Abstract

   Aging of the fetal membranes participates in labor onset. However, the
   underlying mechanism is poorly understood. Here, we identify that the
   classical secretory protein S100 calcium‐binding protein A9 (S100A9),
   upon de‐phosphorylation at Thr 113, translocates to the nuclei of
   amnion fibroblasts of the human fetal membranes, where S100A9 causes
   heterochromatin erosion via segregation of heterochromatin maintenance
   proteins, resulting in Long Interspersed Nuclear Element‐1 (LINE1)
   de‐repression at parturition. Increased LINE1 retrotransposition
   further activates the type I interferon response via the cGAS‐STING
   pathway, thereby leading to amnion fibroblast senescence with
   consequent increased secretion of components associated with
   senescence‐associated secretory phenotype. Mouse studies show that
   intra‐amniotic injection of vector specifically expressing S100A9 in
   the nucleus induces preterm birth along with LINE1 de‐repression and
   increased cellular senescence in the fetal membranes, which is blocked
   by inhibition of LINE1 reverse‐transcription. Together, these findings
   highlight that nuclear‐translocated S100A9 acts as a heterochromatin
   disruptor to de‐repress LINE1 which subsequently triggers amnion
   fibroblast senescence at parturition.

   Keywords: fetal membranes, heterochromatin, LINE1, parturition, S100A9,
   senescence
     __________________________________________________________________

   This study shows that the classical secretory protein S100
   calcium‐binding protein A9 (S100A9) can translocate to the nucleus upon
   de‐phosphorylation at Thr 113 in human amnion fibroblasts at
   parturition, where S100A9 induces heterochromatin erosion through
   segregation of the heterochromatin maintenance protein, resulting
   inLong Interspersed Nuclear Element‐1 (LINE1) de‐repression. Increased
   LINE1 retrotransposition causes cellular senescence via activation of
   the cGAS‐STING‐type I interferon response pathway in human amnion
   fibroblasts.

   graphic file with name ADVS-12-2414682-g007.jpg

1. Introduction

   Cellular senescence is an irreversible biological process that
   progresses with age, heralding the end of lifespan in almost all
   multicellular species.^[ [40]^1 ^] The placenta and fetal membranes are
   unique temporary organs in pregnancy. They fulfill their mission once
   pregnancy terminates. Accumulating evidence indicates that toward the
   end of gestation cells in the placenta and fetal membranes undergo
   senescence, a process which is believed to be an integral event of
   labor onset.^[ [41]^2 ^] Cellular senescence and senescence‐associated
   secretory phenotype (SASP) provoke inflammatory reactions in these
   tissues leading to overwhelming production of pro‐labor factors.^[
   [42]^3 ^] In the fetal membranes, cellular senescence may also weaken
   the tensile strength of the membranes thus enabling membrane rupture
   which further enhances the production of pro‐labor factors.^[ [43]^3e
   ^] Therefore, understanding the cause of cellular senescence in the
   fetal membranes is of utmost importance for elucidation of the
   mechanism whereby the fetal membranes initiate parturition in humans,
   which may help develop effective strategies for the prevention of
   preterm birth, a major cause of perinatal mortality and morbidity.^[
   [44]^4 ^]

   Among the identified causes of cellular senescence, epigenomic
   instability, especially heterochromatin disorganization, is emerging as
   a prominent trigger.^[ [45]^5 ^] Heterochromatin is defined as a
   densely packed chromatin, rendering access of DNA and RNA polymerases
   to the associated gene difficult, thereby preventing DNA replication
   and gene transcription.^[ [46]^6 ^] Maintenance of heterochromatin is
   normally achieved through enrichment of trimethylated histone H3 at
   lysine 9 (H3K9me3) as well as heterochromatin maintenance proteins,
   such as CBX5 (Chromobox protein homolog 5, also known as
   heterochromatin protein 1α, HP1α), KRAB‐associated protein 1 (KAP1) and
   the nucleosome remodeling and deacetylase (NuRD) complex.^[ [47]^6 ^]
   Heterochromatin is crucial for silencing transposable elements (TEs),
   especially Long Interspersed Nuclear Element‐1 (LINE‐1), which are the
   most abundant and only active autonomous retrotransposon in humans,
   accounting for nearly 17% of the human genome.^[ [48]^7 ^] Tight
   repression of LINE‐1 expression is essential for the prevention of DNA
   damage, which is normally required to maintain genomic integrity.^[
   [49]^7e ^] Emerging evidence indicates that increased LINE1
   transcription and retrotransposition as a result of heterochromatin
   erosion are associated with cellular senescence in a variety of
   non‐gestational tissues.^[ [50]^8 ^] It is recognized that increased
   LINE1 transcription and subsequent retrotransposition trigger cellular
   senescence through activation of the innate immune response such as the
   type I interferon (IFN) response.^[ [51]^8 ^] However, whether the
   de‐repression of LINE1 as a result of heterochromatin disorganization
   participates in cellular senescence of the fetal membranes remains
   unknown. If so, it is crucial to identify the inducer of
   heterochromatin erosion in the fetal membranes so that novel
   therapeutic targets can be developed for the treatment of preterm
   premature rupture of membranes and preterm birth.

   S100 calcium‐binding protein A9 (S100A9, also known as MRP14) is a
   member of the S100 family of calcium‐binding proteins, which is
   normally secreted to the extracellular space as a cytokine or as a
   protein of damage‐associated molecular patterns in response to
   inflammatory stimuli, tissue damage, and stress, etc.^[ [52]^9 ^] By
   re‐examination of our previously published transcriptomic data,^[
   [53]^10 ^] we found that the abundance of S100A9 was significantly
   increased in the amnion layer of the human fetal membranes at
   parturition. Serendipitously, we found that S100A9 was localized in the
   nuclei of mesenchymal fibroblasts of the human amnion (human amnion
   fibroblasts, hAFs), a distinct cell type of the amnion layer that not
   only provides pro‐labor factors but also extracellular matrix (ECM)
   components.^[ [54]^11 ^] More intriguingly, we found that S100A9 could
   translocate to the nucleus upon de‐phosphorylation at Thr 113 in hAFs,
   where S100A9 caused heterochromatin disruption through segregation of
   heterochromatin maintenance proteins such as CBX5, thus leading to
   LINE1 de‐repression. LINE1 retrotransposition, in turn, stimulated the
   type I IFN response through activation of the cGAS (cyclic GMP‐AMP
   synthase)‐STING (stimulator of interferon genes) signaling pathway,
   resulting in cellular senescence in hAFs. Mouse studies showed that
   intra‐amniotic administration of vector specifically expressing S100A9
   in the nucleus induced preterm birth along with increased LINE1
   activation and cellular senescence in the fetal membranes, which lent
   further support for the crucial role of nuclear S100A9 in the induction
   of cellular senescence in the fetal membranes at parturition. Here, we
   described the identification of this novel crucial role of
   nuclear‐translocated S100A9 in the induction of cellular senescence in
   the fetal membranes at parturition.

2. Results

2.1. Increased S100A9 in the Nuclei of hAFs at Term and Preterm Labor

   Reanalysis of our previously published transcriptomic sequencing data
   (NCBI GEO accession number [55]GSE166453)^[ [56]^10 ^] revealed a
   significant increase in S100A9 mRNA in the amnion layer of the fetal
   membranes obtained from pregnant women undergoing spontaneous labor at
   term (designated as term labor, TL) when compared with that obtained
   from pregnant women undergoing elective cesarean section without labor
   at term (designated as term no labor, TNL) (Figure [57] 1A). The
   increase in S100A9 in the TL amnion was subsequently confirmed at mRNA
   and protein levels with quantitative real‐time polymerase chain
   reaction (qRT‐PCR), Western blotting, and enzyme immunoassay (ELISA)
   respectively (Figure [58]S1A–C, Supporting Information). Intriguingly,
   despite S100A9 being a classical secretory protein,^[ [59]^9b ^]
   immunofluorescent and immunohistochemical staining of the human amnion
   showed an intense distribution of S100A9 in the nuclei of hAFs in the
   TL group (Figure [60]1B; Figure [61]S1D, Supporting Information).
   However, this phenomenon was not observed in epithelial cells of the
   human amnion (hEPCs). We further compared the abundance of nuclear
   S100A9 between the TL and TNL groups in the nuclear protein extracted
   from the TL and TNL amnion as well as from hAFs isolated from the TL
   and TNL amnion with Western blotting, which revealed an increased
   abundance of nuclear S100A9 in both the amnion tissue and hAFs at
   parturition (Figure [62]1C–E). Immunofluorescent staining also showed
   an intense staining of S100A9 in the nuclei of hAFs isolated from the
   TL amnion (Figure [63]1F). The identity of hAFs in the amnion and
   isolated hAFs was confirmed with immunofluorescent staining for
   vimentin, a mesenchymal cell marker (Figure [64]1B,F; Figure [65]S1E,
   Supporting Information). Further, a significant increase in nuclear
   S100A9 abundance was also observed in the amnion obtained from
   spontaneous preterm labor with no indication of infection (designated
   as preterm labor, PL) as compared to preterm elective cesarean section
   without labor (designated as preterm no labor, PNL) (Figure [66]1G).
   These data indicate that S100A9 nuclear translocation in hAFs may play
   a role in both term and preterm labor.

Figure 1.

   Figure 1
   [67]Open in a new tab

   Increased nuclear S100A9 in hAFs at parturition. A) Volcano plot of
   transcriptomic sequencing data displaying the differentially expressed
   genes of the human amnion in the TL group (n = 3) as compared to the
   TNL group (n = 3). TL, term labor; TNL, term no labor. B)
   Representative images of immunofluorescence staining showing intense
   S100A9 staining (green color) in the nuclei of hAFs from the TL amnion.
   n = 3. White arrows indicate hAFs, which are positive for vimentin
   staining (red color), a classical mesenchymal fibroblast marker. Nuclei
   were counterstained with DAPI (blue color). Scale bars, 25 µm. C)
   Diagram illustrating the procedure of nuclear protein extraction from
   human amnion tissue and isolated hAFs. D) Western blot analysis showing
   the difference of nuclear S100A9 protein abundance in the human amnion
   between TL (n = 6) and TNL (n = 6) groups. The left panel is the
   Western blot and the right panel is the average data. α‐tubulin and
   Lamin A/C were used as internal controls for cytoplasmic and nuclear
   protein, respectively. E) Western blotting analysis showing the
   difference of S100A9 protein abundance in isolated hAFs between TL (n =
   6) and TNL (n = 6) groups. The left panel is the Western blot and the
   right panel is the average data. α‐tubulin and Lamin A/C were used as
   internal controls for cytoplasmic and nuclear protein, respectively. F)
   Representative images of immunofluorescence staining showing intense
   S100A9 staining (red color) in the nuclei of hAFs isolated from the TL
   amnion. n = 3. hAFs were positive for vimentin staining (green color).
   Nuclei were counterstained with DAPI (blue color). Scar bar, 25 µm. G)
   Western blot analysis showing the difference of nuclear S100A9 protein
   abundance in the human amnion between PL (n = 6) and PNL (n = 6)
   groups. The left panel is the Western blot and the right panel is the
   average data. α‐tubulin and Lamin A/C were used as internal controls
   for cytoplasmic and nuclear protein, respectively. PL, preterm labor;
   PNL, preterm no labor. Data are mean ± SEM. Two‐tailed unpaired
   Student's t‐test (D, E, G). *p < 0.05, **p < 0.01.

