Human embryo implantation involves Syncytin-2/MFSD2A-mediated heterokaryon formation with maternal endometrium

18 Human embryo implantation involves attachment of the blastocyst to the endometrial 19 epithelium to subsequently gain access to the underlying stromal compartment. The 20 blastocyst is believed to cross the epitheli um either by migration through, or upon 21 apoptosis of, the endometrial epithelial cell layer. Yet, how the blastocyst exactly traverses 22 the endometrial epithelium remains unknown. Here, we describe an in vitro implantation 23 model of human blastoids and hormonally matured endometrial organoids amenable to 24 high-resolution live imaging. We demonstrate that the initial step of implantation is 25 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint mediated by the direct fusion of blastoid cells with endometrial epithelial cells. Blastoids 26 express the f usion proteins Syncytin-1 and -2, while the endometrial epithelium mainly 27 expresses the fusion co-receptor for Syncytin-2, called MFSD2A. CRISPR-induced loss of 28 MFSD2A in the endometrial epithelium prevents blastoids from attaching and abolishes 29 fusion. Together, these findings support a model in which fetal-maternal cell fusion 30 constitutes the critical initiating mechanism of human embryo implantation, with 31 endometrial MFSD2A playing an indispensable role in this process. 32 33 The early stages of human embryo implantation are critical for the successful establishment of 34 pregnancy. Compared to most mammalian species, human fertility is relatively low, with 35 approximately 60% of conception attempts failing, often for reasons that remain poorly understood. 36 A key factor in achieving a successful early pregnancy is the proper attachment and invasion of 37 the blastocyst into the endometrium, which happens b etween days 5 and 7 post -fertilization. 38 During this period, t he blastocyst enters the uterine cavity and aligns its polar trophectoderm 39 towards the endometrium. Here, trophoblast cells within the polar trophectoderm differentiate and 40 fuse to generate the multinucleated syncytiotrophoblast (STB)2. This cell fusion is mediated by 41 the upregulation of the fu sion proteins Syncytin -1 and Syncytin -2, which interact with their 42 respective receptors, ASCT2 ( SLC1A5) and MFSD2A 3–7. The formation of the STB marks the 43 initiation of implantation, as it coincides with breaching of the endometrial epithelium and 44 subsequent invasion into the underlying stromal compartment - a crucial step in anchoring the 45 embryo and establishing a functional placental interface. Two main mechanisms have been 46 proposed to explain how the STB breaches the endometrial epithelium. One would involve 47 transepithelial penetration, in which epithelial cells are displaced and engulfed8,9. The alternative 48 hypothesis posits that the STB induces apoptosis of epithelial cells, thereby creating a gateway 49 for entry into the underlying stroma10. However, current knowledge is largely derived from static 50 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint observations, as methods enabling real -time analysis of human embryo implantation have not 51 been available. 52 53 To investigate the early events of human embryo implantation, we sought to develop a co-54 culture system amenable to high-resolution imaging that could recapitulate the interaction 55 between the blastocyst and endometrial epithelium (Fig.1a). To be able to distinguish the 56 embryonic and endometrial components during imaging, we developed reporter lines with non-57 overlapping fluorescent reporters (Fig.1a). We established genetically engineered human 58 endometrial organoids expressing Lifeact-mScarlet11 (Fig. 1b). To optimize imaging resolution 59 and ensure an apical-out orientation, the endometrial organoids were transferred to a 2D culture 60 format by placing the cells onto a thin coat (~30µm) of hydrogel consisting of a mixture of collagen 61 I and Matrigel® in a glass-bottom imaging chamber (Fig. 1c-d). This hydrogel mimics the collagen 62 composition and stiffness of the in vivo stromal compartment12,13. 63 The 2D organoids were stimulated with estrogen and progesterone to induce the secretory 64 phase of the menstrual cycle14,15 (Fig. 1a). Similar to in vivo secretory endometrial epithelium, we 65 found that this induced cells to become elongated with a height of ~17,5 µm16,17 (Fig. 1e). The 66 visualization of actin within these 2D organoids revealed distinct morphological changes typical 67 of hormonal differentiation: the remodeling of cell-cell junctions and the emergence of elongated 68 cells with sporadic apical protrusions resembling pinopodes18 (Extended data Fig. 1d, e, Video 1). 69 Moreover, gene expression analysis and immunostainings revealed the presence of ciliated and 70 secretory cells upon hormone stimulation of endometrial cells , essential for recreating the 71 appropriate environment for implantation (Extended data Fig. 1a -c). The differentiation towards 72 secretory endometrium yielded ~6% of multi-ciliated cells in 2D endometrial organoids, similar to 73 what has been found in receptive endometrium in vivo19 (Fig. 1f, Extended Fig. 1c). Importantly, 74 Lifeact-mScarlet expression did not appear to affect differentiation , nor did we observe any 75 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint morphological differences compared to the original 2D endometrial organoid line (Extended data 76 Fig. 1f). 77 Separately, we generated human embryonic stem cells (hESCs) expressing Lifeact-GFP 78 and H2B -iRFP, which were used to create blastoids to model the blastocyst as described 79 previously20,21. Lifeact-GFP visualized cell borders and finer structures like apical microvilli, while 80 H2B-iRFP marked nuclei (Fig. 1h, i, Extended data Fig. 2a). Expression of Lifeact-GFP and H2B-81 iRFP in blastoids did not appear to affect their development and morphology (Extended data Fig. 82 2a-d). The blastoids were cultured on the 2D monolayer of matured endometrial organoid cells. 