Stress-inducible phosphoprotein 1 (STIP1) is a critical stemness regulator in mouse embryonic stem cells and early mammalian development

Abstract

Proteostasis, maintained by a network of molecular chaperones, plays a central role in cell biology, and has emerged as a critical mechanism underlying pluripotency and development. The stress-inducible phosphoprotein 1 (STIP1) is a co-chaperone essential for proteostasis. STIP1 knockout causes embryonic lethality in mice, but its precise function during embryogenesis remains poorly understood. Here, we investigate the role of STIP1 in early development using in silico and cell-based approaches. Single-cell RNA sequencing data reveals that Stip1 is co-expressed with pluripotency genes in mouse embryos, suggesting a role in stem cell maintenance. To test this, we generated mouse embryonic stem cells (mESCs) from genetically modified mice with altered Stip1/STIP1 expression. STIP1 depletion in mESCs decreases the expression of pluripotency markers, reduces proliferation, and induces apoptosis and genomic instability, whereas its overexpression enhances pluripotency markers expression, promotes proliferation, and confers protection against cellular stress. Moreover, proteins involved in cell cycle progression and DNA damage response are differentially regulated in mESCs, depending on STIP1 levels. Our findings highlight STIP1 as a pivotal regulator of the pluripotent phenotype, early embryogenesis, and cellular resilience, advancing our understanding of proteostasis in stem cell biology and organismal development, with potential implications for disease modeling and regenerative medicine.

Introduction

Embryogenesis is orchestrated by molecular processes that determine cell fate. Since their derivation, embryonic stem cells (ESCs) have been the gold-standard in vitro model for early mammalian development^1,2,3. Within this complex landscape, protein homeostasis (proteostasis) and its regulators have recently emerged as important, albeit underexplored, controllers of ESCs⁴. Proteostasis declines with aging, and its disruption is a hallmark of degenerative diseases^5,6,7. Conversely, ESCs exhibit enhanced proteostasis mechanisms intrinsically linked to pluripotency maintenance^4,8. Pluripotency refers to the sustained undifferentiated state of ESCs, with unlimited self-renewal and potential to generate all somatic cell types⁹. Chaperones and co-chaperones, as key components of proteostasis, play crucial roles in maintaining protein integrity across cell types, including ESCs^10,11,12.

Stress-inducible phosphoprotein 1 (STIP1/STI1/p60), also known as heat-shock organizing protein (HOP), is a co-chaperone that interacts with heat shock proteins of 70 kDa and 90 kDa (HSP70 and HSP90)^{13,14,15,16,17}. STIP1 assists in substrate recognition, transfer, and activation of HSP70/HSP90 client proteins, a role evolutionarily conserved across species^18,19,20,21. Although recent evidence shows that HSP70 and HSP90 can bind independently of STIP1 in human cells²², mouse models indicate that STIP1 downregulation can acutely disturb chaperone activity²³. Nonetheless, increasing evidence suggests STIP1 may have additional functions beyond acting as a co-chaperone. Its versatility is evidenced by its nucleus-cytoplasm shuttling, cell surface localization, and interactions with diverse protein complexes in multiple compartments²⁴. For example, STIP1 forms an extracellular signaling complex with the cellular prion protein, modulating neural development and plasticity^25,26,27,28 and cancer stem cell proliferation and migration^29,30. Moreover, STIP1 expression varies with and across tumor types³¹, suggesting complex regulation according to cell state.

STIP1 knockout (KO) results in embryo degeneration around embryonic day 7 to 10.5 (E7-E10.5)³². KO embryos display impaired limb budding, neural tube defects, and reduced Mendelian frequency at E3.5, indicating a previously unexplored function for STIP1 in early development, particularly in pre-implantation³². These findings align with human genetic data, showing that STIP1 loss-of-function is poorly tolerated²³.

Although prior studies implicate STIP1 in mammalian development and stemness, its underlying molecular mechanisms remain unclear. Here, we combined in-depth bioinformatics analysis of pre- and post-implantation mouse embryos with in vitro studies in mouse ESCs (mESCs) expressing varying STIP1 levels to elucidate its developmental role. We demonstrate that regular STIP1 levels and activity are critical for maintaining the expression of pluripotency factors, self-renewal, proliferation, and differentiation potential in mESCs. Moreover, STIP1 influences essential pathways such as cell cycle progression and DNA damage response. Our results implicate STIP1 and the HSP70/HSP90 machinery as regulators of early development via proteostasis control of stemness-related pathways.

Results

Stip1 is co-expressed with pluripotency transcription factors in the blastocyst

Our group previously showed that STIP1 KO is lethal in mice, with embryos presenting prominent neural tube degeneration, establishing a crucial role for STIP1 in early development³². To better understand Stip1 gene expression dynamics during mouse development, we mined publicly available single-cell RNA sequencing (scRNA-seq) data from embryos at E6.5 to E8.5³³. Stip1 was broadly expressed across all identified cell clusters, though slightly enriched in the notochord (Supplementary Fig. 1A). This subtle increase was reproduced in a second scRNA-seq dataset from E8.25 embryos³⁴ (Supplementary Fig. 1B). Our previous data also indicated that Stip1 functions as a maternal-effect gene, suggesting that maternally inherited STIP1 may be required in early embryos before zygotic gene activation, consistent with the observation that STIP1 KO embryos occur at significantly lower-than-expected Mendelian frequency than WT controls³². These findings prompted us to investigate Stip1 dynamics during pre-implantation by exploring scRNA-seq data from zygote to late blastocyst stages³⁵.

Cell clusters were identified using gene signatures from the original dataset and visualized by principal component analysis (PCA) (Fig. 1A). Stip1 gradually increased from the zygote stage, peaking at 8- and 16-cell stages, followed by a slight decrease during transition to the blastocyst (Fig. 1B).

Fig. 1: Expression profile of Stip1 and associated genes during early mouse development evidenced by single-cell RNA sequencing.

A Cells from the zygote to the late blastocyst projected onto a principal component analysis (PCA) plot and colored according to their developmental stage. B Violin plots with boxplot display Stip1 expression (normalized to zygote levels) across pre-implantation stages. Wilcoxon test, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001 for specific stages relative to all others. Boxplots depict the quartiles and extend from the minimum and maximum data points within 1.5 times the interquartile range. C PCA shown in (A), colored to indicate the scaled expression of Stip1, Sox2, Pou5f1/Oct4, Nanog and Klf4. D Pearson correlation plot highlighting genes that are positively (red) or negatively (blue) correlated with Stip1, while non-significant correlations are shown in light gray. E Gene Ontology overrepresentation analysis of genes positively (red) and negatively (blue) correlated with Stip1. F Pearson’s correlation coefficient between Stip1 and key pluripotency factors: Pou5f1 (R = 0.41), Sox2 (R = 0.23) and Nanog (R = 0.47).

Remarkably, Stip1 expression correlates with the expression of the key pluripotency transcription factors Sox2, Pou5f1/Oct4, Nanog and Klf4, especially during early- and intermediate blastocyst stages (Fig. 1C; Supplementary Fig. 1C). Stip1 expression was also significantly correlated with heat shock protein genes, including its main partner HSP90 (Hsp90aa1, Hsp90ab1, Supplementary Fig. 1D). Moreover, Pearson correlation analysis specifically in the blastocyst revealed that expression of numerous genes was either positively or negatively correlated with Stip1 (Fig. 1D; Supplementary Data 1). Overrepresentation analysis (ORA) of these genes showed enrichment for DNA and cell cycle-related terms, including DNA repair, DNA replication, histone modification, G1/S and G2/M transitions, as well as stem cell maintenance, proliferation, and p53 signaling among positively correlated genes (Fig. 1E; Supplementary Data 1). Proteasomal protein catabolism, Wnt signaling, and cell fate commitment were enriched among both positively and negatively correlated genes, suggesting that Stip1 may exert bidirectional regulatory effects on these pathways (Fig. 1E; Supplementary Data 1). Core pluripotency genes also exhibited marked positive correlation with Stip1, with Nanog among the top 20 rank (Fig. 1D, F; Supplementary Data 1). Altogether, these findings suggest a link between Stip1 expression and pluripotency maintenance, offering initial insight into its finely tuned regulation during pre-implantation development, correlating with key developmental genes and pathways.

Stip1 decrease and overexpression induce distinct transcriptional profiles in mESCs

To expand the scRNA-seq findings, we derived mESC lines from blastocysts of genetically modified mice with differential Stip1/Stip1 expression levels^23,32,36. The 16-day derivation process involved in vitro maturation of blastocysts and isolation of the first passage of mESCs from the inner cell mass (Supplementary Figs. 2A, B). A total of 33 mESC lines were established (Supplementary Table 1): wild-type (WT) mESCs, exhibiting regular STIP1 expression levels; STIP1^+/− mESCs, harboring a heterozygous KO of Stip1 gene, resulting in a 50% reduction in STIP1 expression compared to WT; and STIP1^TgA mESCs, a BAC transgenic line carrying 20 extra Stip1 copies, resulting in Stip1/STIP1 overexpression (Fig. 2A). These cell lines were cultivated in vitro, and pluripotency was validated by assessing the expression of germ layers markers after stochastic differentiation (Supplementary Fig. 2C).