2.2. Correlation of Cellular Senescence with Nuclear S100A9 in hAFs at
Parturition

   Previous studies have shown that cellular senescence of the amnion at
   parturition is largely an epithelial phenomenon.^[ [68]^2b–d ^]
   However, it is not known whether hAFs also undergo cellular senescence
   at parturition. By staining the amnion with senescence‐associated
   β‐galactosidase (SA‐β‐gal) and p21, a classical marker of cellular
   senescence,^[ [69]^1 , [70]^12 ^] we observed that the proportion of
   SA‐β‐gal‐ and p21‐positive cells were significantly higher in both
   hEPCs and hAFs of the TL amnion versus TNL amnion (Figure [71] 2A,B).
   Cellular senescence in hAFs was further confirmed with Western
   blotting, which showed significantly increased p53 and p21 protein
   abundance in hAFs isolated from the TL amnion versus TNL amnion
   (Figure [72]2C–E). Additionally, ELISA assay showed that the level of
   prostaglandin E2 (PGE2), a well‐recognized pro‐labor eicosanoid that is
   crucial for cervical ripening and myometrial contraction in
   parturition,^[ [73]^3a ^] was significantly higher in hAFs isolated
   from the TL amnion than that from the TNL amnion (Figure [74]2F).
   Correspondingly, the inducible cyclooxygenase‐2 (COX‐2, encoded by
   PTGS2), a rate‐limiting enzyme in prostaglandin synthesis, was also
   significantly increased in hAFs isolated from the TL amnion as compared
   to that from the TNL amnion (Figure [75]2G,H). Moreover, correlation
   analysis showed that the abundance of COX‐2, PGE2, p21, p53, an
   upstream stimulator of p21, and nuclear S100A9 correlated positively
   with each other in hAFs (Figure [76]2I‐N; Figure [77]S1F–I, Supporting
   Information). These data suggest that hAFs, in addition to hEPCs, also
   undergo cellular senescence at parturition. Moreover, cellular
   senescence and increased PGE2 production in hAFs may be associated with
   nucleus‐translocated S100A9.

Figure 2.

   Figure 2
   [78]Open in a new tab

   Increased cellular senescence and PGE2 production in hAFs at
   parturition. A) SA‐β‐gal staining showing an increased percentage of
   cells positive for SA‐β‐gal staining (blue color) in the human amnion
   collected from TL (n = 4) compared to TNL (n = 4). Nuclei were
   counterstained with nuclear fast red (red color). Red and blue arrows
   indicate SA‐β‐gal‐positive hAFs and hEPCs respectively. The left panel
   is the representative image and the right panel is the average data.
   Scale bars, 50 µm. B) Immunohistochemical staining of p21 showing an
   increased percentage of cells positive for p21 (red color) in the human
   amnion collected from TL (n = 4) compared to TNL (n = 4). The tissue
   section was counterstained with hematoxylin (blue color). Red and blue
   arrows indicate p21‐positive hAFs and hEPCs respectively. Left panel is
   the representative image and the right panel is the average data. Scale
   bars, 50 µm. C–E) Western blot analysis showing increased p21 and p53
   protein abundance in hAFs isolated from TL (n = 6) compared to TNL (n =
   6). (E) is the Western blot and (C‐D) is the average data. F) ELISA
   showing increased PGE2 concentration in the conditioned culture medium
   of hAFs isolated from TL (n = 6) compared to that from TNL (n = 6).
   G–H) Western blot analysis showing increased COX‐2 protein abundance in
   hAFs isolated from TL (n = 6) compared to that from TNL (n = 6). (H) is
   the Western blot and (G) is the average data. I) Spearman analysis
   showing a positive correlation between p21 and p53 in isolated hAFs. n
   = 12. J,K) Spearman analysis showing a positive correlation between
   nuclear‐localized S100A9 and p21 (J) or p53 (K) in isolated hAFs. n =
   12. L) Spearman analysis showing a positive correlation between COX‐2
   and PGE2 in isolated hAFs. n = 12. M,N) Spearman analysis showing a
   positive correlation between nuclear‐localized S100A9 and PGE2 (M) or
   COX‐2 (N) in isolated hAFs. n = 12. Data are mean ± SEM. Two‐tailed
   unpaired Student's t‐test (A, B, D, F, G), or Mann–Whitney U test (C).
   *p < 0.05, **p < 0.01, ***p < 0.001.

2.3. Induction of Cellular Senescence by Nuclear S100A9 in hAFs

   To investigate the exact role of nuclear S100A9 in hAFs, we transfected
   hAFs isolated from the TNL amnion with a constructed adenoviral vector
   carrying a nuclear localization sequence (NLS) to the C‐ and N‐termini
   of S100A9 (NLS‐S100A9) to specifically express S100A9 in the nucleus.
   Nuclear overexpression was confirmed with immunofluorescent staining
   and Western blotting (Figure [79]S2A–C, Supporting Information).
   However, ELISA assay showed that extracellular S100A9 was not altered
   by transfection of NLS‐S100A9 in hAFs (Figure [80]S2D, Supporting
   Information). To screen for pathways affected by nuclear S100A9 in
   hAFs, transcriptomic sequencing was performed in hAFs with
   overexpression of S100A9 in the nuclei. Compared to transfection with
   an empty adenoviral vector, there were 136 down‐regulated and 41
   up‐regulated genes (fold change ≥ 1.5 and p < 0.05) after nuclear
   overexpression of S100A9 in hAFs (Figure [81]S2E, Table [82]S1,
   Supporting Information). The gene set enrichment analysis (GSEA) showed
   that the down‐regulated gene sets seen with nuclear S100A9
   overexpression were associated with cell cycle and DNA replication
   (Figure [83] 3A; Figure [84]S2F, Supporting Information). Analysis of
   the differentially expressed genes with the Kyoto Encyclopedia of Genes
   and Genomes (KEGG) revealed that cell cycle and cellular senescence
   ranked first and sixth respectively among the enriched KEGG pathways
   (Figure [85]3B), which coincides largely with the Gene Ontology (GO)
   analysis showing that cell division and mitotic cell cycle were the top
   two enriched GO terms (Figure [86]3C). The altered genes associated
   with the cell cycle GO term were illustrated in the heatmap
   (Figure [87]S2G, Supporting Information), which included CCNE2 (cyclin
   E2), CDK1 (cyclin dependent kinase 1), E2F1 (E2F transcription factor
   1), and CDKN1A (cyclin dependent kinase inhibitor 1A, also known as
   p21). These alterations were subsequently confirmed with qRT‐PCR
   (Figure [88]3D). Subsequent Western blotting analysis showed that
   nuclear overexpression of S100A9 increased the protein abundance of p21
   and p53 in hAFs (Figure [89]3E), which was confirmed by dual
   immunofluorescent staining showing increased p21 abundance in hAFs with
   nuclear S100A9 overexpression (Figure [90]S3A, Supporting Information).
   Consistently, SA‐β‐gal staining revealed an increased ratio of
   SA‐β‐gal‐positive cells in hAFs with nuclear overexpression of S100A9
   (Figure [91]3F) along with an increased abundance of γH2AX, a sensitive
   marker of the DNA damage response,^[ [92]^13 ^] as demonstrated with
   immunofluorescence staining (Figure [93]3G). Moreover, nuclear
   overexpression of S100A9 increased PGE2 production (Figure [94]3H) and
   COX‐2 expression (Figure [95]S3B,C, Supporting Information) in hAFs. By
   contrast, the expression of COX‐1, a constitutive counterpart of the
   inducible cyclooxygenase in prostaglandin synthesis, was not affected
   by nuclear overexpression of S100A9 in hAFs (Figure [96]S3D, Supporting
   Information). Conversely, small interfering RNA (siRNA)‐mediated
   knockdown of S100A9 in hAFs isolated from the TL amnion significantly
   decreased nuclear S100A9 abundance along with decreased p21 and p53
   abundance (Figure [97]3I). To examine the effect of extracellular
   S100A9 on cellular senescence in hAFs, hAFs isolated from the TNL
   amnion were treated with recombinant human S100A9 (rhS100A9; 0.05, 0.5
   and 5 µg mL^−1) for 72 h or with rhS100A9 (0.5 µg mL^−1) for 24, 48,
   and 72 h. These time course and concentration‐dependent studies showed
   that extracellular S100A9 had no effect on the abundance of p53 and p21
   protein in hAFs (Figure [98]S3E,F, Supporting Information). These data
   indicate that nuclear overexpression of S100A9 can trigger cellular
   senescence in hAFs, whereas extracellular S100A9 had no such effect.

Figure 3.

   Figure 3
   [99]Open in a new tab

   Induction of cellular senescence by nuclear S100A9 in hAFs. A) Gene Set
   Enrichment Analysis (GSEA) in Kyoto encyclopedia of genes and genomes
   (KEGG) enrichment showed that the cell cycle pathway was downregulated
   in hAFs by overexpression of S100A9 in the nuclei (n = 3 per group).
   NES, normalized enrichment scores; FDR, false discovery rate; NC,
   control with transfection of empty vector; OE, overexpression of S100A9
   in the nuclei. B,C) KEGG (B) and Gene ontology (GO) biological process
   (C) analyses of the transcriptomic sequencing data obtained from hAFs
   with nuclear S100A9 overexpression (n = 3 per group) showing the
   pathways that the differentially expressed genes were enriched in. D)
   qRT‐PCR analysis showing the difference of CDKN1A, CCNE2, CDK1, and
   E2F1 mRNA abundance between hAFs with and without nuclear S100A9
   overexpression. n = 4. E) Western blot analysis showing the difference
   of p21 and p53 protein abundance between hAFs with and without nuclear
   S100A9 overexpression. n = 4. The left panel is the representative blot
   and the right panel is the average data. F) SA‐β‐gal staining showing
   the percentage of SA‐β‐gal positive cells (blue color) in cultured hAFs
   with and without nuclear S100A9 overexpression. The left panel is the
   representative image and the right panel is the average data. Scale
   bars, 25 µm. n = 5. G) Immunofluorescence staining of γH2AX, a
   sensitive marker of DNA damage response, showing the percentage of
   γH2AX‐positive cells (red color) in hAFs with and without nuclear
   S100A9 overexpression. Nuclei were counterstained with DAPI (blue
   color). Dashed lines indicate the nuclear boundaries of the
   γH2AX‐positive (> 2 red dots) cells. The left panel is the
   representative image and the right panel is the average data. Scale
   bars, 25 µm. n = 3. H) The difference of PGE2 concentrations between
   the conditioned culture medium obtained from hAFs with and without
   nuclear S100A9 overexpression as measured with ELISA. n = 5. I) Western
   blot analysis showing the protein abundance of p21 and p53 in hAFs with
   or without siRNA‐mediated knockdown of S100A9. The left panel is the
   representative blot and the right panel is the average data. NC,
   negative scrambled siRNA; GAPDH, α‐tubulin, and Lamin A/C were used as
   internal controls for total cellular, cytoplasmic, and nuclear protein,
   respectively. si‐1 and si‐2 are two separated siRNA against S100A9. n =
   4. Data are mean ± SEM. Two‐tailed paired Student's t‐test (D‐H), or
   One‐way ANOVA followed by Tukey's post hoc tests (I). *p < 0.05, **p
   < 0.01, ***p < 0.001.