83 The three different fluorescent proteins allowed us to distinguish the blastoids from the 84 endometrial cells (Fig. 1j, k). As blastocysts are believed to require 2-3 days to implant in vivo22, 85 we co -cultured blastoids on the differentiated 2D endometrium for 72 hours. We then gently 86 washed the monolayer with medium and counted the number of attached blastoids. We thus 87 noted that between 20 - 40% of blastoids remained firmly attached (Fig. 1l, m), similar to what 88 has been reported by other in vitro implantation experiments20,23. 89 Surprisingly, when examining the attached blastoids, we consistently observed the 90 presence of large, multinucleated cells co-expressing Lifeact-mScarlet, Lifeact-GFP and H2B -91 iRFP (Fig. 2a, b). After 72 hours of co-culture, all attached blastoids contained such triple-positive, 92 multinucleated cells. Given that (1) these cells contain ed Lifeact originating from both the 93 blastoids and the endometrial cells and (2) that blastocysts are known to contain cells with 94 fusogenic capacity, this suggested that blastoid cells may have fused with endometrial cells. 95 By comparing triple-positive cells in attached blastoids, we observed two morphologies: 96 In ~30 % of the attached blastoids, the triple-positive cells were located directly underneath the 97 blastoid (group #1), while in the remaining ~70 %, the triple-positive cells were located at the 98 periphery of the interaction site between blastoid and endometrial monolayer , covering a larger 99 area of the hydrogel (group #2) (Extended Fig. 3a-c). To understand the dynamics of this fusion 100 event and the relationship between the two morphologies, we performed over-night live imaging 101 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint with a time resolution of 30 minutes to capture the moment of blastoid attachment and the 102 subsequent events. Attachment appeared to be initiated by an acute fusion event between the 103 blastoid and endometrium (Fig. 2c, d, Video 2). First, the blastoid fused with one endometrial cell, 104 which was rapidly followed by fusion with multiple neighboring endometrial cells. This resulted in 105 the formation of a syncyti al cell that quickly increase d in size over the following six hours, after 106 which further expansion slowed down (Extended data Fig. 4a, b). After 21 hours of imaging, we 107 observed that further blastoid cells penetrated through th e syncytial cell layer without further 108 fusion, resulting in triple-positive cells moving towards the periphery (Fig. 2d). After an extended 109 co-culture of 8 days, the triple-positive cells were retained at the interface of the blastoid and the 110 endometrium (Extended data Fig. 3 d). It thus appeared that the first morphology (group #1) 111 represented the early stage of blastoid attachment, while the second (group #2) represented the 112 stage in which the blastoid had attached via fusion and started to spread further on the hydrogel. 113 We confirmed the fusion event using endometri al organoids from a n independent donor 114 (Extended data Fig. 3e). 115 Notably, following fusion between blastoids and endometrium, nuclei started appearing 116 that were originally negative for the blastoid iRFP histone mark (and therefore were of endometrial 117 origin), yet became progressively positive for the mark (Extended data Fig. 5a, b, Video 3). This 118 observation indicated that histones encoded in the blastoid nuclei integrated into endometrial 119 nuclear chromatin and further supported the notion that endometrial and blastoid nuclei resided 120 in the same cells. To confirm these observations, we generated blastoids stably expressing H2B-121 mNeon and endometrial organoids expressing H2B-iRFP (Extended data Fig. 5c). After 72 hours 122 of co-culture, we observed clusters of nuclei that were positive for both H2B -mNeon and H2B-123 iRFP (Extended data Fig. 5d, e). F-actin staining confirmed that the double-positive nuclei resided 124 within the same cell ( Extended data Fig. 5 f). Within these fused cells, variable fluorescence 125 intensities were observed per nucleus. Nuclei with higher H2B-mNeon intensity were presumably 126 blastoid-derived, while those with higher H2B -iRFP intensity were endometrium -derived 127 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint (Extended data Fig. 5g). Of note, individual syncytia consistently contained multiple trophoblast 128 and endometrial nuclei. 129 The core machinery driving STB formation in the blastocyst consists of four proteins: the 130 fusion ligands Syncytin-1 (ERVW-1) and Syncytin-2 (ERVFRD-1), and their respective receptors 131 ASCT2 (SLC1A5) and MFSD2A (MFSD2A)6,24 (Fig. 3a). Immediately after formation of blastoids, 132 in the absence of endometrium , Syncytin proteins were expressed at low levels , while higher 133 expression levels were detected for the two receptors (Fig. 3b). After 2 days of expanded blastoid 134 culture in the absence of endometrium , qPCR gene expression analysis revealed a significant 135 upregulation of the ligands ERVW-1 and ERVFRD-1, while the receptor SLC1A5 expression 136 remained stable and MFSD2A showed a modest increase (Fig. 3b). Notably, the expression of 137 the STB marker genes OVOL1, GCM1 and GCB were increased after two days, and we confirmed 138 the presence of multinucleated cells within the blastoids (Fig. 3c, d). These findings indicate that 139 under the given culture conditions, blastoid cells were capable of maturing and differentiating into 140 STB cells. 141 The upregulation of STB markers coincided with the time window during which blastoid–142 endometrial fusion was observed in our previous experiments. Since STB formation and epithelial 143 fusion appeared to occur concurrently, it was plausible that the Syncytin-1:ASCT2 and/or the 144 Syncytin-2:MFSD2A interactions could also mediate blastoid-endometrial fusion. To explore this, 145 we examined the expression of these receptors in vivo using the single -cell RNA-seq dataset 146 generated by Garcia-Alonso et al.25. This analysis revealed that MFSD2A is strongly expressed 147 in endometri um epithelial cells, specifically in luminal and SOX9+ cells, whereas SLC1A5 is 148 predominantly localized to the stromal compartment of the endometrium (Fig. 3 e). 