Fig. 2: STIP1 modulation alters the transcriptional profile of mESCs.

A Schematic representation showing wild-type (WT), STIP1^+/− (50% expression) and STIP1^TgA (overexpression) mESCs lines derived from the inner cell mass of the blastocyst of genetically modified Stip1 mice. Total RNA extracted from these mESCs was subjected to bulk RNA sequencing followed by functional profiling. B Volcano plot of differentially expressed genes (DEGs) in STIP1^TgA versus (vs) WT. DEGs are classified as non-significant (gray), up- (red) or downregulated (blue), with genes of interest highlighted, p < 0.05. C Overrepresentation analysis (ORA) of Gene Ontology (GO) terms enriched among the DEGs of STIP1^TgA mESCs. D Volcano plot of non-significant (gray), up- (red) or downregulated (blue) DEGs of STIP1^+/− vs WT, p < 0.05. E ORA of GO terms enriched among the DEGs of STIP1^+/−, sorted by p-value. F Volcano plot of non-significant (gray), up- (red) and downregulated (blue) DEGs of STIP1^TgA vs STIP1^+/−, p < 0.05. G ORA of GO terms enriched among the DEGs of STIP1^TgA relative to STIP1^+/−, sorted by p-value.

To determine the cellular and molecular consequences of altered Stip1 levels, we performed bulk RNA sequencing (RNA-seq) on cultured mESCs (Fig. 2A). Differential expression analysis revealed broad transcriptional changes in STIP1^TgA cells compared to WT counterparts (Fig. 2B, Supplementary Data 2). As expected, Stip1 was significantly upregulated among the differentially expressed genes (DEGs) in STIP1^TgA cells, confirming the overexpression. Functional profiling revealed enrichment in pathways associated with cell cycle (regulation of cell cycle arrest, p53-mediated signal transduction, G1/S phase transition, G1 DNA damage checkpoint), apoptotic signaling, DNA damage response, and cell fate determination in STIP1^TgA mESCs (Fig. 2C, Supplementary Data 2). Conversely, Stip1 expression was downregulated in STIP1^+/− mESCs, along with pathways associated with development (reproductive system and placenta, epithelial cell migration, trophectodermal differentiation) and stress response mechanisms (stress-activated MAPK cascade, oxidative stress) (Fig. 2D, E, Supplementary Data 3).

To explore pathways bidirectionally regulated by STIP1, we compared STIP1^TgA versus (vs) STIP1^+/− mESCs. As expected Stip1 was upregulated in very significant levels (Fig. 2F, Supplementary Data 4). Notably, HSP70 isoforms Hspa1a and Hspa1b were also upregulated (Fig. 2F, Supplementary Data 4). Again, we found enrichment in cell cycle regulation and development pathways (Fig. 2G, Supplementary Data 4). Interestingly, neural tube development was also upregulated, which is in line with the result of STIP1 KO embryos showing impaired neural tube formation³². These findings strongly establish an association between Stip1 levels and several developmental processes.

To assess the correlation between STIP1 levels in mESCs and the expression of early developmental stage markers, we compared the RNA-seq results with public scRNA-seq data from pre-implantation³⁵ and post-implantation³³ mouse embryos, which were previously analyzed in our study. After retrieving the gene signatures of the clusters depicted in Fig. 2A, we compared them with our DEGs in STIP1^+/− and STIP1^TgA mESCs, analyzing their similarity through a Jaccard index (Supplementary Fig. 2D). DEGs in both cell lines presented similarity with all stages of pre-implantation, with a higher prevalence in the mid- and late blastocyst stages (Supplementary Fig. 2D). Particularly, STIP1^+/− DEGs exhibited an increased similarity with the late blastocyst stage, while STIP1^TgA mESC profile was closer to the mid-blastocyst stage (Supplementary Fig. 2D). This differential distribution might suggest that STIP1 influences the fine-tuned timing of gene expression changes during early development, potentially modulating cell state transitions. Regarding the post-implantation stage, it was particularly interesting to observe that the DEGs from STIP1^+/− mESCs did not align with any cell clusters identified during mouse gastrulation and early organogenesis. This finding is consistent with the downregulation of pathways related to developmental processes in this group. Conversely, DEGs from STIP1^TgA mESCs aligned with clusters originating from different embryonic lineages, including extraembryonic (ExE) tissues (Supplementary Fig. 2E), suggesting that increased STIP1 levels enhance differentiation potential and may broaden lineage competency.

Stip1 levels modulate the stemness phenotype of mESCs

To further characterize the impact of the transcriptional changes on the pluripotent phenotype, we validated the expression of Stip1 and pluripotency markers in WT, STIP1^+/− and STIP1^TgA mESCs in vitro (Fig. 3A). In agreement with our previous findings, Stip1 gene expression levels decreased in STIP1^+/− and increased in STIP1^TgA mESCs. Additionally, STIP1^+/− exhibited reduced Sox2 expression, while STIP1^TgA showed elevated expression of the pluripotency markers Pou5f1, Sox2, Klf4, Nanog, and Tfc2pl1 (Fig. 3A). Western blot analysis confirmed decreased expression of STIP1 in STIP1^+/− and a significant increase in STIP1, OCT4, SOX2 and NANOG in STIP1^TgA relative to WT and STIP1^+/− (Fig. 3B, C). In accordance, STIP1^+/− mESCs colonies presented decreased levels of the pluripotency marker alkaline phosphatase (AP), as shown by the increased proportion of AP negative (AP−) colonies in this population (Supplementary Figs. 3A, B), as well as a reduction of AP positive (AP+) colonies relative to STIP1^TgA (Supplementary Fig. 3C). Additionally, STIP1^TgA display an increase in AP+ colonies compared also to WT mESCs (Supplementary Fig. 3C). Moreover, although the localization of pluripotency transcription factors did not change in any of the studied mESCs lines (Fig. 3D), we observed a pronounced decrease in NANOG colocalizing with mitotic figures in all studied cell lines (Supplementary Fig. 3D, arrowheads). In addition, some, but not all, STIP1^+/− mESCs colonies exhibited preferential nuclear localization of STIP1 (Supplementary Fig. 3E). Furthermore, the pluripotency marker stage-specific embryonic antigen-1 (SSEA1), located on mESCs membrane, presented a marked “punctate” pattern in STIP1^+/− that was not observed in controls and STIP1^TgA cell lines (Fig. 3E).

Fig. 3: Modulation of Stip1 levels alters stemness phenotypes of mESCs.

A Gene expression (relative to WT) by RT-qPCR (2^-ΔΔCt method, Tbp and Gapdh as reference genes). One-way ANOVA, with Tukey’s post-hoc. #: p-values of STIP1^+/− or STIP1^TgA vs WT; *: p-values of STIP1^+/− vs STIP1^TgA. Stip1 (^#p = 0.0016, ***p = 0.0008), Pou5f1/Oct4 (*p = 0.0187), Sox2 (^#p = 0.0150, **p = 0.0011), Klf4 (^###p = 0.0004, ***p = 0.0002), Nanog (^####p < 0.0001 ****p < 0.0001), and Tfc2pl1 (^###p = 0.0008, ***p = 0.0004). Mean ± standard deviation (SD), n = 3. B Representative western blot image for WT, STIP1^+/−, and STIP1^TgA mESCs genetically independent biological triplicates. C Protein expression (relative to WT) of STIP1 (*p = 0.0379, **p = 0.0030, ***p = 0.0003), HSP70, OCT4 (*p = 0,0295, **p = 0.0050), SOX2 (*p = 0.0173, **p = 0.0036) and NANOG (*p = 0.0198, **p = 0.0047), normalized by β-tubulin. One-way ANOVA, with Tukey’s post-hoc. Mean ± SD, n = 3. D Representative immunofluorescence panels of cells stained with STIP1 (green), SOX2 (yellow), and OCT4 (red). Merged images in the panel include DAPI-stained nuclei (blue). Scale bar = 10 µm. E Representative immunofluorescence panels of cells stained with DAPI/Nuclei (red) and SSEA1 (green). Scale bar = 10 µm. F Representative immunofluorescence panels of cells stained with DAPI/Nuclei (red) and Ki67 (green). Scale bar = 10 µm. G Percentage of Ki67 positive (Ki67+) mESCs. Welch’s ANOVA test with Dunnett’s T3 multiple comparisons test of WT (n = 11), STIP1^+/− (n = 12) and STIP1^TgA (n = 14), *: p-values of STIP1^+/− or STIP1^TgA vs WT; ^#: STIP1^+/− vs STIP1^TgA, **p = 0.0068, ****p < 0.0001, ^####p < 0.0001. Mean ± SD. H Growth curve of WT, STIP1^+/− and STIP1^TgA mESCs. One-way ANOVA with Tukey’s post-hoc analysis was performed for each time-point. *: p-values of STIP1^+/− or STIP1^TgA vs WT; #: p-values of STIP1^+/− vs STIP1^TgA. 24 h: p = non-significant; 48 h: *p ≤ 0.0330, ^#p < 0.0001; 72 h: *p = 0.0340, **p = 0.0062, ^#p < 0.0001; 96 h: *p = 0.0102, **p = 0.0035, ^#p < 0.0001. Mean ± SD, n = 6. I Representative immunofluorescence panels of cells stained with DAPI/Nuclei (blue) and CC3 (red). Scale bar = 10 µm. J Quantification of the percentage of CC3 positive (CC3+) cells. Welch’s ANOVA test with Dunnett’s T3 multiple comparisons test of WT (n = 29), STIP1^+/− (n = 28) and STIP1^TgA (n = 8). *p ≤ 0.0251, ****p < 0,0001, mean ± SD. K Representative immunofluorescence panels of cells stained with DAPI/Nuclei (blue) and pH2AX (green). Scale bar = 10 µm. L Quantification of the number of pH2AX foci each 10 µm. Welch’s ANOVA test with Dunnett’s T3 multiple comparisons test of WT (n = 17), STIP1^+/− (n = 12) and STIP1^TgA (n = 11). *p = 0.0210; **p = 0.0019, mean ± SD.

