A new human cell model for studying normal aging and finding anti-aging treatments

The trophoblast development of CTBs to STBs in healthy placenta exhibits aging features, including an increase in senescence‐associated beta‐galactosidase (SA‐β‐gal) staining, high levels of cyclin kinase inhibitors and mTOR, telomeres attrition, and chronic inflammation (Cox and Redman 2017; Ciampa et al. 2023; Gal et al. 2019; Velicky et al. 2018). These data collectively position the placenta as a potential model for investigating human aging. The published aging phenotype of trophoblast was primarily identified at the relatively mature placenta (Cox and Redman 2017; Gal et al. 2019; Qi et al. 2022). Here, we systematically profiled cellular aging signatures in CTB‐STB development at single‐cell resolution, focusing on the early placental stage (Figure 1A). We re‐analyzed published single‐cell RNA sequencing (scRNAseq) data in various placental stages, encompassing peri‐implantation (refer to PI, 4139 cells from Zhou et al. (2019), accession GSE109555, Figure 1B,C), first‐trimester (refer to T1, 13,777 cells from Vento‐Tormo et al. (2018) in Figures 1D,E and S1A, accession E‐MTAB‐6701, 3854 cells from Shannon et al. (2022) in Figure S1B,C, accession GSE174481), and term‐stage placenta (averaged count for CTB of 3 stages and STB from Pavlicev et al. (2017) in Figure S1D, accession GSE87720). Trajectory analyses on reduced dimensions revealed the primary development paths from CTBs into STBs and EVTs (Figures 1B,D and S1A,B).

We first cross‐examined a curated high‐fidelity senescence gene set (refer to Saul et al. 2022, Table S1), which was generated by an in‐depth literature search and validated in human biopsies, along the trophoblast differentiation. The senescence scores progressively increased during STB development across all stages of pregnancy (Figures 1B,D and S1B,D). This was paralleled by a shift from the G2M/S phase in progenitor cells to G1/G0 arrest in STBs, indicating cell cycle arrest and a transition to a senescent state (Figures 1C,E and S1C,D). Cellular senescence markers, including CDK inhibitor expression and low expression of proliferation marker Ki67 (MKI67), could be observed in the STBs across the stages (Figures 1C,E and S1A,C).

We particularly deconvoluted the well‐documented aging hallmarks within the scRNAseq data of the early placental stages in detail (Data of T1 showed here, PI not shown). Using comprehensive gene signature scores, STBs exhibited lower DNA repair gene expression compared to CTBs (Figure S1A), including base excision repair BER, nucleotide excision repair NER, non‐homologous end joining repair NHEJ, Fanconi anemia DNA repair Fanc, and homologous recombination repair HR. Lamin B1 (LMNB1), a marker of nuclear integrity and genomic stability (Lopez‐Otin et al. 2023; de Magalhaes 2024; Gil 2023), was also reduced in STBs (Figure S1A). Regarding histone modifications, STBs exhibited a trend toward decreased levels of H3K27me3 and H3K9me3, accompanied by declined methyltransferase EZH2 (regulate H3K27me3) and SUV39H1 (regulate H3K9me3) and increased KDM6B (also known as JMJD3, H3 lysine demethylase) (Figure S1A), which plays a central role in inflammation and heterochromatin alterations during aging (Lopez‐Otin et al. 2023; Fogarty et al. 2015; Salminen et al. 2014; Djeghloul et al. 2016). In addition, previously underexplored aging features in early trophoblasts (Cheung et al. 1999; Schaetzlein et al. 2004), telomere attrition, could be corroborated by a decrease in the expression of telomeric repeat‐binding factors (TERF1, TERF2; Figure S1A). Other aging‐related changes, including increased SASP secretion and metabolism alterations, were also observed in STBs (Figure S1A). In this analysis, EVTs have been similarly identified with senescence features (marked by damaged DNA repair, SASP, and metabolic shifts) (Figures 1D and S1A, Table 1). The scRNAseq revealed more prominent aging signatures (e.g., downregulated MKI67, CDK1, SUV39H1, and TERF1/2; LMNB1 loss) in the CTB‐to‐STB pathway (Figures 1E and S1A, Table 1), underscoring replicative decline, telomeres dysfunction, and genomic instability uniquely, prioritizing it for subsequent investigations.

All tissue‐specific stem cells, such as muscle stem cells (MuSCs) (Kedlian et al. 2024) and hematopoietic stem cells (HSCs) (Lee et al. 2023), undergo aging. To assess the relevance of the CTB‐STB development to cellular aging beyond the placenta, we applied a linear regression model to simplistically perform global‐profile cross‐comparisons with MuSCs and HSCs. First, we trained the model on the whole transcriptome from 1902 MuSCs and 3568 (should be same as methods 4.23) HSCs from donors of various ages, spanning from young to old (Kedlian et al. 2024; Lee et al. 2023) (Figure S1E). Remarkably, the predicted donor ages of MuSCs and HSCs showed a consistent correlation with the CTB to STB development across the placental stages (PI and T1, Figure 1F). In contrast, MuSCs and HSCs themselves did not exhibit a strong correlation during aging (data not shown). Cellular aging features observed in other tissue contexts can, therefore, be effectively modeled by the CTB‐STB system. These results highlight the potential to extend insights from the CTB‐STB development to the aging processes of other systems, offering new perspectives on organismal aging in general.

Our re‐analysis of placental trophoblast scRNAseq data revealed key aging hallmarks, prompting us to establish an in vitro hTSC‐to‐STB development system that mirrors native trophoblast aging. The model will overcome the limitations of in vivo studies, such as scarce tissue access, ethical barriers, and static sampling.

