Epigenetic Changes in the Mature Uterus Following Neonatal Feeding of Tamoxifen in Female ICR Mice

Animal Welfare: All experiments involving animals as shown in this manuscript have been demonstrated to be ethically acceptable and where relevant conform to appropriate national guidelines for animal usage in research. The study protocol was approved by the institutional ethics review board of Fujian Maternity and Child Health Hospital (2025KY176).

Natural Science Foundation of Fujian Province of China (2022J011041 to Y.C.), Startup Fund for Scientific Research of Fujian Medical University (2020QH1195 to Y.C., and 2024QH1203 to S.G.), and Joint Funds for the Innovation of Science and Technology, Fujian Province (2023Y9392 to R.C.), and Sponsored by Fujian Provincial Health Technology Project (2023QNA058 to S.G).

All mice survived the experiment. Immunohistochemical (IHC) analysis assessed the immunoexpression of histone modification‐related proteins (Hdacs, Kat2a, Ezh2, Suz12, Ehmt2, Setdb1, Jmjd3), DNA modification‐related proteins (Dnmts and Tets), and the estrogen‐responsive protein Six1 in the uterine endometrium and myometrium of neonatal (PND5) and adult (PND42) mice. The results are shown in Figure  3 . All markers were detectable in endometrial epithelial cells and myometrial smooth muscle cells and showed two localization patterns. Nuclear‐only staining was seen for Hdac2, Ezh2, and Suz12 (all ++ in both tissues) and for Six1 (endometrium ++, myometrium −), whereas nuclear‐cytoplasmic staining was seen for Hdac1, Hdac3, Kat2a, Ehmt2, Dnmt1, and Dnmt3a (endometrium +++, myometrium ++), Jmjd3 and Tet1‐3 (endometrium ++, myometrium +), and Setdb1 (+ in both tissues). The grading reflected the predominant immunoreactive scores, as seen in the representative images in Figure  3 . At PND42, focal and superficial adenomyosis‐like lesions were identified in 3 of 8 mice (37.5%) in TAM‐exposed mice, whereas no such lesion was found in control mice based on available H&E sections (Figure  S2 ; Table  S3 ), consistent with previous reports [ 16 ]. The lesions observed at PND42 consisted of focal glandular extensions beyond the endometrial‐myometrial interface into the superficial myometrium. Based on the semi‐quantitative depth classification, all identified lesions corresponded to Grade 1 involvement, and no mid‐myometrial or deep myometrial involvement was observed. These findings suggest that PND42 represents a time point with variable and predominantly superficial lesion appearance in this cohort. Representative photomicrographs of immunostaining and histochemistry analysis of epigenetically relevant proteins in the endometrium and myometrium. Immunohistochemical images showing the expression of epigenetically relevant proteins in the endometrium and myometrium of neonatal/infancy (PND5) and adult (PND42) ICR mice exposed to neonatal tamoxifen (TAM group) or not (CTL group). In all figures, magnification: ×400. Scale bar = 50 μm. With the only exception of Setdb1, staining levels of all 14 markers were significantly higher in the endometrium than in the myometrium, irrespective of time point (all p ‐values ≤ 0.0003; Table  2 and Figure  4 ). Across time and exposure groups, endometrial staining levels correlated positively with those of myometrium (all r 's ≥ 0.47, all p ‐values ≤ 0.0063; Table  2 ). Overall, except for Hdac2, Tet2, and Tet3, neonatal TAM feeding significantly altered 12 markers in both tissue types at different time points (Table  2 and Figure  4 ). Results of multiple linear regression analyses on staining levels incorporating the effect of neonatal feeding of tamoxifen, time of tissue samples collection, tissue type (endometrium vs. myometrium), and tamoxifen × tissue type interaction. The Pearson's correlation coefficient between endometrium and myometrium also is displayed. Note: Effect of TAM (vs. vehicle): ↑ = increased staining levels; ↓ = reduced staining levels. Time: ↑ = increased over time (from neonate to adulthood); ↓ = decreased over time. Endometrium versus myometrium: ↑ = higher in endometrium than that of myometrium; ↓ = lower in endometrium than that of myometrium. TAM × Tissue interaction: Interaction between tamoxifen and tissue. Symbols of statistical significance: * p  < 0.05; ** p  < 0.01; *** p  < 0.001; NS, not significant ( p  ≥ 0.05). Expression changes of epigenetic enzymes in the endometrium and myometrium of mice. Line graphs illustrate the expression patterns of target epigenetic enzymes in the control (blue) and tamoxifen (red) groups at the neonatal (PND5) and adult (PND42) stages. Solid lines represent protein expression in the endometrium, while dashed lines denote that in the myometrium. Data are shown as mean ± standard deviation. For each marker, MOD values from five representative fields per mouse were averaged to generate one animal‐level MOD value before statistical comparison. Staining levels (the Y ‐axis) are quantified as the MOD values derived from immunohistochemical staining. The short segment (green) in each subplot indicates the neonatal TAM feeding period (PND1–5). Symbols of statistical significance: * p  < 0.05; ** p  < 0.01; *** p  < 0.001; NS = not significant ( p  ≥ 0.05). More specifically, at PND5, mice neonatally exposed to TAM exhibited significantly elevated endometrial staining of Hdac1, Setdb1, Ehmt2, Suz12, Dnmt3a, and Six1, but reduced staining of Ezh2, Jmjd3, and Tet1 (all p ‐values ≤ 0.015; Figure  4 and Table  3 ). In contrast, at the same time point, only Hdac1, Dnmt1, and Dnmt3a staining levels were significantly elevated while Setdb1 staining was significantly reduced in myometrium (all p ‐values ≤ 0.0498; Figure  4 and Table  3 ). Changes in staining levels at PND5 and PND42 in endometrium and myometrium resulting from the neonatal feeding of tamoxifen as compared with the control group. Note: ↑ = elevated staining; ↓ = reduced staining (TAM vs. control). Symbols of statistical significance: * p  < 0.05; ** p  < 0.01; *** p  < 