreover, knockdown of HIF1α by transfecting trophoblast cells with its siRNAs reduced the mRNA and protein levels of NCOA4; and overexpression of HIF1α increased its levels in hypoxic trophoblast cells ( Fig. 2 N-Q). ChIP assays showed that the promoter region of NCOA4 could be enriched by HIF1α in hypoxic trophoblast cells ( Fig. 2 R). Luciferase reporter assays further indicated that HIF1α performed transcription using wild-type but not mutant HRE of NCOA4 in hypoxic trophoblast cells ( Fig. 2 S). Collectively, these data indicated that HIF1α acted as a transcription factor of NCOA4 and hypoxia induced ferroptosis through HIF1α/NCOA4 pathway in hypoxic trophoblast cells. It has been reported that the SUMOylation of HIF1α increases its transcriptional capability possibly by increasing HIF1α protein stability [ 32 ]. However, whether the SUMOylated HIF1α (termed as HIF1α-SUMO) itself could directly act as a stronger transcription factor than HIF1α is still unexplored. Our data showed that, in IP assays using identical but limited amounts of HIF1α antibody, the protein levels of HIF1α-SUMO IPed by HIF1α, together with the protein levels of NCOA4 in Input, were consistently increased upon hypoxia but were decreased with treatment with 2-D08 (a broad-spectrum inhibitor of SUMOylation [ 64 ]) in human trophoblast cells ( Fig. 3 A, S3A, B ), indicating that HIF1α-SUMO might increase NCOA4 expression levels in hypoxic trophoblast cells. It has been reported that HIF1α was SUMOylated at its K391 and K477 residues [ 65 ]. When both lysines were mutated into arginines, the SUMOylation of HIF1α was greatly inhibited as shown in IP assays using identical but limited amounts of HIF1α antibody ( Fig. S3C ). Furthermore, overexpression of wild-type HIF1α-WT, but not mutant HIF1α-MT, significantly up-regulated the mRNA and protein levels of NCOA4; and this up-regulation caused by HIF1α-WT was fully abolished after treatment with 2-D08 ( Fig. 3 B and C). Collectively, these data indicated that HIF1α could be SUMOylated at K391 and K477 in human trophoblast cells, and this HIF1α-SUMO up-regulated NCOA4 expression levels in hypoxic trophoblast cells. Fig. 3 The SUMOylated HIF1α primarily performed NCOA4 transcription in hypoxic trophoblast cells. (A) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by HIF1α antibody in normoxic or hypoxic Swan 71 or HTR-8/SVneo cells treated with 2-D08 (50 μM), with NCOA4 protein levels in input. (B and C) The mRNA and protein levels of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT (wild type) or HIF1α-MT (mutant) and treatment with 2-D08 (50 μM). (D) Scheme of ChIP-re-ChIP assays. (E) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) or HIF1α in hypoxic Swan 71 cells in ChIP-re-ChIP assays. (F) The levels of NCOA4 promoter region enriched by HIF1α-SUMO in hypoxic Swan 71 cells with overexpression of HIF1α-WT or HIF1α-MT and treatment with 2-D08 (50 μM) in ChIP-re-ChIP assays. (G) Scheme of DNA pulldown assays. (H and I) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 promoter region in hypoxic Swan 71 or HTR-8/SVneo cells in DNA pulldown assays. (J) The relative luciferase activity of HIF1α-WT or HIF1α-MT binding with HRE of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells treated with 2-D08. Data in (A-C, E, F, I) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (E, I), Two-way ANOVA followed by Tukey's multiple comparisons test analysis for (B, C, F, J), and one-way AVOVA followed by Dunnett's multiple comparisons test analysis for (A). P  < 0.05 was considered as significant difference. Fig. 3 The SUMOylated HIF1α primarily performed NCOA4 transcription in hypoxic trophoblast cells. (A) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by HIF1α antibody in normoxic or hypoxic Swan 71 or HTR-8/SVneo cells treated with 2-D08 (50 μM), with NCOA4 protein levels in input. (B and C) The mRNA and protein levels of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT (wild type) or HIF1α-MT (mutant) and treatment with 2-D08 (50 μM). (D) Scheme of ChIP-re-ChIP assays. (E) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) or HIF1α in hypoxic Swan 71 cells in ChIP-re-ChIP assays. (F) The levels of NCOA4 promoter region enriched by HIF1α-SUMO in hypoxic Swan 71 cells with overexpression of HIF1α-WT or HIF1α-MT and treatment with 2-D08 (50 μM) in ChIP-re-ChIP assays. (G) Scheme of DNA pulldown assays. (H and I) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 promoter region in hypoxic Swan 71 or HTR-8/SVneo cells in DNA pulldown assays. (J) The relative luciferase activity of HIF1α-WT or HIF1α-MT binding with HRE of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells treated with 2-D08. Data in (A-C, E, F, I) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (E, I), Two-way ANOVA followed by Tukey's multiple comparisons test analysis for (B, C, F, J), and one-way AVOVA followed by Dunnett's multiple comparisons test analysis for (A). P  < 0.05 was considered as significant difference. To further investigate whether HIF1α-SUMO might have higher NCOA4 transcription activity than the unmodified HIF1α, we performed ChIP-re-ChIP assays using identical but excessive amounts of HIF1α antibody or SUMO antibody. In the first ChIP assays, HIF1α antibody could pull down NCOA4 promoter region that bound with total HIF1α (including HIF1α and HIF1α-SUMO); and in the re-ChIP assays, SUMO antibody further pulled down NCOA4 promoter region that bound with only HIF1α-SUMO ( Fig. 3 D). It was found that NCOA4 promoter region was primarily bound with HIF1α-SUMO, which accounted for about 55 %-63 % of total HIF1α ( Fig. 3 E, S3D ). Moreover, the binding of HIF1α-SUMO on NCOA4 promoter region was further enhanced by overexpressing HIF1α-WT, but not HIF1α-MT; and this enhancement caused by HIF1α-WT was abolished after treatment with 2-D08 ( Fig. 3 F, S3E ). Furthermore, we also used biotin-labeled dsDNA containing NCOA4 promoter region to pull down HIF1α-SUMO and total HIF1α in hypoxic trophoblast cells ( Fig. 3 G). Analysis of HIF1α and HIF1α-SUMO by Western blotting confirmed that this dsDNA region was primarily bound with HIF1α-SUMO because more HIF1α-SUMO than HIF1α was enriched by this dsDNA probe ( Fig. 3 H and I). Finally, luciferase reporter assays showed that overexpression of HIF1α-WT, but not HIF1α-MT, enhanced the transcription activity on HRE of NCOA4 promoter; however, this enhancement caused by HIF1α-WT overexpression was diminished with 2-D08 treatment ( Fig. 3 J). Taken together, it was HIF1α-SUMO that primarily performed NCOA4 transcription in hypoxic trophoblast cells. Having known that HIF1α-SUMO primarily performed NCOA4 transcription, whether this process was regulated by lncRNA was further explored. Recently, we have identified a group of novel lncRNAs that play important roles in regulation of trophoblast cell dysfunctions and the occurrence of miscarriage [ [41] , [42] , [43] , [45] , [46] , [47] , [66] , [67] , [68] ], including lnc-HZ06 (located in Chr 22, 46,082,078–46,082,284, submitted to NCBI with accession number of MT874978 ). Lnc-HZ06 was highly expressed in