2.4. Disruption of Heterochromatin by Nuclear S100A9 in hAFs

   To screen for potential partners that nuclear S100A9 might interact
   with in the induction of cellular senescence in hAFs,
   immunoprecipitation (IP) was performed following transfection of a
   vector expressing Flag‐tagged S100A9 in the nuclei of hAFs. The
   proteins precipitated by the anti‐Flag antibody were then subject to
   analysis with liquid chromatography‐tandem mass spectrometry (LC‐MS)
   (Figure [100] 4A). A list of candidate transcription factors and
   cofactors that S100A9 might interact with to modulate heterochromatin
   were identified in the nucleus, including CBX5, and methyl‐CpG‐binding
   domain protein 3 (MBD3), a component of the NuRD complex^[ [101]^14 ^]
   (Figure [102]4B,C; Figure [103]S4A and Table [104]S2, Supporting
   Information). These interactions between S100A9 and CBX5 or MBD3 were
   further validated with co‐immunoprecipitation (Co‐IP) assay in hAFs
   with nuclear S100A9 overexpression (Figure [105]4D–G). Interestingly,
   protein‐protein docking prediction with ZDOCK SERVER or HDOCK SERVER
   showed that the key sequence in CBX5 that might interact with S100A9
   ranged from residues 134 to 177 (Docking Score: 1387.324 for ZDOCK and
   ‐228.52 for HDOCK) (Figure [106]4H; Figure [107]S4B, Supporting
   Information), which overlaps with the reported sequence (residues
   121–179, C‐terminal chromoshadow domain) that interacts with
   KRAB‐associated protein 1 (KAP1),^[ [108]^15 ^] an essential scaffold
   for the formation of the repressor complex in heterochromatin
   formation.^[ [109]^16 ^] This overlap indicates that nuclear
   overexpression of S100A9 may intervene in the interaction between CBX5
   and KAP1, resulting in heterochromatin disruption. This notion was
   supported by further Co‐IP assay showing decreased CBX5 and KAP1
   binding upon nuclear overexpression of S100A9 in hAFs (Figure [110]4I).
   The disruption of heterochromatin by nuclear S100A9 overexpression was
   further supported by the finding of decreased H3K9me3 level, a
   representative hallmark of heterochromatin erosion, in the nuclei of
   hAFs (Figure [111]4J–L; Figure [112]S4C,D, Supporting Information). By
   contrast, overexpression of CBX5 not only counteracted the decrease in
   H3K9me3 abundance (Figure [113]S4C,D, Supporting Information), but also
   attenuated the increases in p21 and p53 abundance (Figure [114]4M;
   Figure [115]S4E, Supporting Information) and SA‐β‐gal‐positive cells
   (Figure [116]4N), which were induced by nuclear S100A9 overexpression
   in hAFs. Conversely, siRNA‐mediated knockdown of CBX5 increased the
   abundance of p21 and p53 (Figure [117]S4F, Supporting Information).
   These data indicate that nuclear overexpression of S100A9 can induce
   cellular senescence by disrupting heterochromatin via segregation of
   the heterochromatin maintenance protein in hAFs.

Figure 4.

   Figure 4
   [118]Open in a new tab

   Disruption of heterochromatin by nuclear S100A9 in hAFs. A) Schematic
   diagram illustrating the workflow of mass spectrometry for
   identification of S100A9‐interacting proteins in hAFs with or without
   nuclear S100A9 overexpression. n = 4 per group. B,C) Dot plot (B) and
   chord plot (C) of the GO cellular component enrichment analysis showing
   S100A9‐interacting proteins identified with mass spectrometry. D–G)
   Co‐IP analysis showing CBX5 and MBD3 binding with S100A9 in hAFs with
   and without nuclear S100A9 overexpression. (D) is immunoprecipitation
   with anti‐Flag‐conjugated beads, (E) is immunoprecipitation with
   anti‐S100A9 antibody, (F) is immunoprecipitation with anti‐CBX5
   antibody, and (G) is immunoprecipitation with anti‐MBD3 antibody. Input
   and pre‐immune IgG served as positive and negative controls
   respectively. n = 3. NC, control with empty vector; OE, nuclear S100A9
   overexpression. H) Protein‐protein docking prediction by Z‐DOCK showing
   the overall co‐complex structure of S100A9 (cyan color) with CBX5
   (yellow color). The panel in the large box is the enlarged view of
   S100A9‐CBX5 binding site. I) Co‐IP analysis showing CBX5 binding with
   KAP1 in hAFs with and without nuclear S100A9 overexpression. Input and
   pre‐immune IgG served as positive and negative controls respectively. n
   = 3. J) 3D reconstruction of z‐stack immunofluorescence images of
   H3K9me3 (red color) in the nucleus of hAFs with or without nuclear
   S100A9 overexpression. Scale bars, 2 µm. The mean fluorescence
   intensity of H3K9me3 (right) was related to Figure S4C. n = 4. K)
   Western blot analysis showing a decreased abundance of H3K9me3 in hAFs
   with nuclear S100A9 overexpression as compared to that without nuclear
   S100A9 overexpression. n = 3. The upper panel is the representative
   blot and the lower panel is the average data. L) Representative images
   of transmission electron microscopy (TEM) showing peripheral
   heterochromatin in hAFs with or without nuclear S100A9 overexpression.
   The percentages of cells with compromised nuclear peripheral
   heterochromatin are presented at the bottom of the TEM images in each
   group. Scale bars, 2 µm. M) Western blot analysis showing the abundance
   of p53 and p21 in hAFs with or without nuclear S100A9 overexpression in
   the presence or absence of CBX5 overexpression. n = 3. The left panel
   is the representative blot and the right panel is the average data. N)
   SA‐β‐gal staining showing the difference in the percentage of cells
   positive for SA‐β‐gal (blue color) in hAFs with or without nuclear
   S100A9 overexpression in the presence or absence of CBX5
   overexpression. n = 5. Scale bars, 25 µm. The left panel is the
   representative image and the right panel is the average data. Data are
   mean ± SEM. Two‐tailed paired Student's t‐test (J, K), or One‐way ANOVA
   followed by Tukey's post hoc tests (M, N). *p < 0.05, **p < 0.01, ***p
   < 0.001.

2.5. De‐Repression of LINE1 by Nuclear S100A9 via Heterochromatin Disruption
in hAFs

   Given that constitutive heterochromatin is crucial for silencing TEs
   across diverse evolutionarily distant organisms,^[ [119]^7e,f ^]
   Cleavage Under Targets and Tagmentation (CUT&Tag) analysis using an
   H3K9me3 antibody was conducted to examine whether the enrichment of
   H3K9me3 at TE regions was altered by nuclear overexpression of S100A9
   in hAFs, followed by high‐throughput sequencing. As expected, nuclear
   overexpression of S100A9 in hAFs significantly reduced the overall
   enrichments of H3K9me3 at whole genome regions (Figure [120]S5A,
   Supporting Information), especially at TE regions, including LINE1,
   SINE (short interspersed nuclear elements) and LTR (long terminal
   repeats) elements (Figure [121] 5A–D; Figure [122]S5B–D, Supporting
   Information). Since LINE1 are the most abundant autonomous
   retrotransposons that are normally silenced in primates,^[ [123]^7b,f
   ^] and increased LINE1 retrotransposition is associated with cellular
   senescence in a variety of non‐gestational tissues,^[ [124]^8 ^] we
   subsequently investigated whether nuclear S100A9 overexpression
   de‐repressed LINE1 via segregation of the heterochromatin maintenance
   protein from LINE1 elements to promote cellular senescence in hAFs with
   chromatin immunoprecipitation (ChIP) assay. This showed that the
   enrichment of heterochromatin maintenance proteins including CBX5, MBD3
   and KAP1 at LINE1 elements was all significantly decreased along with
   decreased H3K9me3 enrichment upon nuclear S100A9 overexpression in hAFs
   (Figure [125]5E; Figure [126]S5E, Supporting Information).
   Consistently, qRT‐PCR and Western blotting showed that LINE1
   transcripts (Figure [127]5F; Figure [128]S5F, Supporting Information)
   and the coded proteins ORF1p and ORF2p (Figure [129]5G) were markedly
   up‐regulated by nuclear S100A9 overexpression in hAFs. Because aberrant
   LINE1 expression is linked to altered retrotransposon activity,^[
   [130]^8b–d ^] we measured the de novo LINE1 retrotransposon efficiency
   in hAFs by using a well‐characterized EGFP‐based reporter system
   (Figure [131]5H).^[ [132]^17 ^] As expected, a higher LINE1
   retrotransposon activity was detected in hAFs with nuclear S100A9
   overexpression (Figure [133]5H) along with increased LINE1 cDNA in
   total and extrachromosomal DNA contents (Figure [134]5I,J;
   Figure [135]S5G, Supporting Information). Conversely, siRNA‐mediated
   knockdown of S100A9 reduced LINE1 mRNA and cDNA contents significantly
   in hAFs isolated from the TL amnion (Figure [136]S5H‐J, Supporting
   Information). Moreover, LINE1 mRNA, total and extrachromosomal LINE1
   cDNA contents were also significantly increased in hAFs isolated from
   the TL amnion as compared to the TNL amnion (Figure [137]5K,L;
   Figure [138]S5K, Supporting Information). In addition to LINE1, we also
   examined the effect of nuclear overexpression of S100A9 on human
   endogenous retroviruses (HERV)‐K, an evolutionarily young subfamily of
   HERVs that has been reported to be associated with aging.^[ [139]^18 ^]
   The result showed that HERV‐K expression was not affected by nuclear
   overexpression of S100A9 in hAFs (Figure [140]S5L, Supporting
   Information). Taken together, these results revealed a novel role of
   nuclear S100A9 in the relief of LINE1 repression through disruption of
   heterochromatin associated with LINE1 elements via segregation of the
   heterochromatin maintenance protein in hAFs (Figure [141]5M).

Figure 5.

   Figure 5
   [142]Open in a new tab

   Activation of LINE1 elements by nuclear S100A9 in hAFs. A, B) The
   metaplot (A) and violin plot (B) show H3K9me3 signals in
   H3K9me3‐enriched TEs in hAFs with or without nuclear S100A9
   overexpression. RPKM, Reads Per Kilobase per Million mapped reads. NC,
   control with empty vector transfection; OE, nuclear S100A9
   overexpression. C, D) The metaplot (C) and violin plot (D) showing
   H3K9me3 signals in H3K9me3‐enriched LINE1 in hAFs with or without
   nuclear S100A9 overexpression. E) ChIP‐qRT‐PCR assay showing the
   enrichment of H3K9me3, CBX5, MBD3, and KAP1 at indicated regions of
   LINE1 in hAFs with or without nuclear S100A9 overexpression. Pre‐immune
   IgG served as the negative control, which was shown in Figure S5E
   (Supporting Information). Top panel showing the regions of the
   evolutionarily young LINE‐1 family (L1HS), aligned in qRT‐PCR, with the
   following details: 1# and 2# are aligned to the 5′UTR (1326‐1598 bp and
   1614–1886 bp, respectively), 3# is aligned to ORF1 (2234‐2485 bp), and
   4# is aligned to ORF2 (3130‐3385 bp). n = 4. F) LINE1 mRNA abundance in
   hAFs with or without nuclear S100A9 overexpression as measured with
   qRT‐qPCR with primers aligned to ORF1p. n = 4. Data with primers
   aligned to ORF2p is shown in Figure S5F (Supporting Information). G)
   Western blot analysis showing the protein abundance of ORF1p and ORF2p
   in hAFs with or without nuclear S100A9 overexpression. n = 5. H)
   Quantification of the de novo LINE1 retrotransposition efficiency in
   hAFs with or without nuclear S100A9 overexpression. n = 4. The upper
   panel is the schematic diagram showing the structure of LINE1
   retrotransposition efficiency reporter pEGFP‐LRE3. I) Schematic diagram
   showing the procedure of extrachromosomal DNA extraction for qRT‐PCR.
   J) Extrachromosomal LINE1 cDNA contents in hAFs with or without nuclear
   S100A9 overexpressing as determined with qRT‐PCR. n = 4. MT‐ND5, a
   mitochondrial gene, was used as the reference gene. K,L) LINE1 mRNA (n
   = 8 each group, K) and extrachromosomal LINE1 cDNA contents (n = 6 each
   group, L) in hAFs isolated from TNL and TL groups as determined with
   qRT‐PCR. TL, term labor; TNL, term no labor. M) Diagram illustrating
   nuclear S100A9‐mediated heterochromatin erosion in hAFs. Increased
   S100A9 in the nucleus interacts with heterochromatin proteins CBX5 and
   MBD3, resulting in their dissociation from heterochromatin, thereby
   leading to heterochromatin erosion and subsequent de‐repression of
   LINE1 elements in hAFs at parturition. Data are mean ± SEM. Two‐sided
   Wilcoxon signed‐rank test (B, D), two‐tailed paired (E‐H, J), or
   unpaired Student's t‐test (K, L). *p < 0.05, **p < 0.01, ***p < 0.001.