0.001; NS, not significant ( p  ≥ 0.05). At adulthood (PND42) when mice were sexually mature, mice neonatally exposed to TAM exhibited significantly elevated endometrial staining of Hdac1, Setdb1, Suz12, Dnmt1, Dnmt3a, and Six1, but significantly reduced staining of Hdac3, Kat2a, Ezh2, Jmjd3, and Tet1 in endometrium (all p ‐values ≤ 0.0498; Figure  4 and Table  3 ). In the myometrium, however, only Dnmt1, Dnmt3a, and Six1 staining was significantly elevated, whereas Hdac3, Kat2a, and Jmjd3 staining was significantly decreased (all p ‐values ≤ 0.021; Figure  4 and Table  3 ). Overall, more proteins exhibited aberrant immunoexpression in adulthood than at PND5 in both endometrium and myometrium (Figure  4 and Table  3 ). Notably, decreased staining of Hdac3, Kat2a, and Jmjd3, along with elevated staining of Dnmt1, Dnmt3a, and Six1, was observed in both compartments in adulthood (Figure  4 and Table  3 ). Of particular interest, Six1 staining was significantly elevated in TAM‐exposed mice in both endometrium and myometrium at infancy and adulthood (Figure  4 and Table  3 ), suggesting possibly aberrant estrogen signaling in both uterine layers. The consistent changes in staining across multiple epigenetic markers, time points, and uterine compartments suggest coordinated, module‐like shifts in the immunoexpression of select epigenetic‐regulatory proteins rather than isolated stochastic variations in individual markers. To characterize coordinated epigenetic expression patterns in the adult uterus, we performed hierarchical clustering of persistently altered epigenetic regulators and Six1 in the endometrium and myometrium (Figure  5 ). In the endometrium, the clustering heatmap revealed a clear module‐like pattern: Ezh2, Kat2a, Tet1, Hdac3, and Jmjd3 were all decreased in the TAM group, whereas Six1, Hdac1, Dnmt1, Suz12, Dnmt3a, and Setdb1 were elevated. This pattern is compatible with a TAM exposure‐induced imprint, characterized by upregulation of chromatin‐silencing‐associated regulators (e.g., Dnmt1/3a, Hdac1, Setdb1, Suz12) alongside downregulation of factors generally linked to transcriptional activation and/or demethylation (e.g., Tet1, Kat2a, Jmjd3), with concomitant Six1 upregulation indicating estrogen‐responsive reprogramming. In the myometrium, a similar overall separation pattern was observed, though less pronounced than in the endometrium. Hierarchical clustering heatmaps of epigenetic‐related proteins and Six1 in adult uterine tissues. Standardized immunoexpression levels ( Z ‐scores) of histone modification–related proteins, DNA methylation/demethylation–related proteins, and the estrogen‐responsive protein Six1 are shown for the endometrium (left) and the myometrium (right) in mice neonatally exposed to tamoxifen or vehicle. Rows represent individual proteins and columns represent two different treatment groups. Hierarchical clustering reveals tissue‐specific coexpression patterns and module organization in the adult uterus, highlighting persistent epigenetic reprogramming following neonatal tamoxifen exposure. To see whether the occurrence of adenomyosis or not has any influence on immunoexpression of the above‐evaluated proteins, we evaluated the difference, if any, in staining levels of these proteins at PND42 between mice with adenomyosis and without within the TAM group. We found that only Hdac2, Ezh2, and Six1 staining levels were significantly different (Figure  6 ). In particular, while both mice with and without adenomyosis had comparable Hdac2 staining levels (both p ‐values = 0.09), mice neonatally exposed to TAM and with adenomyosis had significantly higher Hdac2 staining than those without ( p  = 0.036; Figure  6A ). For Ezh2, while mice without adenomyosis had comparable staining levels as controls ( p  = 0.35), those with adenomyosis had significantly lower staining than controls ( p  = 0.012) and those without adenomyosis ( p  = 0.049; Figure  6B ). TAM mice with and without adenomyosis both had significantly higher staining of Six1 (both p ‐values ≤ 0.012) but those with adenomyosis had even higher staining than those without ( p  = 0.036; Figure  6C ). Thus, the difference in Ezh2 staining between TAM and the control groups appears to be attributable to the occurrence of adenomyosis, while for Six1 the occurrence of adenomyosis further elevated its immunoexpression. Summary of difference in endometrial immunoexpression of three proteins between TAM mice with and without adenomyosis at post‐natal day 42 (PND42). Boxplot showing the staining levels of Hdac2 (A), Ezh2 (B) and Six1 (C) in mice without neonatal feeding of TAM (controls), with neonatal feeding of TAM but had no adenomyosis (TAM/No AD), and with neonatal feeding of TAM and had adenomyosis (TAM/AD). AD: Adenomyosis; TAM, Tamoxifen. Symbols for statistical significance: NS: p  > 0.05; * p  < 0.05; ** p  < 0.01 (all comparisons were against the control group). # p  < 0.05 (comparison against the TAM/No AD group). To functionally annotate the epigenetic markers that are significantly altered at adulthood, we performed KEGG and GO analyses to summarize shared biological roles, if any, to see whether collectively these genes/proteins converge on the same functional pathways, and to distill a concise summary of affected biological functions [ 43 , 44 ]. The KEGG analysis indicated endometrial enrichment in “Polycomb repressive complex”, “thyroid hormone signaling pathway”, and “microRNAs in cancer”, whereas no significant KEGG enrichment was detected in the myometrium (Figure  7A ). GO analysis further indicated that endometrial enriched terms were mainly related to chromatin/histone and transcriptional regulation (Figure  7B,C ). Consistently, the g:Profiler GO Biological Process analysis returned terms related to chromatin remodeling, methylation, response to BPA, and epithelium development, supporting the use of these analyses as functional annotation of the selected marker panel. Because these analyses were based on a preselected