unexplained RM vs HC villous tissues; and GO analysis of the differentially expressed mRNAs in lnc-HZ06-silenced Swan 71 cells showed that knockdown of lnc-HZ06 might alter iron ion binding capability ( Fig. S4A ) [ 57 ]. This inspired us to explore whether lnc-HZ06 might regulate ferroptosis in human trophoblast cells. Firstly, we found that lnc-HZ06 was highly expressed with incubation period in hypoxic trophoblast cells ( Fig. 4 A), but unchanged in normoxic trophoblast cells ( Fig. S4B ). Subsequently, overexpression of lnc-HZ06 up-regulated the levels of MDA, Fe 2+ ion, LDH, COX-2, and ROS, and down-regulated ferritin, GPx activity and cell viability in hypoxic trophoblast cells ( Fig. 4 B, C, S4C-G), indicating that lnc-HZ06 promoted hypoxic trophoblast ferroptosis. However, the promotion effects were diminished after treatment with Fer-1 ( Fig. 4 B, C, S4D, F, G ). In contrast, knockdown of lnc-HZ06 down-regulated the levels of MDA, Fe 2+ ion, LDH, COX-2, and ROS, and up-regulated ferritin, GPx activity and cell viability in hypoxic trophoblast cells ( Fig. S4H -M). Therefore, these data indicated that lnc-HZ06 promoted ferroptosis in hypoxic trophoblast cells. Fig. 4 Lnc-HZ06 promoted HIF1α SUMOylation by suppressing SENP1 in hypoxic trophoblast cells. (A) RT-qPCR analysis of the relative expression levels of lnc-HZ06 in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 6, 12, or 24 h. (B and C) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and treatment with Fer-1 (10 μM) for 2 h. (D and E) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by identical but limited amount of HIF1α antibody in normoxic or hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06, with total HIF1α protein levels in input. (F and G) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of lnc-HZ06, with total HIF1α protein levels in input. (H and I) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and HIF1α-WT or HIF1α-MT, with HIF1α protein levels in input. (J and K) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of SENP1. (L and M) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of SENP1 and overexpression of HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08, with HIF1α protein levels in input. (N and O) The protein levels of SENP1 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression or knockdown of lnc-HZ06, with Tubulin as internal standard. (P and Q) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of SENP1, with HIF1α protein levels in input. Data in (A-Q) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, K, N, O), One-way AVOVA followed by Tukey's multiple comparisons test analysis for (B-F, P, Q), and two-way ANOVA followed by Sidak's multiple comparisons test analysis for (G-I, L, M). P  < 0.05 was considered as significant difference. Fig. 4 Lnc-HZ06 promoted HIF1α SUMOylation by suppressing SENP1 in hypoxic trophoblast cells. (A) RT-qPCR analysis of the relative expression levels of lnc-HZ06 in hypoxic Swan 71 or HTR-8/SVneo cells for 0, 6, 12, or 24 h. (B and C) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and treatment with Fer-1 (10 μM) for 2 h. (D and E) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by identical but limited amount of HIF1α antibody in normoxic or hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06, with total HIF1α protein levels in input. (F and G) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of lnc-HZ06, with total HIF1α protein levels in input. (H and I) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and HIF1α-WT or HIF1α-MT, with HIF1α protein levels in input. (J and K) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of SENP1. (L and M) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of SENP1 and overexpression of HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08, with HIF1α protein levels in input. (N and O) The protein levels of SENP1 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression or knockdown of lnc-HZ06, with Tubulin as internal standard. (P and Q) The protein levels of HIF1α-SUMO that was immunoprecipitated by identical but limited amount of HIF1α antibody in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of SENP1, with HIF1α protein levels in input. Data in (A-Q) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, K, N, O), One-way AVOVA followed by Tukey's multiple comparisons test analysis for (B-F, P, Q), and two-way ANOVA followed by Sidak's multiple comparisons test analysis for (G-I, L, M). P  < 0.05 was considered as significant difference. Since HIF1α-SUMO primarily performed NCOA4 transcription, whether lnc-HZ06 might also regulate HIF1α-SUMO levels was explored. In IP assays using identical but limited amounts of HIF1α antibody, lnc-HZ06 overexpression increased, whereas lnc-HZ06 knockdown reduced, HIF1α-SUMO protein levels in hypoxic trophoblast cells ( Fig. 4 D-G). In details, lnc-HZ06 enhanced the SUMOylation of HIF1α-WT but had little effect on that of HIF1α-MT ( Fig. 4 H and I). It was reported that SENP1 was a deSUMO-specific protease of HIF1α in MEF cells [ 69 ]. In trophoblast cells, we found that knockdown of SENP1 up-regulated the level of HIF1α-SUMO in IP assays ( Fig. 4 J, K, S4N, O). Moreover, the enhancement of SUMOylation of HIF1α-WT, but not HIF1α-MT, caused by silencing SENP1 was diminished with 2-D08 treatment in hypoxic trophoblast cells ( Fig. 4 L and M). Furthermore, overexpression of lnc-HZ06 down-regulated, whereas knockdown of lnc-HZ06 up-regulated, the protein levels of SENP1 in hypoxic trophoblast cells ( Fig. 4 N and O). Co-transfection assays showed that the promotion in SUMOylation of HIF1α caused by lnc-HZ06 overexpression was further enhanced by silencing SENP1 in hypoxic trophoblast cells ( Fig. 4 P and Q). Taken together, these data demonstrated that lnc-HZ06 up-regulated the level of HIF1α-SUMO by suppressing SENP1-mediated deSUMOlylation. Subsequently, we explored how lnc-HZ06 regulated NCOA4 expression in hypoxic trophoblast cells. Overexpression of lnc-HZ06 increased the mRNA and protein levels of NCOA4 in hypoxic trophoblast cells ( Fig. 5 A, S5A ). In contrast, knockdown of lnc-HZ06 decreased NCOA4 levels ( Fig. 5 B, S5B ). However, the increase in NCOA4 expression levels caused by lnc-HZ06 overexpression was abolished by silencing HIF1α ( Fig. 5 C, S5C ). Conversely, this increase caused by lnc-HZ06 overexpression was further enhanced by overexpressing HIF1α-WT but not affected by HIF1α-MT; and this increase was abolished by treating cells with 2-D08 ( Fig. 5 D, S5D ). Therefore, these results suggested that lnc-HZ06 up-regulated NCOA4 expression levels through the SUMOylated HIF1α. To further confirm this, ChIP assays showed that overexpression of lnc-HZ06 enhanced the enrichment of HIF1α on NCOA4 promoter region, and this enhancement was abolished with 2-D08 treatment ( Fig. 5 E, S5E ). ChIP-re-ChIP assays (scheme in Fig. 3 D) further showed that NCOA4 promoter region could be enriched by HIF1α-SUMO, and this enrichment was further enhanced with lnc-HZ06 overexpression ( Fig. 5 F, S5F ). In details, the percentage of the NCOA4 promoter bound