2.6. De‐Repression of LINE1 Causes Cellular Senescence Via Elicitation of the
Type I IFN Response in hAFs with Nuclear S100A9 Overexpression

   Accumulating evidence indicates that intermediates of activated LINE1
   transposition, i.e., cytosolic LINE1 cDNA, can be recognized by cGAS to
   activate the cGAS‐STING pathway, which can subsequently elicit the type
   I IFN response with consequent cellular senescence in non‐gestational
   tissues.^[ [143]^8 ^] In support of this, GSEA analysis of the
   transcriptomic sequencing data obtained from hAFs showed that the type
   I IFN response (i.e., IFN α/β signaling) was indeed activated by
   nuclear S100A9 overexpression (Figure [144]S6A, Supporting
   Information). Reactome enrichment analysis of the up‐regulated genes by
   nuclear S100A9 overexpression also revealed that many of the top ten
   enriched pathways were related to the type I IFN response (Figure [145]
   6A). Correspondingly, the expression of genes associated with the type
   I IFN response and SASP, including IFNA, IFNB, IFIT1
   (interferon‐induced protein with tetratricopeptide repeats 1 and 3),
   OAS1 (2′‐5′‐Oligoadenylate synthetase 1), ISG15 (interferon‐stimulated
   gene 15), CCL4 (C‐C motif chemokine ligand 4), CXCL10 (C‐X‐C motif
   chemokine ligand 10), etc., were all significantly upregulated by
   nuclear S100A9 overexpression in hAFs (Figure [146]6B; Figure [147]S6B,
   Supporting Information), as well as in hAFs isolated from the TL amnion
   as compared to the TNL amnion (Figure [148]6C). In line with the
   activation of cGAS‐STING pathway following LINE1 de‐repression,^[
   [149]^8 ^] the concentration of 2′3′‐cGAMP (2′3′‐cyclic GMP‐AMP), a
   cGAS downstream nucleotide messenger,^[ [150]^19 ^] was significantly
   increased by nuclear S100A9 overexpression in hAFs (Figure [151]6D),
   along with increased phosphorylation of STING, TANK‐binding kinase 1
   (TBK1) and IFN regulatory factor 3 (IRF3), the three molecules
   downstream of the cGAS‐STING pathway^[ [152]^19 ^] (Figure [153]6E).
   Conversely, inhibition of cGAS and STING activities with RU521 and H151
   respectively blocked not only the increases in the phosphorylation of
   STING, TBK1 and IRF3 (Figure [154]6E), but also the increases in
   transcripts associated with the type I IFN response and SASP, and
   protein abundance of p21 and p53 (Figure [155]S6C,D, Supporting
   Information) induced by nuclear S100A9 overexpression in hAFs. This
   indicates that the cGAS‐STING signaling pathway is involved in the
   induction of cellular senescence by nuclear overexpression of S100A9 in
   hAFs. Moreover, Tenofovir and Lamivudine (3TC), the two nucleoside
   analog reverse‐transcriptase inhibitors (NRTIs) that have been used to
   inhibit LINE1 reverse transcription,^[ [156]^20 ^] effectively
   attenuated LINE1 retrotransposition (Figure [157]6F; Figure [158]S6E,
   Supporting Information), type I IFN response and SASP (Figure [159]6G;
   Figure [160]S6F, Supporting Information) induced by nuclear
   overexpression of S100A9 in hAFs along with decreased p21 and p53
   abundance (Figure [161]6H; Figure [162]S6G, Supporting Information),
   reduced proportion of SA‐β‐gal‐positive cells (Figure [163]6I), and
   decreased COX‐2 expression and PGE2 production (Figure [164]6J;
   Figure [165]S6H, Supporting Information). In contrast, Etravirine, a
   non‐nucleoside analog reverse‐transcriptase inhibitor (NNRTI) used in
   anti‐human immunodeficiency virus (HIV) therapy with no effect on LINE1
   reverse transcription,^[ [166]^20c ^] failed to alleviate the type I
   IFN response and cellular senescence‐induced by nuclear overexpression
   of S100A9 in hAFs (Figure [167]S6I‐K, Supporting Information). These
   data substantiate that nuclear overexpression of S100A9 promotes
   cellular senescence and intensifies SASP and PGE2 production via
   elicitation of the type I IFN response following LINE1 de‐repression in
   hAFs (Figure [168]6K).

Figure 6.

   Figure 6
   [169]Open in a new tab

   Activation of LINE1 by nuclear S100A9 induces type I IFN response and
   cellular senescence in hAFs. A) Reactome pathway enrichment analysis
   showing the pathway enrichment of up‐regulated expressed genes as
   revealed with transcriptomic sequencing in hAFs with and without
   nuclear S100A9 overexpression (n = 3 per group). B) qRT‐PCR analysis
   showing the abundance of mRNA associated with type I IFN responses and
   SASP in hAFs with or without nuclear S100A9 overexpression. n = 4. NC,
   control with empty vector transfection; OE, nuclear S100A9
   overexpression. C) qRT‐PCR analysis showing the abundance of mRNA
   associated with type I IFN response and SASP in hAFs isolated from TL
   (n = 8) and TNL (n = 8). TL, term labor; TNL, term no labor. D)
   Measurement with ELISA showing the difference of 2′3’‐cGAMP
   concentration in hAFs with or without nuclear S100A9 overexpression. n
   = 5. E) Western blot analysis showing the difference in protein
   abundance of p‐STING, p‐IRF3, and p‐TBK1 in hAFs with or without
   nuclear S100A9 overexpression in the presence or absence of cGAS
   inhibitor RU521 and STING inhibitor H151. n = 4. The left panel is the
   average data and the right panel is the representative blot. F, G) The
   difference of LINE1 cDNA contents (F) and abundance of mRNA associated
   with type I IFN response and SASP (G) in hAFs with or without nuclear
   S100A9 overexpression in the presence or absence of Tenofovir (Tenof).
   n = 4. H) Western blot analysis showing the difference in protein
   abundance of p21 and p53 in hAFs with or without nuclear S100A9
   overexpression in the presence or absence of Tenof. n = 4. Left panel
   is the representative blot and the right panel is the average data. I)
   SA‐β‐gal staining of hAFs with or without nuclear S100A9 overexpression
   in the presence or absence of Tenof. n = 5. The Left panel is the
   representative image and the right panel is the average data. Scale
   bars, 25 µm. J) ELISA showing PGE2 production in hAFs with or without
   nuclear S100A9 overexpression in the presence or absence of Tenof. n =
   4. K) Schematic diagram showing the activation of cGAS‐STING pathway,
   type I IFN response, and SASP by LINE1 in hAFs with nuclear S100A9
   overexpression. Data are mean ± SEM. Two‐tailed paired (B, D) or
   unpaired Student's t‐test (C), or One‐way ANOVA followed by Tukey's
   post hoc tests (E‐J). *p < 0.05, **p < 0.01, ***p < 0.001.

2.7. De‐Phosphorylation of S100A9 at Thr 113 is Required for Nuclear
Translocation in hAFs

   Given the crucial role of the nuclear S100A9 in the induction of
   cellular senescence in hAFs, we next explored the molecular mechanism
   underlying S100A9 nuclear translocation in hAFs. Interestingly, we
   found a significant reduction in S100A9 phosphorylation at Thr 113
   (p‐S100A9) in hAFs isolated from the TL amnion as compared to that from
   the TNL amnion (Figure [170] 7A–D). Moreover, the abundance of p‐S100A9
   correlated negatively with the abundance of S100A9 in the nucleus and
   the levels of p21 or p53 in hAFs (Figure [171]S7A‐C, Supporting
   Information). Furthermore, p‐S100A9 was found to be exclusively
   localized in the cytosol but not in the nuclei of hAFs (Figure [172]7E;
   Figure [173]S7D, Supporting Information), indicating that
   de‐phosphorylation of S100A9 at Thr 113 is required for its nuclear
   translocation. To confirm this, vectors carrying Flag‐tagged wild‐type,
   T113A (Thr→Ala) de‐phosphorylation‐mimic mutant, or T113D (Thr→Asp)
   phosphorylation‐mimic mutant S100A9 were constructed and transfected
   into hAFs. Immunofluorescent staining and Western blotting analyses
   showed that T113D phosphorylation‐mimic mutant S100A9 was mainly
   localized in the cytosol (Figure [174]7F,G), while T113A
   de‐phosphorylation‐mimic mutant S100A9 had higher nuclear accumulation
   than wild‐type S100A9 or T113D phosphorylation‐mimic mutant S100A9
   (Figure [175]7F). Accordingly, overexpression of T113A
   de‐phosphorylation‐mimic mutant S100A9 led to LINE1 de‐repression
   (Figure [176]7H), elicitation of the type I IFN responses, and
   consequently cellular senescence in hAFs (Figure [177]7I–K), while
   overexpression of T113D phosphorylation‐mimic mutant S100A9 had no such
   effects (Figure [178]S7E,F, Supporting Information). Moreover, Co‐IP
   assay showed that there was abundant binding between CBX5 and KAP1, but
   almost no binding of CBX5 and S100A9 in hAFs with T113D
   phosphorylation‐mimic mutant S100A9 overexpression (Figure [179]S7G,H,
   Supporting Information). These results indicate that de‐phosphorylation
   of S100A9 at Thr 113 is required for its nuclear translocation, where
   it de‐represses LINE1 via segregation of the heterochromatin
   maintenance protein, which further induces cellular senescence in hAFs.

Figure 7.

   Figure 7
   [180]Open in a new tab

   De‐phosphorylation of S100A9 at Thr 113 enables its nuclear
   translocation in hAFs. A–D) Western blotting analysis showing the
   differences of total S100A9 and phosphorylated S100A9 at Thr 113
   (p‐S100A9) in hAFs isolated from TL (n = 6) and TNL (n = 6). (A) is the
   Western blot and (B–D) is the average data. TL, term labor; TNL, term
   no labor. E) Western blot analysis showing the abundance of S100A9 and
   p‐S100A9 in the cytosolic and nuclear fractions of hAFs isolated from
   TL and TNL. α‐tubulin and Lamin A/C were used as internal controls for
   cytoplasmic and nuclear protein, respectively. n = 3. F and G) Western
   blot analysis (F) and immunofluorescent staining (G) of Flag (green
   color) showing the difference of cytosolic and nuclear distributions of
   wild‐type S100A9, T113D mutant S100A9 and T113A mutant S100A9 in hAFs.
   The left panel of (F) is the representative blot, and the right panel
   of (F) is the average data. n = 3. Nuclei and cytoskeleton were
   counterstained with DAPI (blue color) and phalloidin (red color)
   respectively. Scar bar, 25 µm. H and I) Measurement with qRT‐qPCR
   showing the difference of LINE1 transcripts (H), and the mRNA
   associated with type I IFN response and SASP (I) in hAFs with or
   without overexpression of S100A9 carrying T113A mutant. n = 4. J)
   Western blot analysis showing the difference of p53 and p21 protein in
   hAFs with or without overexpression of S100A9 carrying T113A mutant. n
   = 4. The left panel is the representative blot and the right panel is
   the average data. K) SA‐β‐gal (blue) staining of hAFs with or without
   overexpression of S100A9 carrying T113A mutant. Scale bars, 25 µm. n =
   5. The left panel is the representative image and the right panel is
   the average data. Data are mean ± SEM. Mann–Whitney U test (C),
   Two‐tailed unpaired (B, D) or paired Student's t‐test (H‐K), or One‐way
   ANOVA followed by Tukey's post hoc tests (F). *p < 0.05, **p < 0.01,
   ***p < 0.001.