We investigated additional characteristics important for stemness maintenance^37,38,39. While STIP1^+/− mESCs exhibited reduced proliferation, STIP1^TgA exhibited a marked increase in proliferation relative to STIP1^+/− and WT controls (Fig. 3F–H). Consistently, STIP1^+/− cells displayed increased apoptosis compared to WT and STIP1^TgA (Fig. 3I, J). In contrast, STIP1^TgA mESCs presented reduced expression of the apoptosis marker cleaved caspase-3 (CC3) when compared to both WT and STIP1^+/− (Fig. 3I, J). To determine whether the observed decrease (STIP1^+/−) or increase (STIP1^TgA) in pluripotency factor expression (Fig. 3A–C) was a consequence of concomitant changes in apoptosis, we further analyzed CC3 expression at a single-cell level within colonies, correlating it with the expression of the pluripotency factors OCT4 and SOX2. We classified mESCs into four groups based on expression thresholds: OCT4/SOX2^High or OCT4/SOX2^Low and CC3^High or CC3^Low. The majority of WT and STIP1^+/− mESCs, stained for OCT4, fell into the OCT4^High-CC3^Low category, suggesting that pluripotent cells generally exhibit low apoptosis levels (Supplementary Figs. 3F, G). However, we also identified a small subpopulation of OCT4^High-CC3^High cells, demonstrating that cells undergoing apoptosis can still maintain high OCT4 expression (Supplementary Figs. 3F, G). These results suggest that while low apoptosis is commonly associated with pluripotency, high pluripotency factor expression does not strictly correlate with low apoptosis. We performed a similar analysis with the STIP1^TgA group, using SOX2 as a marker. Notably, this group exhibited a higher proportion of SOX2^Low-CC3^Low cells (Supplementary Fig. 3H). This finding suggests that, at the single-cell level, STIP1 overexpression may shift cells into an alternative state in which certain pluripotency markers decrease without a concomitant increase in apoptosis. Conversely, STIP1^TgA cells also displayed a higher proportion of SOX2^High-CC3^High cells, indicating that under conditions of elevated STIP1, cells undergoing apoptosis can still retain high pluripotency factor expression (Supplementary Fig. 3H).

Moreover, STIP1^+/− cells displayed a higher number of phospho-H2A.X (pH2AX) foci per nucleus, indicating increased spontaneous DNA double-stranded break points, even in the absence of any DNA damage-inducing treatments. On the other hand, STIP1^TgA cells presented markedly reduced levels of pH2AX (Fig. 3K, L). Therefore, our in vitro experiments confirmed the association between STIP1 levels and the expression of pluripotency factors as well as the maintenance of stemness-associated pathways, consistent with RNA-seq (single-cell and bulk) results. Our findings further indicate that reduced STIP1 levels impair stemness, diminishing expression of pluripotency markers and proliferation, while enhancing sensitivity to spontaneous DNA damage and apoptosis. Conversely, elevated levels of STIP1 are associated with an opposing phenotype, characterized by a marked reduction in pH2AX and CC3 levels.

Stip1 knockdown and knockout partially recapitulate the STIP1^+/− phenotype

To confirm that the observed phenotypes in our genetically modified mESC lines were due to the modulation of STIP1 expression, we performed STIP1 knockdown (KD) in the commercial mESC line ES-E14TG2a (Supplementary Figs. 4A–D). Consistent with STIP1^+/− cells, STIP1 KD mESCs also exhibited reduced proliferation (Supplementary Fig. 4E, F). Immunofluorescence staining further revealed that STIP1 KD mESCs displayed decreased protein levels of STIP1 and NANOG (Supplementary Fig. 4G–K).

Furthermore, while Stip1 KO is lethal for mouse embryos³², it remained unclear whether pluripotent cells tolerate the complete depletion of this protein. To address this, we generated Stip1 KO in ES-E14TG2a mESCs using the CRISPR/Cas9 system (Supplementary Fig. 4L, M). Stip1 depletion acutely impaired mESCs proliferation and survival after only a few passages, hindering long-term functional experiments, and confirming our previous results in mouse embryos³². To assess the differentiation potential of Stip1 KO cells, we performed teratoma assays in nude mice, following tumor growth for 42 days. Compared to parental cells, Stip1 KO-derived teratomas were significantly smaller in volume (Supplementary Fig. 4N, O) and exhibited only mesodermal differentiation (Supplementary Fig. 4P), indicating impaired pluripotency in the absence of STIP1. These results recapitulate some of the phenotypes observed in STIP1^+/− cells and STIP1 KO mouse embryos, further supporting STIP1’s critical role in maintaining pluripotency and stemness in early development.

Low expression of a hypomorphic Stip1 allele impairs pluripotency

Given that complete KO severely compromises viability while haploinsufficiency (STIP1^+/−) induces prominent phenotypes in mESCs, we derived mESCs from genetically modified mice with a targeted deletion of Stip1 exons 2 and 3 to further examine how acute reduction in STIP1 levels impact stemness. Specifically, exons 2 and 3 encode the TPR1 domain, which is crucial for HSP70 binding^16,17. This deletion leads to reduced expression of a truncated dysfunctional (hypomorphic) STIP1 variant, named here as STIP1^∆TPR1 (Fig. 4A), previously characterized by our group in other cell types^23,32.

Fig. 4: Low levels of truncated-STIP1 lead to changes in transcriptional profile and altered expression of pluripotency markers in mESCs.

A Schematic diagram illustrating the generation of STIP1^∆TPR1 mESCs lines derived from Stip1 genetically modified mice. B Volcano plot showing differentially expressed genes (DEGs) in STIP1^∆TPR1 vs WT, with WT as the control group. Genes are classified as non-significant (gray), up- (red) or downregulated (blue), and DEGs of interest are highlighted. p < 0.05. C Overrepresentation analysis of Gene Ontology terms enriched among the DEGs of STIP1^∆TPR1 mESCs. D Gene expression (relative to WT) by RT-qPCR (2^-ΔΔCt method, Gapdh as reference gene). Student’s t test for each individual gene. Stip1 - exons 7/9 (*p = 0.0189), Stip1 - exons 2/3 (***p = 0.0002), Hsp90β, Pou5f1/Oct4, Sox2 (***p = 0.0001), Nanog, Tfc2pl1, Klf4 (**p = 0.0045), Stat3 and Esrrb (*p = 0.0316) and Ssea1. Mean ± SD, n = 3. E Representative western blot image for WT and STIP1^∆TPR1 mESCs for HSP90, STIP1, the truncated for of STIP1 (STIP1-∆TPR1) and Actin from genetically independent biological triplicates. F Protein expression (relative to WT) by band densitometry of HSP90 and STIP1, normalized by Actin. Student’s t test for each individual protein. **p = 0.0012, *p = 0.0186. Mean ± SD, n = 3. G Representative western blot image for WT and STIP1^∆TPR1 mESCs for OCT4, NANOG, SOX2 and β-tubulin from genetically independent biological triplicates. H Protein expression (relative to WT) by band densitometry of OCT4, NANOG, SOX2, normalized by β-tubulin. Student’s t test for each individual protein. ***p = 0.0006; *p ≤ 0.0204. Mean ± SD, n = 3. I Representative immunofluorescence panels of cells stained with STIP1 (green), OCT4 (red), and SOX2 (yellow). Merged images in the panel include DAPI-stained nuclei (blue). Scale bar = 10 µm. Scale bar = 10 µm. J Total immunofluorescence quantification (in arbitrary units) normalized by WT. Student’s t test for each individual protein relative to control was used for STIP1 and OCT4 expression. STIP1: ***p = 0.0002; OCT4: **p = 0.0020. Mann–Whitney test was used for SOX2 expression. SOX2: *p = 0.0411. Mean ± SD, n = 5. K Representative immunofluorescence panels of cells stained with DAPI/Nuclei (blue) and NANOG (red). Scale bar = 10 µm. Yellow arrowheads: cells in mitotic figures with observed NANOG reduction. L Representative immunofluorescence panels of cells stained with DAPI/Nuclei (red) and SSEA1 (green). Scale bar = 10 µm.