Under the published protocol (Okae et al. 2018; Ruan et al. 2022), hTSCs (hTSC‐M1 and hTSC‐em), which were derived from human expanded potential stem cells (M1 hEPSC and hEPSC‐em) (Gao et al. 2019; Chen et al. 2023) and highly similar to other reported hTSC lines molecularly, matured into multinucleated STBs with villi in approximately 6 days (Figure 2A,B). We collected cells on day 0 (D0, hTSC) and differentiated days (STB D2, STB D4, STB D6) for molecular, transcriptomic, and proteomics analysis (Mass spectrometry analysis (MS) also included STB D8 samples) (accession GSE190432, PXD071811, Table S2). The development of hTSCs to STBs was accompanied by high levels of mature STB markers (Figures 2C,D and S2A) and elevated secretion of human chorionic gonadotropin (hCG) (Figure S2B). Conversely, genes encoding trophoblast transcription factors GATA2 and GATA3 exhibit minimal changes (Figure 2C). Further analysis revealed strong transcriptomic parallels between the hTSC‐STB in vitro and those in vivo in the T1 and PI stages (Figure S2C).

Similar to in vivo CTB‐STB data analysis, the in vitro development of hTSCs to STBs exhibited a progressively higher senescence score (Figures 2E and S2D), referring to Saul et al. (2022) (Table S1). Increased apoptosis resistance (characterized by low apoptosis level and activation of anti‐apoptotic proteins in the Bcl2 family) and high necroptosis are well‐established features of cellular senescence (Lopez‐Otin et al. 2023; Di Micco et al. 2021). At the transcriptomic level, the transition from hTSCs to STBs showed a clear upward trend in necroptosis and apoptosis resistance (Figure 2E). Adopting the same regression model for cross‐comparisons, RNAseq data revealed that the development of hTSCs to STBs positively correlated with the aging trajectory observed in HSCs and MuSCs (Kedlian et al. 2024; Lee et al. 2023) (Figure 2F), consistent with in vivo STBs (Figure 1F).

We next experimentally interrogated cellular senescence in in vitro STBs. hTSCs could self‐renew robustly, as indicated by the healthy cell proliferation rate (Figure S2E). However, their proliferation capacity rapidly decreased, compromising stemness maintenance upon STB induction (Figure S2E). Notably, a greater number of SA‐β‐gal‐positive cells were seen in STBs (Figures 2G and S2F). Transcriptomically, the G2M score and S phase score both decreased during hTSC‐STB development (Figure 2H), along with the reduced cell cycle and DNA replication scores in the proteomics (Figure S2G). Permanent cell cycle arrest in STBs was further confirmed by the absence of Ki67 (Figure S2H,I) and proteomics‐validated reductions in proliferating cell nuclear antigen (PCNA) and minichromosome maintenance (MCM) (Jurikova et al. 2016) (Figure S2J).

Senescence is typically mediated by the tumor suppressors TP53 and CDKN2A/p16^INK4a, and their downstream effectors CDKN1A/p21 and RB‐E2F, respectively (Gorgoulis et al. 2019). Along the hTSC‐STB development axis, the expression of cyclin‐dependent kinases (CDKs) gradually decreased, while CDK inhibitors (CDKN1A/p21, CDKN2A/p16^INK4a, CDKN2B/p15, CDKN1C/p57kip2) exhibited an opposing trend (Figure 2I–K). The higher levels of CDK inhibitors and other negative regulators of the cell cycle, such as p38MAPK (Anerillas et al. 2023) and p27 (Bastians et al. 1998), were further supported by western blotting and immunostaining (Figures 2J,K and S2K,L). Altogether, the accumulated senescence score and evidence of cell cycle arrest in STBs indicate the potential of the hTSC‐STB system to serve as an in vitro cellular aging model.

To delineate detailed senescence characteristics during hTSC‐STB development and evaluate its application potential, we investigated its aging‐related signals in multiple aspects. First, senescent cells involve impaired DNA repair and unstable genomes (Schumacher et al. 2021). In the hTSC‐STB development, a general decline in the scores of DNA repair pathways, including NER, BER, HR, NHEJ, and Fanc, at both transcriptomic (Figure 3A) and proteomic levels (Figure S3A), was evident, in line with the in vivo data (Figure S1A). Consequently, STB cells exhibited more DNA damage, as evidenced by elevated γH2AX and p53‐binding protein 1 (53BP1, marker of DNA double‐strand breaks (DSBs)) foci (Figure 3B,C). Furthermore, STB cells showed higher levels of phospho‐ATM (pATM) and reduced levels of key HR DNA repair regulator breast cancer gene 1 (BRCA1) (Figure S3B–E). To further assess DNA repair capacity, we treated hTSCs and hSTBs with etoposide (Eto) to induce DNA damage, followed by a 24 h recovery. Results showed hTSCs repaired DNA damage (low 53BP1 post‐recovery), whereas hSTBs failed to recover efficiently (Figure 3D,E). We also stained BRCA1 in this experiment, BRCA1 levels remained consistently higher in hTSCs than STBs, but were unaffected by Eto treatment or recovery in either cell type (Figure 3E, quantification not shown here). We concluded that the hTSC‐STB system is accompanied by defective DNA repair capacity and accumulated DNA damage.