2.8. Induction of Preterm Birth by Nuclear S100A9 in the Mouse

   The role of nuclear S100A9 in parturition was further investigated in
   pregnant mice. The mouse fetal membranes are comprised of the amnion
   and yolk sac membranes.^[ [181]^21 ^] Immunofluorescent staining showed
   an intense staining of S100A9 in the nuclei in both the amnion and yolk
   sac membranes at 19 days post‐coitum (dpc) (Figure [182] 8A;
   Figure [183]S8A, Supporting Information). Consistently, Western
   blotting analysis showed that the abundance of nuclear S100A9 was
   significantly higher at 19 dpc than at 17.5 dpc (Figure [184]8B).
   Correspondingly, transcripts of LINE1 (Figure [185]8C) and those
   related to the type 1 IFN response and SASP (Ifna, Ifnb, Isg15, Ifit1,
   Ccl4, Cxcl10) (Figure [186]8D) were significantly increased along with
   increased p21 protein abundance (Figure [187]8E) in the fetal membranes
   at 19 dpc as compared to 17.5 dpc. These data suggest that nuclear
   translocation of S100A9, LINE1 de‐repression, and cellular senescence
   are increased in the mouse fetal membranes toward the end of gestation.

Figure 8.

   Figure 8
   [188]Open in a new tab

   Nuclear overexpression of S100A9 in the fetal membranes causes preterm
   birth in the mouse. A) Representative images of immunofluorescence
   staining showing intense S100A9 staining (green color) in the cell
   nuclei of the mouse amnion layer at 19 dpc. Nuclei were counterstained
   with DAPI (blue color). Scale bars, 25 µm. n = 3. B) Western blot
   analysis showing increased abundance of S100A9 in the nuclear fraction
   extracted from the mouse fetal membranes at 19 dpc as compared to 17.5
   dpc. n = 5 per group. The left panel is the blot and the right panel is
   the average data. C,D) Measurement with qRT‐PCR showing increased
   abundance of LINE1 (C), mRNA associated with type I IFN response, and
   SASP (D) in the mouse fetal membranes at 19 dpc as compared to 17.5
   dpc. n = 5 per group. E) Western blot analysis showing increased p21
   protein abundance in the mouse fetal membranes at 19 dpc as compared to
   17.5 dpc. n = 6 per group. The left panel is the blot and the right
   panel is the average data. F) Time‐line illustrating the procedure of
   S100A9 administration in pregnant mice. G,H) Effects of intra‐amniotic
   injection of scAAV vector expressing S100A9 (1 × 10^10 TU/10 µl/ per
   gestational sac at 15.5 dpc) in the nucleus in the presence or absence
   of Tenofovir (Tenof; 2 nmol/10 µl/per gestational sac) on delivery time
   (G) and rates of preterm and post‐term birth (H). n = 11 per group. NC,
   control empty vector. OE, scAAV vector expressing S100A9 in the
   nucleus. I) Gestational lengths in mice with an intra‐amniotic
   injection of scAAV vector expressing S100A9 in the nucleus in the
   presence or absence of Tenof (2 nmol/10 µl/per gestational sac). n = 11
   per group. J) Measurement with qRT‐PCR showing the abundance of LINE1
   transcripts and mRNA associated with type I IFN response and SASP in
   the mouse fetal membranes with intra‐amniotic injection of scAAV vector
   expressing S100A9 in the nucleus in the presence or absence of Tenof (2
   nmol/10 µl/per gestational sac). n = 6 per group. K) Western blot
   analysis showing the protein abundance of p21 in the mouse fetal
   membranes with an intra‐amniotic injection of scAAV vector expressing
   S100A9 in the nucleus in the presence or absence of Tenof (2 nmol/10
   µl/per gestational sac). n = 6 per group. The left panel is the
   representative blot and the right panel is the average data. L)
   SA‐β‐gal (blue color) staining of the amnion layer of the mouse fetal
   membranes with an intra‐amniotic injection of scAAV vector expressing
   S100A9 in the nucleus in the presence or absence of Tenof (2 nmol/10
   µl/per gestational sac). Nuclei were counterstained with nuclear fast
   red (red color). n = 3 per group. Scale bars, 50 µm. The Left panel is
   the representative image and the right panel is the average data. Data
   are mean ± SEM. Two‐tailed unpaired Student's t‐test (B‐E),
   Gehan‐Breslow‐Wilcoxon test (I), or One‐way ANOVA followed by Tukey's
   post hoc tests (G, J‐L). *p < 0.05, **p < 0.01, ***p < 0.001.

   Intra‐amniotic injection of the self‐complementary adeno‐associated
   virus (scAAV) vector (1 × 10^10 transducing units (TU)/10 µL/per
   gestational sac) expressing S100A9 specifically in the nucleus at 15.5
   dpc (Figure [189]8F; Figure [190]S8B,C, Supporting Information) had no
   effect on fetal mortality (Figure [191]S8D), but led to a 81.8% preterm
   birth rate (Figure [192]8G–I), along with increased transcripts of
   LINE1 and those associated with the type I IFN response and SASP
   (Figure [193]8J) in the mouse fetal membranes. Meanwhile, the abundance
   of p21 protein (Figure [194]8K), and the proportion of p21‐ and
   SA‐β‐gal‐positive cells (Figure [195]8L; Figurse [196]S8E
   and [197]S9A‐C, Supporting Information) were also significantly
   increased in the mouse fetal membranes. Notably, intra‐amniotic
   injection of the reverse transcription inhibitor Tenofovir (0.2 nmol/10
   µL/per gestational sac) not only tended to prolong gestational age by
   itself, but also delayed preterm birth induced by intra‐amniotic
   injection of scAAV vector expressing S100A9 in the nucleus
   (Figure [198]8G–I), along with decreased transcripts of genes
   associated with the type I IFN response and SASP (Figure [199]8J), p21
   protein (Figure [200]8K), and the proportion of cells positive for p21‐
   and SA‐β‐gal (Figure [201]8L; Figures [202]S8E and [203]S9A–C,
   Supporting Information) in the mouse fetal membranes. Moreover,
   intra‐amniotic injection of scAAV vector expressing S100A9 in the
   nucleus also significantly increased PGE2 levels in the mouse fetal
   membranes, an effect also attenuated by Tenofovir as well
   (Figure [204]S8F, Supporting Information).

3. Discussion

   There are three major cell types in the human fetal membranes, i.e.,
   amnion epithelial cells, chorion trophoblasts, and mesenchymal
   fibroblasts.^[ [205]^11a,c ^] A series of studies have identified
   cellular senescence in amnion epithelial cells and chorionic
   trophoblasts.^[ [206]^2b–d ^] However, it is not known whether
   mesenchymal fibroblasts of the fetal membranes, an important source of
   ECM, pro‐inflammatory and pro‐labor factors, also undergo cellular
   senescence toward the end of gestation to participate in parturition.
   In this study, we found that cellular senescence in hAFs was an
   important feature of parturition as well. We demonstrated that nuclear
   translocation of S100A9 triggered cellular senescence in hAFs through
   segregation of the heterochromatin maintenance protein with resultant
   heterochromatin erosion and LINE1 de‐repression. De‐repression of LINE1
   further activated the cGAS‐STING pathway and elicitation of the type I
   IFN response, which ultimately led to cellular senescence in hAFs. Of
   interest, increased nuclear translocation of S100A9 was observed only
   in hAFs but not in hEPCs despite cellular senescence occurring in both
   cell types at parturition, suggesting differential mechanisms may
   underlie cellular senescence in these two cell types. Previous studies
   have shown that telomere erosion and oxidative stress may underlie
   cellular senescence in hEPCs.^[ [207]^2b–d,f ^] In the present study,
   we demonstrated that nucleus‐translocated S100A9 may be a cause of
   cellular senescence in hAFs. As hAFs are an important source of ECM
   components and pro‐inflammatory factors of the fetal membranes, the
   induction of cellular senescence by nucleus‐translocated S100A9 in hAFs
   may be crucial to ECM remodeling and inflammatory reactions required
   for membrane rupture and labor onset. This crucial role of
   nucleus‐translocated S100A9 in parturition was illustrated by induction
   of preterm birth following intra‐amniotic injection of scAAV vector
   expressing S100A9 in the nucleus in the mouse.

   It is known that S100A9 is predominantly expressed in myeloid cells,
   such as neutrophils and monocytes, and functions primarily as a
   secretory chemoattractant or pro‐inflammatory cytokine that interacts
   with the toll‐like receptor (TLR), receptor for advanced glycation end
   products or CD68, either on its own or by forming heterodimers or
   heterotetramers with S100A8.^[ [208]^9 , [209]^22 ^] Interestingly,
   there is one study reporting that S100A9 may induce bone marrow stromal
   cell senescence as an extracellular cytokine via activation of the TLR4
   receptor.^[ [210]^23 ^] However, in our study, we found that nuclear
   rather than extracellular S100A9 took part in the induction of cellular
   senescence in hAFs. We do not currently have a plausible explanation
   for this discrepancy. More studies are required to confirm the effect
   of extracellular S100A9 on cellular senescence in different cell types.
   In line with our findings, there are two studies reporting the presence
   of S100A9 in the nuclei of other cell types.^[ [211]^24 ^] Schonthaler
   et al. have found that S100A9 is localized in the nuclei of
   keratinocytes where it modulates the expression of complement factor C3
   via chromatin binding and remodeling at a region upstream to the C3
   transcription start site rather than as a transcription factor.^[
   [212]^24a ^] Song et al. have demonstrated that S100A9 in the nuclei of
   MCF‐7 breast cancer cells may function as a coactivator for
   transcription factors, such as c‐Fos, CCAAT/enhancer‐binding protein β,
   Signal transducer and activator of transcription 3, etc.,.^[ [213]^24b
   ^] To the best of our knowledge, the present study is the first to show
   that S100A9 translocated to the nucleus may function as a
   heterochromatin disruptor via segregation of the heterochromatin
   maintenance protein CBX5 from the interaction with other
   heterochromatin maintenance proteins such as KAP1. Of interest, the
   predicted CBX5 sequence responsible for binding with S100A9 overlaps
   with the reported sequence which binds with KAP1,^[ [214]^15 ^]
   suggesting that nuclear overexpression of S100A9 may compete with KAP1
   for binding with CBX5, resulting in the loss of CBX5 from the
   heterochromatin maintenance protein complex. It has been reported CBX5
   usually serves as an adapter to recruit effector molecules to
   heterochromatin, including H3K9 methyltransferases (SUV39H1, SETDB1,
   G9a, etc.) for methylation of H3K9.^[ [215]^25 ^] Alternatively, CBX5
   may also enhance H3K9 methylation by stabilizing H3K9
   methyltransferases or preventing H3K9 de‐methylation.^[ [216]^26 ^]
   Thus, the loss of CBX5 from the heterochromatin maintenance complex may
   decrease the enrichment of H3K9me3 resulting in the disruption of
   heterochromatin integrity, as has been demonstrated in several previous
   studies.^[ [217]^8c,d ^] However, it is uncertain whether
   methyltransferases or demethylases involved in H3K9 methylation may be
   affected directly by nuclear S100A9 in hAFs.

   Previous studies have shown that phosphorylation of S100A9 at Thr 113
   not only contributes to microtubule reorganization in monocytes,^[
   [218]^27 ^] but also facilitates S100A9 secretion into the
   extracellular space in neutrophils.^[ [219]^28 ^] Here, we identified
   that de‐phosphorylation of S100A9 at Thr 113 enabled its nuclear
   translocation. Similarly, de‐phosphorylation has also been shown to
   mediate the nuclear import of a variety of other proteins, including
   nuclear factor of activated T cells, nuclear xenobiotic receptor CAR2,
   Y‐box‐binding protein 1, cofilin, etc.^[ [220]^29 ^] Activation of p38
   mitogen‐activated protein kinase is reported to be responsible for
   S100A9 phosphorylation at Thr 113 in monocytes and neutrophils.^[
   [221]^27 , [222]^30 ^] However, the factor driving de‐phosphorylation
   of S100A9 in hAFs in parturition remains to be determined.