to HIF1α-SUMO in the total amount of NCOA4 promoter bound to total HIF1α was increased from 61 % to 84 % and from 54 % to 74 % after lnc-HZ06 was overexpressed in both hypoxic trophoblast cells ( Fig. 5 G, H, S5G , H). Using biotin-labeled dsDNA probe containing NCOA4 promoter region, it was primarily HIF1α-SUMO that was pulled down by dsDNA probe, and the pulled HIF1α-SUMO level was further increased with lnc-HZ06 overexpression ( Fig. 5 I). Luciferase reporter assays further showed that HIF1α-WT showed higher transcription activity than HIF1α-MT using NCOA4 HRE sequence, and this transcription activity of HIF1α-WT was further increased with lnc-HZ06 overexpression ( Fig. 5 J, S5I ). Collectively, the results indicated that lnc-HZ06 enhanced HIF1α-SUMO-mediated NCOA4 transcription in hypoxic trophoblast cells. Fig. 5 Lnc-HZ06 promoted HIF1α-SUMO-mediated NCOA4 transcription in hypoxic trophoblast cells. (A and B) The protein levels of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression or knockdown of lnc-HZ06. (C) The protein levels of NOCA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and knockdown of HIF1α. (D) The protein levels of NOCA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08. (E) The levels of NCOA4 promoter region enriched by HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in the presence of DMSO or 2-D08 in ChIP assays using identical but excessive amount of HIF1α antibody; and the agarose gel image of the PCR product of NCOA4 promoter region enriched by HIF1α. (F) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays. (G)The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) or total HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays using identical but excessive amount of HIF1α antibody and SUMO antibody. (H) The levels of NCOA4 promoter region enriched by HIF1α-SUMO or HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays. (I) The protein levels of HIF1α and HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 promoter region in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in DNA pulldown assays. (J) The relative luciferase activity of HIF1α-SUMO bind onto NCOA4 HRE in hypoxic Swan 71 cells with overexpression of HIF1α-WT or HIF1α-MT or co-overexpression of lnc-HZ06. Data in (A-G, I, J) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, B), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (C-G, I, J). P  < 0.05 was considered as significant difference. Fig. 5 Lnc-HZ06 promoted HIF1α-SUMO-mediated NCOA4 transcription in hypoxic trophoblast cells. (A and B) The protein levels of NCOA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression or knockdown of lnc-HZ06. (C) The protein levels of NOCA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and knockdown of HIF1α. (D) The protein levels of NOCA4 in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 and HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08. (E) The levels of NCOA4 promoter region enriched by HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in the presence of DMSO or 2-D08 in ChIP assays using identical but excessive amount of HIF1α antibody; and the agarose gel image of the PCR product of NCOA4 promoter region enriched by HIF1α. (F) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays. (G)The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) or total HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays using identical but excessive amount of HIF1α antibody and SUMO antibody. (H) The levels of NCOA4 promoter region enriched by HIF1α-SUMO or HIF1α in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in ChIP-re-ChIP assays. (I) The protein levels of HIF1α and HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 promoter region in hypoxic Swan 71 cells with overexpression of lnc-HZ06 in DNA pulldown assays. (J) The relative luciferase activity of HIF1α-SUMO bind onto NCOA4 HRE in hypoxic Swan 71 cells with overexpression of HIF1α-WT or HIF1α-MT or co-overexpression of lnc-HZ06. Data in (A-G, I, J) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, B), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (C-G, I, J). P  < 0.05 was considered as significant difference. In hypoxic trophoblast cells, lnc-HZ06 promoted trophoblast ferroptosis, as indicated by the up-regulation of the levels of MDA, Fe 2+ ion, COX-2, ROS, and LDH, and the down-regulation of cell viability and GPx activity; however, these changes were all abolished by silencing HIF1α ( Fig. 6 A-D, S6A ). Moreover, these changes were all further enhanced by overexpressing HIF1α-WT but less affected by overexpressing HIF1α-MT ( Fig. 6 E-H, S6B ). These results indicated that lnc-HZ06 promoted trophoblast ferroptosis through up-regulating the SUMOylated HIF1α. Furthermore, the promotion in ferroptosis caused by lnc-HZ06 overexpression was also diminished by silencing NCOA4 in hypoxic trophoblast cells ( Fig. 6 I-L, S6C). Collectively, these results showed that lnc-HZ06 promoted ferroptosis through HIF1α-SUMO/NCOA4 pathway in hypoxic trophoblast cells. Fig. 6 Lnc-HZ06 promoted ferroptosis by enhancing HIF1α-SUMO-mediated NCOA4 transcription in hypoxic trophoblast cells. (A-D) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of HIF1α. (E-H) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-overexpression of HIF1α-WT or HIF1α-MT. (I-L) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of NCOA4. Data in (A-L) show mean ± SD, n = 5 independent experiments. Two-way AVOVA followed by Tukey's multiple comparisons test analysis for (A-L); P  < 0.05 was considered as significant difference. Fig. 6 Lnc-HZ06 promoted ferroptosis by enhancing HIF1α-SUMO-mediated NCOA4 transcription in hypoxic trophoblast cells. (A-D) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of HIF1α. (E-H) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-overexpression of HIF1α-WT or HIF1α-MT. (I-L) Analysis of MDA and free Fe 2+ levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of lnc-HZ06 or co-knockdown of NCOA4. Data in (A-L) show mean ± SD, n = 5 independent experiments. Two-way AVOVA followed by Tukey's multiple comparisons test analysis for (A-L); P  < 0.05 was considered as significant difference. Since lnc-HZ06 was highly expressed in hypoxic trophoblast cells, it underlying mechanism was investigated. There was also a HRE in the promoter region of lnc-HZ06 (sequence in Table S1 ), implying that HIF1α or HIF1α-SUMO might act as its transcription factor. To explore this possibility, knockdown of HIF1α by its specific siRNAs down-regulated lnc-HZ06 levels in hypoxic trophoblast cells ( Fig. 7 A); and treatment of cells with 2-D08 that reduced HIF1α-SUMO levels also reduced lnc-HZ06 expression levels in hypoxic trophoblast cells ( Fig. 7 B). Furthermore, overexpression of HIF1α-WT, but not HIF1α-MT, up-regulated lnc-HZ06 levels, and this up-regulation caused by HIF1α-WT was diminished with 2-D08 treatment ( Fig. 7 C), indicating that HIF1α-SUMO promoted lnc-HZ06 expression. ChIP-re-ChIP assays (scheme in Fig. 3 C) further showed that WT HIF1α-SUMO was bound onto the promoter region of lnc-HZ06 in hypoxic