The loss of nuclear scaffold proteins lamin A and lamin B has been identified as molecular markers of aging in progeria syndrome, skin decrepitude, and hippocampal degeneration (Gil 2023; Wang et al. 2017; Bin Imtiaz et al. 2021). The in vitro STBs exhibited low expression of lamin A/C (LMNA, LMNC) and lamin B1 (LMNB1) (Figure 3F–H), indicating an unstable nuclear skeleton and increased susceptibility of the genome to stress. Lamins play an important role in chromatin remodeling by interacting with the transcriptional repressor HP1 (Fischle et al. 2005). Altered HP1 distribution on chromosomes occurs in aged Drosophila and premature human aged cells (Wood et al. 2010; Zhang et al. 2015), whereas HP1 overexpression extends lifespan (Larson et al. 2012). The hTSCs to STBs system exhibited a clear decline of HP1α (CBX5) and HP1γ (CBX3) (Figure 3I–K). HP1 functions as part of the KRAB‐KAP1‐SETDB1 complex in the maintenance of constitutive heterochromatin (Schultz et al. 2002; Kalousi et al. 2015). Expression of the components of this repression complex was also decreased in STBs versus hTSCs (Figure 3L,M).

Emerging research emphasizes the role of epigenetic shifts, including epigenetic clocks (DNA methylation patterns that predict biological age) (Horvath 2013; Wu et al. 2024; Pal and Tyler 2016). In the development of hTSCs to STBs, we noticed a gradual decline in the repressive histone marks (H3K9me3, H3K27me3) and a rise in the active mark H3K4me3 (Figures 3N,O and S3F, accession GSE243059), coinciding with decreased lamins and heterochromatin loss seen in aged or progeria cells (Zhang et al. 2015; Wu et al. 2024; Pal and Tyler 2016; Lee, Salamah, et al. 2024). Transcriptomic analysis of histone‐modifying enzymes (writers/erasers) (Hyun et al. 2017) revealed coordinated changes aligning with these shifts in the hTSC‐STB system (Figure S3G–I). Specifically, H3K9me3 writers decreased globally (no eraser change; Figure S3G). H3K27me3 writers declined, while erasers rose significantly (Figure S3I). H3K4me3 writers and erasers showed the inverse trends (Figure S3H), consistent with the observed histone mark dynamics. Canonical histone levels, which are reported to decline with aging across species (Yi and Kim 2020; Hu et al. 2014), were significantly lower in STBs than hTSCs (mass spectrometry PXD071811; Figure S3J).

Global DNA methylation typically decreases with age in organs and tissues (Horvath 2013; Horvath and Raj 2018; Seale et al. 2022). So we checked regulators of DNA methylation, including methyltransferases and demethylases in the trophoblast system. A reduction in the expression of DNA methyltransferases DNMT1 and DNMT3B was observed in STBs (Figure S3K,L), consistent with the reported age‐related changes in MSCs and skin fibroblasts (Xie et al. 2017; Lopatina et al. 2002; Zhou et al. 2020). The expression of another member, DNMT3A, moderately increased in STBs (Figure S3K), mirroring high DNMT3A in aged human blood cells (Bond et al. 2019). For DNA demethylases, the downregulation of TET1/2/3 (10–11 translocation family) in STB cells (Figure S3M) mirrors patterns seen in premature reproductive aging (Huang et al. 2020; Unnikrishnan et al. 2018). Additionally, the thymine DNA glycosylase (TDG), which removes TET‐mediated DNA oxidation products, was also reduced (Figure S3M). These observations suggest a shared epigenetic mechanism in the hTSC‐STB system and other tissue or cell aging.

mRNA surveillance is also implicated in senescence (Kwon et al. 2023; Harries 2023). A marked mRNA surveillance pathway suppression was seen in the hTSC‐STB system (Figure S3N). N6‐methyladenosine (m6A), the predominant mammalian mRNA modification, is regulated by a core catalytic complex (including methyltransferases like 3 and 14: METTL3, METTL14) linked to aging (Liu et al. 2021). In the trophoblast model, METTL3, METTL14, and m6A complex components (RBM15, WTAP, ZC3H13) were downregulated (Figure S3O,P). Pharmacological METTL3 inhibition (STM2457 treatment) (Yankova et al. 2021) in hTSCs induced STB features with upregulated markers (CGA, CGB3, ERVW1) and concomitant suppression of proliferation‐associated genes (MKI67, PCNA) (Figure S3Q). These findings indicate that m6A methylation safeguards hTSC stemness, restraining premature differentiation and senescence.

The human genome harbors abundant transposable elements (TEs), enriched in constitutive heterochromatin and epigenetically silenced in healthy cells (Slotkin and Martienssen 2007). Transcriptional silencing collapses (e.g., lamin depletion, heterochromatin erosion) enable TEs derepression, triggering interferon‐driven inflammation, which drives cellular senescence and tissue aging (Autio et al. 2020; Liu et al. 2023).

During STB differentiation alongside epigenetic shifts, the expression of TEs was significantly elevated (Figures 4A and S4A). Retrotransposons are linked to cellular senescence and are detectable in aged human serum (Autio et al. 2020; Liu et al. 2023). STBs expressed considerably higher levels of key retrotransposon HERVs (Figure 4B). This coincided with the heightened cGAS‐STING pathway score (Figure S4B), increased expression of its components pTBK and pIRF3 (Figure 4C), and the downstream interferon‐related molecules (Figure 4D).