epigenetic‐regulatory marker panel rather than an unbiased omics dataset, these results should be interpreted as exploratory, intended for hypothesis‐generating purpose. Functional annotation of select epigenetic‐regulatory markers altered in the uterus in maturity (PND42) in female ICR mice. Markers that are significantly different between TAM and Control at adulthood were analyzed separately for endometrium and myometrium. (A) KEGG pathway enrichment shown as a bubble plot, with bubble size indicating the gene count and color indicating the magnitude of p value. (B) GO enrichment results summarized for biological process (BP), molecular function (MF), and cellular component (CC), shown as bar plots of −log10 ( p ‐value). (C) GO biological process network analysis illustrating gene‐overlap similarity among enriched terms. Only terms with p  < 0.05 are shown. Taken together, neonatal TAM exposure induced early (PND5) and adulthood (PND42) changes in IHC staining for epigenetic regulators and Six1, predominantly in the endometrium rather than the myometrium. Six1 remained elevated at PND42 and cooccurred with module‐like changes across multiple epigenetic regulators, consistent with alterations in estrogen‐responsive features occurring in the context of remodeling of the uterine epigenetic regulatory markers.

In this study, we demonstrated that neonatal TAM exposure induces changes in immunoexpression of epigenetic‐regulatory proteins in the uterus from sexually mature female ICR mouse, most prominently in the endometrium. In particular, the estrogen signaling, as reflected by the overexpression of Six1, appears to be consistently altered in the reproductive tract. These findings raise the possibility that neonatal TAM exposure may influence the mature uterine state and contribute to the development of adenomyosis in adult female CD‐1/ICR mice [ 14 ]. Our findings align with a voluminous body of research demonstrating seemingly permanent and lasting epigenetic alterations in the reproductive tract of mice developmentally exposed to xenobiotic estrogens and EDCs. Developmental exposure to EDCs follows a universal mechanism of exposure in the developmental window, epigenetic reprogramming, and uterine dysfunction in adulthood, with downstream reproductive phenotypes emerging long after the exposure window has closed. Specifically, our observation of elevated endometrial Dnmt1 expression alongside reduced Ezh2 expression is consistent with Swiss mice neonatally exposed to estrogen and DES [ 34 ]. In addition, the elevated expression of Hdac1 and reduced expression of Hdac3—but not Hdac2—mirrors similar findings in eutopic endometrium from adenomyosis patients [ 45 , 46 ]. The activation of these proteins may underlie the reported hypermethylation of PR‐B in adenomyosis [ 47 ]. Of note, the increased immunostaining of the repressive epigenetic writers Hdac1, Dnmt1, Dnmt3A, Setdb1, and Suz12 was accompanied by reduced immunoexpression of epigenetic activators and erasers including Ezh2, Kat2a, Tet1, Jmjd3, and Hdac3. In particular, the elevated immunoexpression of Dnmt1 and Dnmt3A may augment both maintenance and de novo DNA methylation, while reduced Tet1 immunoexpression could compromise DNA demethylation, collectively driving genome‐wide DNA hypermethylation [ 48 ] in the endometrium. Of course, more definite proof of this would await the evaluation of functional validation and of cell‐type and temporal specificity, global DNA methylation/hydroxymethylation profiling, histone landscape mapping, chromatin accessibility assessment, and chromatin organization evaluation. Elevated Hdac1 alongside decreased Kat2a and Hdac3 expression disrupts physiological histone acetylation homeostasis, favoring a globally condensed, transcriptionally repressive chromatin state. Similarly, upregulated Setdb1 enhances widespread H3K9me3 heterochromatin formation, whereas the reciprocal expression pattern of elevated Suz12 and reduced Ezh2 remodels PRC2 complex composition and genome‐wide H3K27me3 targeting [ 49 ]. Concomitant downregulation of the H3K27me3 demethylase Jmjd3 may further stabilize repressive histone marks across the epigenome [ 50 ]. This seemingly coordinated perturbation of DNA methylation, histone acetylation, and histone methylation machinery is consistent with a non‐gene‐specific global epigenetic alteration , rather than isolated locus‐specific chromatin changes [ 51 ]. Remarkably, aside from Ezh2, these aberrations were not influenced by the presence of adenomyosis, although the condition did further elevate Six1 immunoexpression in the endometrium (Figure  6 ). But despite the apparent influence of adenomyosis on the Ezh2 staining, it would be still premature to conclude that the presence of adenomyosis suppresses Ezh2 since at best this is merely an association—the reverse also could be true, especially in view of the fact that Ezh2 staining was significantly lower at PND5 already (Figure  4 ). Previous studies have reported that neonatal feeding of TAM causes adenomyosis in 100% of exposed mice by the end of the 3rd month [ 14 ], disentangling which aberration(s) are responsible cause adenomyosis—or vice versa—would be difficult, if not impossible, if we had chosen PND120. Nevertheless, based on our findings (Figure  6 ), it appears that all epigenetic aberrations except Ezh2 are already observable before adenomyotic lesions can be detected. The epigenetic aberrations resulting from neonatal exposure of TAM and perhaps also to other xenobiotic estrogens such as DES [ 39 ] raise the possibility of potentially rectifying or reversing the aberration through pharmacological means, such as specific HDAC inhibitors and demethylation agents, for prevention or therapeutic purposes. Curiously, promoter hypermethylation of progesterone receptor isoform B [ 47 ], LIM and SH3 protein 1 (LASP1) [ 52 ], Klotho and PPARγ [ 53 ], and PIEZO1 [ 54 ] has been reported in adenomyosis, and the use of HDAC inhibitors has been shown to