Graphical abstract

   graphic file with name fx1.jpg
   [40]Open in a new tab

Highlights

     * •
       Toti is the first-in-field multi-omics resource focused on
       totipotency
     * •
       Toti enables intuitive and comparative in silico exploration of
       totipotency
     * •
       Toti provides interactive single-cell transcriptomic analysis of
       totipotent states
     * •
       Toti offers motif- and pathway-based analyses
     __________________________________________________________________

   In this article, Sheng, Fu, and colleagues present Toti, the first
   comprehensive multi-omics database focused on totipotency, enabling in
   silico exploration of its transcriptional and epigenetic regulation.
   Toti integrates data from in vivo, in vitro, and genome-edited models
   across 8,284 human and mouse embryonic samples, allowing flexible
   search, visualization, and analysis of totipotent and related stem cell
   states.

Introduction

   Totipotent stem cells (TSCs), emerging in early embryogenesis with the
   broadest cellular plasticity in the mammalian body, harbor enormous
   potential for regenerative medicine and reproductive technology
   ([41]Cai et al., 2022). Relative to the developmentally more restricted
   pluripotent stem cells (PSCs), TSCs are capable of producing all of the
   differentiated cell types in both embryo and extraembryonic components
   and forming an entire organism ([42]Wu and Schöler, 2016). Totipotency
   is limited to early-stage blastomeres. In mice, only the zygotes and
   blastomeres from 2-cell embryos are bona-fide TSCs and can give rise to
   the blastocyst, which is composed of the inner cell mass (ICM) and
   outer trophectoderm ([43]Figure 1A) ([44]Sotomaru et al., 1998). As
   cells develop into distinct cell lineages by the blastocyst stage,
   totipotency is lost and cellular plasticity is gradually reduced.
   Currently, due to the scarcity of embryonic TSCs, the gene regulatory
   logic underlying totipotency is still incompletely understood.

Figure 1.

   [45]Figure 1
   [46]Open in a new tab

   Toti provides multi-omics data to investigate epigenetic and
   transcriptomic underpinning of totipotency

   (A) Toti contains multi-omics data in in vivo, in vitro, and
   genome-edited embryonic TSCs, TSC-like cells, PSCs, and embryos
   spanning preimplantation stages from 4,680 human and 3,604 mouse
   samples. FSHD, facioscapulohumeral muscular dystrophy; PSCs,
   pluripotent stem cells; TSCs, trophoblast stem cells; 8CLCs, 8-cell
   like cells; EPSCs, expanded pluripotent stem cells; EPSs, extended
   pluripotent stem cells; H9-e4CL, H9 hESC cultured in e4CL (stepwise)
   for 5 days; TBLCs, totipotent blastomere-like cells; ESCs, embryonic
   stem cells; H9-4CL, H9 hESC cultured in 4CL (stepwise) for 12 days;
   C2C12, cultured mouse myoblasts; MEF, mouse embryonic fibroblasts;
   2CLCs, 2-cell-like cells; TPSs, totipotent potential stem cells; TLSCs,
   totipotent-like stem cells; ciTotiSCs, chemically induced totipotent
   stem cells; ZGA, zygotic genome activation.

   (B) A Ciros plot showing read coverage for gene expression (orange),
   H3K4me3 (blue), and H3K27ac (green) histone-modified peaks in mESC,
   2CLC, and 2-cell embryo. The black lines in the center represent
   genomic gaps or low-mappability regions, such as centromeres and other
   repetitive sequences.

   (C) Read coverage represents gene expression (orange), H3K4me3 (blue),
   and H3K27ac (green) modifications around totipotent signature genes,
   Zfp352, Zscan4b, and Zscan4c, and pluripotent signature gene Klf4, in
   2-cell embryo, 2CLC, and mESC. mESC, mouse embryonic stem cell.

   In vitro cellular models are critical for understanding the molecular
   architectures of cell stemness. For instance, embryonic stem cells
   (ESCs), derived from the ICM of blastocysts, are classic pluripotent
   cellular models and greatly facilitate the exploration of pluripotency
   ([47]Rossant and Tam, 2009). Intriguingly, approximately 1%–5% of ESCs
   resembling the blastomeres of 2-cell embryos, referred to as
   “2-cell-like cells” (2CLCs), arise spontaneously in ESC culture in
   vitro. They show greater cellular plasticity and downregulated protein
   levels of pluripotency factors ([48]Rodriguez-Terrones et al., 2018).
   Unlike pluripotent ESCs, 2CLCs retain the totipotent-like state to
   generate both embryonic and extraembryonic tissues when reintroduced
   into early embryos ([49]Yang et al., 2020), highlighting the promising
   application of these cells in understanding totipotency. Other in vitro
   totipotent-like cellular counterparts with biological relevance,
   including totipotent blastomere-like cells ([50]Shen et al., 2021),
   totipotent potential stem cells (TPSs) ([51]Xu et al., 2022), and
   chemically induced totipotent stem cells (ciTotiSCs), have been
   actively pursued afterward ([52]Malik and Wang, 2022) and widely used
   in totipotency study as well.