149 Immunohistochemical staining of primary human endometrial tissue confirmed these findings: 150 MFSD2A was consistently expressed in epithelial cells throughout the proliferative and secretory 151 phases, while ASCT2 was confined to stromal cells during the proliferative phase and to glandular 152 epithelium during the secretory phase (Fig. 3f). Encouragingly, MFSD2A expression occurred in 153 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint the 2D endometrial epithelial system and was upregulated upon maturation towards the secretory 154 phase (Fig.4a). 155 To functionally assess the requirement of MFSD2A for blastoid -endometrial fusion, we 156 generated clonal endometrial organoid lines lacking functional MFSD2A by introducing a 157 premature stop codon using CRISPR base-editing (Extended Data Fig. 6a). Successful mutation 158 was confirmed by Sanger sequencing (Fig. 4 b). MFSD2AKO organoids displayed similar 159 morphology and long -term growth to their wildtype counterparts (Fig. 4 c). Gene expression 160 analysis further showed that MFSD2AKO organoids retained normal differentiation capacity, 161 displaying upregulation of lineage markers such as PAEP (secretory cells), TPPP3 (ciliated cells), 162 and PGR upon hormone -induced differentiation (Fig. 4 d, Extended Data Fig. 6 b). Next, we 163 cultured blastoids for 72 hours on 2D endometrial epithelium derived from either wildtype or clonal 164 MFSD2AKO organoids. Strikingly, whereas blastoids readily attached to the wildtype endometrium, 165 no stable adhesion was observed in the MFSD2A KO condition (Fig. 4e). Blastoids cultured on 166 mutant endometrial 2D organoids did not remain adherent upon medium flushing, indicating a 167 loss of stable interaction with the endometrium (Fig.4f). We concluded that fusion did not occur 168 between blastoids and the MFSD2AKO endometrial epithelium, and that fusion is a prerequisite for 169 stable attachment. Together, these results identified MFSD2A as an essential epithelial receptor 170 required for both attachment and fusion with blastoids (Fig. 4g). 171 Taken together, b y combining fluorescently labeled human blastoids and endometrial 172 organoids in an in vitro implantation model amenable for high spatiotemporal -resolution live 173 imaging, we have visualized an unexpected, highly dynamic interaction between human blastoids 174 and the endometrial epithelium. The described fusion mechanism is distinct from earlier models 175 that suggest that the endometrial epithelium undergoes apoptosis or passively retracts to permit 176 embryo invasion. While these processes may coexist, our findings highlight cell fusion as a 177 necessity for stable attachment, emphasizing the active role of the endometrium in early 178 implantation. Of note, it has recently been shown that stromal cells can fuse with the STB in vitro, 179 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint suggesting that multiple maternal cell types can contribute to the placenta23. Several observations 180 support that fusion does occur in vivo. For instance, the primary syncytium contains differently 181 sized nuclei: larger ones formed by endore duplication, and smaller ones, po tentially originating 182 from the endometrial epithelium or stroma. Moreover, the concept of fusion of the blastocyst with 183 the endometrial epithelium is not entirely new, as an early study in 1971 has reported that rabbit 184 blastocysts may fuse with the endometrial epithelium26. Similar to what we have observed, fusion 185 with the endometrial epithelium occurred in a localized fashion and did not spread along the entire 186 epithelium. Interestingly, rabbits express only one Syncytin, called Syncytin-Ory1, orthologous to 187 human Syncytin-1, which binds to the orthologous receptor ASCT227. Our findings raise a plethora 188 of questions . It will be important to understand whether fusion with endometrial cells indeed 189 ensures proper anchoring to the uterus and whether the maternal cells contribute to the furt her 190 invasion of the primary syncytium. An attractive hypothesis is that fusion with maternal cells could 191 play a role in the evasion of the immune system . Finally, our findings imply that the endometrial 192 epithelium is not just a barrier, but that it may play an active role during the initiation of pregnancy. 193 Future studies of this atypical fusion ma y highlight why blastocysts fail to implant and could 194 provide new therapeutic insights into human infertility. 195 196 197 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint

198 1. Detailed Clinic Summary Report. 199 https://www.sartcorsonline.com/rptCSR_PublicMultYear.aspx?ClinicPKID=2001091#patient-200 cumulative. 201 2. Malhotra, S. S., Banerjee, P. & Gupta, S. K. Regulation of trophoblast differentiation during 202 embryo implantation and placentation: Implications in pregnancy complications. J. Reprod. 203 Health Med. 2, S26–S36 (2016). 204 3. Blaise, S., de Parseval, N., Bénit, L. & Heidmann, T. Genomewide screening for fusogenic 205 human endogenous retrovirus envelopes identifies syncytin 2, a gene conserved on primate 206 evolution. Proc. Natl. Acad. Sci. U. S. A. 100, 13013–13018 (2003). 207 4. Blond, J. L. et al. An envelope glycoprotein of the human endogenous retrovirus HERV-W is 208 expressed in the human placenta and fuses cells expressing the type D mammalian 209 retrovirus receptor. J. Virol. 74, 3321–3329 (2000). 210 5. Esnault, C. et al. A placenta-specific receptor for the fusogenic, endogenous retrovirus-211 derived, human syncytin-2. Proc. Natl. Acad. Sci. U. S. A. 105, 17532–17537 (2008). 212 6. Azar, C. et al. RNA-Seq identifies genes whose proteins are transformative in the 213 differentiation of cytotrophoblast to syncytiotrophoblast, in human primary villous and BeWo 214 trophoblasts. Sci. Rep. 8, 5142 (2018). 