have promising therapeutic effects in treating adenomyosis in preclinical and clinical studies [ 21 , 22 , 55 , 56 ]. Future studies are warranted to determine the precise timing and the best intervention approach to rectify or reverse the epigenetic aberration through specific inhibitors and/or activators. Our results also concur with reports that Six1, an estrogen‐responsive protein in the mouse female reproductive tract, is normally present in the cervical and vaginal stratified squamous epithelium but absent in the endometrium [ 39 , 57 ]. Following neonatal DES exposure, however, Six1 seems to be steadily and persistently upregulated in the mouse endometrium and further induced by exposure to endogenous estrogens and observed in various adult‐onset pathologies [ 39 , 57 , 58 , 59 ]. These findings raise the possibility that aberrant uterine expression of Six1 following neonatal exposure to TAM could serve as a functional link between early disruption of cellular differentiation and later development of adenomyosis in this model. Beyond its role as an estrogen‐responsive marker [ 60 ], Six1 is a core component of the DACH/SIX/EYA transcriptional network and a transcription factor essential in development and tissue homeostasis through regulating cellular proliferation, differentiation, survival, migration, and invasion [ 61 , 62 , 63 ]. It also acts as a coregulator of transforming growth factor‐β (TGF‐β) signaling [ 64 ], directly involved in epithelial‐mesenchymal transition (EMT) [ 65 ]. It is interesting to note that the DNA methylation level of the Six1 promoter core region is found to be negatively correlated to Six1 gene expression in vivo [ 66 ]. Combined with our findings, sustained upregulation of Six1 could promote endometrial epithelial cells to acquire an EMT phenotype and enhance their invasive capacity—a process closely linked to the key pathological feature of endometrial invasion into the myometrium in adenomyosis [ 7 ]. This provides important clues for understanding the epigenetic basis of this pathological process. One unstated assumption underlying the use of the TAM‐induced adenomyosis mouse model is that all pathological manifestations observed in the mouse stem from the induced adenomyosis (Figure  1 , scenario A). However, in light of our findings presented in this study, this assumption appears to be questionable and likely invalid. A more likely scenario would be B or C in Figure  1 . Neonatal TAM exposure may result in a uterus marked by persistent epigenetic changes in adulthood and possibly sustained alteration in estrogen‐responsive features, which not only contribute to the occurrence of adenomyosis but could also account for much, if not all, of the associated pathological manifestations. For example, stable downregulation of Hdac3 in endometrium has been implicated in impaired endometrial receptivity in endometriosis [ 67 ] and in heavy menstrual bleeding in women with adenomyosis due to blunted inflammation required for proper endometrial repair [ 46 ]. Our data show that neonatal TAM exposure induces Hdac3 suppression (Figure  4 ). Thus, extreme caution should be exercised when attempting to extrapolate findings from this mouse model of adenomyosis to humans. Distinguishing effects of neonatal TAM exposure from consequences attributable to adenomyosis per se is essential for interpreting results from this model. Notably, TAM‐induced epigenetic changes exhibit a distinct endometrial preference, contrasting with the pan‐uterine changes induced by other classic EDCs. Our data consistently show that a larger proportion of epigenetic regulators and Six1 exhibited possibly persistent alterations in the endometrium (80%) than in the myometrium (53%); in the myometrium, the upregulation was largely restricted to Dnmt1 and Dnmt3a. This study may advance our understanding of adenomyosis pathogenesis. Although the neonatal TAM feeding model has been extensively used, its precise mechanisms remain largely unknown, with the only exception of “disordered stromal differentiation” [ 14 ] and the upregulation of Ngf in the uterus [ 14 ]. In retrospect, the observed disordered stromal differentiation in the TAM model may be related to Six1 upregulation concomitant with global epigenetic aberrations induced by Hdac1‐3, Dnmt1, Dnmt3a, Tet1‐3, Jmjd3, Setdb1, Ehmt2, Ezh2, and Suz12 in the endometrium, possibly alongside elevated expression of ERβ and reduced expression of progesterone receptor, as seen in mice neonatally exposed to estrogen and DES [ 34 ]. Directly related to TAM's role as a SERM, its ability to specifically target ER highly expressed in the endometrium results in regulatory effects concentrated in this layer. Based on our IHC findings, neonatal TAM exposure is associated with altered immunoexpression of epigenetic‐regulatory proteins and of Six1 in the endometrium. These changes could lead to exposure‐induced epigenetic regulatory alterations that may contribute to enhanced endometrial invasiveness and possible breach of the endometrial‐myometrial interface, ultimately inducing adenomyosis lesions. By comparison, its direct regulatory effects on the myometrium are limited and weak. In reproductive research, the impact of this model‐intrinsic imprinting, unbeknownst to its users, may be even more pronounced. Prior studies indicate that neonatal TAM exposure can reshape local hormonal responsiveness and the endometrial microenvironment, potentially accompanied by downregulation of implantation‐related genes [ 68 ] (e.g., HOXA10) and perturbations in folliculogenesis [ 19 ], estrous cyclicity, and ovarian reserve [ 69 ]. Consequently, “reproductive phenotypes” in this model may reflect multi‐level influences spanning both uterine and ovarian compartments. While developmental EDC exposure paradigms often exert some degree of reproductive interference, a notable feature of the neonatal TAM model is that, in addition to inducing an adenomyosis‐like phenotype, it may superimpose exposure‐induced