   Recent advances in totipotency cellular model and sequencing technology
   lead to the ever-growing influx of multi-omics data on totipotent-like
   cells. These results have rapidly forwarded our understanding of
   epigenomic and transcriptomic features governing the establishment and
   exit of totipotency. For example, previous studies have identified
   transcription factor (TF) Dux as the master inducer of the 2C-like
   transition in mouse embryonic stem cells (mESCs) ([53]Hendrickson et
   al., 2017), based on RNA sequencing (RNA-seq) and chromatin
   immunoprecipitation sequencing (ChIP-seq) data detected from
   CRISPR-based genome-edited embryonic cells. Supported by RNA-seq data,
   2CLCs are typically characterized by transient activation of major
   satellites ([54]Kresoja-Rakic and Santoro, 2019) and endogenous
   retroviral elements that significantly contribute to promoting zygotic
   genome activation (ZGA) ([55]Lu and Zhang, 2015). Particularly, as
   shown by comprehensive epigenetic information measured by ATAC-seq,
   bisulfite sequencing, and ChIP-seq, endogenous retrovirus, MERVL,
   activated many 2C-specific genes, including the Zscan4 cluster genes,
   likely through epigenetic variations, such as DNA methylation and
   histone modifications associated with DNMT ([56]Eckersley-Maslin et
   al., 2016) and LSD1 (KDM1A) ([57]Wasson et al., 2016). Other features
   of 2-cell embryos, such as their chromatin accessibility landscape
   ([58]Hendrickson et al., 2017) and increased global histone mobility
   ([59]Bošković et al., 2014), are also found to be recapitulated in
   2CLCs. Moreover, the remodeling of the totipotency-specific broad
   H3K4me3 domains could help stabilize totipotency in vitro ([60]Yang et
   al., 2022), highlighting the importance of characterizing epigenetic
   underpinning related to totipotency. Besides, benefiting from scRNA-seq
   data, our previous studies identified a novel intermediate state during
   the 2C-like transition ([61]Fu et al., 2019, [62]2020). Therefore,
   extensive investigation of epigenetic and transcriptomic architectures
   in totipotent cells is of paramount importance to pinpoint molecular
   mechanisms controlling cellular plasticity and lineage segregation in
   early development.

   Integrative analysis of multi-omics data in both embryonic cells and
   their in vitro counterparts makes it possible to comprehend the
   molecular basis of in vivo cell fate differentiation and in vitro cell
   fate engineering. Currently, arising databases mainly contain
   information on pluripotent cells instead of totipotent cells and few
   information on CRISPR-based genome-edited embryonic cells. For example,
   EMAGE ([63]Christiansen et al., 2006), DBTMEE ([64]Park et al., 2015),
   EmExplorer ([65]Hu et al., 2019), StemCellDB ([66]Mallon et al., 2013),
   LifeMap Discovery ([67]Edgar et al., 2013), FunGenES ([68]Schulz et
   al., 2009), and StemMapper ([69]Pinto et al., 2018) provide gene
   expression patterns during mouse or human embryo development and in
   induced pluripotent stem cells. ScRNASeqDB ([70]Cao et al., 2017)
   provides single-cell gene expression profiles in 200 human early
   embryonic cells at different developmental stages. MethBank ([71]Li et
   al., 2018), DevMouse ([72]Liu et al., 2014), iHMS ([73]Gan et al.,
   2017), GED ([74]Bai et al., 2017), and MetaImprint ([75]Wei et al.,
   2014) are mainly focused on epigenetic modifications, such as DNA
   methylation and histone modifications during embryogenesis in human and
   mouse. EpiDenovo ([76]Mao et al., 2018), ESCAPE ([77]Xu et al., 2013),
   DevOmics ([78]Yan et al., 2021), and dbEmbryo ([79]Huang et al., 2022)
   integrated genomic, epigenomic, and transcriptomic data from human and
   mouse early embryos without TSCs. However, none of the existing
   databases have provided multi-omics data on in vivo, in vitro, and
   genome-edited totipotent cells, greatly hindering the interpretation of
   underlying mechanisms that contribute to totipotency and embryo
   development.