215 7. Mi, S. et al. Syncytin is a captive retroviral envelope protein involved in human placental 216 morphogenesis. Nature 403, 785–789 (2000). 217 8. Ruane, P. T. et al. Trophectoderm differentiation to invasive syncytiotrophoblast is promoted 218 by endometrial epithelial cells during human embryo implantation. Hum. Reprod. Oxf. Engl. 219 37, 777–792 (2022). 220 9. Bentin-Ley, U. et al. Ultrastructure of human blastocyst-endometrial interactions in vitro. J. 221 Reprod. Fertil. 120, 337–350 (2000). 222 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint 10. Tamura, N., Sugihara, K., Akama, T. O. & Fukuda, M. N. Trophinin-mediated cell adhesion 223 induces apoptosis of human endometrial epithelial cells through PKC-δ. Cell Cycle Georget. 224 Tex 10, 135–143 (2011). 225 11. Riedl, J. et al. Lifeact: a versatile marker to visualize F-actin. Nat. Methods 5, 605–607 226 (2008). 227 12. Abbas, Y. et al. Tissue stiffness at the human maternal-fetal interface. Hum. Reprod. Oxf. 228 Engl. 34, 1999–2008 (2019). 229 13. Shi, J.-W. et al. Collagen at the maternal-fetal interface in human pregnancy. Int. J. Biol. Sci. 230 16, 2220–2234 (2020). 231 14. Turco, M. Y. et al. Long-term, hormone-responsive organoid cultures of human 232 endometrium in a chemically defined medium. Nat. Cell Biol. 19, 568–577 (2017). 233 15. Boretto, M. et al. Patient-derived organoids from endometrial disease capture clinical 234 heterogeneity and are amenable to drug screening. Nat. Cell Biol. 21, 1041–1051 (2019). 235 16. Sarani, S. A., Ghaffari-Novin, M., Warren, M. A., Dockery, P. & Cooke, I. D. Morphological 236 evidence for the ‘implantation window’ in human luminal endometrium. Hum. Reprod. Oxf. 237 Engl. 14, 3101–3106 (1999). 238 17. Nikas, G. et al. Uterine pinopodes as markers of the ‘nidation window’ in cycling women 239 receiving exogenous oestradiol and progesterone. Hum. Reprod. Oxf. Engl. 10, 1208–1213 240 (1995). 241 18. Murphy, C. R. Junctional barrier complexes undergo major alterations during the plasma 242 membrane transformation of uterine epithelial cells. Hum. Reprod. Oxf. Engl. 15 Suppl 3, 243 182–188 (2000). 244 19. Masterton, R., Armstrong, E. M. & More, I. A. The cyclical variation in the percentage of 245 ciliated cells in the normal human endometrium. J. Reprod. Fertil. 42, 537–540 (1975). 246 20. Kagawa, H. et al. Human blastoids model blastocyst development and implantation. Nature 247 601, 600–605 (2022). 248 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint 21. Okubo, T. et al. Hypoblast from human pluripotent stem cells regulates epiblast 249 development. Nature 626, 357–366 (2024). 250 22. Niakan, K. K., Han, J., Pedersen, R. A., Simon, C. & Pera, R. A. R. Human pre-implantation 251 embryo development. Dev. Camb. Engl. 139, 829–841 (2012). 252 23. Shibata, S. et al. Modeling embryo-endometrial interface recapitulating human embryo 253 implantation. Sci. Adv. 10, eadi4819 (2024). 254 24. Roberts, R. M. et al. Syncytins expressed in human placental trophoblast. Placenta 113, 8–255 14 (2021). 256 25. Garcia-Alonso, L. et al. Mapping the temporal and spatial dynamics of the human 257 endometrium in vivo and in vitro. Nat. Genet. 53, 1698–1711 (2021). 258 26. Enders, A. C. & Schlafke, S. Penetration of the uterine epithelium during implantation in the 259 rabbit. Am. J. Anat. 132, 219–230 (1971). 260 27. Heidmann, O., Vernochet, C., Dupressoir, A. & Heidmann, T. Identification of an 261 endogenous retroviral envelope gene with fusogenic activity and placenta-specific 262 expression in the rabbit: a new ‘syncytin’ in a third order of mammals. Retrovirology 6, 107 263 (2009). 264 28. Pleguezuelos-Manzano, C. et al. Establishment and Culture of Human Intestinal Organoids 265 Derived from Adult Stem Cells. Curr. Protoc. Immunol. 130, e106 (2020). 266 29. Krueger, D. et al. Epithelial tension controls intestinal cell extrusion. Science 389, eadr8753 267 (2025). 268 30. Celotti, M. et al. Protocol to create isogenic disease models from adult stem cell-derived 269 organoids using next-generation CRISPR tools. STAR Protoc. 5, 103189 (2024). 270 31. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for 271 RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 272 273 274 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Acknowledgments 275 Research reported in this publication was supported by the European Research Council under an 276 ERC starting grant agreement no. 850554 and by ZonMW under PSIDER grant number 277 10250022120003 (K.F.S, T.N. and L.S.) and the Hubrecht Institute . We thank the Hubrecht 278 imaging center for the maintenance of equipment. We thank Bas van Rijn for providing tissue to 279 establish endometrial organoid lines. We thank Daniel Krueger for providing the Lifeact -GFP, 280 Lifeact-mScarlet and H2B-iRFP expression vectors. We thank Yasuhiro Takashima for providing 281 us with the H9 cell line (WiCell Research Institute). 282 283 Author contributions 284 T.N, M.C, H.C. and K. F.S. conceived the study; H.C. and K. F.S. supervised the project; J.v.E. 285 and K.F.S. daily supervision; T.N and M.C. designed and performed the experiments; L.S., R.v.E., 286 A.M.S. analysed data; H.E, F.d.J., H.B., J.K., E.B., G.S. helped with experiments; N.R. hosted 287 T.N. to learn how to culture blastoids; T.N, M.C, H.C. and K.F.S. wrote the manuscript. 288 289 Conflict of interest 290 H.C. is inventor on several patents related to organoid technology. H.C.’s full disclosure is given 291 at https://www.uu.nl/staff/JCClevers/. 292 293 294