HERVK (HML‐2) (Figure S4C), the youngest integrations in humans, are highly expressed in aged cells (Autio et al. 2020; Liu et al. 2023). STB cells (notably STB D2) showed robust RNA‐level expression of HERVK elements (LTR5‐Hs, HERVK‐int; Figures 4E and S4D) and elevated HERVK‐gag protein levels (Figure 4F). Given the reported role of HERVK in aging (Liu et al. 2023), we targeted LTR5‐Hs in hTSCs using an inducible dCas9‐KRAB system (Zhang et al. 2022; Fuentes et al. 2018) (LTR5‐Hs KD), which suppressed the HERVK transcripts (gag, pol, env) and protein (HERVK‐gag) (Figures 4G,H and S4D). LTR5‐Hs KD impaired STB differentiation: STBs displayed fewer villi, reduced fusion index (Figure S4E,F), and lower expression of maturation markers (CGA, CGB, ERVW1) (Figure S4G,H).

We subsequently measured the total senescence score as in Figure 2E. The LTR5‐Hs KD STBs displayed a moderately reduced senescence score (Figure 4I), with suppressed cGAS‐STING activation (Figure 4J) and downregulated interferon signaling (Figure S4I). SA‐β‐gal staining confirmed the less senescent LTR5‐Hs KD STBs (Figure 4K,L). The senescence‐associated genes in the LTR5‐Hs KD STBs exhibited three expression patterns: downregulation (e.g., mTOR, CDKN2B, Bcl2), no change (e.g., TERT, CDKN1A), or upregulation (e.g., PI3K) (Figure S4J). However, γH2AX levels remained unchanged (Figure S4K), indicating incomplete senescence suppression with HERVK knockdown. This aligns with other studies showing HERVK repression mitigates senescence or aging to a limited extent (Autio et al. 2020; Liu et al. 2023; Hurme and Pawelec 2021), underscoring its role as a partial driver in the hTSC‐STB senescence process.

Young, healthy cells maintain dynamic networks of organelles (e.g., mitochondria, telomeres) that degrade with aging (Sahin and Depinho 2010; Heiby and Ori 2022). In particular, telomere attrition drives genome instability, impairs cellular health, and directly contributes to aging and age‐linked diseases (Lopez‐Otin et al. 2023; Rossiello et al. 2022). However, telomere length did not differ between hTSCs and STBs (Figure S5A), likely because differentiation time and cell divisions in the system were limited (Figure S2E), preventing measurable telomere shortening. We thus analyzed TERT (telomerase) and shelterin complex components (TRF1, TRF2, POT1, TIN2, TPP1, RAP1), all of which showed gradual downregulation in STBs compared to hTSCs (Figure 5A–D). Telomere damage in STBs was further supported by γH2AX‐TRF1/2 colocalization (Figure S5B,C). Additionally, proteomics analysis revealed reduced expression of telomere regulators (TELO2, TE2IP, TCAB1, etc.) in STBs (Figure 5D). These results demonstrate a progressive failure of telomere maintenance mechanisms during hTSC development into STBs.

Senescent cells exhibit abnormal mitochondria biogenesis and increased mass (Gorgoulis et al. 2019; Palikaras et al. 2015; Lopez‐Lluch et al. 2008). STBs showed an elevated mitochondrial mass (MitoTracker staining, Figure S5D) and structural defects, including fragmented cristae and hollow centers, via electron microscopy (Figure 5E), mirroring mitochondrial ultrastructure in other aging models (Brandt et al. 2017). Besides morphology changes, STBs also displayed impaired respiratory function, with reduced oxidative phosphorylation (OXPHOS) and glycolysis activity (Figure 5F) and decreased ATP production from both pathways (Figure 5G), highlighting the mitochondria dysfunction from hTSCs to STBs. Senescent cells are known to accumulate reactive oxygen species (ROS) (Gorgoulis et al. 2019; Sahin and Depinho 2010; Sreedhar et al. 2020). While cytoplasmic ROS levels were similar between hTSCs and STBs, nuclear ROS intensity was markedly higher in STBs (Figure S5E). This nuclear‐specific increase likely reflects elevated DNA damage (Figure 3B,C) and diminished antioxidant capacity in the nucleus, as reported (Yakes and Van Houten 1997; Paardekooper et al. 2019; Murphy et al. 2022).

Senescent cells secrete inflammatory factors (SASPs), including cytokines, growth factors, and proteases. While STBs naturally produce pregnancy‐related factors, they showed higher SASP and inflammation scores than hTSCs (Figure 5H). RNAseq showed that most SASP components (e.g., GDF‐15) (Moon et al. 2020; Pence 2022) were upregulated in STBs (Figures 5I and S5F,G), but the classic IL‐6 (Childs et al. 2015; Tyrrell and Goldstein 2021) remained unchanged (Figure 5I). Surprisingly, several other assays (RT‐qPCR, immunostaining, ELISA) revealed elevated IL‐6 in STBs (Figures 5J and S5F,H), suggesting potential mRNA‐protein discordance in aging cells (Hipp et al. 2019). We further confirmed the enrichment of SASP‐linked pathways (mTOR, TNF, JAK–STAT, etc.) in proteomics of the hTSC‐STB system (Figure 5K). In summary, STBs display organelle dysfunction and robust SASP activation.