   Emerging evidence indicates that genomic instability is a prominent
   contributor to cellular senescence, and condensed heterochromatin is
   required for safeguarding genomic integrity, at least in part, through
   LINE1 silencing.^[ [223]^5 , [224]^31 ^] The expression and
   retrotransposition of LINE1 elements have been reported to be
   associated with a number of physiological events, including zygotic
   genome activation, early embryonic development, neurological disorders,
   and cancers.^[ [225]^32 ^] A series of studies have demonstrated that
   LINE1 is de‐repressed and aberrantly transcribed to induce cellular
   senescence via activation of the cGAS‐STING‐type I IFN response in
   non‐gestational tissues.^[ [226]^8 ^] It is known that the cGAS‐STING
   pathway is primarily activated by the presence of cytosolic DNA, which
   may derive from damaged DNA, retrotransposons, or viral infections.
   Upon detection of cytosolic DNA, cGAS synthesizes cGAMP, which binds
   and activates STING. Activated STING further induces the production of
   type I IFNs and other inflammatory cytokines, which are part of the
   innate immune response,^[ [227]^33 ^] and are well‐recognized to play a
   crucial role in the initiation and maintenance of cellular
   senescence.^[ [228]^34 ^] Our study also revealed that the induction of
   cellular senescence by nuclear overexpression of S100A9 was accompanied
   by activation of the cGAS‐STING pathway and the type I IFN response in
   hAFs. Moreover, we also found that COX‐2 expression and PGE2 production
   were significantly increased by nuclear overexpression of S100A9, which
   could be blocked by a reverse transcription inhibitor in hAFs,
   indicating that the induction of COX‐2 expression and PGE2 production
   is a consequence of LINE1 retrotransposition‐sponsored cellular
   senescence in hAFs. Given that senescent cells can release a variety of
   SASP factors, including pro‐inflammatory cytokines that are
   well‐recognized inducers of COX‐2 expression and PGE2 production,^[
   [229]^35 ^] we believe that the induction of COX‐2 expression and PGE2
   production by nuclear S100A9 is attributed to SASP factors released
   from senescent cells.

   Notably, apart from LINE1, other TEs, such as the recently integrated
   HERV‐K subgroup, were also reported to be involved in aging.^[ [230]^18
   ^] Autio, et al.^[ [231]^18b ^] found a slight elevation in the
   expression levels of HERV‐K in peripheral blood mononuclear cells in
   older individuals compared to younger counterparts. A recent elegant
   study from Liu's lab reported that the activation of endogenous
   retroviruses such as HERV‐K could also activate innate immunity and
   trigger cellular senescence in human mesenchymal progenitor cells.^[
   [232]^18c ^] However, HERV‐K was not affected by nuclear S100A9 in our
   cell system. Whether other repetitive elements are also involved in the
   induction of cellular senescence by nuclear overexpression of S100A9 in
   hAFs awaits further studies.

   NRTIs are the most widely used drugs for HIV treatment. A series of
   studies have demonstrated that NRTIs, such as Tenofovir and 3TC, can
   mitigate the type I IFN and SASP responses and counteract cellular
   senescence triggered by LINE1 cDNA accumulation in multiple cell types
   and animals.^[ [233]^8b–e ^] Here we found that Tenofovir and 3TC were
   not only capable of ameliorating the aging phenotype of hAFs but also
   delayed the time of delivery in pregnant mice with nuclear
   overexpression of S100A9 in the mouse fetal membranes. Considering that
   Tenofovir use in pregnancy has generally been thought to be safe for
   both mother and fetus,^[ [234]^36 ^] there may be potential in
   developing Tenofovir as a therapeutic drug for the prevention of
   preterm birth. In addition to the inhibition of de‐repression of LINE1
   with Tenofovir, the molecular mechanism identified that regulates
   nuclear translocation of S100A9 may also serve as a therapeutic target
   for the prevention of preterm birth.

4. Conclusion

   We have discovered a novel role for nuclear‐translocated S100A9 as a
   heterochromatin disruptor to de‐repress LINE1, which subsequently
   activates the cGAS‐STING pathway resulting in the activation of the
   type I IFN response and consequent senescence of hAFs at parturition.
   We have also demonstrated that de‐phosphorylation of S100A9 at Thr 113
   is required for its nuclear translocation. Our findings not only expand
   knowledge on the function of nuclear‐translocated S100A9, but also
   provide new insights into the mechanism underlying cellular senescence
   in hAFs of the fetal membranes at parturition.

5. Experimental Section

Collection of Human Fetal Membranes and Processing of the Amnion Tissue

   This study was approved by the Ethics Committee of Ren Ji Hospital,
   Shanghai Jiao Tong University School of Medicine (approval code:
   KY2023‐007‐B). All ethical regulations relevant to participants of
   human research were followed. Human fetal membranes were collected from
   uncomplicated pregnancies at delivery with written informed consent.
   Participating subjects were divided into four groups: elective cesarean
   section without labor at term (designated as term no labor, TNL);
   spontaneous labor at term (designated as term labor, TL); preterm
   elective cesarean section without labor for maternal or fetal
   conditions including placenta previa, vasa previa, and fetal distress
   (designated as preterm no labor, PNL); and spontaneous preterm labor
   with no indication of infection (designated as preterm labor, PL).
   Pregnancies with complications such as preeclampsia, fetal growth
   restriction, gestational diabetes, and chorioamnionitis were excluded
   from this study. The demographic features of recruited women are listed
   in Tables [235]S3 and [236]S4, (Supporting Information).

   The amnion layer was collected by peeling it off the fetal membranes
   immediately after delivery for processing as follows. Several amnion
   tissue pieces were snap‐frozen in liquid nitrogen, and then sectioned
   for SA‐β‐gal staining, or processed for extraction of total RNA
   extraction, and total and nuclear protein for subsequent analysis with
   qRT‐PCR and Western blotting respectively. Parts of the amnion tissue
   were rolled prior to fixation in 10% formalin and then embedded in
   paraffin wax for subsequent sectioning and immunohistochemical or
   immunofluorescence staining. For isolation of primary hAFs, the entire
   amnion layer from the reflected membranes was used and processed as
   follows.

Isolation of Primary hAFs

   The amnion tissue was washed with sterile cold normal saline containing
   antibiotics (Life Technologies Inc.; #15640055) to remove residual
   blood, and then subject to consecutive digestions with 0.125% trypsin
   (Life Technologies Inc.; #9002‐07‐7) and 0.1% collagenase
   (Sigma–Aldrich; #C0130) to isolate hAFs following a protocol described
   previously.^[ [237]^37 ^] The identity of cells was verified by
   immunofluorescent staining for the mesenchymal cell marker vimentin and
   the epithelial marker cytokeratin‐7. More than 95% of the cells were
   vimentin‐positive (Figure [238]S1E, Supporting Information), which was
   consistent with the previous report.^[ [239]^37 ^]

   To compare the abundance of S100A9, p‐S00A9, COX‐2, p21, and p53
   protein in hAFs isolated from the TNL and TL groups, nuclear protein or
   total protein was extracted from isolated hAFs for analysis with
   Western blotting. To compare the LINE1 transcripts, LINE1 cDNA and
   those transcripts associated with the type I IFN response and SASP in
   hAFs between the TL and TNL groups, total RNA, cellular and
   extrachromosomal DNA were extracted from isolated hAFs for analysis
   with qRT‐PCR. To compare the PGE2 production in hAFs between TL and TNL
   groups, the conditional culture medium of hAFs was collected for
   analysis with ELISA.

Reagents and Plasmids

   Puromycin (#S9361), RU521 (#S6841), H‐151 (#S6652), Lamivudine/3TC
   (#S1706), Tenofovir (#S1401), and Etravirine (#S3080) were purchased
   from Selleckchem. rhS100A9 was purchased from R&D systems
   (#9254‐S9‐050). The adenovirus vectors carrying human genes including
   full‐length S100A9 in the nucleus (pDC315‐3NLS‐S100A9‐3NLS‐3Flag),
   full‐length CBX5 (pDC315‐CBX5‐SV40), wide type full‐length S100A9
   (pDC315‐S100A9‐3Flag‐SV40‐EGFP), site‐directed mutagenesis (T113A) of
   full‐length S100A9 (pDC315‐S100A9T113A‐3Flag‐SV40‐EGFP), site‐directed
   mutagenesis (T113D) of full‐length S100A9
   (pDC315‐S100A9T113D‐3Flag‐SV40‐EGFP) were prepared by Shanghai
   GeneChem, Co. Ltd. The “empty” vectors (pDC315‐3NLS‐3Flag, or pDC315,
   or pDC315‐3Flag‐SV40‐EGFP) were used for negative control where
   applicable. The scAAV vector expressing mouse full‐length S100A9
   (posCAAV‐CMV‐S100A9‐3NLS‐3Flag‐tWPA) in the nucleus was constructed by
   Shanghai OBiO Technology. A corresponding “empty” vector
   (pCsCAAV‐CMV‐3NLS‐3Flag‐tWPA) was used as a negative control. NLS
   sequences (DPKKKRKV) were copied from Invitrogen's pCMV/myc/nuc. The
   efficiency of overexpression and subcellular localization was examined
   with western blotting or immunofluorescence staining in the
   corresponding experiment. The information on antibodies used in this
   study is given in Table [240]S5, (Supporting Information).

Adenovirus Infection and siRNA Transfection in hAFs

   The hAFs isolated from the TL amnion were used for the S100A9 knockdown
   experiment, whereas the hAFs isolated from the TNL amnion were used for
   all other mechanistic studies to avoid the confounding effects of
   labor. For adenovirus infection, isolated hAFs were cultured for 24 h
   in Dulbecco's Modified Eagle Medium (DMEM; Life Technologies Inc.;
   #21063029) containing 10% fetal bovine serum (FBS; Biological
   Industries; #04‐121‐1A) and antibiotics at 37 °C in 5% CO[2]/95% air.
   Then, the cells were inoculated with adenovirus (1 × 10^9 TU/mL DMEM)
   with packaging plasmids carrying NLS‐S100A9, CBX5, wild‐type S100A9, or
   T113A mutant S100A9, T113D mutant S100A9 for 24 h. Afterward, the
   solution was removed, and the cells were washed with PBS and incubated
   with complete DMEM for 48 h in the presence or absence of RU521 (1 µM),
   H151 (1 µM), 3TC (10 µM), Tenofovir (5 µM), or Etravirine (5 µM) before
   harvesting. For siRNA transfection, isolated hAFs were immediately
   incubated with 50 nM of two sets of siRNA against CBX5 (1#:
   5′‐GCCGAUGACAUCAAAUCUATT‐3′; 2#: 5′‐GGUGAUUUAAUGUUCCUAATT‐3′) or two
   sets of siRNA against S100A9 (1#: 5′‐GCUUCGAGGAGUUCAUCAUTT‐3′; 2#:
   5′‐GCAACAUAGAGACCAUCAUTT‐3′) or randomly scrambled siRNA
   (5′‐UUCUCCGAACGUGUCACGUTT‐3′; GenePharma) in Opti‐MEM (Life
   Technologies Inc.; #31985062) using an electroporator (Nepa Gene) at
   165 V for 5ms. The cells were then incubated in complete DMEM for 72 h
   before harvesting for further processing and analysis. To investigate
   the effect of extracellular S100A9 on cellular senescence in hAFs, the
   cells were treated with rhS100A9 (0.05, 0.5, and 5 µg mL^−1) for 24,
   48, and 72 h 3 days after plating.

RNA Extraction and qRT‐PCR

   Total RNA was extracted from the snap‐frozen amnion tissue or hAFs with
   a total RNA isolation Kit (Foregene; #RE‐03113) according to the
   protocol provided by the manufacturer. After examination of RNA quality
   and concentration, reverse transcription was carried out using a
   Prime‐Script RT Master Mix Kit (TaKaRa; RR036B). The mRNA abundance of
   S100A9, IFNA, IFNB, CDK1, E2F1, CCNE2, CDKN1A, PTGS1, PTGS2, OAS1,
   IFIT1, ISG15, CCL4, CXCL10, CBX5, HERV‐K, and LINE1 was determined with
   qRT‐PCR using the above reverse‐transcribed cDNA and MagicSYBR Mixture
   (CoWin Biosciences; #CW3008). The housekeeping gene GAPDH
   (Glyceraldehyde 3‐phosphate dehydrogenase) was amplified in parallel as
   an internal control. The relative mRNA abundance was quantitated using
   the 2‐^△△Ct method, which was normalized to GAPDH. The primer sequences
   used for qRT‐PCR are illustrated in Table [241]S6, (Supporting
   Information).