Bulk RNA-seq revealed significant downregulation of pluripotency markers Lefty1 and Lefty2⁴⁰, Klf2⁴¹, Foxo4⁴², and Otx2⁴³ in STIP1^∆TPR1, alongside marked upregulation of Hspa1a and Hspa1b (HSP70 isoforms), and the early differentiation markers Gata4 and Gata6⁴⁴ (Fig. 4B, Supplementary Data 5). Functional profiling showed that, while morphogenesis of epithelial tube and branching structure are downregulated, recapitulating what was observed in KO embryos³², pathways associated with cell cycle regulation (negative regulation of cell cycle, cell cycle arrest) and DNA repair are upregulated in STIP1^∆TPR1 mESCs (Fig. 4C, Supplementary Data 5). Notably, pathways related to proteostasis, including positive regulation of proteolysis, proteasome-mediated ubiquitin-dependent protein catabolic process, and autophagy, were also upregulated in STIP1^∆TPR1 (Fig. 4C, Supplementary Data 5).

STIP1^+/− and STIP1^∆TPR1 mESCs showed distinct transcriptomic profiles since they present different disruptions in Stip1 gene. While STIP1^+/− cells experience a quantitative reduction in wild-type Stip1/STIP1 levels, STIP1^∆TPR1 cells not only have reduced Stip1/STIP1 expression but also produce a truncated protein lacking the TPR1 domain, potentially acting as a dominant-negative mutant. We hypothesize that this structural alteration leads to qualitative changes in molecular interactions, triggering distinct transcriptional responses. Despite this distinction, many molecular pathways are recapitulated between the two mESCs lines (Figs. 2D, E, 4C). We also compared our RNA-seq dataset of STIP1^∆TPR1 cells (Fig. 4B) with scRNA-seq data analyzed previously³³. STIP1^∆TPR1 DEGs were enriched in a broad range of clusters from mouse gastrulation and early organogenesis, including visceral and parietal endoderm, notochord, ExE endoderm and ectoderm, and endothelium (Supplementary Fig. 5A). These findings suggest that alterations in STIP1 levels, whether increase (Supplementary Fig. 2E) or acute decrease of a mutated form (Supplementary Fig. 5A), might modulate differentiation trajectories.

We subsequently focused on assessing the expression of STIP1 and pluripotency markers in STIP1^∆TPR1 mESCs in vitro. As expected, STIP1^∆TPR1 lacked expression of Stip1 exons 2-3 and showed a concomitant reduction in exons 7-9 (Fig. 4D). Additionally, we observed decreased expression of Sox2, Klf4 and Esrrb (Fig. 4D). Moreover, protein levels of HSP90, STIP1, OCT4, SOX2 and NANOG were also reduced in STIP1^∆TPR1 mESCs (Fig. 4E–J). Interestingly, we again observed acute NANOG decrease colocalizing with mitotic figures in both WT and STIP1^∆TPR1 (Fig. 4K, arrowheads). Furthermore, SSEA1 cellular localization was altered in the mESCs membranes to a “punctate” pattern (Fig. 4L), similar to what was observed in STIP1^+/−. STIP1^∆TPR1 also exhibited a reduced number of AP+ colonies, as well as marked increase in AP- colonies compared to WT mESCs (Supplementary Fig. 5B–D). These results emphasize the importance of STIP1 in sustaining the levels of pluripotency factors, highlighting molecular pathways that appear to be consistently enriched with STIP1 modulation, such as cell cycle control and DNA damage response.

STIP1 levels affect the differentiation potential of mESCs

The ability of mESCs to differentiate into cell types of the three embryonic germ layers is an essential hallmark of pluripotency. While LIF + 2i culture conditions preserve the naïve state of pluripotency, LIF-only is discussed to sustain pluripotency but promotes a primed phenotype^45,46,47. To assess whether Stip1 expression varies across the pluripotency spectrum (naïve to primed), we cultured WT, STIP1^∆TPR1, STIP1^+/− and STIP1^TgA mESCs under LIF-only conditions for 1 day and 10 days. After 10 days, STIP1^+/− and STIP1^∆TPR1 mESCs exhibited a marked reduction in Stip1 levels compared to their 1-day counterparts (Supplementary Fig. 6A). This decrease was not significant in WT or STIP1^TgA cells. This progressive Stip1 downregulation in cells with endogenous STIP1 reduction suggests that its expression is modulated during the transition from a naïve to a more primed-like state, exclusive in a context of already low STIP1 expression levels. Together with our previous results, this evidence may suggest that reduced STIP1 impact key signaling pathways required to stabilize the pluripotent phenotype, potentially making cells more vulnerable during naïve to primed transition.

Moreover, mESCs were grown as embryoid bodies (EBs) and allowed to differentiate stochastically for 10 days. EBs derived from STIP1^∆TPR1 and STIP1^+/− exhibited a smaller diameter than WT and STIP1^TgA EBs (Supplementary Fig. 6B, C). After 13 days of EBs culture, we observed cells still expressing the transcription factors OCT4 and SOX2, but others showing no expression of these markers in all genotypes (Supplementary Fig. 6D, dashed and arrowheads), highlighting the heterogeneity of pluripotency markers expression within colonies during differentiation^41,47,48. Interestingly, we also observed that the pattern of STIP1 expression appears to change with differentiation. Specifically, STIP1 appears to co-localize with actin filaments in cells that detached from the core colonies and are prone to differentiation (Supplementary Fig. 6E, pink arrowheads). Additionally, cells with no expression of pluripotency markers express more nuclear STIP1, this being more frequent in STIP1^∆TPR1 and STIP1^+/− EBs, compared to WT and STIP1^TgA (Supplementary Fig. 6E, blue arrowheads). Together, our findings indicate a potential role for STIP1 in regulating cellular differentiation dynamics.

To further contextualize our findings, we investigated whether STIP1 role in early development is evolutionarily conserved. We analyzed publicly available scRNA-seq datasets of human embryonic models^{49,50,51,52,53} (Supplementary Fig. 7). Our analysis demonstrates that STIP1 is expressed in the human epiblast and primitive endoderm from embryonic days 3-7 (Supplementary Figs. 7A, B), closely mirroring its expression in the inner cell mass of mouse blastocyst-stage and primitive endoderm. Additionally, STIP1 expression persists in the post-implantation epiblast and trophectoderm, reinforcing its potential role in pluripotency maintenance and lineage specification (Supplementary Fig. 7B, C). At the onset of gastrulation, we observed STIP1 expression in human primitive streak, axial mesoderm, and caudal mesoderm, which aligns with our findings in mouse gastrulation, where Stip1 is slightly elevated in the notochord (Supplementary Figs. 7D, 1A, B). The notochord plays a crucial role in embryonic axis formation and patterning, suggesting that STIP1 may contribute to mesodermal differentiation and early developmental organization in multiple species, which is in line with our previous results (Supplementary Figs. 2E, 4R, 5A). Furthermore, we also found STIP1 expression reported in human amnion-like and mesoderm-like cells, aligning with its detection in mouse trophoblast and mesodermal derivatives, indicating a potential conserved role in ExE tissue development (Supplementary Figs. 7E, 2E and 5A).

STIP1 is essential for proliferation, DNA damage response and cell cycle progression in mESCs

We next sought to shed light on the molecular mechanisms that might contribute to the altered pluripotency and stemness in response to varying STIP1 expression. To further characterize the similarities between STIP1^∆TPR1 and STIP1^+/− phenotypes, we assessed STIP1^∆TPR1 proliferation compared to WT mESCs, finding a substantial decrease (Fig. 5A–C). Furthermore, STIP1^∆TPR1 cells exhibited reduced viability (Fig. 5D) and increased number of pH2AX foci per nucleus, indicating DNA damage (Fig. 5E, F), further supporting the relationship between these phenotypes and STIP1 loss-of-function.