Most anti‐aging agents, like caloric restriction mimetics (CRMs), target the IGF/IGF1R, mTOR, and AMPK pathways (Lopez‐Otin et al. 2023; Madeo et al. 2019) (Figure S6A). In hTSC‐STB differentiation, transcriptomics revealed increased insulin resistance but reduced AMPK, IGF1, and insulin signaling (Figure S6B). Proteomics validated higher insulin resistance in STBs alongside elevated insulin secretion, AMPK, mTOR, and MAPK pathway activities (Figure S6C). Sirtuins are core nutrient‐sensing proteins that regulate aging processes and interventions (e.g., caloric restriction) and are typically reduced in aged cells (Lopez‐Otin et al. 2023; Madeo et al. 2019). While their members showed stable mRNA levels during hTSC‐STB differentiation (Figure S6D), the protein levels of SIRT1, SIRT3, and SIRT6 dropped significantly in STBs (Figure S6E–G). In contrast, the levels of SIRT2 and SIRT5, which are implicated in age‐related diseases and models (Madeo et al. 2019; Diaz‐Perdigon et al. 2020; Maxwell et al. 2011; Mao et al. 2023; Greene et al. 2019), increased in STBs (Figure S6F). The differences here in mRNA and proteomics again indicate the decoupling of mRNA‐protein in senescent cells. These shifts in nutrient‐sensing pathways align with other aging hallmarks observed in both the in vitro hTSC‐STB and the in vivo CTB‐STB systems (PI and T1) (Table S1).

The above findings demonstrate that the hTSC‐STB system recapitulates core aging hallmarks, positioning it as a bona fide human cellular model for aging studies with potential broad applicability. To test its utility in anti‐aging drug discovery, we evaluated three agents: mTOR inhibitors (rapamycin, INK128) and senolytic fisetin (Lopez‐Otin et al. 2023; Mannick and Lamming 2023; Yousefzadeh et al. 2018; Kaeberlein et al. 2023) (Table 2). All three compounds sharply reduced hCG levels in STBs (Figure 6A). Decreased hCG is an indicator of defective STB development. hCG produced by trophoblasts sustains pregnancy via several mechanisms, including induction of progesterone production, trophoblast fusion, vascular remodeling, and immune tolerance. Its dysregulation impacts fetal development and disease risk. Further experiments indicated that the treatments also lowered γH2AX level (Figure 6B) and modestly boosted Ki67, HP1γ, and laminB1 expressions in STBs (Figure 6B), confirming their cellular‐level anti‐aging effects.

To leverage the system for anti‐aging molecule screening, we proposed hCG dynamics track established aging hallmarks in this unique hTSC‐STB system to some extent. CGA combines with CGB to form intact hCG. Without CGA, functional hCG cannot be secreted, impairing placental signaling. CGA may independently regulate hTSC proliferation/differentiation (via LH/CG receptor interactions triggering cyclic AMP (cAMP)‐dependent signaling cascades) and immune tolerance at the maternal‐fetal interface (Riccetti et al. 2017). We established the component CGA expression as a marker for rapid assessment of aging activity in the hTSC‐STB system. We engineered a reporter hTSC line by inserting an EGFP cassette into the CGA locus (Figure 6C). The marked difference in CGA levels between hTSCs and STBs enabled sensitive, quantitative detection of anti‐aging effects. Tagged CGA allows real‐time tracking of differentiation efficiency and drug effects on syncytialization. We observed a progressive increase in EGFP intensity during hTSC‐to‐STB with no difference at D4 and D6 (Figure 6D–F). Treatment with known anti‐aging molecules reduced EGFP signals in the hTSC‐to‐STB transition (Figure 6G,H, Table 2), indicating detection of potential anti‐aging activities. We subsequently screened a natural product library (MedChemExpress HY‐L021; 2869 compounds) with rapamycin and INK128 as the anti‐aging positive controls (Figures 6I,J and S6H,I). Reporter hTSCs were differentiated into STBs (Day 6) in the presence of the compounds, and EGFP signals (normalized to DAPI) were quantified. Most library compounds did not have a detectable effect on STB development (strong EGFP signals), while a subset significantly reduced EGFP intensity, including known anti‐aging agents and novel candidates (Figures 6I,J and S6H,I). Interestingly, the reporter system also identified putative pro‐aging candidates (Figure 6I,J).

Among the molecules identified in the screen, anisomycin is known to alleviate neuroinflammation and Aβ/tau pathology in Alzheimer's models (Jiao et al. 2024), another candidate, nitidine, which scavenges ROS, alleviates inflammation/senescence and inhibits PI3K/Akt/mTOR signaling to induce apoptosis (Lin et al. 2022; Zhang et al. 2024). The two selected candidates were assessed for reducing β‐gal activity in STBs (Figure 6K,L). Similar experiments were performed on the HFF‐1 skin fibroblasts, where Eto (VP‐16) was used to induce cell cycle arrest. INK128, anisomycin, and nitidine all attenuated β‐gal signals in the induced aged fibroblasts (Figure 6M,N). This proof‐of‐principle screen demonstrates the hTSC‐STB system's utility for the high‐throughput discovery of novel compounds for aging interventions.

While the hTSC‐STB system showcases aging hallmarks and offers a novel platform, several key limitations should be acknowledged: Placenta‐Specific Features. The placenta is a transient organ with a lifespan of approximately 10 months. Although STBs display core aging characteristics, their biology may not fully mirror the aging processes occurring in tissues or organs that persist for decades, which may limit direct extrapolation. Scope of Reporter. The current anti‐aging screening relies on reductions in hCG and CGA level within the STB culture system, which indicate disrupted maturation. The CGA‐GFP reporter is sensitive for primary screening and captures STB maturation, but does not encompass the full spectrum of aging hallmarks. Consequently, candidates identified through this reporter line could also include agents for hydatid mole/choriocarcinoma treatment and compounds that may pose risks during pregnancy. Incorporating additional reporters, such as those targeting sirtuins, cell cycle regulators, and lamins, could enable more comprehensive tracking of specific aging pathways and broaden the system's applicability. Validation Needs. As a novel tool, anti‐aging candidates identified in this system require validation in complementary models such as organoids or humanized animals to confirm relevance beyond placental contexts. This step is essential to ensure translational potential and to establish that findings are applicable to other tissues and aging processes in humans.