alteration in epigenetic‐regulatory protein expression and hormone‐response resetting—thereby limiting its interpretability as a purely “lesion‐driven” model for dissecting reproductive mechanisms. More importantly, few investigators have fully appreciated the inherent limitations or the constraints of this model on the generalizability of findings. Many cellular, molecular, or epigenetic abnormalities reported in this model may plausibly arise, in fact, from possibly persistent epigenetic alterations and aberrant estrogen‐response programming induced by neonatal TAM exposure, rather than being direct consequences of adenomyosis per se. If so, extrapolating such findings to human disease presents substantial challenges in biological plausibility. These limitations underscore the necessity of cross‐validation across multiple models. Compared with the TAM model, the mechanical injury model [ 8 ] more closely recapitulates the etiologic pathway implicated clinically, namely EMID following uterine procedures. Independently validated by different injury modes [ 70 , 71 ], this model can reproducibly induce adenomyotic lesions and exhibits progressive features such as lesion enlargement and increasing fibrosis. The mechanical injury model simulates the disruption of the endometrial‐myometrial interface caused by clinical uterine procedures, more closely approximating human adenomyosis pathology. Studies have demonstrated that lesion volume increases and fibrotic content elevates as the disease progresses [ 72 ]. Moreover, this model avoids the confounding epigenetic alterations, estrogen response pattern remodeling, and progesterone resistance induced by TAM, and no interference with ovarian function or epigenetic changes has been observed, making it more suitable for investigating the mechanisms and efficacy of drugs related to adenomyosis and reproductive disorders [ 73 ]. Furthermore, our unpublished data also support the use of an ultrasound‐guided mechanical injury model in Sprague–Dawley rats, which has a high success rate and minimal traumatic stress (without abdominal wounds), making it an ideal option for future research. Our study has several strengths. First and foremost, we selected proteins known to be altered by developmental exposure to EDCs, especially xenobiotic estrogens, such as Ezh2, Hdac1, and Dnmt1. Second, we evaluated changes in Six1 immunoexpression, an estrogen‐responsive gene critically involved in endometrial function. Finally, by assessing both endometrial and myometrial immunoexpression of proteins involved in shaping the epigenome in infancy and adulthood, we obtained a comprehensive view of the effects of neonatal TAM exposure on the epigenome. This study also has limitations. First, our data are based solely on IHC analyses of a select protein panel and therefore demonstrate altered immunoexpression of preselected epigenetic‐regulatory proteins, rather than direct demonstration of a genome‐wide epigenomic remodeling. In addition, these immunoexpression merely reflects protein levels, not activities. Functional and epigenomic validation studies are needed to define the underlying mechanisms. Second, we evaluated only endometrial and myometrial tissues, excluding the ovaries, fallopian tubes, and other parts of the reproductive tract. Third, for sexually mature mice, we only evaluated our markers at one single time point (PND42). To see whether these changes are permanent and long‐lasting, later time points should be evaluated as well. Fourth, while our data suggest seemingly coordinated DNA hypermethylation and a possible globally condensed, transcriptionally repressive chromatin state, we did not directly evaluate epigenomic states such as DNA methylation landscapes, histone marks, and chromatin accessibility, nor did we provide further evidence for a durable reprogramming. Fifth, PND42 is a time point when all mice were sexually mature, and the expression levels of several evaluated proteins, especially Six1, likely depend on the estrous status due to their dependence on estrogen‐responsive signaling. Regrettably, this study did not stage estrous‐cycle and controlled for this viable. As such, the observed group differences may partly reflect uneven estrous‐stage distribution between the two groups. While the randomization procedure that we used may render the extreme distributions (e.g., all estrus in one group but all diestrus in another) practically unlikely, the failure to control for the estrous cycle could obscure the true signal, making the detection of the true difference either too difficult (because of the noise) or distorted. Future studies that control for estrous cycles are warranted. Sixth, while multiple tests were performed, we did not adjust for multiplicity due to technical difficulty. This may inflate type I errors. Lastly, given the changes in Six1 immunoexpression, deeper evaluation of estrogen signaling—such as expression of ERα and ERβ—is warranted. Future studies are warranted to address these issues. In sum, neonatal tamoxifen exposure induces early and adulthood remodeling of uterine epigenetic regulators in female ICR/CD‐1 mice, especially in the endometrium. In particular, Six1, as an estrogen‐responsive readout, remains persistently elevated in adulthood, coinciding with coordinated, module‐like aberrations in multiple epigenetic regulators. Collectively, these findings indicate that the neonatal TAM exposure in ICR/CD‐1 mice not only induces adenomyosis but also sears a likely lasting exposure imprint on the reproductive tract in maturity. While the adenomyosis model is easy and inexpensive to use and may help to gain insights into the adenomyosis pathogenesis, extreme caution should be exercised when using this model for research purposes since sound interpretation of the data and the assessment of the relevance to human adenomyosis require a clear definition of modeling boundaries and cross‐validating key conclusions across multiple adenomyosis models.