Fig. 5: mESCs with differential STIP1 expression present changes in proliferation, DNA damage response and cell cycle profile and proteins.

A Representative immunofluorescence panel of cells stained with DAPI/Nuclei (red) and Ki67 (green). Scale bar = 10 µm. B Quantification of Ki67-positive (Ki67+) mESCs. Student’s t test with Welch’s correction, ****p < 0.0001, WT: n = 85; STIP1^ΔTPR1: n = 36. Mean ± SD. C Growth curve of WT and STIP1^∆TPR1 mESCs was analyzed per day by Student’s t test. Day 1 (24 h after seeding): **p = 0.0094; Day 3 (72 h): ***p = 0.0005; Day 5 (120 h): ***p = 0.0004; and Day 7 (168 h): ***p = 0.0004. Mean ± SD, n = 3. D Colorimetric analysis (corrected absorbance) of cell viability. Student’s t test, ***p = 0.009, Mean ± SD, n = 4. E Representative immunofluorescence panel of cells stained with DAPI/Nuclei (blue) and pH2AX (green). Scale bar = 10 µm. F Quantification of the number of pH2AX foci in each 10 µm. Student t test, ***p = 0.0009. Mean ± SD, n = 15. G Flow cytometry for WT and STIP1^∆TPR1 shown as representative scatter plots, staining for BrdU (S-phase) and 7-AAD (total DNA content). H Quantification of the percentage of cells for each cell cycle phase. Student’s t test, **p = 0.0085. Mean ± SD, n = 3. I Flow cytometry of cells labeled for RB1 (p < 0.0001, all comparisons), p53 (p < 0.0001, all comparisons), p21 (p < 0.0001, WT vs STIP1^TgA, STIP1^ΔTPR1 vs STIP1^TgA), CDC2 (p < 0.0001, all comparisons), cyclin B1 (p < 0.0001, all comparisons), and cyclin D1 (p < 0.0001, all comparisons). One-way ANOVA with Tukey’s post-hoc. Between 13.000–30.000 events were measured, and the mean and standard deviation of those events were used for statistics. The color pattern represents the respective groups across all charts.

Several cell cycle processes and DEGs were enriched in mESCs with varying levels of STIP1 (Figs. 2, 4A–C). Proper cell cycle progression is a well-established mechanism governing pluripotency, and it is intrinsically linked to proliferation and DNA damage sensing and response^54,55. Therefore, we investigated the cell cycle as a potential mechanism to explain the observed phenotypes in the pluripotency of our mESC models. The upregulation of the cell-cycle related DEGs Cdc25c, Rb1 and Cdkn1b⁵⁴ was confirmed by RT-qPCR (Supplementary Fig. 8A). In line with this, flow cytometry analysis revealed an altered cell cycle profile in STIP1^∆TPR1 mESCs, with a significantly higher proportion of cells in the G2/M phase when compared to WT (Fig. 5G-H). Notably, this increase was not observed in the other mESC models (WT, STIP1^+/− and STIP1^TgA; Supplementary Fig. 8B), suggesting that it is exclusive to more drastic loss-of-function due to the truncated STIP1 product.

To deepen the investigation of cell cycle progression in our models, searching for any bidirectional effect due to STIP1 levels, we analyzed a panel of cell cycle-related proteins (involved in G1/S, G2/M and DNA damage response) in WT, STIP1^∆TPR1 and STIP1^TgA mESCs (Fig. 5I). Despite transcriptional upregulation, RB1 protein levels were reduced in STIP1^∆TPR1 and increased in STIP1^TgA. p53, on the other hand, was augmented in STIP1^∆TPR1 mESCs and diminished in STIP1^TgA compared to WT (Fig. 5I). Interestingly, p21 levels were elevated in STIP1^TgA. Moreover, CDC2 was slightly downregulated in both models compared to WT, with a more pronounced decrease in STIP1^TgA. Cyclin B1 and Cyclin D1 were subtly increased in both STIP1^TgA and STIP1^∆TPR1 (Fig. 5I). p27 expression was also analyzed, but no differences were observed (Supplementary Fig. 8C). In addition, immunofluorescence imaging revealed that STIP1 partners HSP70 and HSP90 co-localized with OCT4 and SOX2 in all studied mESCs lines going through mitosis (Supplementary Fig. 8D–G). Together, these data emphasize the crucial role of STIP1 expression in preserving stemness and demonstrate that cell cycle and DNA damage-related molecules are altered upon STIP1 modulation, underlining these as potential mechanisms to explain the disrupted mESCs phenotypes.

Discussion

Our study uncovers a key role for STIP1 in regulating pluripotency and early development. By combining bioinformatics and in vitro approaches based on mESCs, we show that STIP1 levels impact the phenotype and undifferentiated state of pluripotent cells. STIP1 depletion reduces the expression of core pluripotency factors, proliferation, and cell viability, while increasing DNA damage. Conversely, its overexpression enhances stemness-related features, highlighting a bidirectional effect of STIP1 on pluripotent stem cell state. These phenotypes align with molecular pathways essential for sustaining pluripotency and developmental plasticity (Fig. 6). While our study is limited by the intrinsic variability of highly plastic pluripotent stem cells and the use of genetically modified mouse models, the integration of multiple complementary approaches strengthens the robustness of the findings reported here.

Fig. 6: STIP1 controls early embryonic development and pluripotency in mESCs.

STIP1 forms a complex with HSP70 and HSP90 in the cytoplasm of different cell types, assisting in proteostasis. STIP1 knockdown (STIP1^+/−) in mESCs results in decreased (blue arrowheads) proliferation and increased (red arrowheads) apoptosis. Furthermore, nuclear localization of the protein can be observed, concomitant with reduced expression of the pluripotency marker SOX2, and increased DNA damage. Low expression levels of a truncated form of the protein (STIP1^∆TPR1) recapitulate some of these phenotypes, including increased DNA damage, decreased SOX2, decreased cell proliferation and viability. We also observed altered cell cycle profile in STIP1^∆TPR1 mESCs, which presented G2 arrest. Increased expression of STIP1 (STIP1^TgA) leads to increased proliferation, decreased apoptosis and DNA damage and increased expression of pluripotency markers. In cells prone to differentiation, we see a strong expression of STIP1 in actin filaments, and a preferential localization of STIP1 in the nucleus in all studied cell lines, but more frequently in STIP1^+/− and STIP1^∆TPR1. This could indicate a location- and stage-specific function of STIP1 in the process of differentiation. In accordance, STIP1 decrease in STIP1^+/− and STIP1^∆TPR1 results in impaired embryoid body formation, while we propose that increased STIP1 leads to late differentiation as it increases the expression of pluripotency markers in mESCs.

STIP1 is a conserved co-chaperone^13,18,21,56 that mediates the transfer of client proteins between HSP70 and HSP90^15,16,31,57. Our previous in vivo studies showed that STIP1 KO causes embryonic lethality as early as E7, with defects in the formation of neural tube and limb buds³². Here, analysis of single-cell data of mouse gastrulation and organogenesis indicates that STIP1 is enriched in the notochord and other embryonic tissues, where morphogen signaling is active^58,59. These findings suggest that STIP1 may influence cell states during early development. This is consistent with prior links between STIP1 dysregulation and conditions such as autism spectrum disorder^60,61,62 and cancer³¹. Moreover, our analysis evidenced regular STIP1 levels in several cell types and structures related to pre- and post-implantation development in human embryo models, indicating possible conserved roles for this protein in the developmental context.

By investigating single-cell data from pre-implantation mouse embryos, we found that Stip1 expression increases from the zygote stage, peaking at the 8- and 16-cell stage, and is co-expressed with pluripotency transcription factors at the blastocyst stage, in line with our in vitro findings in mESCs. Stip1 is also co-expressed with proteostasis-related genes in the blastocyst, and changes in STIP1 levels alter the transcriptional landscape of proteostasis regulation in mESCs. Although studies are still limited, compelling evidence implicates chaperones and co-chaperones in stemness maintenance^4,8,12,63. Particularly, HSP90 inhibition leads to downregulation of OCT4, NANOG, and disrupts STAT3 signaling⁶⁴. STIP1 cooperates with HSP90 to facilitate STAT3 phosphorylation and nuclear translocation, sustaining pluripotency gene expression⁶⁵. More recently, STIP1 was shown to interact with AGO1 in an RNA-independent manner, enhancing protein quality control and maintaining ESC identity⁶⁶. Building upon these insights, we reveal how STIP1 levels impact pluripotency and stemness phenotypes in mESCs.