Samples of hTSCs and STBs were fixed with 1 mL of 1X Fixative Solution (Cell Signaling Technology (CST), #9860) to each 35 mm well, for 10–15 min at room temperature (RT). All the subsequent steps precisely referred to the instructions of the vendor (CST, #9860). Samples were checked under Nikon Eclipse Ti2 Inverted Microscope Systems, and images were further processed and analyzed with ImageJ 1.47v.

Fresh hTSC and STB medium were added to hTSCs and STBs at around 60%–70% confluency. The cultured medium of hTSCs and STBs was collected after 4 h. After centrifuging, the cultured medium, together with sterile non‐cultured hTSC and STB medium, was used to perform ELISA tests, referring to the instructions of the Human IL‐6 ELISA Kit (ABclonal, RK00004) and the Human hCG ELISA Kit (Abcam, ab100533). Data were read with a plate reader and processed under Prism 9 (GraphPad).

Based on the template plasmid pCD31‐5HA‐T2A‐H2B‐EGFP‐loxp‐EF1a‐puro‐loxp‐3HA and pSpCas9(BB)‐2A‐Puro (PX459) V2 (Addgene: #62988) in our lab, we picked the gene CGA for reporter cell line system generation due to the huge difference in CGA expression between hTSCs and STBs. First, we got the full sequences of the human CGA gene from the website Ensembl genome browser 110. We designed the gRNAs for the gene right after the stop codon (less than 100 bp away from the stop codon) and inserted the gRNAs into the pSpCas9(BB)‐2A‐Puro (PX459) V2 construct by digestion and T4 ligation method. Then we got around 1000 bp homologous arms, both 5′HA and 3′HA parts, right in front of the stop codon and right after the gRNA‐targeted site, respectively. With the homologous recombination method in the ClonExpress II One Step Cloning Kit, we inserted both 5′HA and 3′HA parts of each gene into the pCD31‐5HA‐T2A‐H2B‐EGFP‐loxp‐EF1a‐puro‐loxp‐3HA to get the CGA‐T2A‐H2B‐EGFP reporter constructs. After that, we transfected the gene reporter plasmid together with the paired gRNA plasmid into the hEPSC lines. With puromycin selection for 7 days, we picked the hEPSC colonies and confirmed through genotyping for the reporter insertion in the genome. Then we expanded the right hEPSC colonies and differentiated to CGA‐T2A‐H2B‐EGFP hTSC reporter cell line for further study.

For the experiment, at least 1 million cells per sample of hTSC and STB (on Days 2, 4, 6, and 8) were collected. Sample processing and liquid chromatography‐mass spectrometry (LC–MS/MS) analysis were performed at LKS Faculty of Medicine, Proteomics and Metabolomics Core Facility (PMcore), Centre for PanorOmic Sciences (CPOS), the University of Hong Kong. Raw mass spectrometry data were processed using Spectronaut. Direct DIA data analysis was performed on Spectronaut v.14 using default settings without N‐acetyl variable modification enabled. Trypsin specificity was set to two missed cleavages and a protein and PSM false discovery rate of 1%, respectively. Data filtering was set to Q‐value, and normalization was set to global normalization. Results were then exported to an Excel file (Microsoft) for further processing. For proteome quantification, the Excel file was processed in R Studio v4.2.2. Excel files with normalized MS data were read in. Average log2 ratios comparing to the control (baseline) were normalized and standardized protein‐wise to a range from −1 to 1 using Z‐score and rescale function built in the "scale" R Package across samples. Proteins of MS with any missing data were excluded from further visualization. Protein expression levels along time were illustrated using the "heatmap.2" function from the "gplots" R Package, with the baseline treated as 0.

The raw data were trimmed by Trim galore (v0.6.10) (Krueger and Galore 2015) with default parameters. Trimmed reads were mapped to the human reference genome (GRCh38) using Bowtie 2 (v2.5.2) (Langmead and Salzberg 2012) with default parameters. SAMtools (v1.18) (Li et al. 2009) was used to retain the paired and unique mapped reads and to remove PCR duplicated reads. bamCoverage provided by deepTools (v3.5.2) (Ramirez et al. 2016) was employed with parameters "‐bs 200‐normalized using RPKM" to produce the BW file. MACS3 (v 3.0.0b3) (Zhang et al. 2008) was used for peak calling with parameters "‐f BAMPE ‐g hs—nomodel ‐q 0.05", with an additional parameter "—broad" for H3K9me3. For the profile plot, peaks of the same histone modification from all samples were merged using BEDTools (v2.31.1) (Quinlan and Hall 2010). The peak summits were further refined by MAC3. plotProfile provided by deepTools was then used to produce a profile plot.

Electroporation of M1 hEPSC and hEPSC‐em were performed with the Invitrogen NeonTransfection System. hEPSCs were dissociated into single cells with TrypLE or 0.05% Trypsin–EDTA. M10 medium (DMEM with 10% FBS) was added to neutralize the trypsin. After centrifugation, 20–40 μL R buffer in the Neon kit was used to resuspend hEPSC cells and mixed with plasmids. The cells were then seeded with a recovery medium (EPSC medium + 10% FBS + 10 μM Y27632). After incubation overnight, cells were switched to a typical hEPSC medium. 48 h later, 1 mg/mL Puromycin was used for selection. After 5–7 days, single colonies were picked, genotyped, and expanded for further experiments.