295 Ethical approval 296 The human embryonic stem cell line used in this study (H9W1) was previously derived and 297 distributed with informed consent for research use, including genetic modification and 298 differentiation. Fluorescent reporter lines were generated from the parental H9W1 line by random 299 genomic integration of transgenes, in accordance with the approved protocol. This study does not 300 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint involve the derivation of new human embryonic stem cell lines, the use of any newly obtained 301 human biological samples, fetal or embryonic tissues, or donated human blastocysts. 302 Furthermore, no in utero transfer of human -derived cells, tissues or embryo models into any 303 species was performed. All research activities were conducted in full compliance with the ethical 304 regulations of the Hubrecht Institute, Dutch legislation governing human embryonic stem cell 305 research and the International Society for Stem Cell Research (ISSCR) Guidelines for Stem Cell 306 Research and Clinical Translation (2021). 307 308 Endometrial organoids culture 309 Human endometrial organoids used in this study were established from tissue biopsies as 310 previously described and cryopreserved15. Established Organoids were maintained in expansion 311 medium that contains: advanced DMEM -F12 medium (adDMEM/F12; Life Technologies), 312 supplemented with 1 × GlutaMAX (11574466, ThermoFisher Scientific), 100 U/mL Penicillin –313 streptomycin (11548876, ThermoFisher Scientific), and 10 mM HEPES (11560496, 314 ThermoFisher Scientific) [referred hereafter as AdDMEM+++]. AdDMEM+++, 2% final volume 315 B27 Supplement (11530536, ThermoFisher Scientific), 1.25 mM N-acetylcysteine (A9165, Sigma-316 Aldrich), 10mM Nicotinamide (N0636, Sigma -Aldrich), 0.25% Noggin conditioned medium (U -317 Protein Express), 10% R -Spondin 1 conditioned medium (produced as described in 318 Pleguezuelos-Manzano et al.28), 50 ng/ml EGF (AF-100-15, Peprotech), 100 ng/mL FGF10 (100-319 26, Peprotech), 1 µM A83 -01 (2939, Tocris), 1 mM Prostaglandin E2 (2296, Tocris), Wnt3A 320 surrogate (U-Protein Express), 10 mM ROCK inhibitor Y-27632 (M1817, Ab- mole) and 10 mg/mL 321 Primocin (ant-pm-1, Invivogen). Organoids were maintained into a 37 °C, 5% CO2 incubator. The 322 medium was changed every two to three days. Every 7 days organoids were dissociated through 323 mechanical dissociation and passaged with a ratio 1:4. 324 325 2D culture and differentiation 326 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Organoids were incubated in dispase for 30 min. After that, they were harvested and washed 327 using AdDMEM+++. The organoid pellet was resuspended in 1 ml TrypLE Express (11568856, 328 Thermo Fisher Scientific) and incubated at 37 °C for 5 min. This was followed by mechanical 329 dissociation with P1000 until organoids were disrupted into single cells. These steps were 330 repeated if required. Subsequently, 200,000 cells were seeded in 200 µL human endometrium 331 expansion medium in an Ibidi well (8-well ibitreat high, Ibidi) on top of a gel mix of collagen (70%) 332 (PureCol EZ Gel, Advanced Biomatrix) and Matrigel (30%). After 3 days, when cells had reached 333 at least 70% confluency, differentiation was started: expansion medium was supplied with 10nM 334 β-Estradiol (E1127, Sigma Aldrich) for 2 days to mimic the proliferative phase. After these 2 days, 335 the medium was switched to differentiation medium for 5 days to mimic the secretory phase. 336 Human endometrium differentiation medium consists of: AdDMEM+++, 2% final volume B27 337 Supplement (11530536, ThermoFisher Scientific), 1.25 mM N -acetylcysteine (A9165, Sigma -338 Aldrich), 10mM Nicotinamide (N0636, Sigma -Aldrich), 0.25% Noggin conditioned medium (U -339 Protein Express), 50 ng/ml EGF (AF-100-15, Peprotech), 100 ng/mL FGF10 (100-26, Peprotech), 340 1 µM A83 -01 (2939, Tocris), 1mM Prostaglandin E2 (2296, Tocris), 1 µM cyclic AMP (B7880, 341 Sigma Aldrich) , 200 ng/ml Progesterone (P0130, Sigma Aldrich) , 10nM β-Estradiol (E1127, 342 Sigma Aldrich) and 10 mg/mL Primocin (ant-pm-1, Invivogen). 343 344 Generation of stable genetically modified organoids 345 To construct the Lifeact -mScarlet-Puro plasmid, the mScarlet coding sequence was amplified 346 from Addgene plasmid #190658, with the Lifeact peptide sequence (MGVADLIKKFESISKEE) 347 incorporated into the forward primer and inserted into the vector backbone (Addgene #240804) 348 digested with XhoI and SmaI (Promega). The mT2TP transposase and H2B-iRFP-Blast plasmids 349 were generated as previously described29. To establish the Lifeact organoid line, 5 µg of mT2TP 350 transposase plasmid was co-transfected with 5 µg of Lifeact-mScarlet-Puro plasmid. Similarly, for 351

of H2B -iRFP into organoid lines, 5 µg of mT2TP transposase plasmid was co -352 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint transfected with 5 µg of H2B-iRFP-Blast plasmid. Organoids were dissociated into single cells as 353 described above and electroporated in cuvettes (BTX) using NEPA electroporator (Nepa Gene). 354 For selection of Life act-expressing organoids, cultures were treated with Puromycin (2 µg/mL; 355 Invivogen). For selection of H2B -expressing organoids, Blasticidin (100 µg/mL; Invivogen) was 356 used instead. Regarding the generation of Knockouts, t he empty sgRNA plasmid backbone for 357 SpCas9 was a kind gift from Keith Joung (BPK1520, Addgene plasmid #65777). MFSD2A W134* 358 sgRNA sequence: 3’ -CACGAACCAGATGAGGAAGT-5’. MFSD2A sgRNA was cloned as 359 previously described 30. Organoids were engineered using a C>T NGG base editor plasmid 360 (SpCas9-CBE6b, Addgene plasmid #215280 ), and cells were electroporated as previously 361 described30. One week after electroporation edited organoids were selected with addition of 362 Hygromycin B Gold (100 µg/ml; Invivogen) to the medium . One week after selection, surviving 363 organoids were picked and dissociated to single cells, and clonal lines were generated. To confirm 364 the gene editing , DNA was extracted using Quick -DNA Microprep Kit (Zymo Research 365 Corporation) and genotyped using gene -specific primer pairs . Forward 5’ -366 TCCCAGTTCCCATCTGCCAT-3', Reverse 5’ -CACGATAGGCGGTGGCAGAA-3' (sanger 367 sequencing performed by Macrogen Europe BV). Successfully edited clonal lines were passed 368 and cryopreserved. 