trophoblast cells; and this binding was diminished with 2-D08 treatment ( Fig. 7 D and E). DNA pulldown assays showed that biotin-labeled dsDNA probe containing lnc-HZ06 promoter region mainly enriched HIF1α-SUMO ( Fig. 7 F). Luciferase reporter assays also showed that HIF1α-WT, but not HIF1α-MT, showed transcription activity using the HRE of lnc-HZ06. However, this transcription activity was abolished by treating cells with 2-D08 ( Fig. 7 G). Taken together, all these data showed that HIF1α-SUMO primarily performed lnc-HZ06 transcription in hypoxic trophoblast cells. Fig. 7 HIF1α-SUMO promoted lnc-HZ06 transcription in hypoxic trophoblast cells. (A) The RNA levels of HIF1α and lnc-HZ06 in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of HIF1α. (B) Lnc-HZ06 levels in hypoxic Swan 71 cells treated with 2-D08. (C) Lnc-HZ06 levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08. (D and E) The levels of lnc-HZ06 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT in ChIP-re-ChIP assays using identical but excessive amount of HIF1α antibody and SUMO antibody. (F) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing lnc-HZ06 promoter region in hypoxic Swan 71 or HTR-8/SVneo cells in DNA pulldown assays. (G) The relative luciferase activity of HIF1α on lnc-HZ06 HRE in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT and co-treatment with 2-D08. Data in (A- G) show mean ± SD, n = 5 independent experiments. One-way AVOVA followed by Dunnett's multiple comparisons test analysis for (A), Student's t -test analysis for (B, F), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (C-E, G). P  < 0.05 was considered as significant difference. Fig. 7 HIF1α-SUMO promoted lnc-HZ06 transcription in hypoxic trophoblast cells. (A) The RNA levels of HIF1α and lnc-HZ06 in hypoxic Swan 71 or HTR-8/SVneo cells with knockdown of HIF1α. (B) Lnc-HZ06 levels in hypoxic Swan 71 cells treated with 2-D08. (C) Lnc-HZ06 levels in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT in the presence of DMSO or 2-D08. (D and E) The levels of lnc-HZ06 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT in ChIP-re-ChIP assays using identical but excessive amount of HIF1α antibody and SUMO antibody. (F) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing lnc-HZ06 promoter region in hypoxic Swan 71 or HTR-8/SVneo cells in DNA pulldown assays. (G) The relative luciferase activity of HIF1α on lnc-HZ06 HRE in hypoxic Swan 71 or HTR-8/SVneo cells with overexpression of HIF1α-WT or HIF1α-MT and co-treatment with 2-D08. Data in (A- G) show mean ± SD, n = 5 independent experiments. One-way AVOVA followed by Dunnett's multiple comparisons test analysis for (A), Student's t -test analysis for (B, F), two-way ANOVA followed by Tukey's multiple comparisons test analysis for (C-E, G). P  < 0.05 was considered as significant difference. Lnc-HZ06 could up-regulate HIF1α-SUMO levels by suppressing its SENP1-mediated deSUMOlylation. Reversely, HIF1α-SUMO could also act as a transcription factor to promote lnc-HZ06 transcription in hypoxic trophoblast cells. Thus, both lnc-HZ06 and HIF1α-SUMO formed a positive auto-regulatory feedback loop in hypoxic trophoblast cells. Moreover, this lnc-HZ06/HIF1α-SUMO loop was up-regulated in hypoxic trophoblast cells, which further promoted the transcription of NCOA4 and induced trophoblast cell ferroptosis. To correlate the occurrence of unexplained RM with the hypoxia-induced ferroptosis of human trophoblast cells and also to verify whether the regulatory mechanisms were consistent in both hypoxic trophoblast cells and villous tissues, we detected the expression levels of lnc-HZ06, HIF1α-SUMO, and NCOA4 in RM and HC villous tissues (each n = 30). Their expression levels were all higher in unexplained RM vs HC villous tissues ( Fig. 8 A-E). After normalization of their relative expression levels, the mRNA and protein levels of NCOA4 were linearly and positively correlated with the levels of lnc-HZ06 in unexplained RM tissues ( Fig. 8 F and G). ChIP-re-ChIP assays showed that more NCOA4 promoter region was enriched by HIF1α-SUMO in unexplained RM vs HC villous tissues ( Fig. 8 H). The levels of HIF1α-SUMO that was pulled down by biotin-labeled dsDNA probe containing NCOA4 promoter region were also higher in unexplained RM vs HC tissues ( Fig. 8 I). These data showed that HIF1α-SUMO-mediated NCOA4 transcription was enhanced in RM vs HC tissues. Moreover, the levels of MDA and Fe 2+ ion were also positively correlated with the levels of lnc-HZ06 in unexplained RM tissues ( Fig. 8 J and K), implying that lnc-HZ06 might promote ferroptosis in unexplained RM tissues. Taken together, combined with the cellular results, we proposed that lnc-HZ06 might promote ferroptosis by enhancing HIF1α-SUMO-mediated NCOA4 transcription in unexplained RM tissues. Fig. 8 Lnc-HZ06/HIF1α-SUMO feedback loop was up-regulated in unexplained RM vs HC villous tissues. (A) The expression levels of lnc-HZ06 in HC and unexplained RM villous tissues (each n = 30). (B) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by HIF1α antibody in HC and unexplained RM tissues (each n = 4) in IP assays using identical but limited amount of HIF1α antibody, with HIF1α protein levels in Input. (C-E) The mRNA and protein levels of NCOA4 in HC and RM tissues (n = 30 for mRNA, n = 12 for protein, and n = 5 for IHC assays). The relative intensity of NCOA4 protein band or NCOA4 protein (brown color) in IHC images was quantified and plotted. (F and G) Person correlation analysis of the expression levels of lnc-HZ06 and the mRNA or protein levels of NCOA4 in HC and RM villous tissues (n = 30 for mRNA and n = 12 for protein). (H) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in HC and RM villous tissues (n = 4). (I) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 or lnc-HZ06 promoter region in HC and RM villous tissues in DNA pulldown assays (n = 4). (J and K) Person correlation analysis of the levels of lnc-HZ06 and MDA or free Fe 2+ in RM and HC villous tissues (n = 30). (L) The levels of lnc-HZ06 promoter region enriched by HIF1α-SUMO in RM and HC villous tissues in ChIP-re-ChIP assays (n = 4). Data in (A-D, H, I, L) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, C, D, H, I, L). Pearson correlation analysis for (F, G, J, K). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 8 Lnc-HZ06/HIF1α-SUMO feedback loop was up-regulated in unexplained RM vs HC villous tissues. (A) The expression levels of lnc-HZ06 in HC and unexplained RM villous tissues (each n = 30). (B) The protein levels of SUMOylated HIF1α (HIF1α-SUMO) that was immunoprecipitated by HIF1α antibody in HC and unexplained RM tissues (each n = 4) in IP assays using identical but limited amount of HIF1α antibody, with HIF1α protein levels in Input. (C-E) The mRNA and protein levels of NCOA4 in HC and RM tissues (n = 30 for mRNA, n = 12 for protein, and n = 5 for IHC assays). The relative intensity of NCOA4 protein band or NCOA4 protein (brown color) in IHC images was quantified and plotted. (F and G) Person correlation analysis of the expression levels of lnc-HZ06 and the mRNA or protein levels of NCOA4 in HC and RM villous tissues (n = 30 for mRNA and n = 12 for protein). (H) The levels of NCOA4 promoter region enriched by SUMOylated HIF1α (HIF1α-SUMO) in HC and RM villous tissues (n = 4). (I) The protein levels of HIF1α or HIF1α-SUMO pulled down by biotin-labeled DNA probe containing NCOA4 or lnc-HZ06 promoter region in HC and RM villous tissues in DNA pulldown assays (n = 4). (J and K) Person correlation analysis of the levels of lnc-HZ06 and MDA or free Fe 2+ in RM and HC villous tissues (n = 30). (L) The levels of lnc-HZ06 promoter region enriched by HIF1α-SUMO in RM and HC villous tissues in ChIP-re-ChIP assays (n = 4). Data in (A-D, H, I, L) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (A, C, D, H, I, L). Pearson correlation analysis for (F, G, J, K). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) To explore the roles of HIF1α-SUMO in lnc-HZ06 transcription in villous tissues, ChIP-re-ChIP assays showed that more lnc-HZ06 promoter region was enriched by HIF1α-SUMO in unexplained RM vs HC villous tissues ( Fig. 8 L). The levels of HIF1α-SUMO pulled down by biotin-labeled dsDNA probe containing lnc-HZ06 promoter region were also higher in unexplained RM vs HC tissues ( Fig. 8 I). Combined with the cellular results, we proposed that HIF1α-SUMO might act as lnc-HZ06 transcription factor and primarily perform its transcription in unexplained RM villous tissues. Taken together, high possibly, hypoxia might induce ferroptosis through lnc-HZ06/HIF1α-SUMO-mediated NCOA4 transcription in unexplained RM villous tissues. To evaluate whether hypoxia might directly induce miscarriage in vivo , we constructed hypoxic pregnant mouse model, as the methods described previously [ [55] , [56] , [57] ]. Pregnant C57BL/6 mice were divided into two groups (each n = 15): the hypoxia group that were fed in a hypoxic chamber containing 10 % O 2 and 90 % N 2 for 8 h and then with normal oxygen for 16 h every day; and the normoxia group that were always in a normoxic environment. To confirm hypoxia in mouse model, placental tissues were collected and HIF1α protein levels were detected. The mRNA and protein sequences of HIF1α were all conservative in human, mouse, dog, monkey, chicken, and zebrafish ( Figs. S7A and B ). The protein levels of murine Hif1α were higher in hypoxia group relative to normoxia group ( Fig. 9 A-C), indicating the successful construction of a hypoxic pregnant mouse model. Subsequently, the effect of hypoxia on miscarriage was investigated. Compared with normoxia group, hypoxia caused embryo adsorption (as indicated by red arrows, Fig. 9 D) and elevated the average miscarriage rates ( Fig. 9 E). Therefore, these assays confirmed that hypoxia induced mouse miscarriage. Fig. 9 Hypoxia induced miscarriage by promoting ferroptosis in hypoxic mouse model. (A) The protein levels of murine Hif1α, Gpx4, Fth1, and Ncoa4 in placental tissues of normoxia- or hypoxia-treated pregnant mice (n = 15, 6 representatives were shown), with Actin as internal standard. (B) The relative intensity of each protein band of Hif1α in normoxia- or hypoxia-treated mouse placental tissues was quantified and plotted (n = 15). (C) IHC image and relative quantification of murine Hif1α protein levels (brown color) in normoxia- or hypoxia-treated mouse placental tissues. (D) Representative uterus morphology on gestational day 13 in normoxia- or hypoxia-treated pregnant mice with embryo absorption indicated by red arrows. (E) Average miscarriage rates in normoxia- or hypoxia-treated pregnant mice (n = 15). (F) The relative expression levels of murine lnc-Hz06 in normoxia- or hypoxia-treated mouse placental tissues (n = 15). (G-K) Analysis of the levels of Ncoa4 protein, MDA, free Fe 2+ , Gpx4 protein, and Fth1 protein in placental tissues of normoxia- or hypoxia-treated pregnant mice (n = 15). (L) IHC image and relative quantification of murine Ncoa4 protein levels (brown color) in placental tissues in normoxia- or hypoxia-treated pregnant mice. (M-O) Person correlation analysis of the levels of murine lnc-Hz06 and Ncoa4 protein, MDA, or free Fe 2+ in placental tissues in normoxia- or hypoxia-treated pregnant mice (n = 15). (P) The protein levels of SUMOylated Hif1α (Hif1α-SUMO) that was immunoprecipitated by Hif1α antibody in placental tissues in normoxia- or hypoxia-treated pregnant mice (n = 6). (Q and S) The levels of Ncoa4 or lnc-Hz06 promoter region enriched by Hif1α-SUMO in placental tissues in normoxia- or hypoxia-treated pregnant mice in ChIP-re-ChIP assays (n = 6). (R and T) The protein levels of Hif1α and Hif1α-SUMO pulled down by biotin-labeled DNA probe containing Ncoa4 or lnc-Hz06 promoter region in placental tissues in normoxia- or hypoxia-treated pregnant mice in DNA pulldown assays (n = 6). Data in (B-L, P-T) show mean ± SD, n = 15 independent samples. Student's t -test analysis for (B, C, E, F-L, P-T). Pearson correlation analysis for (M-O). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 9 Hypoxia induced miscarriage by promoting ferroptosis in hypoxic mouse model. (A) The protein levels of murine Hif1α, Gpx4, Fth1, and Ncoa4 in placental tissues of normoxia- or hypoxia-treated pregnant mice (n = 15, 6 representatives were shown), with Actin as internal standard. (B) The relative intensity of each protein band of Hif1α in normoxia- or hypoxia-treated mouse placental tissues was quantified and plotted (n = 15). (C) IHC image and relative quantification of murine Hif1α protein levels (brown color) in normoxia- or hypoxia-treated mouse placental tissues. (D) Representative uterus morphology on gestational day 13 in normoxia- or hypoxia-treated pregnant mice with embryo absorption indicated by red arrows. (E) Average miscarriage rates in normoxia- or hypoxia-treated pregnant mice (n = 15). (F) The relative expression levels of murine lnc-Hz06 in normoxia- or hypoxia-treated mouse placental tissues (n = 15). (G-K) Analysis of the levels of Ncoa4 protein, MDA, free Fe 2+ , Gpx4 protein, and Fth1 protein in placental tissues of normoxia- or hypoxia-treated pregnant mice (n = 15). (L) IHC image and relative quantification of murine Ncoa4 protein levels (brown color) in placental tissues in normoxia- or hypoxia-treated pregnant mice. (M-O) Person correlation analysis of the levels of murine lnc-Hz06 and Ncoa4 protein, MDA, or free Fe 2+ in placental tissues in normoxia- or hypoxia-treated pregnant mice (n = 15). (P) The protein levels of SUMOylated Hif1α (Hif1α-SUMO) that was immunoprecipitated by Hif1α antibody in placental tissues in normoxia- or hypoxia-treated pregnant mice (n = 6). (Q and S) The levels of Ncoa4 or lnc-Hz06 promoter region enriched by Hif1α-SUMO in placental tissues in normoxia- or hypoxia-treated pregnant mice in ChIP-re-ChIP assays (n = 6). (R and T) The protein levels of Hif1α and Hif1α-SUMO pulled down by biotin-labeled DNA probe containing Ncoa4 or lnc-Hz06 promoter region in placental tissues in normoxia- or hypoxia-treated pregnant mice in DNA pulldown assays (n = 6). Data in (B-L, P-T) show mean ± SD, n = 15 independent samples. Student's t -test analysis for (B, C, E, F-L, P-T). Pearson correlation analysis for (M-O). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Subsequently, we investigated whether hypoxia might induce ferroptosis in this hypoxia-miscarriage mouse model. UCSC-BLAT analysis showed the presence of murine lnc-Hz06 (93.0 % sequence similarity to human lnc-HZ06) ( Fig. S7C ), and this murine lnc-Hz06 does not have the protein encoding capability ( Fig. S7D ). Moreover, the mRNA and protein sequences of NCOA4 were also conservative in human, mouse, dog, monkey, chicken, and zebrafish ( Figs. S7E and F ). In hypoxia group, lnc-Hz06 was highly expressed ( Fig. 9 F), indicating that hypoxia up-regulated lnc-Hz06 expression levels in mouse placenta. Moreover, the levels of Ncoa4, MDA, and Fe 2+ were all higher, whereas the levels of Gpx4 and Fth1 were lower, in hypoxia group than those in normoxia group ( Fig. 9 A, G-L), indicating that hypoxia induced ferroptosis. Pearson correlation analysis showed that the relative levels of lnc-Hz06 were positively correlated with those of Ncoa4 protein, MDA, and free Fe 2+ ( Fig. 9 M − O), implying that lnc-Hz06 might positively regulate Ncoa4-mediated ferroptosis in hypoxic mouse placenta. IP assays using identical but limited amounts of HIF1α antibody showed that the levels of Hif1α-SUMO were higher in hypoxic vs normoxic mouse placenta ( Fig. 9 P). ChIP-re-ChIP assays showed that more Ncoa4 promoter region was enriched by Hif1α-SUMO in hypoxia group relative to normoxia group ( Fig. 9 Q). The levels of Hif1α-SUMO that was pulled down by biotin-labeled dsDNA probe containing Ncoa4 promoter region were higher in hypoxia group relative to those in normoxia group ( Fig. 9 R). These data showed that hypoxia induced ferroptosis by up-regulating Hif1α-SUMO-mediated Ncoa4 transcription in hypoxia mouse placenta. To explore whether Hif1α-SUMO might regulate lnc-Hz06 transcription in hypoxia mouse placenta, ChIP-re-ChIP assays showed that more lnc-Hz06 promoter region was enriched by Hif1α-SUMO in hypoxia vs normoxia group ( Fig. 9 S). The levels of Hif1α-SUMO pulled down by biotin-labeled dsDNA probe containing lnc-hz06 promoter region were also higher in hypoxia vs normoxia group ( Fig. 9 T). Therefore, agreed with the results in hypoxic trophoblast cells, Hif1α-SUMO might act as lnc-Hz06 transcription factor and primarily perform lnc-Hz06 transcription in hypoxic mouse placenta. Taken together, high possibly, hypoxia could induce miscarriage by promoting ferroptosis through up-regulating lnc-Hz06/Hif1α-SUMO-mediated Ncoa4 transcription in hypoxic mouse placenta. Having known the mechanism underlying hypoxia-induced miscarriage, we next explored whether lnc-Hz06 and Ncoa4 might be used as targets to alleviate miscarriage in hypoxic mouse model. For this aim, we constructed a miscarriage intervention model in which the antisense locked oligonucleotide of lnc-Hz06 (AS-Hz06, 10 nmol) or Ncoa4 (AS-Ncoa4, 10 nmol) was intraperitoneally injected into hypoxia-treated pregnant mice once per three days, with AS-NC as control ( Fig. 10 A). Injection with AS-Hz06 or AS-Ncoa4 could efficiently reduce the expression levels of lnc-hz06 or NCOA4, respectively, in pregnant mouse placenta relative to control group ( Fig. 10 B and C). Moreover, in control group, hypoxia led to obvious embryo adsorption and high miscarriage rates; however, treatment with AS-Hz06 or AS-Ncoa4 reduced embryo adsorption and alleviated miscarriage rates ( Fig. 10 D and E). Therefore, these results showed that knockdown of murine lnc-Hz06 or Ncoa4 could efficiently alleviate miscarriage in the hypoxic mice. Fig. 10 Knockdown of lnc-Hz06 or Ncoa4 efficiently alleviated miscarriage in hypoxia-treated pregnant mice. (A) Schematic diagram of miscarriage intervention by injecting hypoxia-treated mice with AS-Hz06 or AS-Ncoa4. (B and C) The RNA levels of murine lnc-Hz06 and Ncoa4 in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (D) Representative uterus morphology on gestational day 13 in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice with embryo absorption indicated by red arrows. (E) Average miscarriage rates in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (F-I) The protein levels of murine Gpx4, Fth1, and Ncoa4 in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15), with Actin as internal standard. (J and K) The levels of MDA and free Fe 2+ in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (L) Lnc-HZ06 and Hif1α-SUMO formed a positive auto-regulatory feedback loop, which was up-regulated in hypoxic trophoblast cells and in RM villous tissues. Hif1α-SUMO further promoted the transcription of NCOA4 and induced trophoblast cell ferroptosis and the occurrence of RM. Data in (B, C, E, G-K) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (B, C). Two-way ANOVA followed by Tukey's multiple comparisons test for (E, G-K). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 10 Knockdown of lnc-Hz06 or Ncoa4 efficiently alleviated miscarriage in hypoxia-treated pregnant mice. (A) Schematic diagram of miscarriage intervention by injecting hypoxia-treated mice with AS-Hz06 or AS-Ncoa4. (B and C) The RNA levels of murine lnc-Hz06 and Ncoa4 in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (D) Representative uterus morphology on gestational day 13 in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice with embryo absorption indicated by red arrows. (E) Average miscarriage rates in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (F-I) The protein levels of murine Gpx4, Fth1, and Ncoa4 in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15), with Actin as internal standard. (J and K) The levels of MDA and free Fe 2+ in placental tissues in AS-Hz06-or AS-Ncoa4-treated hypoxic pregnant mice (n = 15). (L) Lnc-HZ06 and Hif1α-SUMO formed a positive auto-regulatory feedback loop, which was up-regulated in hypoxic trophoblast cells and in RM villous tissues. Hif1α-SUMO further promoted the transcription of NCOA4 and induced trophoblast cell ferroptosis and the occurrence of RM. Data in (B, C, E, G-K) show mean ± SD, n = 5 independent experiments. Student's t -test analysis for (B, C). Two-way ANOVA followed by Tukey's multiple comparisons test for (E, G-K). P  < 0.05 was considered as significant difference. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Subsequently, ferroptosis was also detected in these mouse placental tissues. The protein levels of Gpx4 and Fth1 were higher, whereas those of Ncoa4 were lower, in AS-Hz06-or AS-Ncoa4-treated groups than those in hypoxic control group ( Fig. 10 F-I). The levels of MDA and free Fe 2+ were lower in AS-Hz06-or AS-Ncoa4-treated groups than those in hypoxic control group ( Fig. 10 J and K). Collectively, knockdown of either lnc-Hz06 or Ncoa4 could efficiently suppress ferroptosis and alleviate miscarriage in hypoxic mouse placenta. These data also indicated that knockdown of lnc-HZ06 or NCOA4 might be a promising therapeutic strategy to prevent against hypoxia-induced miscarriage.

All unique reagents generated in this study, such as mRNA and cell lines are available from the Lead Contact with a completed Material Transfer Agreement.