The proteins were extracted with RIPA lysis (Thermo Fisher Scientific, 89901) and Pierce Protease Inhibitor Mini Tablets (Thermo Fisher Scientific, A32953). Protein concentrations were measured with Pierce BCA Protein Assay Kits (Thermo Fisher Scientific, 23227). Then samples with a certain amount of NuPAGE LDS sample buffer (4×) (Thermo Fisher Scientific, NP0008) were boiled at 95°C for 10 min. Boiled samples were loaded with the same amount and separated in 12% polyacrylamide gels (Thermo Fisher Scientific, NP0343BOX) and transferred to PVDF membranes using the Bio‐Rad transblot turbo system according to the manufacturer's guidance. Images were developed by the ChemiDoc Imaging System and processed with ImageJ 1.47v. Primary antibodies were used at a dilution ratio of 1:1000, while secondary antibodies were incubated at a dilution ratio of 1:2000.

Samples were seeded in a μ‐slide 8‐well dish (ibidi, 80826) pre‐coated with Matrigel/Geltrex at a density of 1–2 × 10⁴ cells per well and fixed at room temperature with 4% paraformaldehyde (PFA) for 15–30 min. Afterwards, 0.05% Triton X‐100 in PBS was utilized to permeabilize for 10–15 min, and 2% FBS in PBS was used for blocking for 1–2 h. Followed by incubation with primary antibodies in a 4°C fridge overnight. Then wash 3–5 times, 5–10 min per time on the shaker. Secondary antibodies were incubated at room temperature for 1–2 h. Another 3–5 rounds of washing in PBS were done and stained with DAPI or Hoechst for 10–15 min to mark nuclei. At last, samples were imaged under a Nikon Eclipse Ti2 NIS elements AR fluorescence microscope.

hTSCs and STBs were seeded in a μ‐Slide 8‐well dish pre‐coated with Matrigel/Geltrex. 50 nM MitoTracker Green^FM (Invitrogen, M7514) and 5 μM DCFH‐DA (MedChem Express, HY‐D0940) were added to the dish, respectively. Wait for 30 min and wash the cells with sterile PBS. Then fix the cells with 4% PFA for 10–15 min at RT and stain with DAPI after permeabilization for 10–15 min with 0.05% Triton X‐100 in PBS. The slides were imaged and processed similarly to immunofluorescence staining samples.

The DNA damage response test was performed as described previously (Fu et al. 2021). In brief, TSCs and day 4 STBs were exposed to 2.5 μM etoposide for 2 h to induce DNA damage. Afterwards, cells were released from the treatment after washing with PBS twice. Untreated cells (0 h), cells under exposure to etoposide for 2 h (2 h), and cells after 24 h releasing from etoposide (24 h) were collected for immunofluorescence microscopy of 53BP1 (#4937, Cell Signaling Technology, 1:300) and BRCA1 (sc‐6954, Santa Cruz, 1:300). More than 50 cells were counted for quantification of each group.

The total RNA of each sample was extracted with Trizol (Life Technology, 391,501). The concentration of the RNA samples was measured with Nanodrop, and 1 μg of each was reverse transcribed into cDNA with PrimeScript RT Master Mix (Takara: RR036A) on a thermal cycler. The PowerUpSYBR Green Master Mix was used for amplifying genes. Primer sequences are listed in Table 3. All qPCR experiments were performed on The StepOnePlusTM Real‐Time PCR System (Applied Biosystems). Gene expression levels were normalized to GAPDH and were analyzed in Prism 9 (GraphPad).

Genome DNA was prepared using the DNeasy Blood & Tissue Kit (QIAGEN, Valencia, CA). Average telomere length was measured from total genomic DNA using a real‐time PCR assay. PCR reactions were performed on the Bio‐Rad CFX Opus 96 System, using telomeric primers, primers for the reference control gene (human 36B4 single‐copy gene), and PCR settings as described (Hastings et al. 2024). For each PCR reaction, a standard curve was made by serial dilutions of known amounts of DNA. The telomere signal was normalized to the signal from the single‐copy gene to generate a T/S ratio indicative of relative telomere length. An equal amount of DNA (20 ng) was used for each reaction. 36B4 primer: 5′‐CAGCAAGTGGGAAGGTGTAATCC‐3′ and 5′‐CCCATTCTATCATCAACGGGTACAA‐3′; Tel primer: 5′‐GGTTTTTGAGGGTGAGGGTGAGGGTGAGGGTGAGGGT‐3′ and 5′‐TCCCGACTATCCCTATCCCTATCCCTATCCCTATCCCTA‐3′.

Human TSCs were seeded in XFe96 culture microplates with a cell density of 2 × 10⁴ two days before the metabolic measuring. Metabolic profiling of STBs was conducted by seeding 2 × 10⁴ human TSCs in XF96 culture microplates and allowed for differentiation for 4 days. On the day of the metabolic profiling, the cell culture medium was replaced by DMEM XFe Medium, pH 7.4 supplemented with 10 mM of glucose, 1 mM of pyruvate, 2 mM of glutamine, and cells were incubated for 45 min at 37°C in a non‐CO₂ atmosphere. Then the medium was replaced with pre‐warmed DMEM XFe Medium, pH 7.4 (supplemented), and cells were subjected to metabolic measuring. The Real‐Time ATP Rate Assay Kit was used to measure the ATP production rate of human TSCs and STBs, with the following reagent concentrations: 1.5 μM for oligomycin and 0.5 μM for rotenone/antimycin A. Reagents were loaded into the sensor cartridge as indicated by the manufacturer during the metabolic measuring. The equipment was calibrated following the manufacturer's instructions. After completing the calibration, the OCR and ECAR were measured before and after the sequential injection of the reagents of the assay kit. ATP production was calculated using the Agilent Seahorse Wave Controller 2.6 software according to the manufacturer's instructions. All the data obtained was normalized to the cell number. The metabolic profiling was performed using the Agilent Seahorse XFe96 Analyzer at the Centre for PanorOmic Sciences (CPOS) at the University of Hong Kong.