369 370 hESC culture 371 Human Naïve Embryonic Stem Cell lines H9W1 , containing a Dox -inducible Gata6 expression 372 vector, were kindly provided by Yasuhiro Takashima. In none of the experiments of this study, 373 Gata6 expression was induced. Cells were cultured in PXGL medium, consisting of N2B27 basal 374 medium including DMEM/F12 (50%, Gibco), neurobasal medium (50%, Gibco), N-2 supplement 375 (Thermo Fisher Science, 17502048), B -27 supplement (Thermo Fisher Science, 17504044), 1x 376 GlutaMAX supplement (Thermo Fisher Science, 35050 -038), 1x non-essential amino acid s 377 (11140050, Gibco), 2 -mercaptoethanol (100 µM, Thermo Fisher Science, 31350010) , 1 x 378 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Pen/Strep 5000 Units/ml, supplemented with PD0325901 (1 µM, MedChemExpress, HY-10254), 379 XAV-939 (1 µM, MedChemExpress, HY-15147), Gö 6983 (2 µM, MedChemExpress, HY-13689) 380 and human recombinant leukemia inhibitory factor (hLIF, 10 ng/ml, Stemcell, 78055). Cells were 381 grown on irradiated mouse embryonic fibroblasts ( MEFs, made in house) with a coat of Geltrex 382 (10mg/mL, Gibco, A14133-02) on top, incubated in low oxygen incubators (5% CO2, 5% 02) and 383 passaged every 3 to 4 days as single cells using Accutase (Stemcell, 07920). 384 385 Generation of stable genetically modified and transfection of hESCs 386 Human ESCs were transfected with Lifeact-GFP, H2B-iRFP and H2B-mNeonGreen. To generate 387 the Life act-GFP-Blast plasmid, GFP -coding sequence was amplified from Addgene plasmid 388 #240804 with the coding sequence for Lifeact (MGVADLIKKFESISKEE) contained in the forward 389 primer and inserted into the vector plasmid (Addgene #240803) digested with XhoI (Promega) 390 and NotI (Promega). The mT2TP transposase and H2B-iRFP-Blast plasmids were generated as 391 previously described 29. H2B-mNeonGreen was obtained from Addgene (H2B-mNeonGreen-392 IRESpuro2 was a gift from Daniel Gerlich (Addgene plasmid #183745). To establish the Lifeact-393 H2B ESCs, 5 µg of mT2TP transposase plasmid was co-transfected with 5 µg of Lifeact-mScarlet-394 Puro plasmid and 5 µg of H2B-iRFP-Blast plasmid. Similarly, for introduction of H2B-mNeonGreen, 395 5 µg of mT2TP transposase plasmid was co -transfected with 5 µg of H2B -mNeonGreen-Blast 396 plasmid. 500.000 hESCs were passaged as single cells, washed twice with OptiMEM (Gibco, 397 12559099), and transfected using the NEPA21 electroporat or (Nepa Gene) . Cells were 398 immediately transferred to PXGL medium with 10 µM Y27632 and incubated on a 6 -well plate 399 with MEF s and Geltrex. 2 days later, cells were either selected using Puromycin (2 µg/mL; 400 Invivogen), Blasticidin (100 µg/mL; Invivogen) or sorted using FACS to create polyclonal lines . 401 Lines were checked for their fluorescent reporter expression every week by checking their 402 fluorescence under the fluorescent microscope . Lines were eventually FACS sorted on their 403 fluorescence expression to exclude negative cells that have silenced the construct. 404 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint 405 Blastoid generation and expanded culture 406 hESCs and underlying MEFs were dissociated by incubation with Accutase (Stemcell, 07920) for 407 5 minutes at 37 °C followed and mechanical separation to create single cells. To exclude MEFs, 408 one well of dissociated cells was cultured on a gelatin-coated, equally sized well for 60 minutes 409 in PXGL medium. Hereafter, supernatant containing the stem cells was removed, spun down and 410 put in N2B27 basal medium containing 10 µM Y27632 (Stemcell, 72304) and counted. The cells 411 were seeded in Aggrewell microwell plates (Stemcell, 34415). Per well, 54.000 cells were seeded 412 into 500 µL N2B27 basal medium with 10µM Y27632 and incubated for 24 hours at 5% O2. Then, 413 medium was replaced by 2 mL N2B27 PALLY medium containing PD0325901(1 µM), A83-01 414 (1 µM, MedChemExpress, HY-10432), 1-oleoyl lysophosphatidic acid sodium salt (LPA) (5 µM, 415 Tocris, 3854), hLIF (10 ng ml−1) and Y-27632 (10 µM). 48 hours later, the medium was replaced 416 with N2B27 + LPA and after another 48 hours the blastoids were formed. For expanded culture 417 without endometrium, blastoids were put on a 9.4 cm non-adherent petri dish (Greiner, 632180) 418 in 4 mL IVC medium , consisting of adDMEM/F12 Life Technologies), supplemented with 419 1 × GlutaMAX (11574466, ThermoFisher Scientific), 20% FBS ( A5256701, ThermoFisher 420 Scientific), 1X ITS-X (Gibco, 10524233), NAC (Sigma-Aldrich, A9165-5G), 1x HEPES (11560496, 421 ThermoFisher Scientific), 1x non-essential amino acids (11140050, Gibco) , 100 U/mL Penicillin–422 Streptomycin (11548876, ThermoFisher Scientific) and 0,22% sodium lactate ( 041529.AK, 423 ThermoFisher Scientific). Blastoids were cultured for an additional 2 days in the incubator on a 424 shaker (20 rpm) to prevent them from sticking to the bottom or to each other. Then, blastoids were 425 used for imaging or gene expression analysis. 426 427 Co-culture of endometrium and blastoids 428 Blastoids and endometrium were co -cultured in misc. medium. To prevent an osmotic shock, 429 blastoids were slowly acclimatized to FBS by increasing the concentration by 1% per 10 minutes 430 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint before adding them to the co -culture. Around 5-10 blastoids were cultured within one well of an 431 8-well ibitreat high dish to prevent aggregation and nutrient depletion of the culture medium. Co-432 cultures were either used for live imaging in a controlled environment (5% CO2, 20% O2, 37 °C) 433 or in an incubator with the same environment for 2-3 days. For the extended co-culture of 8 days, 434 IVC medium was refreshed every 2 days. 435 436 qPCR analyses 437 RNA was isolated using the RNAEasy kit (Qiagen) following the manufacturer’s instructions. 438 Reverse transcription was performed on 250-500 ng RNA using the High-Capacity RNA-to-cDNA 439 kit (ThermoFisher). qPCR analysis was performed using SYBR green (Bio -Rad) on a CFX384 440 Touch Real-Time PCR detection system (Bio -Rad). qPCR data were analyzed using excel and 441 plotted in GraphPad Prism. 