To construct hypoxic trophoblast cell model, human trophoblast cell lines (Swan 71 or HTR-8/SVneo) were incubated with 1 % O 2 , 94 % N 2 , and 5 % CO 2 [ 48 ] or treated with CoCl 2 (Cat. No. 15862; Sigma-Aldrich, St Louis, MO, USA) [ 52 ]. Ferroptosis markers (MDA, GPx activity, LDH, cell viability, ROS, Fe 2+ , COX-2, and Ferritin) were measured. Villous tissues were collected from unexplained RM patients or their matched healthy controls (HC) (n = 30 in either group), as described previously [ [41] , [42] , [43] , [44] , [45] , [47] , [54] ]. Mouse model of hypoxia-induced miscarriage (n = 15 in each group) was constructed as described previously [ [55] , [56] , [57] ]. The experimental details of the cellular assays, tissue assays, and mouse model assays were fully described in Supporting Information.

In this study, we find that hypoxia causes trophoblast cell ferroptosis to induce miscarriage by up-regulating HIF1α-SUMO-mediated NCOA4 transcription, which is further regulated by lnc-HZ06. Lnc-HZ06 promotes the SUMOylation of HIF1α by suppressing its SENP1-mediated deSUMOylation. Meanwhile, HIF1α-SUMO also acts as a transcription factor to promote lnc-HZ06 transcription in hypoxic trophoblast cells. Thus, both lnc-HZ06 and HIF1α-SUMO form a positive auto-regulatory feedback loop. This loop is up-regulated in hypoxic trophoblast cells, which further promotes the transcription of NCOA4 and induces ferroptosis and the occurrence of miscarriage. Collectively, this study provides new insights in understanding the regulatory roles of lnc-HZ06/HIF1α-SUMO/NCOA4 axis among hypoxia, ferroptosis, and miscarriage, and also offers an effective approach for treatment against miscarriage.

Association and regulatory mechanisms among hypoxia, ferroptosis, and miscarriage. Hypoxia may be associated with adverse pregnancy outcomes, such as preeclampsia and fetal intrauterine restriction [ 11 ]. Ferroptosis might be associated with preeclampsia and endometrium [ [11] , [12] , [13] ]. Recently, we have found that BPDE exposure induces ferroptosis of trophoblast cells [ 14 ]. However, whether hypoxia may induce trophoblast cell ferroptosis is completely unknown; and whether hypoxia or ferroptosis might further induce unexplained RM is also unknown. In this study, it was the first time that we associated hypoxia, ferroptosis, and unexplained RM. Hypoxia may induce trophoblast cell ferroptosis and miscarriage by up-regulating HIF1α-SUMO-mediated NCOA4 transcription. In the proposed mechanism ( Fig. 10 L), lnc-HZ06 promotes the SUMOylation of HIF1α by suppressing its SENP1-mediated deSUMOylation. Reversely, HIF1α-SUMO also acts as a transcription factor to promote lnc-HZ06 transcription in hypoxic trophoblast cells. Thus, both lnc-HZ06 and HIF1α-SUMO form a positive auto-regulatory feedback loop. This loop is up-regulated in hypoxic trophoblast cells, which further promotes the transcription of NCOA4 and induces ferroptosis and the occurrence of RM. HIF1α-SUMO primarily performed NCOA4 transcription. Whether HIF1α-SUMO could primarily act as a transcription factor of NCOA4 to promote its transcription is unclear. A previous study has showed that the SUMOylation of HIF1α increases its transcription activity possibly by increasing its protein stability [ 32 ]. In this study, we identified that HIF1α-SUMO could primarily act as a transcription factor of NCOA4; and it is HIF1α-SUMO that plays a predominant role in NCOA4 transcription. Moreover, it has been reported that SUMOylation may regulate cell apoptosis and autophagy in U87MG and U251 cells [ [70] , [71] , [72] , [73] ]. In this study, we discovered that SUMOylation may also regulate cell ferroptosis through HIF1α-SUMO in human trophoblast cells. In details, HIF1α-SUMO acts as a transcription factor to promote NCOA4 transcription and further induced trophoblast cell ferroptosis and miscarriage. LncRNAs regulate HIF1α. Several lncRNAs have been reported to regulate HIF1α. LncRNA HITT inhibits HIF1α mRNA expression in human colorectal cancer cells. Reversely, HIF1α promotes the degradation of HITT; and both of which form a negative feedback loop [ 50 ]. LincRNA-p21 suppresses HIF1α ubiquitination in Hela cells [ 49 ]. In this work, we identified a new regulatory mode in which lnc-HZ06 up-regulates HIF1α-SUMO levels by suppressing its deSUMOylation in hypoxic trophoblast cells, enriching the epigenetic regulatory mechanisms in hypoxic cells. LncRNAs regulate ferroptosis. It has been reported that several lncRNAs regulate ferroptosis. LncRNA PVT1 regulates ferroptosis through miR-214-mediated TFR1 and p53 in brain ischemia/reperfusion [ 74 ]. Knockdown of lncRNA MEG8 induces ferroptosis of hemangioma endothelial cells by regulating miR-497-5p/NOTCH2 axis [ 75 ]. However, in these studies, p53 and NOTCH2 do not directly lead to ferroptosis, suggesting that these lncRNAs might regulate ferroptosis in an indirect mode. In this study, we elucidated that lnc-HZ06 directly induces ferroptosis by promoting HIF1α-SUMO-mediated NCOA4 transcription. Moreover, it was the first time that we found that lncRNA could simultaneously regulate hypoxia, trophoblast cell ferroptosis, and miscarriage. HIF1α-SUMO/lnc-HZ06 positive feedback loop. In our previous studies, we have found several positive feedback loops in human trophoblast cells and in miscarriage villous tissues, such as lnc-HZ01 and p53 [ 76 ], lnc-HZ01 and MXD1 [ 41 ] [41] , [77] , lnc-HZ03 and miR-hz03 [ 42 ]. In this study, we found that lnc-HZ06 and HIF1α-SUMO also form a positive feedback loop in hypoxic trophoblast cells and in RM villous tissues. HIF1α-SUMO promotes lnc-HZ06 transcription; and lnc-HZ06 also promotes the SUMOylation of HIF1α in hypoxic trophoblast cells. Once this feedback loop is initiated upon hypoxia, the downstream NCOA4 might be automatically transcribed in an auto-regulatory mode, which further induces trophoblast ferroptosis. Thus, it seems that hypoxia may act as a trigger to transiently activate the feedback loop. Upon activation, hypoxia might possibly become less important compared with its downstream ferroptosis; and it might be the trophoblast ferroptosis that ultimately induces miscarriage. Miscarriage treatment. It has been reported that lncRNAs could be used as targets for cancer therapy. For example, LINC00301 is highly expressed in non-small cell lung cancer (NSCLC) tumor cells; and knockdown of LINC00301 by injecting its antisense Locked Nucleic Acid can efficiently reduce tumor size and volume in mouse model, providing a promising strategy for treatment against NSCLC [ 79 ]. In this study, we discovered that knockdown of murine lnc-Hz06 or Ncoa4 in mice could efficiently alleviate hypoxia-induced miscarriage, providing a new approach for treatment against unexplained miscarriage. Limitation and prospect. In cells, functional HIF1 is consisted of HIF1α and HIF1β [ 80 ]. Whether the SUMOylated HIF1α might affect its interactions with HIF1β is still unexplored. The correlation and regulatory mechanisms between HIF1α-SUMO and HIF1α protein levels are still largely elusive and should be further explored. Furthermore, the mechanisms how lnc-HZ06 down-regulates SENP1 or HIF1α remains unknown. As for HIF1α-SUMO/lnc-HZ06 auto-feedback loop, transient hypoxia treatment of cells might be a good choice to explore whether trophoblast ferroptosis might be initiated and enlarged in an auto-regulatory mode.