The authors have nothing to report.

Adenomyosis is a common uterine disease affecting women of reproductive age [ 1 ] and recognized as a contributing factor to dysmenorrhea, pelvic pain, heavy menstrual bleeding, and subfertility [ 2 , 3 , 4 , 5 ]. Despite its prevalence, it is an under‐researched condition [ 6 ], and its pathogenesis and pathophysiology have traditionally been viewed as enigmatic [ 7 ]. In the last few years, endometrial‐myometrial interface disruption (EMID) resulting from iatrogenic uterine procedures has been identified as an etiological factor for adenomyosis [ 8 ], especially for Kishi's type I adenomyosis [ 9 ]. However, the pathogenesis of other types of adenomyosis still remains elusive. One approach to investigating its pathogenesis is through animal models [ 10 ]. Notably, unlike endometriosis, adenomyosis can and does occur spontaneously in rodents, even though they do not menstruate [ 10 ]. Thus, insights gained from rodent models should prove invaluable in unraveling the pathogenesis of adenomyosis. Indeed, hyperprolactinemia was shown to induce adenomyosis in mice over four decades ago [ 11 , 12 ] and recently has been validated in another mouse strain [ 13 ]. Nevertheless, whether hyperprolactinemia is a pathogenic factor in causing adenomyosis in humans remains to be investigated. Another widely used model of adenomyosis is the neonatal feeding of tamoxifen (TAM) in CD‐1 mice, established in 2001 [ 14 ]. This model has been extensively adopted in adenomyosis research due to a high induction success rate, ease of induction, and cost‐effectiveness. It has been employed to dissect the pathological mechanisms of adenomyosis [ 15 , 16 , 17 , 18 ], the mechanisms underlying adenomyosis‐impaired fertility [ 19 , 20 ], and preclinical testing of potential therapeutics [ 21 , 22 , 23 , 24 , 25 ]—just to name a few. TAM belongs to the family of selective estrogen receptor modulators (SERMs). An increased risk of adenomyosis has been reported in postmenopausal breast cancer patients treated with TAM [ 26 ]. However, no direct epidemiological data link neonatal TAM exposure to adenomyosis in humans. Elucidating the mechanism underlying the neonatal TAM model may therefore provide much needed insights into adenomyosis pathogenesis. To date, it is known that neonatal TAM feeding causes disorganized uterine morphology and abnormal stromal differentiation in mouse [ 14 ], hinting at potential disruption of estrogen signaling—a key regulator in the female reproductive tract—within the uterus. More importantly, all studies utilizing this mouse model operate under an implicit yet untested assumption: that all the molecular and pathological changes observed in the model are directly attributable to adenomyosis itself and that the resultant findings can be potentially extrapolated to its human counterpart (Figure  1 , Scenario A). However, this assumption requires validation, as alternative scenarios may exist (Figure  1 , Scenarios B&C). If this is true, then the extrapolation to humans would be questionable. Conceptual models for interpreting the neonatal TAM‐induced adenomyosis model. (A) This is the scenario in which neonatal TAM exposure induces adenomyosis, which then gives rise to all the downstream phenotypes; (B) This scenario depicts the case in which neonatal TAM exposure induces lasting and permanent epigenetic changes in the uterus and sustained alteration of estrogen signaling, which may contribute to adenomyosis as well as downstream phenotypes; (C) In this scenario, the downstream phenotypes may arise via adenomyosis‐dependent effects as well as persistent epigenetic and estrogen‐response remodeling resulting from the neonatal TAM exposure. Arrows indicate proposed directions of influence. TAM, Tamoxifen. In both humans and rodents, the neonatal period—like the prenatal period—represents a critical developmental window during which the reproductive organs are exquisitely sensitive to xenobiotic estrogens and exogenous endocrine‐disruptive chemicals (EDCs), particularly those with estrogenic activity. A striking example is observed in honeybees: all larvae are fed royal jelly for the first 3 days of their development, after which worker bees switch to worker jelly, whereas queen bee larvae continue consuming royal jelly into adulthood. This dietary divergency leads to two reproductively distinct castes (infertile workers vs. fertile queens), driven by royal jelly‐induced epigenetic changes [ 27 ]. This phenomenon illustrates that epigenetic alterations induced by early environmental factors can permanently reshape reproductive phenotypes. Extensive studies have demonstrated that prenatal or neonatal exposure to xenobiotic estrogens—including exogenous EDCs such as diethylstilbestrol (DES), bisphenol A (BPA), as well as endogenous 17β‐estradiol (E 2 )—induces long‐lasting and irreversible changes in the reproductive tract in key gene/protein expression, along with epigenetic modifications that specifically perturb estrogen signaling [ 16 , 28 , 29 , 30 , 31 , 32 , 33 ]. Of particular relevance, such early‐life exposures also induce adenomyosis‐related uterine pathological changes. Neonatal DES exposure leads to adenomyosis‐like lesions