442 443 Histology 444 Organoids were dissociated from the Basement Membrane Extract (BME) by washing with 10 mL 445 of ice -cold AdDMEM+++ per well, followed by centrifugation at 500×g for 5 min. Pelleted 446 organoids were fixed in 10% formalin for at least 30 min, then processed for paraffin embedding 447 through serial incubations in 70%, 96%, and 100% ethanol, xylene, and liquid paraffin. Sections 448 of 4 µm thickness were cut, rehydrated, and subjected to hematoxylin and eosin (H&E), periodic 449 acid–Schiff (PAS), or immunohistochemical staining. For immunohistochemistry, antigen retrieval 450 was performed according to the antibody manufacturer’s instructions, followed by blocking in 1% 451 bovine serum albumin (BSA; MP Biomedicals, 160069) in PBS. Primary antibodies included anti-452 ACTUB (sc-23950, Santa Cruz ), anti-PAEP (HPA020108, Sigma A ldrich), anti-MFSD2A (PA5-453 21049, ThermoFisher Scientific) and anti-ASCT2 (ab237704, Abcam) applied overnight at 4 °C. 454 After three PBS washes, sections were incubated with secondary antibody (rabbit anti -goat; 455 Southern Biotech, 6160-01) for 1 h at room temperature, followed by three additional PBS washes. 456 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Signal detection was performed using BrightVision poly-HRP anti-rabbit (Agilent, K400311-2) or 457 poly-HRP anti-mouse (Agilent, K400111-2) and developed with 3,3′-diaminobenzidine (DAB) for 458 10 min. Finally, sections were dehydrated and mounted using Pertex®. Images were acquired 459 with a DM4000 optical microscope (Leica) and processed using Olyvia software. 460 461 Immunofluorescence 462 Fixation of the 2D endometrium and the blastoids was performed by adding 4% PFA 463 (paraformaldehyde, Roth, 4979.1) for 15 minutes, followed by 1 hour incubation with 0,1%Triton 464 X-100 (ThermoFisher, HFH10) + 5% BSA (Sigma-Aldrich, 05470 -5G) in PBS on RT. 465 Subsequently, primary antibodies were added in a solution of 0,01%T riton X-100 + 2% BSA in 466 PBS, and the samples were incubated overnight at 4°C. The next day, secondary antibodies were 467 added and incubated for 2 hours at room temperature. Then, samples were washed with PBS 468 and kept at 4°C until imaging. Samples were imaged using the Leica Stellaris 8 White Light laser 469 system, either with a 93x glycerol or a 20x dry objective. 470 471 List of antibodies used 472 Antibody/fluorescent molecule Catalog number dilution Ac-Tub sc-23950 1:200 MFSD2A PA5-21049, ThermoFisher Scientific 1:100 ASCT2/SLC1A5 Ab237704, Abcam 1:100 Phalloidin 555 8953, Cell signaling 1:1000 DAPI Sigma 5µg/mL PAEP PA020108, Sigma Aldrich 1:100 473 Live imaging 474 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Live imaging of blastoids and endometrial co -culture was performed using a Leica Stellaris 8 475 White Light laser system. Imaging was performed in a humid atmosphere at 5% CO2, 20% O2 476 and 37°C. 488, 569 and 682 nm lasers were used for eGFP/mNeonGreen, m-Scarlet and iRFP, 477 respectively, and images were either taken with a 93x glycerol or a 20x dry objective. For imaging, 478 a field of view was chosen to focus on the interface of the endometrial epithelium and the blastoid. 479 Z-Stacks of approximately 30 µm were taken with an increment of 1µm at 1024 x 102 4 pixels 480 every 30 minutes. 481 482 Image data analyses 483 To measure the height of the endometrial epithelial cells, Z-stacks of differentiated 2D endometrial 484 layers were taken with an increment of 1 µm. With the Fiji plugin orthogonal views, we generated 485 a cross section. The distance from the basal to the apical side within individual cells was 486 measured by drawing a line between the two sides in the middle of the cell. To quantify the H2B 487 expression in blastoid and endometrial nuclei, t he fluorescent intensity of H2B -iRFP and H2B -488 mNeon in nuclei from blastoids and endometrium were quantified by generating a maximum 489 intensity projection, drawing a line through the cells covering the nucleus and measuring the 490 intensity. 491 Surface area of attached blastoids was measured by manually defining a region of interest at the 492 contact site of the endometrium and blastoids, surrounding the entire blastoid structure. 493 Subsequently, the area was measured in Fiji. 494 495 Expression of SLC1A5 and MFSD2A in in vivo endometrium. 496 To make the dot plots of in vivo fusion receptor expression, single cell RNAseq data from the non-497 pregnant uterus was downloaded from the reproductive cell atlas 25 498 (https://www.reproductivecellatlas.org/non-pregnant-uterus.html). The dot plots were made using 499 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Scanpy's (v1.11.4) built-in dotplot function grouping the cells by "Broad cell type" as annotated by 500 the creators of the dataset. 501 502 Bulk mRNA sequencing 503 RNA was isolated from 2D endometrial organoids, as done for qPCR analyses. Bulk mRNA 504 sequencing was performed by Single Cell Discoveries (Utrecht, Netherlands). Briefly, polyA -505 enriched RNA was reverse transcribed and sequenced on an Illumina NextSeq500. Analysis of 506 bulk RNA-seq samples was performed in R using DESeq2 (version 1.26.0)31. 507 508 Statistics 509 Across all figures, error bars represent the standard error of the mean (SEM). Unpaired Student’s 510 t-tests were used for significance testing . Statistical analyses and data visualization were 511 performed using GraphPad Prism v10.6.0. The meaning of individual data points is detailed in the 512 respective figure legends. 513 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Main figures and legend .