Pluripotent cells cope with diverse extrinsic and intrinsic stressors that can potentially influence their stemness function, including their proliferation and differentiation capacities^67,68,69. Chaperones, like STIP1, respond to such stress by regulating proteostasis and DNA damage repair⁷⁰. We observed nuclear translocation of STIP1 in loss-of-function models, similar to its relocalization upon heat shock or genotoxic stress^71,72,73, suggesting a protective role under adverse conditions (Fig. 6). Increased DNA damage marker and reduced proliferation in STIP1-deficient cells support this idea. Previous reports of STIP1-deficient fibroblasts, astrocytes, and HEK293T cells showing similar vulnerability further validate our findings^32,74.

STIP1 may influence DNA damage response through interaction with cell cycle regulators⁷⁵. Pluripotent cells have a unique cell cycle, characterized by rapid progression and specialized checkpoint control⁵⁴. In our models, STIP1 depletion induced G2 arrest, elevated phospho-H2A.X, and altered expression of p53, p21, cyclin B/CDK1, RB1, and cyclin D1, key cell cycle regulators and master controllers of cell cycle checkpoints. In response to DNA damage, p53 activates p21, leading to cell cycle arrest at G1/S or G2/M checkpoints through RB1-E2F complex activation⁷⁶. In our models, p21 upregulation occurred even in the absence of strong p53 induction, indicating an alternative regulatory mechanism, possibly linked to the developmental context^77,78,79. In line with this, p53 is typically inactive in pluripotent cells, which may contribute to the increased proliferation capacity of ESCs⁸⁰. Upon DNA damage, p53 is induced, which in turn decreases OCT4 and NANOG expression, leading to differentiation^81,82. Strikingly, p53 can negatively modulate Stip1 promoter activity, establishing a bidirectional relationship between these molecules⁸³. In addition, HSPs, including those interacting with STIP1, coordinate proteostasis of core regulators of the cell cycle and DNA damage response, such as p53 and its effectors^84,85.

Conversely, STIP1 overexpression (STIP1^TgA mESCs) accelerated cell cycle progression without checkpoint (G1 or G2) accumulation. Reduced p53 levels and increased p21 suggest a p53-independent proliferative control. Increased RB1 expression in STIP1^TgA cells may act as a compensatory mechanism to limit excessive proliferation. In our data, upregulation of cyclin D1 and stable cyclin B1 support enhanced G1/S transition without impairing mitosis. These findings suggest STIP1 helps balance self-renewal and genome integrity by coordinating chaperone activity with cell cycle control. In cells lacking the TPR1 domain, we observed increased DNA damage and prolonged G2 arrest, highlighting the importance of this domain in STIP1’s function and mESCs proper checkpoint control. These results support a model in which STIP1 modulates protein stability, assists in cell cycle transition, and protects against DNA damage, all critical functions for maintaining the pluripotent stem cell state.

Altogether, our findings identify STIP1 as a critical regulator of stemness, helping to explain its essential role during early mouse development³². STIP1 safeguards ESCs from stress, supports the expression of core pluripotency factors, and enables rapid, yet tightly controlled, proliferation. By integrating chaperone activity with cell cycle regulation, STIP1 is suggested to be a key orchestrator of proteostasis in pluripotent states. Further studies are necessary to explore the mechanisms of how chaperone and co-chaperone networks, together with STIP1, coordinate pluripotent cell states and cellular resilience, ultimately advancing our understanding of early development and informing regenerative strategies.

Methods

Shiny apps of publicly available data

The analysis of single-cell RNA sequencing data from mouse gastrulation of E6.5 to E8.5 embryos³³ and organogenesis of E8.25³⁴ was performed using the shiny apps provided by the authors, available online in the following links:

- E6.5-8.5: https://crukci.shinyapps.io/mousegastrulation2018/;

- E8.25: https://crukci.shinyapps.io/organogenesis/.

Public single-cell RNA sequencing data

Raw counts from pre-implantation mouse embryo cells were retrieved from the Gene Expression Omnibus (GEO accession code: GSE45719) through GEOquery in R³⁵. In total, 317 cells were available in the dataset; after filtering out fibroblasts and liver cells, 295 cells ranging from the zygote to the late blastocyst stage remained for downstream analysis. We used CellRouter⁸⁶ to process the single-cell data. Briefly, we normalized and then scaled the count data according to all genes present in the dataset (n = 22958), which was followed by principal component analysis (PCA) and the construction of a k-nearest neighbors (KNN) graph. Specifically for the blastocyst stage (n = 133 cells spanning early, intermediate, and late blastocyst), we calculated the Pearson correlation between the expression of Stip1 and all other genes present in the dataset; a correlation was considered significant when p-value ≤ 0.05. Significantly correlated genes were subjected to functional profiling with clusterProfiler; the enrichGO function was used inside compareCluster with the setting ‘ont’ set to ‘ALL’ to allow all Gene Ontology categories (biological processes, cellular components, and molecular functions) to be considered when performing overrepresentation analysis. A threshold of adjusted p-value ≤ 0.05 was used to define significantly enriched terms. The single-cell data processed by CellRouter was used to create a Seurat object, and Seurat’s FeaturePlot function was used with the ‘blend = TRUE’ setting for the visualization of co-expression between genes.

Ethics statement

C57BL/6J genetically modified mice, obtained from Ozgene (Perth, Australia), were bred, housed, and handled at the University of Western Ontario animal facility and were managed and treated according to the Canadian Council of Animal Care (CCAC) guidelines and approved Animal Use Protocols (2008/127, 2020-162, 2020-163). Nude BALB/c, obtained from the Animal Facility of the International Research and Teaching Center at A.C. Camargo Cancer Center, were housed at the Experimental Animal Facility of the Department of Cell and Developmental Biology of the Institute of Biomedical Sciences of the University of Sao Paulo (ICB-USP) and used in accordance with protocols approved by the ICB Animal Ethics Committee (CEUA-ICB) (Approved under No. 21, Page 3, Book 3, 2013, and No. 16, dated 2013/05/06). Cellular and molecular experiments were conducted at ICB-USP and approved by the Research Ethics Committee (CEP) of this institution (Protocol CEP-ICB 945/2018). Animals were housed under standard specific-pathogen-free conditions in accordance with institutional guidelines (The University of Western Ontario and ICB-USP), with environmental enrichment as per facility protocols. We have complied with all relevant ethical regulations for animal use.

Mouse embryonic stem cells derivation from transgenic mice

Genetically modified mice were generated for previous studies (Ozgene, Perth, Australia) on a C57BL/6 J genetic background^{23,32,36,62,87}. Animals were named WT, STIP1^+/−, STIP1^TgA and STIP1^ΔTPR1, according to wild-type/truncated STIP1 expression levels. The isolation of blastocysts from transgenic animals was performed based on previously established methods⁸⁸. The 16-day process involved the superovulation of females after mating, blastocyst in vitro maturation, and isolation of the first passage of mESCs. Briefly, females of the different genotypes, with a proximate age of 6 to 10 weeks, were superovulated (5 IU of pregnant mare serum gonadotropin, followed 48 h later by 5 IU human chorionic gonadotropin), and mated with genetically modified males of the matching/desired genetic backgrounds. Females were euthanized 3-4 days later, and the blastocysts were collected by uterine flushing using M2 medium (Sigma). Embryos were collected from timed pregnant females and were subsequently genotyped. No procedures known to cause pain or distress beyond routine handling were performed. A total of 33 mESC lines were uniquely generated for this study. A full list of the mESC lines generated in this study, with their identification and genetic background, can be found in Supplementary Table 1.

Cell culture

mESCs were derived from genetically modified mice expressing different levels of STIP1 or purchased from American Type Collection (ES-E14TG2a, ATCC #CRL-1821). Cells were grown in Glasgow Minimal Essential Medium (GMEM, Sigma) supplemented with 20% fetal bovine serum (Gibco), 1X non-essential amino acids (Gibco), 1% L-glutamine (Sigma), 1% Sodium Pyruvate (Gibco), 1% Penicillin-Streptomycin (Gibco), 1% Antibiotic-Antimycotic (Gibco) and 0.1 mM β-mercaptoethanol (Sigma) (referred hereafter as complete GMEM). To sustain pluripotency, complete GMEM was supplemented with 1000U/ml Leukemia Inhibitor Factor (LIF) (Millipore), 3 µM of glycogen synthase kinase 3 (GSK3) inhibitor (CHIR99021, Cayman Chem) and 1 µM of mitogen-activated protein kinase (MEK) inhibitor (PD 0325901, Cayman Chem). The inhibitor combination is named 2i. Cells were seeded in 0.1% gelatin (Millipore) coated flasks or plates. In experiments of LIF-only conditions, 2i factors were not added to the media. For differentiation experiments, no factors were added (LIF+2i). Cells were dissociated every other day with TrypLE™ Express Enzyme (Gibco) and maintained in incubators at 37 ^oC and 5% CO_2.