Trophoblast cell dysfunctions and miscarriage. Miscarriage (abnormal early embryo loss) occurs in approximately 15 %-25 % of pregnant women [ 1 ]; and recurrent miscarriage (RM, twice or more consecutive miscarriage) occurs in 1 %-5 % pregnant women [ 2 ]. Moreover, about 41 % RM women experience anxiety, 9 % suffer from major depression, and 1.4 % result in death [ 3 ]. RM could be induced or caused by several factors, including chromosome abnormalities, genetic causes, endocrine, autoimmune or thrombotic abnormalities. However, almost half of the causes remain unexplained [ 4 ]. Trophoblast cells play important roles in embryonic development [ 5 ]. Dysfunctions of human trophoblast cells may lead to various adverse pregnancy outcomes, including unexplained RM. Therefore, to explore the mechanisms of trophoblast cell dysfunctions may provide effective approaches to understand the pathogenesis of unexplained RM. Ferroptosis. Ferroptosis is a new, caspase-independent and non-apoptotic modality of cell death, which is morphologically and biochemically different from the traditional apoptosis, necrosis, or autophagy [ 6 , 7 ]. Ferroptosis is closely related with villous physiology and pathobiology, stroke, infertility, endometriosis, preeclampsia, or degenerative diseases [ 5 , [8] , [9] , [10] ]. It has been reported that the serum iron levels, many ferroptosis-related key protein levels, and the lipid oxidation levels were all higher in preeclampsia or endometriosis patients [ [11] , [12] , [13] ]. Recently, we have discovered that environmental BPDE exposure induces ferroptosis of human trophoblast cells [ 14 ]. However, whether trophoblast cell ferroptosis might be associated with unexplained RM is completely undiscovered. Hypoxia. Hypoxia, which was contributed to 2019 Nobel Prize in Physiology or Medicine, plays significant physiological roles in normal and different pathological processes, and is also involved in different disease processes, such as placentation [ 15 ], angiogenesis [ 16 ], preeclampsia [ 17 ], liver fibrosis [ 18 ], and lung cancer [ 19 ]. Many unavoidable environmental factors, social or psychological factors, such as air pollution, stress, common obesity or nutritional deficiencies, may induce hypoxia [ 20 ]. Besides, exposure to some drugs (including cocaine and misoprostol) or severe shock may also lead to hypoxia [ 21 ]. Retrospective studies have shown that the incidence of pregnancy complications is increased at high altitude relative to that at low altitude [ 22 ]. It has been reported that, if placenta continuously remains in a hypoxic environment (about 1-3 % O 2 ), it might induce trophoblast cell dysfunctions and miscarriage, fetal intrauterine restriction, or preeclampsia [ 11 , 23 , 24 ]. A retrospective study has shown that the levels of HIF1α (hypoxia inducible factor-1α) in peri-implantation endometrium of RM women are higher than those in the matched healthy control groups [ 25 ]. Another study also shows that HIF1α suppresses the proliferation and invasion of trophoblast cells under hypoxia [ 26 , 27 ]. However, the association and causality between hypoxia and miscarriage should be investigated experimentally. Hypoxia and ferroptosis . HIF1α, as a hypoxic transcription factor, binds to the HRE (Hypoxia-Responsive Element) in the promoter regions of hypoxia-induced genes and activates their transcription [ [28] , [29] , [30] , [31] ]. The SUMOylated HIF1α (termed as HIF1α-SUMO) increases its transcriptional activity possibly by increasing its protein stability [ 32 ]. Meanwhile, NCOA4 (nuclear receptor coactivator 4) is involved in ferroptosis [ [33] , [34] , [35] ]. Recently, a genome-wide ChIP-on-ChIP assay shows that HIF1α protein might bind about 1.5 kb upstream of NCOA4 exon 1 in HepG2 cells [ 36 ]. However, whether HIF1α-SUMO might act as a NCOA4 transcription factor to directly promote its transcription and then cause trophoblast cell ferroptosis is completely unknown and should be fully investigated. LncRNA. LncRNAs (long non-coding RNAs), act as important epigenetic regulatory factors [ 37 ], have been reported to regulate human trophoblast cell functions and miscarriage, such as lncRNA EPB41L4A-AS1 [ 38 ], lncRNA-TCL6 [ 39 ], lncRNA MEG8 [ 40 ]. In our recent studies, we have identified a group of novel lncRNAs, such as lnc-HZ01 [ 41 ], lnc-HZ03 [ 42 ], lnc-HZ04 [ 43 ], lnc-HZ05 [ 44 ], lnc-HZ08 [ 45 ], lnc-HZ09 [ 46 ], and lnc-HZ14 [ 47 ], all of which regulate trophoblast cell functions and the occurrence of miscarriage. As for hypoxia, lnc-HIFCAR promotes oral cancer development by enhancing HIF1α activity [ 48 ]; lincRNA-p21 affects cell glycolysis by regulating HIF1α degradation [ 49 ]; lnc-HITT inhibits HIF1α translation and regulates tumor angiogenesis [ 50 ]. As for ferroptosis, lncRNA LINC00336 serves as an endogenous sponge of miR-6852 to regulate the expression of cystathionine-β-synthase and inhibits ferroptosis in lung cancer cells [ 51 ]. However, to our knowledge, no lncRNA simultaneously regulates hypoxia, ferroptosis, and miscarriage. We expect to identify a lncRNA that might regulate these processes and to explore its regulatory mechanism. In this work, we discovered that hypoxia could result in ferroptosis of human trophoblast cells and then induce unexplained miscarriage, identified a novel lnc-HZ06 that simultaneously regulated hypoxia (HIF1α), ferroptosis, and miscarriage, and also explored their underlying mechanisms, thus providing new understanding in the pathogenesis of unexplained miscarriage. Based on the mechanisms, we also developed an intervention strategy for efficient alleviation of miscarriage, providing an approach for treatment against unexplained RM.

Actin Carbonic anhydrase isoform 9 Cell Counting Kit 8 Chromatin Immunoprecipitation Co-immunoprecipitation Prostaglandin-endoperoxide synthase 2 Erastin Ferrostatin-1 Glutathione peroxidase Hypoxia inducible factor-1alpha Homeobox A10 Hypoxia-responsive element Immunohistochemistry Immunoprecipitation Lactate dehydrogenase Long non-coding RNA Malondialdehyde Proteasome inhibitor MG132 Nuclear receptor coactivator 4 Reactive oxygen species Reverse transcription quantitative real time polymerase chain reaction Sodium dodecyl sulfate polyacrylamide gel electrophoresis Tubulin Vascular endothelial growth factor Western blotting

None.