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Figure 1- In vitro implantation setup amenable for high resolution imaging. (a) Schematic of the implantation setup amenable for live imaging. Top: generation of blastoids from Lifeact-GFP (magenta) and H2B-iRFP (yellow) expressing naïve human embryonic stem cells (hESCs). Bottom: generation of a 2D monolayer and differentiation of Life act-mScarlet (cyan) expressing endometrial organoids. (b) Representative image of a 3D Lifeact-mScarlet (cyan) expressing endometrial organoid. Scale bar, 50 µm. (c) Top view of a 2D Life act- mScarlet (cyan) expressing endometrial epithelial layer. Scale bar, 20 µm. (d) Orthogonal view of the 2D Lifeact-mScarlet (cyan) expressing endometrial layer. (e) Dot plot showing epithelial height from the basal to apical membrane (µm). Each dot represents a single cell; data represent mean ± s.e.m. (n = 2). (f) Quantification of ciliated cells (%). Comparison between expansion medium (EM) and differentiation medium (DM). Data represent mean ± s.e.m. (n = 3). (g) Relative expression of endometrial marker genes measured by RT -qPCR: PAEP (secretory cells), TPPP3 (ciliated cells), and PGR (hormone-responsive gene). Comparison between Wildtype and Lifeact-mScarlet organoids cultured in EM or DM. Data represent mean ± s.e.m.; statistical analysis by unpaired t-test, two tailed, ns= non -significant. (n = 3). (h) Bright-field image of naïve hESCs. Fluorescence image showing Lifeact-GFP (magenta) and H2B-iRFP (yellow) expression. Scale bar, 50 µm. (i) Fluorescence images of a Life act-GFP (magenta) and H2B-iRFP (yellow) expressing blastoid, and merged channels. Scale bar, 50 µm. (j) Top view of day 0 of co-culture showing a Lifeact-GFP (magenta) and H2B-iRFP (yellow) expressing blastoid on a 2D Life act-mScarlet (cyan) expressing endometrial layer. Scale bar, 200 µm. (k) Orthogonal view of j. Scale bar, 50 µm. (l) Dot plot quantifying blastoid attachment (%). Each color represents an independent experiment; data represent mean ± s.e.m. (n = 3). (m) Bright-field images of co-cultures after two days, before and after washing. Red asterisks indicate attached blastoids. Scale bar, 500 µm. .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Figure 2- Blastoids fuse with the endometrial epithelium. (a) Representative images of the co-culture at day 2 showing fusion between the endometrium and blastoid. Top: top-view image showing the attached blastoid, surrounded by a giant cell positive for Lifeact-mScarlet (cyan), Lifeact-GFP (magenta) and H2B-iRFP (yellow). Bottom: orthogonal view of the same region, with a zoom -in (white box) highlighting the giant fused cell (surrounded by a yellow -dotted line). Scale bar, 50 µm. (b) Image representing a triple positive cell at the interface of the endometrium and blastoid (circled in yellow). Top panel: blastoid (magenta and yellow). Second panel: endometrium (cyan). Third panel: merged image showing fusion (purple), blastoid (magenta and yellow), and endometrium (cyan). Scale bar, 50 µm. On bottom, schematic illustration of the fusion (purple) between the endometrium (cyan) and blastoid (magenta and yellow) (c) Live imaging (top and orthogonal views) of blastoid-endometrial fusion . Top panels: blastoid (magenta). Middle panels: endometrium (cyan). Bottom panels: merged images showing the fused cells (circled in yellow). Scale bar, 20 µm. each image represents a z -plane of 1 µm. (d) Schematic representation of in vitro blastoid attachment and fusion with the endometrial epithelium. .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Figure 3- Fusion proteins are present in blastoids and in vivo endometrial tissue. (a) Schematic representation of the fusion machinery in the blastocyst, highlighting fusion ligands (Syncytin-2 (ERVFRD-1) and Syncytin-1 (ERVW-1)) and their receptors (MFSD2A and ASCT2 (SLC1A5)). (b) Relative expression of fusion genes ( ERVW-1, ERVFRD-1, SLC1A5 and MFSD2A) in blastoids at day 0 and day 2 after formation, measured by RT -qPCR. Data represent mean ± s.e.m.; statistical analysis by unpaired t-test, two tailed, ns= non-significant, *** (p -value= 0.0004), ** (p -value=0.0014) (n = 3). (c) Relative expression of syncytiotrophoblast marker genes (OVOL1, CGB3 and GCM1) in blastoids at day 0 and day 2 after formation, measured by RT-qPCR. Data represent mean ± s.e.m.; statistical analysis by unpaired t-test two tailed, ** (p -value< 0.005), **** (p -value<0.0005) (n = 3). (d) Fluorescence image of a Lifeact-GFP (magenta) and H2B-iRFP (yellow) blastoid at the polar side, two days after formation, showing a syncytiotrophoblast (circled in yellow). Scale bar, 20 µm. (e) Dot plot showing MFSD2A and SLC1A5 expression in in vivo endometrium. Bottom right: viridis color scale indicating mean expression per cell group. Bottom left: proportion of cells expressing each gene (percentage). Data analyzed from the dataset by García -Alonso et al. (f) Immunohistochemistry of endometri al tissue for ASCT2 and MFSD2A, comparing proliferative and secretory phases. Arrows indicate regions with SLC1A5 expression. Scale bar, 50 µm. .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint Figure 4 - Fusion mediated by MFSD2A is essential for blastoid attachment to the endometrial epithelium. (a) Expression of MFSD2A (normalized counts) in endometrial organoids cultured in expansion medium (EM) or differentiation medium (DM). Data represent mean ± s.e.m.; statistical analysis by unpaired t-test two tailed, * (p -value= 0.0448) (n = 3 , technical replicates). (b) Validation of MFSD2A knockout in endometrial organoids by Sanger sequencing, showing a tryptophan-to-stop codon mutation in exon 4 (W134*). (c) Bright-field images of Wildtype and MFSD2A knockout (KO) endometrial organoids. Scale bar, 1 mm. (d) Relative expression of endometrial marker genes measured by RT -qPCR: PAEP (secretory cell marker), TPPP3 (ciliated cell marker), and PGR (hormone-responsive gene). Comparison between Wildtype and MFSD2A KO organoids cultured in EM or DM. Data represent mean ± s.e.m.; statistical analysis by unpaired t-test, two tailed, ns= non-significant (n = 3). (e) Blastoid attachment efficiency (%) in co -cultures with Wildtype (pink) or MFSD2A KO (green) endometrial organoids. Data represent mean ± s.e.m. Total of 55 b lastoids analyzed for the knockout condition and 71 for wildtype. (n = 2). (f) Bright-field images of co -cultures with MFSD2A KO organoids before and after washing, showing attached blastoids. (g) Schematic model of the proposed fusion mechanism. .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 24, 2025. ; https://doi.org/10.64898/2025.12.22.695952doi: bioRxiv preprint