Total, Nuclear and Cytoplasmic Protein Extraction

   Total cellular protein was extracted from the snap‐frozen amnion tissue
   or hAFs with an ice‐cold radioimmunoprecipitation assay lysis buffer
   (WeiaoBio; #WB0101) containing protease inhibitor (Roche; #04693124001)
   and phosphatase inhibitor (Roche; #04906837001). Nuclear and
   cytoplasmic protein was extracted from the snap‐frozen amnion tissue,
   or hAFs using a MINUTE Cytoplasmic & Nuclear Extraction Kit (Invent
   Biotechnologies; #NT‐032) or a Nuclear Extract Kit (Active Motif;
   #40010) according to the respective protocol provided by the
   manufacturer.

Western Blotting

   After determination of protein concentration using the Bradford method,
   the abundance of S100A9, p‐S00A9 (Thr 113), p53, p21, COX‐2, H3K9me3,
   Flag, CBX5, ORF1p, ORF2p, IRF3, p‐IRF3 (Ser 396), TBK1, p‐TBK1 (Ser
   172), STING, p‐STING (Ser 366) protein in each sample was determined
   with Western blotting. Briefly, 30 µg of protein from each sample was
   electrophoresed in a sodium dodecyl sulfate (SDS)‐polyacrylamide gel.
   After transferring to a nitrocellulose membrane blot (Millipore;
   #HATF00010), the blot was blocked with 5% non‐fat milk and incubated
   with primary antibodies overnight at 4 °C, followed by incubation with
   corresponding secondary antibodies conjugated with horseradish
   peroxidase (Proteintech). The peroxidase activity was developed with a
   chemiluminescence detection system (Millipore; WBKLS0500), which was
   visualized with the digital imaging equipment eBlot Touch Imager (eBlot
   Tec.). GAPDH, α‐tubulin, and Lamin A/C were used as internal control
   for total protein, cytoplasmic, and nuclear protein, respectively.
   Total H3 was used as an internal control for H3K9me3. The ratio of the
   phosphorylated protein to the corresponding total protein was used to
   indicate the relative abundance of phospho‐proteins. Information on
   antibodies used is given in Table [242]S5 (Supporting Information).

Immunohistochemical Staining

   After deparaffinization of the paraffin‐embedded sections, antigen
   retrieval, and quenching of the endogenous peroxidase activity with
   0.3% H[2]O[2], the sections were blocked with non‐immune serum,
   followed by incubation with the corresponding primary antibody
   overnight at 4 °C. Pre‐immune IgG served as negative control. The
   antibody information is given in Table [243]S5, Supporting Information.
   After washing, a biotinylated secondary antibody (Vector Laboratories;
   #BA‐1300) and the avidin‐biotin complex conjugated with horseradish
   peroxidase (Vector Laboratories; #PK‐7100) were applied to the sections
   for further incubation. Afterward, the peroxidase activity was
   developed as a red color using the substrate 3‐amino‐9‐ethyl carbazole
   (Vector Laboratories; #SK‐4200). The slide was counterstained with
   hematoxylin (blue color) and examined under bright field with a
   microscope (Zeiss; Axio Vert.A1). The percentage of p21‐positive cells
   was counted in five randomly selected visual fields with each field
   containing > 25 cells for each sample with ImageJ software (NIH).

Immunofluorescence Staining

   After deparaffinization of the paraffin‐embedded sections, antigen
   retrieval was performed. The sections were then permeabilized with 0.4%
   Triton X‐100. After blocking with normal goat serum, the corresponding
   primary antibodies were applied at 4 °C for incubation overnight,
   followed by incubation with Alexa Fluor 488‐, Alexa Fluor 594‐, or
   Alexa Fluor 647‐labeled secondary antibodies (Proteintech). Nuclei were
   counterstained with DAPI (blue color). For immunofluorescence staining
   in hAFs, the cells were fixed in 10% paraformaldehyde, and then primary
   antibodies against Vimentin, Cytokertin‐7, γ‐H2AX, Flag, p21, H3K9me3,
   S100A9 and p‐S100A9 (Thr 113) where applicable were applied. Nuclei and
   cytoskeleton were counterstained with DAPI (blue color) and phalloidin
   (red color) respectively. A confocal microscope system (TCS SP8; Leica)
   or a fluorescence microscope (Zeiss; Axio Vert.A1) was used to observe
   immunofluorescence signals. Quantification and statistical analysis of
   fluorescence signals were based on five random visual fields with each
   field containing > 25 cells for each sample with ImageJ software (NIH).
   The 3D reconstruction of H3K9me3 staining was performed by serial
   z‐stack sectioning for up to 80 images at 100‐nm intervals, and further
   processed to perform 3D reconstruction by using a Leica confocal system
   (TCS SP8). Exposure settings were unchanged throughout the acquisition
   of corresponding photomicrographs. The information of antibodies used
   in immunofluorescence staining is listed in Table [244]S5 (Supporting
   Information).

ELISA

   The supernatant collected from the homogenized amnion tissue or the
   conditioned culture medium of hAFs was used for measurement of S100A9
   or PGE2 where applicable with a S100A9 ELISA kit (R&D systems; #DY5578)
   or a PGE2 ELISA kit (Cayman Chemical; #514010) following the protocol
   from the manufacturer. The level of 2′3′‐cGAMP in the cell lysate of
   hAFs was measured using a 2′3′‐cGAMP ELISA Kit (Cayman Chemical;
   #501700) according to the manufacturer's instruction.

Transcriptomic Sequencing

   Total RNA was extracted from hAFs with or without nuclear S100A9
   overexpression (n = 3 each) using Trizol reagent (Thermo Fisher
   Scientific; #15596026). RNA purity and integrity were determined using
   a NanoDropND‐2000 (Thermo Fisher Scientific) and an Agilent 2200
   TapeStation. Transcriptomic sequencing libraries were constructed using
   a TruSeq RNA sample preparation kit (Illumina) according to the
   manufacturer's protocol, followed by sequencing of the libraries on an
   Illumina NovaSeq 6000 system. A computational pipeline was used to
   process the transcriptomic sequencing data. After removing the adaptor
   and low‐quality reads, the clean reads were aligned to the human genome
   (NCBI; GRCh38) using Hisat v.2.1.0 with default options.^[ [245]^38 ^]
   The fragment per kilobase per million mapped reads (FPKM) was obtained
   to indicate the abundance of transcripts. Differential expression was
   analyzed using the DESeq2 algorithm.^[ [246]^39 ^] P value < 0.05 and
   fold Change ≥1.5 or ≤0.67 were set as the threshold for significantly
   differential expression. GO and KEGG pathway enrichment analyses of
   differentially expressed genes were performed using R based on the
   hypergeometric distribution. GSEA (version 4.2.3) was used to
   statistically calculate the difference of a set of genes. Gene sets
   were obtained from the GSEA molecular signature database:
   [247]https://www.gsea‐msigdb.org/gsea/msigdb.

SA‐β‐Gal Staining Assay

   The SA‐β‐gal staining was conducted using a commercial staining kit
   (Beyotime; #C0602). Briefly, after washing with PBS, the cryosections
   or cultured hAFs were fixed with 4% paraformaldehyde for 15 minutes at
   room temperature, and subsequently incubated with a working solution
   containing 0.05 mg mL^−1
   5‐bromo‐4‐chloro‐3‐indolyl‐β‐d‐galactopyranoside at pH 6.0 overnight in
   darkness at 37 °C. For cryosections, nuclear fast red (Beyotime;
   #C0151) counterstaining was conducted for better visualization of
   cellular morphology. A phase contrast microscope system (Zeiss; Axio
   Observer 3) was used to capture the images. The percentage of SA‐β‐gal
   positive cells was counted in five random visual fields with each field
   containing > 25 cells in tissue staining and > 50 cells in cell
   staining for each sample with ImageJ software (NIH).

LC‐MS/MS Analysis

   Total protein was extracted from hAFs with or without nuclear S100A9
   overexpression using IP lysis buffer (Thermo Fisher Scientific; #87787)
   containing protease and phosphatase inhibitors. After determining the
   protein concentration, 500 µg of protein was incubated with Flag
   antibody‐coupled beads (Selleckchem; #[248]B26101) on a rotating
   platform overnight at 4 °C. The beads were then thoroughly washed, and
   the bound protein was eluted using glycine‐HCl (pH 2.5). Then the
   protein was reduced with 10 mM dithiothreitol and alkylated with 50 mM
   iodoacetamide, followed by acetone precipitation overnight. After
   centrifugation and washing with 90% acetone, the cleaned precipitates
   were dissolved with 50 mM ammonium bicarbonate and digested by trypsin
   at 37 °C overnight. The hydrolysates were desalted by C18 stage tips
   and dried using a vacuum drier. After re‐dissolving by 99.9% water with
   0.1% formic acid, the hydrolysates were applied to a sample vial for
   LC‐MS analysis, which was performed on an EASY‐ nLC 1000
   ultra‐high‐pressure system equipped with an Orbitrap Fusion Tribrid
   mass spectrometer (all from Thermo Fisher Scientific). After separation
   at 55 °C using a 25 cm × 75 µm C18 column (Aurora Ultimate;
   #AUR3‐25075C18‐CSI), the resulting peptides were subject to MS analysis
   in data‐independent acquisition (DIA) mode, which consisted of MS1 scan
   ranging from 390 to 980 m/z at 60000 resolution and the automatic gain
   control (AGC) target 1 × 10^6 or a maximum injection time of 100 ms.
   Then, 27 DIA rolling segments with isolation windows of 17, 23, 29, and
   35 m/z were acquired at the resolution of 30000 with an AGC target 5 ×
   10^5 or a maximum injection time of 40 ms. The setting “injections for
   all available parallelizable time” was enabled, and higher‐energy
   collision dissociation fragmentation was performed with a normalized
   collision energy of 24%. The spectra were recorded in profile mode,
   with the default charge state for MS2 set to two. Then, the raw data
   were searched against an in silico predicted spectral library generated
   from UniProt human protein database with a default setting.
   Match‐between‐runs were employed using the spectral library created
   from DIA data, with precursor filtering at 0.01 of false discovery
   rate. Neural network classifier = double‐pass mode, Quantification
   strategy = robust high precision, Cross‐run normalization =
   RT‐dependent.

Molecular Docking

   HDOCK SERVER ([249]http://hdock.phys.hust.edu.cn/) or ZDOCK SERVER
   ([250]http://zdock.wenglab.org) was used for the molecular docking
   study between S100A9 and CBX5. S100A9 (Uniprot ID: [251]P06702) was
   selected as the receptor protein and CBX5 (Uniprot ID: [252]P45973) was
   selected as the ligand‐protein. Pre‐treatment of the protein (deletion
   of water molecules and excess ligands, addition of hydrogen atoms) was
   completed using PyMol 2.4. The model with the lowest binding energy was
   selected as the best docking model, and PyMOL was used to visualize
   protein‐protein interactions.

Co‐IP Assay

   Co‐IP assay was performed as previously described.^[ [253]^40 ^]
   Cellular protein was extracted from hAFs with or without nuclear S100A9
   or T113D mutant S100A9 overexpression using IP lysis buffer (Thermo
   Fisher Scientific; #87787). After the determination of protein
   concentration, 500 µg of protein was incubated with Flag
   antibody‐coupled beads (Selleckchem; #[254]B26101) on a rotating
   platform overnight at 4 °C. In parallel, another 500 µg of protein was
   incubated with antibodies against Flag (Cell Signaling Technology;
   #14793), S100A9 (Abcam; #ab63818), CBX5 (Cell Signaling Technology;
   #2616), or MBD3 (Proteintech; #14258‐1‐AP), or pre‐immune rabbit IgG as
   a negative control, and rotated at 4 °C overnight, followed by
   incubation with protein A agarose beads for 2 h at 4 °C with rotation
   to pull down the antibody/antigen complex. The precipitated beads were
   washed adequately and then denatured in Western blotting loading buffer
   at 100 °C for subsequent analysis with Western blotting.