   To bridge this gap, here we present a novel database, Toti, the unique
   and first-in-field multi-omics database dedicated to comprehensively
   investigating epigenetic and transcriptomic architectures in
   early-stage embryogenesis of mouse and human ([80]Figure 1A). Toti
   encompasses in vivo, in vitro, and genome-edited embryonic cells, with
   the particular focus on totipotent cells, thus providing an
   unprecedented platform for extensive investigation of totipotency in
   silico ([81]Figure 1A). Toti allows facilitated and quick access of
   interested samples/datasets by searching for genes, sequencing types,
   or other relevant keywords. Toti enables comparative analyses not only
   on epigenetic characteristics measured by multiple methods, including
   bisulfite sequencing, ATAC-seq, ChIP-seq, CUT&RUN, and CUT&Tag, but
   also on temporal gene expression patterns supported by RNA-seq and
   single-cell RNA-seq (scRNA-seq) data to dissect molecular
   (dis)similarity across in vivo and in vitro totipotent cells, PSCs, and
   other embryonic cells in human and mouse ([82]Figure 1B). Toti also
   facilitates users to prioritize top enriched TFs, genes and pathways
   under different developmental stages, and genome-edited conditions.
   Taken together, Toti thus serves as a unique, comprehensive, and
   valuable resource to provide insights into molecular characteristics
   shaping totipotency.

Results

Toti overview

   As characterization of the molecular basis in TSCs is the key to
   comprehend mechanisms underlying totipotency, here we design Toti to
   store gene expression and epigenetic modification information covering
   in vivo, in vitro, or CRISPR-based genome-edited totipotent,
   pluripotent, and differentiated embryonic cells ([83]Figure 1A) from
   4,680 human and 3,604 mouse samples ([84]Tables S1 and [85]S2). Toti
   contains: (1) DNA methylation (WGBS, PBAT, and RRBS); (2) genome-wide
   chromatin features, including open chromatin peaks (ATAC-seq),
   TF-binding sites, and histone modifications (ChIP-seq, CUT&Tag, and
   CUT&RUN); and (3) gene expression patterns (RNA-seq and scRNA-seq) from
   different types of cells. According to these holistic multi-omics data
   in Toti, we observed globally similar patterns of peaks in gene
   expression, H3K4me3 and H3K27ac modifications in both mouse 2-cell
   embryos and 2CLCs, but distinct from mESCs ([86]Figure 1B).
   Specifically, regarding totipotent genes, such as Zfp352 ([87]Li et
   al., 2023), Zscan4b, and Zscan4c ([88]Falco et al., 2007), enriched
   peaks in gene expression, H3K4me3, and H3K27ac were shown in both mouse
   2-cell embryos and 2CLCs, but not in mESCs, highlighting their
   totipotency-specific signatures in 2-cell embryos and 2CLCs ([89]Figure
   1C). As a comparison, we also observed mESC-specific peaks in gene
   expression, H3K4me3, and H3K27ac around the pluripotent gene Klf4
   ([90]Figure 1C) ([91]Guo et al., 2009). Hence, by integrative analysis
   of these orthogonal omics data from cells under different developmental
   stages or genome-edited conditions, Toti makes it possible for users to
   comprehensively investigate totipotency in silico. Toti is designed
   with three main functionalities: Search, Browse, and Analysis
   ([92]Figure 2).

Figure 2.

   [93]Figure 2
   [94]Open in a new tab

   The main function of Toti

   (A) “Search” module allows a flexible search of genes, sequencing
   types, or other relevant keywords.

   (B) “Browse” includes two sub-modules. “Genome Browse” visualizes
   analyzed multi-omics data by embedding configured JBrowse. “Featured
   Gene” visualizes expression patterns of top variable signature genes or
   user-selected genes under selected developmental stages or
   genome-edited conditions. ZGA, zygotic genome activation.

   (C) “Analysis” contains three sub-modules. “Motif Enrichment” and
   “Pathway Enrichment” prioritize enriched transcription factors and
   biological pathways associated with interested development stage or
   genome-edited condition, respectively. “scRNA-seq Analysis” facilitates
   the intuitive comparison of gene expression or co-expression patterns
   among embryos under selected developmental stages or genome-edited
   conditions.