hTSCs were seeded on 1% Matrigel/Geltrex‐coated 96‐well Flat Clear Bottom Black Polystyrene TC‐treated Microplates for STB differentiation. For library screening, the library plates (MedChemExpress, HY‐L021, total 2869 compounds) were diluted to working concentration first. DMSO and molecules (library plates or well‐known anti‐aging molecules) were used to treat hTSCs to STBs according to different working concentrations. Waiting for 4–6 days for the STB maturation, plates were fixed with 4% PFA and stained with DAPI/Hoechst. Then Large images were generated and analyzed with the Nikon Spinning Disk Confocal Imaging System. Image processing and quantifications were performed using ImageJ 1.47v.

Around 300,000 cells per mL of trypsinized hTSC and STB cells were fixed with EM fixative (4% PFA or 2.5% GTA) at 4°C overnight. Then put in sharp bottom micro‐centrifuge tubes (1.7 mL) and brought in an ice bath to the Electronic Microscopy laboratory (Queen Mary Hospital) for further sample preparation. Afterwards, samples were checked with the Philips CM100 TEM system, and different magnification images were taken for further analysis.

All produced datasets had performed at least three independent repeats. Datasets were imported into Prism 9 (GraphPad). Statistical graphs with the mean and SD/SEM were plotted via unpaired two‐tailed Student's t‐test, two‐way ANOVA, and one‐way ANOVA with Dunnett's test. No significance (ns), p > 0.05; *p ≤ 0.05, **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001.

scRNA‐seq dataset of peri‐implantation trophoblast from Zhou et al. is publicly available (accession GSE109555). The annotation was obtained from Castel, et al. (https://gitlab.univ‐nantes.fr/E137833T/Castel_et_al_2020/‐/tree/master↗). Among the 5911, 8–14 embryonic day cells, late TE stage to STB differentiation was subsetted, and AME‐like cells were identified using canonical markers and removed, ending up with 3238 cells. scRNA‐seq datasets of the first‐trimester placenta are available from Vento‐Tormo et al. (2018) (13,777 cells from Carnegie stage 14 to 10‐week post‐fertilization human stage, accession E‐MTAB‐6701) and Shannon et al. (2022) (3854 cells from 6 to 7 weeks' gestational age to 11–12 weeks gestational age, accession GSE174481), respectively. For term‐stage placenta, averaged scRNA count across CTB of 3 stages and bulk RNA‐seq of STB from Pavlicev et al. (2017) (accession GSE87720). Bulk RNA‐seq of TSC differentiation is available from Ruan et al. (2022) (accession GSE190432). The CHIP‐seq of H3K9me3, H3K27me3, and H3K4me3 is available from Lee, Salamah, et al. (2024) (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE243059↗). The result file of mass spectrometry was available in Table S2, and all the raw data can be found on PRIDE website (https://www.ebi.ac.uk/pride/↗, Reviewer access: Project accession: PXD071811, Token: otfPRZ0sSICy).

Codes used to generate figures in this paper are available in the GitHub repository (https://github.com/leeyoyohku/Aging↗).

All transcriptomes are concatenated, among which single‐cell data were summed into pseudobulk according to the specific group (stage × cell type) using ADPBulk. The Pearson correlation coefficients were computed between each pair of groups based on all overlapping genes. The cluster maps were then plotted using seaborn, with linkage being computed using the "weighted" method.

Each transcriptome is normalized to millions of counts per cell for downstream processing. Gene lists for aging‐related pathways were obtained through KEGG, REACTOME, and selected manuscripts (See Table S1 for gene lists). The weighted average score of each pathway was then computed for each cell/sample in the transcriptome. More details are available in the Scanpy documentation (scanpy.tl.score_genes: https://scanpy.readthedocs.io/en/latest/generated/scanpy.tl.score_genes.html↗).

1902 TNFRSF12A+ 15+ years old MuSCs from Muscle Aging Cell Atlas (Kedlian et al. 2024) and 3568 15+ years old HSCs from Bone Marrow Atlas (Lee, Hodzic Kuerec, and Maier 2024) were preprocessed, including normalized and log‐transformed, before computing 50 principal components (PCs). A linear regression model was fitted to the 50 PCs of the two‐reference data with the cell donor's chronological age as the response variable. Integration of trophoblast datasets with the MuSC or HSC dataset was performed using ingest (sc.tl.ingest). For MuSC data, we used the mean age from each age range (categorical). The linear regression model could be used to predict the age of each cell from trophoblast datasets using the integrated PC values. 25% of each dataset was used for self‐predictions.

Differentially methylated genes were identified using the Mann Whitney test and filtered with a p‐value cutoff = 0.1 and log fold change > 1. Gene enrichment and ontology analyses were then performed on these differential genes using the gseapy package. Top items with significant enrichment p‐values were visualized using dot plots.

Transposable element annotations were obtained from the UCSC Genome Browser using RepeatMasker. SQuIRE was utilized in 'total' mode to quantify TE expression. Differential expression of TEs was analyzed using DESeq2. TEs with an expression fold change greater than 1.5, and an adjusted p‐value less than 0.05 were considered significantly differentially expressed.