in CD‐1 mice [ 16 , 28 ] and induces adenomyosis in Swiss mice [ 34 ]. Similarly, neonatal estrogen exposure in CD‐1 mice causes numerous histological changes consistent with early‐stage adenomyosis [ 16 ], and can also induce adenomyosis in Swiss mice [ 34 ]. Thus, a substantial body of evidence indicates that early‐life/developmental exposure to xenobiotic estrogens perturbs estrogen signaling via epigenetic mechanisms, thereby constituting a potential pathogenic pathway for adenomyosis. As a SERM, the molecular mechanism by which TAM induces adenomyosis remains unclear. Given the temporal overlap between the sensitive window for neonatal TAM exposure and that for EDC exposure, we hypothesized that neonatal TAM feeding, similar to prenatal/neonatal EDC exposure, could induce epigenetic changes (e.g., DNA methylation, histone modifications) in the mature uterus, disrupt estrogen signaling, and consequently trigger adenomyosis‐related pathological changes. To test this hypothesis, we characterized early and adulthood changes in proteins involved in epigenetic regulation within the endometrium and myometrium following neonatal TAM exposure in ICR mice. In addition, we evaluated the immunoexpression of sine oculis homeobox homolog 1 (Six1), an estrogen‐responsive protein known to regulate endometrial aberrations following developmental exposure to xenobiotic estrogens [ 35 ].

S.W.G. is a member of the Scientific Advisory Board of Heranova, E3A Healthcare, Foraviset, and of FimmCyte A.G., has provided consultancy advice to these companies, as well as ReproNovo and Ziwig, but these activities had no bearing on this work. All other authors have no conflicts of interest.

This study was approved by the institutional ethics review board of Fujian Maternity and Child Health Hospital (2025KY176). Twelve pregnant female ICR mice (gestational days 17–18) were purchased from Shanghai Jie Sijie Laboratory Animal Company (Shanghai, China) and housed individually for the remainder of gestation, as well as during the subsequent birth and nursing period. The day of birth was designated as postnatal day 0 (PND0) [ 36 ]. We used ICR mice since they are closely related to CD‐1 (CD‐1 is an ICR mouse strain that has been rederived and commercialized by Charles River). On PND 1, all pups were sexed, and female pups were selected for the study. Each dam and her litter were housed together in the same cage under controlled conditions, with a 12‐h light/12‐h dark cycle and ad libitum access to food and water. On PND1, 32 neonatal female ICR mice were randomly assigned to 2 groups in equal sizes: the control group (CTL) and the tamoxifen‐treated group (TAM). From PND1 to PND5, the TAM group received tamoxifen treatment at a dose of 1 mg/kg via oral feeding [ 14 , 37 ]. Tamoxifen was suspended in a mixture of peanut oil, lecithin, and condensed milk (2:0.2:3, v/v) and administered at a dose volume of 1 μL/g/day. The CTL group received the same volume of the excipient mixture without tamoxifen, following the identical administration protocol as the TAM group. At two time points—neonatal (PND5) and sexually mature (PND42)—8 mice in each group were sacrificed by cervical dislocation. We chose PND42 as an end time point not only because it represents a tradeoff between PND30 and PND70 that were used in studies evaluating epigenetic regulator expression after neonatal exposure to xenobiotic estrogens [ 38 , 39 ] but also at PND42 adenomyosis started to appear in mice neonatally exposed to TAM [ 16 ]. Uterine tissues were rapidly excised, rinsed with normal saline to remove blood and impurities, and then fixed in 4% neutral‐buffered formalin for further immunohistochemical analysis. The experimental design is shown in Figure  2 . Schematic diagram showing the experimental design of this study. This study aims to assess the short‐ and long‐term epigenetic effects, if any, of neonatal TAM exposure on the uterus. Serial 4‐μm thick sections were cut from each paraffin block. The first section cut from each paraffin block was stained with hematoxylin and eosin (H&E; Jiehao Biotechnology, Shanghai, China) to confirm the morphology of the endometrium and myometrium and to provide a descriptive histological assessment of adenomyosis‐like lesions across the entire uterine cross‐section at PND42 including their presence or absence and the maximum depth of myometrial involvement. Adenomyosis‐like lesions were defined as endometrial glandular structures extending beyond the endometrial–myometrial interface into the myometrium, with or without associated stromal components [ 40 ]. Lesion depth was graded according to the depth of myometrial involvement, as used in previous mouse studies: Grade 0, no lesion; Grade 1, superficial myometrial involvement; Grade 2, mid‐myometrial involvement; and Grade 3, involvement beyond the mid‐myometrium [ 19 , 22 ]. Because this study was not designed for systematic whole‐uterus lesion‐burden assessment, the histological reassessment was considered to be descriptive and exploratory. After routine deparaffinization and rehydration, antigen retrieval was performed by heating the tissue sections in citrate buffer (pH 6.0; Servicebio, Wuhan, China) at 98°C for 30 min, or in