Embryoid bodies formation assay

For labeling with lineage-specific markers, embryoid bodies (EBs) were derived using the hanging drop approach, by seeding 2000 cells in droplets of 30 µL of complete GMEM (without LIF+2i factors) in a 100 mm dish lid. EBs were grown for 4 days in suspension, and subsequently seeded on ibidis^®. Imaging was performed by confocal microscopy (Leica TCS SP8, Leica Microsystems Inc.). For diameter measurements followed by labeling with specific antibodies, EBs were grown in suspension. 0.8 × 10⁶-1 × 10⁶ cells were seeded in low-biding 60 mm dishes, with complete GMEM (without LIF+2i factors). Dishes were placed in the Orbi-Shaker™ CO₂ (Benchmark Scientific) for low agitation (30 rpm) for 24 h to prevent aggregation, and EBs were maintained in suspension for 10 days in incubators at 37 ^oC and 5% CO₂. For each EB, two perpendicular diameters were measured, and their average was used as the final size. A total of 60 EBs were measured individually, derived from three independent experiments, with each experiment contributing 20 EBs per mESCs genetic background. The image was acquired with an AxioVert A1 microscope (Zeiss), and the pool of all measured EBs per group was used for the statistical test. The experiment was independently repeated as described for confirmation of results. Subsequently, EBs were seeded on ibidi^® plates and allowed to grow for 3 more days (13 days total) while attached to the plate. Imaging of the EBs was performed by confocal microscopy.

RNA extraction and bulk RNA sequencing

mESCs, derived from two genetically independent biological replicates (samples coming from different mice within the same group of Stip1 modification), were seeded in 6-well plates and total RNA was extracted from these colonies, using RiboPure RNA Purification (Ambion) kit, following the manufacturer’s instructions. RNA concentration was quantified in BioTek Epoch (Agilent). 4.5 µg of total RNA was provided for library preparation by an outsourced service (Indegene) with the NextSeq 500/550 High Output Kit v2.5 (Illumina). Sequencing depth ranged between 40–45 million reads per sample, and read length is 76bp.

In-house bulk RNA sequencing data of mouse embryonic stem cell lines

For data preprocessing, FastQC was used to check the quality of the raw sequence data and after trimming. Samples with a very low number of reads were discarded. Cutadapt was used for trimming (following the parameters -m 15 # discard minimum length, -q 20 # trim low-quality bases, -e 0.1 # error rate). TopHat and Bowtie were used for the alignment, followed by the estimation of transcript abundance with cufflinks. For differential gene expression analysis and functional profiling, raw counts were used as input in DESeq2⁸⁹ to find differentially expressed genes between the following conditions: STIP1^TgA and WT, STIP1^+/− and WT, and STIP1^ΔTPR1 and WT. In all comparisons, WT samples were used as controls. A comparison was also made between STIP1^TgA and STIP1^+/−, in which STIP1^+/− samples were used as controls. In all DESeq2 analyses, genes with a very low number of raw counts across samples (≤1) were filtered out before doing the statistical tests. In each analysis, the results for the comparison of interest were extracted and log₂(fold change) shrinkage was performed using the shrinkage estimators of the normal distribution. A gene was considered differentially expressed when p-value ≤ 0.05; no log₂(fold change) cut-offs were applied. Differentially expressed genes of each comparison were subjected to functional profiling with clusterProfiler; the enrichGO function was used inside compareCluster with the setting ‘ont’ set to ‘ALL’ to allow all Gene Ontology categories (biological processes, cellular components, and molecular functions) to be considered when performing over-representation analysis. A threshold of adjusted p-value ≤ 0.05 was used to define significantly enriched terms.

Comparative analysis of public single-cell data with in-house generated results

For comparison with pre-implantation data, gene signatures were retrieved from Deng et al.³⁵, and the Jaccard index approach was used to establish the similarity between this and our in-house generated datasets. For post-implantation comparisons, single-cell RNA-seq data from mouse gastrulation and organogenesis^33,34 were analyzed with Seurat. The counts matrix and metadata were downloaded from marionilab.cruk.cam.ac.uk and a Seurat object was created with cell type annotations and UMAP embeddings. Data was normalized, and marker genes were identified using FindAllMarkers and filtered based on detection in at least 25% of cells, an average log2 fold-change > 1 and p-adjusted < 0.05. DEGs were identified from contrasts (STIP1^+/− versus WT, STIP1^TgA versus WT and STIP1^ΔTRP1 versus WT) using DESeq2, applying thresholds of log2 fold-change > 1 and p-adjusted < 0.05. We did gene annotation with biomaRt, mapping gene symbols to their respective Ensembl identifiers. To assess enrichment of differentially expressed genes within cell clusters, Fisher’s exact test was applied to compare cluster-specific markers with differentially expressed genes from each condition. Adjusted p-values were computed to evaluate statistical significance. The heatmap was generated with ggplot2.

Real-time quantitative PCR

Total RNA was used to obtain 2 μg of cDNA by reverse transcription using SuperScript IV VILO Master Mix (Thermo Fisher Scientific) or SuperScript™ III Reverse Transcriptase (Thermo Fisher Scientific). Sample preparation was done using SYBR™ Select Master Mix Green (Thermo Fisher Scientific), and cDNA was diluted 1:20. Reactions were performed in StepOnePlus™ Real-Time PCR System (Applied Biosystems) in technical triplicate. Gene expression levels were calculated using the 2^-∆∆Ct method⁹⁰, normalized to endogenous genes (Gapdh, Tbp and/or Actb) and relative to controls. Statistical analyses were based on results from independent experiments or biological triplicates.

Western blotting

Protein was extracted from whole cell lysates using ice-cold RIPA lysis buffer with protease (10%) and phosphatase inhibitors (5%). Samples were centrifuged for 15 minutes at 14.000 x g at 4 °C, then the supernatant was collected and used for protein quantification using Bradford (Bio-Rad) or BCA (Thermo Fisher Scientific) assays. 5-30 μg of protein was loaded on polyacrylamide gels of variable percentage for SDS-PAGE. Proteins were transferred to nitrocellulose or PVDF membranes (Amersham) using the Trans-Blot Turbo transfer system (Bio-Rad) or wet transfer. 5% bovine serum albumin (BSA) or non-fat dry milk in TBST was used for blocking for 1 h at room temperature (RT). Primary antibodies were incubated in 5% BSA or non-fat dry milk, according to the manufacturer’s recommendations, overnight at 4 °C in agitation. Secondary antibodies (anti-mouse, rabbit or rat HRP) were incubated in 5% BSA or non-fat dry milk for 1 h at RT, in agitation. A full list of antibodies can be found in Supplementary Data 6. HRP signal was developed in Amersham Hyperfilm ECL or Amersham ECL Prime. Protein expression was calculated normalized to β-actin or β-tubulin and relative to controls. Band densitometry was quantified using ImageJ (NIH) or ImageLab for the ChemiDoc system (Bio-Rad). Experiments were performed in genetically independent biological triplicates (samples derived from three distinct animals within each mESC genetic group) and repeated independently for confirmation of results.

Alkaline phosphatase assay

mESCs were seeded on gelatin-coated coverslips and cultured for 72 h in complete GMEM+LIF2i, followed by staining with the Alkaline Phosphatase (AP) Staining Kit (Stemgent), according to the manufacturer’s instructions. Three or more independent coverslips were imaged using a light microscope (AxioVert A1, Zeiss), and multiple random fields were captured per coverslip. All captured images were included in the following analysis. Colonies classified as positive, negative, or mixed for AP staining were quantified using ImageJ software by experimenters blinded to the genetic background of the cells.

Immunofluorescence

Cells were cultured on 12 mm coverslips, coated with 0.1% gelatin and placed in 24-well plates for 24 h. Cells were fixed with 4% paraformaldehyde (PFA) for 20 min at RT. After successive washes with PBS, cells were blocked for 1 h at RT in 5% BSA in PBS and incubated overnight at RT with primary antibodies. Alexa Fluor-conjugated secondary antibodies were used at concentrations recommended by the manufacturers in 0.5% BSA and 0.1% Triton X-100 in PBS. Cells were counterstained with DAPI (1:500), mounted onto slides using ProLong™ Gold Antifade Mountant (Invitrogen) and imaged using Leica TCS SP8 (Leica Microsystems Inc.) confocal system (63X objective) or similar. For immunofluorescence analysis of Ki67, cleaved caspase-3 (CC3), and phospho-Histone H2A.X (pH2AX), three or more gelatin-coated coverslips per mESC genetic group were prepared using genetically independent biological samples. All captured random fields from each coverslip were used for quantification. The percentage of labeled colonies and total fluorescence intensity were assessed using ImageJ software. Manual cell counting was performed by experimenters blinded to the genetic background of the cells. Total fluorescence signal was normalized to background intensity, and values were expressed in arbitrary units. Additionally, the open-source software CellProfiler 4.2.5 (Broad Institute; https://cellprofiler.org) was also used for image quantification, specifically for analysis of the expression of CC3 and pluripotency markers (OCT4/SOX2). The complete pipeline is available in the paper GitHub page (https://github.com/marilenehohmuth/stip1-in-early-devepment). In brief, image segmentation was performed with the RunCellpose module, adjusting the expected cell diameter for each group for better fitting. Resulting objects were then measured in intensity for channels depicting OCT4 or SOX2 and CC3 staining using the MeasureObjectIntensity module, and further classified according to the relative intensity of each marker (high or low) via the ClassifyObjects module. The resulting classification into four groups was exported in spreadsheets and processed into graphics using GraphPad Prism software for statistical analysis. A full list of antibodies can be found in Supplementary Data 6.