TEM Examination

   TEM examination was performed to examine the heterochromatin disruption
   in hAFs with or without nuclear S100A9 overexpression. After rapid
   fixation with 2.5% glutaraldehyde at 4 °C for 2 h, and post‐fixation
   with 1% osmium tetroxide solution for 2 h at 4 °C, the cells were
   processed for dehydration, infiltration, and routine embedding in
   resin. After solidification, ultrathin sections (70 nm) were cut and
   stained with lead citrate, which was then observed for heterochromatin
   at the nuclear periphery with a transmission electron microscope (FEI
   Talos L 120C; Thermo Fisher Scientific).

CUT&Tag

   A Hyperactive Universal CUT&Tag Assay kit (Vazyme; #TD904) was used to
   construct the CUT&Tag libraries of hAFs with or without nuclear
   S100A9‐overexpression according to the manufacturer's protocol. In
   brief, 0.1 million cells for each sample were collected, washed, and
   resuspended in 100 µL wash buffer, and then 10 µL of activated
   concanavalin A‐coated magnetic beads were added to each sample and
   incubated at room temperature for 10 min. The cells were then incubated
   with 2 µg H3K9me3 antibody (Epigentek; #A‐4036‐025) on a rotating
   platform overnight at 4 °C, followed by incubation with an anti‐rabbit
   IgG antibody at room temperature for 1 h. After washing with Dig‐Wash
   buffer, the samples were incubated with pA/G‐Tnp, and resuspended in
   tagmentation buffer for fragmentation. Afterward, 2 µL 10% SDS and 1 pg
   Spike‐in DNA were added to each sample for normalization and incubated
   at 55 °C for 10 min. The fragmented DNA in the supernatant was then
   extracted with DNA Extract Beads Pro, and DNA libraries were
   constructed using TruePrep Index Kit V2 for Illumina (Vazyme; #TD202).
   All CUT&Tag libraries were sequenced using the NovaSeq X Plus platform
   with a PE150 sequencing strategy.

   For the processing of H3K9me3 CUT&Tag data, raw reads were trimmed with
   Trim Galore software (version 0.4.5). Trimmed reads were mapped to the
   human genome (UCSC; hg38) using Bowtie2 (version 2.5.4).^[ [255]^41 ^]
   Duplicated reads were discarded by MarkDuplicates.jar program in Picard
   tools (version 1.119), and filtered quality reads were sorted with
   SAMtools (version 1.2.0).^[ [256]^42 ^] The sequencing reads were
   aligned to the spike‐in sequences using Bowtie2, and a normalization
   factor calculated as 10000 divided by the number of reads was applied
   to adjust all samples. MACS2 (version 1.1) was used to call peak with
   the parameters set to “‐f BAMPE –broad ‐q 0.05 –nolambda”. TE
   annotation file was downloaded from
   [257]http://hgdownload.cse.ucsc.edu/goldenPath/hg38/database/rmsk.txt.g
   z. For the classification of H3K9me3‐enriched TEs regions, the H3K9me3
   signals for each TE region were calculated. Then TEs regions were
   partitioned into “H3K9me3‐enriched TEs” and “other TEs” by robust
   k‐means analysis with two clusters using R package pheatmap (version
   1.0.12). Analysis of differential H3K9me3 signals enriched in TEs
   regions were performed by Manorm (version 1.3.0), and the Log2 ratios
   of H3K9me3 signals for each H3K9me3‐enriched TEs were visualized by R
   package RIdeogram (version 0.2.2).

ChIP Assay

   The enrichments of H3K9me3, CBX5, MBD3, and KAP1 at LINE1 regions were
   determined with ChIP assay as described previously.^[ [258]^40 ^]
   Briefly, after crosslinking with 1% formaldehyde, hAFs were lysed with
   SDS lysis buffer containing a protease inhibitor cocktail (Roche). The
   lysed cells were sonicated to shear the chromatin DNA to an optimal
   size of around 500 bp, which was then immunoprecipitated with
   antibodies against H3K9me3, CBX5, MBD3, or KAP1. An equal amount of
   pre‐immune IgG served as negative control. Sheared DNA without
   immunoprecipitation served as input control. The immunoprecipitate was
   pulled down with Magna ChIP Protein A+G agarose Magnetic Beads
   (Millipore; #16‐663) on a magnetic stand. After reverse cross‐linking
   with 5 M NaCl, RNA and protein were digested with RNase A and
   proteinase K, respectively. Afterward, the sheared DNA was extracted
   using a DNA purification kit (Cwbiotech; #CWY012) for subsequent
   qRT‐PCR. The antibody information and primer sequences for ChIP‐qRT‐PCR
   are illustrated in Tables [259]S5 and [260]S7 (Supporting Information),
   respectively. The data were analyzed and presented as follows: percent
   input = 100% × 2^[(Ct (input sample)‐Ct (IP sample).

LINE1 Retrotransposition Assay

   LINE1 retrotransposition assay in hAFs was conducted following a
   modified protocol published previously.^[ [261]^17 ^] In brief, hAFs
   with or without nuclear S100A9 overexpression were transfected with
   2 µg of the pEGFP‐LRE3 vector (a gift from Dr. Zhiyong Mao) using
   Lipofectamine 3000 (Thermo Fisher Scientific; L3000015). An empty
   vector served as the negative control. After transfection for 48 h the
   cells were incubated in DMEM containing 2 µg mL^−1 puromycin for
   another 72 h. Then the surviving cells were harvested for flow
   cytometry analysis (BD Biosciences). The percentage of GFP‐positive
   cells was determined with the FlowJo software (Ashland).

Extrachromosomal DNA Extraction

   Extrachromosomal DNA was extracted using the modified Hirt protocol.^[
   [262]^43 ^] After digestion from the culture plate, hAFs were
   resuspended in Buffer 1 (50 mM Tris‐HCl, 10 mM EDTA, containing
   100 mg/mL RNase A), and then mixed and lysed with Buffer 2 (1.2% SDS)
   by inverting at room temperature for 5 min to ensure complete lysis.
   Afterward, cellular debris and high‐molecular weight chromosomal DNA
   were precipitated with addition of Buffer 3 (3 M CsCl, 1 M potassium
   acetate, and 0.67 M acetic acid) on ice for 15 min. The
   extrachromosomal DNA‐containing supernatant was then collected by
   centrifugation, and extrachromosomal DNA was purified using a DNA
   purification kit (Cwbiotech; #CWY012) according to the protocol
   provided by the manufacturer. Total cellular DNA was extracted from
   hAFs directly using the same DNA purification kit. The amount of LINE1
   DNA content was determined with qRT‐PCR using the above extracted
   extrachromosomal DNA or total DNA as a template. Mitochondrial gene
   MT‐ND5 (NADH dehydrogenase 5) and 5s rDNA were amplified in parallel as
   internal control for extrachromosomal DNA and total DNA, respectively.

Animal Study

   The mouse study was carried out following ARRIVE guidelines and the
   relevant ethical regulations for animal use, which was approved by the
   Institutional Review Board of Ren Ji Hospital, Shanghai Jiao Tong
   University School of Medicine (approval code: RJ‐2024‐28A). C57BL/6
   mice (Charles River) of 10 to 13 weeks of age were mated overnight. The
   presence of a vaginal plug the next morning was counted as 0.5 dpc.

   To investigate gestational age‐dependent changes in nuclear S100A9 and
   p21 protein, transcripts of LINE1 and mRNA associated with the type I
   IFN response and SASP, mouse fetal membranes were collected at 17.5 and
   19 dpc (n = 5–6 per group). The abundance and distribution of nuclear
   S100A9 and p21 in the fetal membranes were determined with Western
   blotting and immunofluorescence staining using antibodies against mouse
   S100A9 and p21. The mRNA abundance of LINE1 and genes associated with
   the type I IFN response and SASP were determined with qRT‐PCR. The
   primers and antibodies information is illustrated in Tables [263]S6
   and [264]S5 (Supporting Information), respectively.

   To examine whether nuclear S100A9 was able to activate LINE1 resulting
   in the activation of the type I IFN response and cellular senescence in
   the fetal membranes thereby leading to preterm birth, intra‐amniotic
   injection of scAAV vectors expressing nuclear S100A9 (1 × 10^^10 TU/10
   µl/per gestational sac) in the presence or absence of Tenofovir (2
   nmol/10 µl/per gestational sac) was performed at 15.5 dpc under
   laparotomy after anesthesia with 65 mg kg^−1 Zoletil 50 (Virbac,
   #83887905). The same amount of empty scAAV vector served as a negative
   control. Some of the mice (n = 11 per group) were allowed to deliver
   spontaneously for observation of delivery time, and some (n = 6 per
   group) were sacrificed at 18 dpc (60 h after surgery) for collection of
   fetal membranes to examine the abundance of nuclear S100A9 and
   senescence phenotype. The abundance and distribution of nuclear S100A9
   in the fetal membranes were determined with Western blotting and
   immunofluorescence staining respectively. The abundance of p21 in the
   fetal membranes was examined with Western blotting and
   immunohistochemical staining, and SA‐β‐gal staining was performed to
   detect the senescent cells. The transcripts of LINE1, mRNA associated
   with the type I IFN response, and SASP in the fetal membranes were
   measured with qRT‐PCR. The primers and antibodies information are
   illustrated in Tables [265]S6 and [266]S5 (Supporting Information),
   respectively. The abundance of PGE2 in the fetal membranes was measured
   with a PGE2 ELISA kit (Cayman; #514010) after extraction with ethyl
   acetate.

Statistical Analysis

   Data are presented as mean ± SEM. All the experiments were performed at
   least three times independently. The n number in each experiment
   provided in the figure legends represents independent experiments using
   samples from individual pregnant women or mice. The normality of the
   data was examined by Shapiro‐Wilk test. Based on the normality test, a
   two‐tailed paired or unpaired Student's t‐test or Mann‐Whitney U‐test
   was used to compare two groups where appropriate. One‐way ANOVA
   followed by Tukey's post hoc tests was performed when assessing the
   normally distributed data among three or more groups. Spearman
   correlation analysis was performed to test the correlation between
   groups. Methods of statistical analysis employed to determine the
   significance of difference in each experiment are also presented in
   related figure legends. Statistical significance was defined as P
   < 0.05.

Ethics Approval Statement

   The human study was conducted under a protocol approved by the Ethics
   Committee of Ren Ji Hospital, Shanghai Jiao Tong University School of
   Medicine with written informed consent from the participating patients
   (approval code: KY2023‐007‐B). All ethical regulations relevant to
   participants of human research were followed. The mouse study was
   carried out following ARRIVE guidelines and the relevant ethical
   regulations for animal use, which was approved by the Institutional
   Review Board of Ren Ji Hospital, Shanghai Jiao Tong University School
   of Medicine (approval code: RJ‐2024‐28A).

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   W.S.W, F.Z., and K.S. conceived the project and designed experiments.
   W.S.W and K.S. supervised the study. F.Z., M.D.L, F.P., and W.J.L.
   performed experiments. Y.X. contributed to LC‐MS/MS data collection and
   analysis. F.Z. and W.S.W. contributed to data analysis and
   interpretation. W.J.L. and L.J.L. collected samples from patients and
   analyzed clinical information. W.S.W, F.Z., L.M., and K.S. wrote the
   manuscript.

Supporting information

   Supporting Information
   [267]ADVS-12-2414682-s002.docx^ (35.4MB, docx)

   Supplemental Table 1
   [268]ADVS-12-2414682-s001.xlsx^ (297.9KB, xlsx)

Acknowledgements