ethylenediaminetetraacetic acid (EDTA) buffer (pH 9.0; Jiehao Biotechnology) at 95°C for 20 min. After cooling to room temperature, slides were blocked with goat serum (Jiehao Biotechnology) and incubated overnight at 4°C with primary antibodies (Table  1 ). Based on findings from a previous study [ 39 ], we selected primary antibodies that targeted key epigenetic regulators, including Histone Deacetylase 1–3 (Hdac1‐3), Lysine Acetyltransferase 2A (Kat2a/Gcn5), Enhancer of Zeste Homolog 2 (Ezh2), Suppressor of Zeste 12 (Suz12; a core subunit of polycomb repressive complex 2 (PRC2)), Euchromatic Histone Lysine Methyltransferase 2 (Ehmt2/G9a), SET Domain Bifurcated Histone Lysine Methyltransferase 1 (Setdb1/Eset), Lysine Demethylase 6B (Kdm6b/Jmjd3), DNA Methyltransferase 1 and 3A (Dnmt1 and Dnmt3a), and Ten‐Eleven Translocation Methylcytosine Dioxygenase 1–3 (Tet1‐3), together with Six1. Slides were then incubated with secondary antibody (Jiehao Biotechnology) for 30 min at room temperature. Signals were visualized with diaminobenzidine (DAB; Jiehao Biotechnology) for 3–5 min, followed by hematoxylin counterstaining and mounting. The rationale for the selection of these proteins is provided in Table  S1 . The photomicrographs for positive and negative controls are shown in Figure  S1 . Names and catalog numbers of antibodies used in immunohistochemistry analysis. Images were captured using an Olympus BX51 microscope (Olympus, Tokyo, Japan) fitted with an Olympus DP70 digital camera. For each marker, five images per mouse were randomly selected from each sample at 400× magnification. Quantitative scoring for all statistical comparisons was based on mean optical density (MOD), calculated using Image‐Pro Plus 6.0 software (Media Cybernetics Inc., Rockville, MD, USA). The MOD values from the five randomly selected fields from the same mouse were averaged to generate one MOD value for each marker per mouse. For descriptive presentation, staining was additionally graded using an immunoreactive score (IRS). Staining intensity (SI) was scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong), and the percentage of positive cells (PP) was scored as 1 (1%–10%), 2 (11%–50%), 3 (51%–80%), or 4 (81%–100%). The immunoreactive score was calculated as SI × PP (range 0–12) and categorized as − (0–2), + (3, 4), ++ (5–8), and +++ (9–12) [ 41 , 42 ]. All statistical analyses were based on MOD. The grading was summarized in Table  S2 . All statistical analyses were performed using R software (Version 4.5.2), with p  < 0.05 considered as statistically significant. All analyses of immunostaining intensity data were performed using animal‐level MOD values, with each mouse as one data point. For independent samples, the t ‐test was used for comparisons of immunostaining intensity between the CTL and TAM groups. Multiple linear regression analysis was conducted to evaluate the main effects of tamoxifen exposure, time (PND5 vs. PND42), and tissue type (endometrium vs. myometrium) as well as the interaction effect between tamoxifen and tissue type on epigenetic enzyme expression. Pearson's correlation coefficient was calculated to assess associations between endometrial and myometrial staining levels. For adult uterine samples (PND42), hierarchical clustering and KEGG/GO enrichment analyses were performed to summarize coexpression patterns and functional annotations.

Figure S1: Negative and positive controls for immunohistochemical staining. Representative negative and positive tissue controls are shown for all antibodies used in this study. Negative controls were performed on mouse endometrial tissue by replacing primary antibodies with phosphate‐buffered saline, and showed no specific brown staining. Positive control tissues included mouse liver for Hdac1, Hdac2, Hdac3, Ehmt2, Tet2, Tet3, and Jmjd3; mouse testis for Kat2a, Dnmt3a, Suz12, Tet1, and Setdb1; mouse colon for Ezh2; and mouse stomach for Dnmt1 and Six1. Figure S2: Representative hematoxylin and eosin (H&E) and immunohistochemical staining images of select markers grouped by exposure status and presence or absence of adenomyosis at PND42. Representative H&E staining and immunohistochemical staining for Hdac1, Hdac2, Ezh2, and Six1 are shown in control mice (no tamoxifen exposure), tamoxifen‐exposed mice without adenomyosis (TAM/No AD), and tamoxifen‐exposed mice with adenomyosis (TAM/AD). H&E images show uterine morphology and focal superficial adenomyosis‐like glandular extensions in TAM/AD mice. In the TAM‐exposed adenomyosis H&E image, the arrow indicates a focal superficial adenomyosis‐like glandular extension beyond the endometrial–myometrial interface. Immunohistochemical images show representative staining patterns of selected markers included in the lesion‐status analysis. Hdac2, Ezh2, and Six1 were chosen due to difference in staining levels between mice with the presence of adenomyosis and without, all were neonatally exposed to tamoxifen. Hdac1 was selected as a comparison. Scale bars = 100 μm. Table S1: Functional relevance of markers used in our study for various experiments. Table S2: Descriptive staining grade and subcellular localization. Table S3: Descriptive histological characterization of adenomyosis at PND42.