Growth curve

10,000 cells were seeded in 24- or 12-well plates, previously treated with 0.1% gelatin. Cells were dissociated and subsequently counted at 24 h, 72 h, 96 h and 120 h (if possible, by cell confluency) after seeding, with a Neubauer chamber or Countess™ 3 Automated Cell Counter (Invitrogen). All counts were done at the same time of the day, excluding dead cells by trypan blue staining. Growth analysis was based on data obtained from three or more biological replicates. The resulting cell numbers were plotted as a scatter graph to visualize growth dynamics over time and presented as bar graphs for individual analysis.

Stip1 knockdown by shRNA

Five units of ready-to-use lentiviral particles containing shRNA sequences specific for mouse Stip1 were obtained from MISSION^® shRNA Transduction Particles (Sigma). shRNA non-target (control) lentiviral particles were also acquired to validate knockdown cell line generation (reference number and sequences can be consulted in Supplementary Data 6). From the five sequences, two were selected for the experiments reported here. Briefly, 1.6 × 10⁴ ES-E14TG2a cells were seeded in 96-well plates with complete GMEM + LIF2i and incubated at 37 °C with 5% CO₂ for 18 hours until reaching 50–80% confluence. Medium was then replaced with fresh medium supplemented with polybrene (8 µg/mL). Under these conditions, lentiviral transductions were carried out at a multiplicity of infection (MOI) of 0.2 and 0.5. Puromycin (0.8 µg/mL) selection of transduced populations was performed over two weeks.

BrdU incorporation assay

Cell proliferation was evaluated using the Cell Proliferation ELISA, BrdU (colorimetric) kit (Sigma/Roche), according to the manufacturer’s protocol. Briefly, mESCs were seeded in flat-bottom 96-well plates in technical triplicates or more, and maintained under standard conditions (GMEM+LIF2i, 37 °C, 5% CO₂). After 48 h, a pulse of BrdU (final concentration: 100 µM) was added, and cells were incubated for 2 h. After the pulse, the labeling medium was removed, and cells were fixed and denatured with the DNA-denaturing solution. Incorporated BrdU was detected using a peroxidase-conjugated anti-BrdU antibody for 1 h, followed by incubation with a tetramethylbenzidine substrate for 20 min. Absorbance was measured at 370 nm with a reference wavelength of 492 nm, using a microplate reader. Absorbance values were used for proliferative activity. Statistical analysis was performed based on the results of independent experiments.

Flow cytometry

Cells (1 × 10⁶) were harvested by enzymatic dissociation and incubated with 5% BSA in phosphate-buffered saline (PBS) for 30 min at 4 °C. Primary and secondary antibodies were incubated in 1% BSA in PBS for 30 min to 1 h at 4 °C. For cell cycle analysis, the Phase-Flow^TM Kit (BioLegend) was used, as instructed by the manufacturer. For permeabilized flow cytometry, the kit CytoFix/Cytoperm Flow (BD Biosciences) was used, according to the manufacturer’s instructions. Flow Cytometry reading was performed at A.C. Camargo Cancer Center Facility using FACSymphony (BD Biosciences). Data was analyzed and generated using FlowJo v9 (FlowJo, LLC) and/or FCS Express (De Novo Software). A full list of antibodies can be found in Supplementary Data 6. Between 13.000-35.000 events were detected for each sample. For analysis, we performed negative and secondary antibody controls for each studied cell line. Statistical tests were performed using the total cell percentage of cells in different independent experiments, or the mean, counts and standard deviation of each event.

Stip1 deletion by CRISPR/Cas9

The CRISPR/Cas9 methodology was standardized to generate mouse STIP1 knockout (KO) in mESCs (ES-E14TG2a, ATCC). Exon 2 was selected for the design of single guide RNA (sgRNA) sequences. Sequences were confirmed based on guide score versus off-target analysis using the Optimized CRISPR Design Tool (http://crispr.mit.edu) and the CRISPR Oligo Design Tool. Once selected, the guide sequences (available at Supplementary Data 6) were synthesized and cloned into the pX330 plasmid, which was modified to co-express Puromycin-GFP to facilitate selection of transfected cells. The sgRNAs were inserted following digestion with the BbsI restriction enzyme. The insertion of the guide sequences into the plasmid was then verified by sequencing. Once full insertion of the sgRNAs into the plasmid was confirmed, lipofection was performed (Lipofectamine 2000; Thermo Fisher Scientific) in parental ES-E14TG2a cells, following the manufacturer’s instructions. Cells were selected with puromycin (0.8 μg/mL) for 4 days. Clonal populations were isolated by limiting dilution and expanded for confirmatory analyses.

Teratoma formation assays

Approximately 1 × 10⁶ mESCs in a 1:1 Geltrex® solution (Thermo Fisher Scientific) (final volume = 100 μL) were subcutaneously injected into the dorsal flank of immunodeficient (nude) BALB/c mice, aged 2–4 months. Subcutaneous injection of mESCs for teratoma assays was conducted under approved protocols. Tumor growth was monitored for around 42 days. After this time, the animals were euthanized in a CO₂ chamber, and the tumors were surgically dissected, measured, and fixed in 4% PFA. The material was then embedded in paraffin and stained with hematoxylin and eosin (HE) to identify the different tissues formed. Images were captured using the AxioVert A1 microscope (Zeiss). Teratoma measurements were performed based on three independent biological replicates. Given that one of the tumors derived from Parental mESCs did not grow, this specific value was excluded from the analysis.

Cell viability assays

Cell viability was quantified using the PrestoBlue™ Cell Viability Reagent (Thermo Fisher Scientific), following the manufacturer’s instructions. mESCs were seeded in 96-well plates and cultured under standard conditions. After proper adhesion, 10 μL (10%) of PrestoBlue™ reagent was added directly to each well and incubated for 1 h at 37 °C. Absorbance was measured at 570 nm, using the measurement at 600 nm as the reference wavelength, in Epoch (BioTek). The background signal from media-only wells was subtracted, and the corrected values were used to calculate the percentage reduction of PrestoBlue™ reagent, indicating cell metabolic activity. Statistical analyses were performed in biological replicates, and the experiment was independently repeated for confirmation of results.

Single-cell RNA sequencing analysis of human embryo models

Single-cell RNA sequencing datasets from multiple studies^{49,50,51,52,53} were processed using Seurat and Scanpy. The count matrices were obtained from h5ad files and converted into Seurat objects. Gene annotations and sample metadata were integrated from GEO, ArrayExpress, and study-specific metadata files to ensure consistency across datasets. For each dataset, gene expression was normalized, and metadata fields such as developmental stage, lineage, and cell type were assigned to the corresponding cells. The expression of STIP1 was specifically analyzed across different conditions, and its distribution was visualized using boxplots. Comparative analyses between studies were performed by aligning metadata categories, such as developmental time points and inferred lineages. The results were visualized using ggplot2, highlighting expression patterns across biological contexts.

Statistics and reproducibility

Statistical analyses were performed using GraphPad Prism (versions 8–10). Normality was assessed prior to hypothesis testing. When data did not meet the assumptions of normal distribution, non-parametric tests were applied. For comparisons between two groups, unpaired two-tailed Student’s t test or Mann–Whitney U test was used, as appropriate. For comparisons among multiple groups, one-way ANOVA followed by Tukey’s or Dunnett’s multiple comparisons test, or Kruskal–Wallis test with Dunn’s post hoc test, was applied as appropriate. Welch’s correction was applied to account for unequal sample sizes. A p-value < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation (SD), unless otherwise stated in figure legends. Sample sizes (n), statistical tests used, exact p-values, and definitions of error bars are provided in the respective figure legends. No statistical methods were used to predetermine sample size. n refers to genetically independent biological samples (samples derived from different animals within a mESC group with the same Stip1 modulation), biological replicates within the same mESC line, or independent experiments. Where applicable, quantifications were performed by scorers blinded to the genetic background.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.