Search

   Toti enables a flexible search that provides quick access to
   totipotency-related studies/datasets of interest by querying genes,
   sequencing types, species, or other relevant keywords, such as histone
   modification type, cell type, library strategy, article name, and GEO
   accession ID. All the queried tables recording meta information of data
   stored in Toti are sortable and downloadable. Users can also click the
   hyperlink of each GEO accession ID to obtain further experimental
   details or acquire the corresponding raw data ([95]Figure 2A).

Browse

   A systematic scrutiny of epigenetic and transcriptomic architectures,
   specifically for signature genes representing for totipotency or
   pluripotency, is extremely important to understand key factors
   governing the establishment and exit of totipotency in embryos. Toti
   ensures both orthogonally and parallelly comparative investigations on
   molecular basis with “Genome Browse” and “Featured Genes” ([96]Figure
   2B).

Genome browse: Allows for orthogonally and parallelly comparative analysis

   Toti provides extensive investigation into gene expression and
   epigenetic variations, such as open chromatin peaks, histone
   modifications, and DNA methylation, across in vivo, in vitro, and
   genome-edited totipotent, pluripotent, and differentiated cells by
   embedding our well-configured JBrowse ([97]Figure 2B). Data are
   hierarchically organized based on species, sequencing type, epigenetic
   modification, and gene expression. The interface permits the selection
   of one or multiple tracks to visualize analyzed multi-omics data. By
   specifying interested species, cell type, gene, or chromosome region,
   tracks for the corresponding regions are displayed. And it is of
   particular convenience to share the queried result to others for
   further discussion by clicking the “Share” button on the top right of
   the Genome Browser. Users can also add custom data tracks for
   personalized comparison. All the resulting images, such as plots for
   gene structure, read coverage, and histone modification peaks, are
   easily to be exported.

Featured genes: Provides insights into expression patterns of signature genes

   The developmental trajectory of totipotent cells during early
   development is governed by highly coordinated transcriptomic
   remodeling. After fertilization, the maternal genes inherited from
   oocytes drive two waves of ZGA, including minor ZGA and major ZGA
   ([98]Lu and Zhang, 2015). These ZGA genes modulate the entry and exit
   of totipotency in embryos. Understanding the intricate interplay of
   these signature genes is critical for comprehending the biological
   nuances of totipotency. Here, we designed “Featured Genes” ([99]Figure
   2B) sub-module to allow users to scrutinize expression patterns of
   signature genes, including maternal, minor ZGA, major ZGA, totipotent,
   and pluripotent genes defined from [100]Yang et al. (2022), across both
   in vivo and in vitro TSCs and cells under different developmental
   stages or genome-edited circumstances. Users can customize their search
   by choosing interested featured genes and specifying cells to compare.
   Toti prioritizes development- or genome-edited-condition-associated
   genes based on the standard deviation of gene expression among selected
   samples. Subsequently, interactive heatmaps are rendered, providing a
   visual representation of top variable signature genes across chosen
   samples. Alternatively, user can also flexibly select any gene of
   interest and visualize their expression patterns across selected
   samples. The expression matrix of each heatmap can be downloaded and
   further tailored for customized use.

Analysis

   It is of fundamental importance to pinpoint top enriched TFs, genes,
   and pathways under different totipotent-associated development stages
   or genome-edited conditions, as these analyses could shed light on gene
   regulatory logics contributing to totipotency. Here, we designed three
   sub-modules, including “Motif Enrichment,” “Pathway Enrichment,” and
   “scRNA-seq analysis” in this “[101]analysis” section ([102]Figure 2C)
   to provide insights into the top development- or
   genome-edited-condition-associated TFs, genes, or pathways and
   facilitate the dissection of gene expression patterns at single-cell
   resolution in totipotency-related samples.

Motif enrichment: Prioritizes top associated TFs

   As TFs play important roles in modulating cellular gene expression, we
   developed a “Motif Enrichment” sub-module to prioritize top associated
   TFs under different genome-edited conditions, histone modifications, or
   developmental stages ([103]Figure 2C). By selecting interested
   epigenetic modifications, such as open chromatin peaks (ATAC-seq),
   TF-binding sites, or histone modifications (ChIP-seq, CUT&Tag, or
   CUT&RUN), in cells under certain conditions, Toti provides enriched
   TF-binding motifs analyzed by HOMER
   ([104]http://homer.ucsd.edu/homer/). Further details for motifs are
   accessible via clicking links, such as “motif file” and “Known Motif
   Enrichment Results.”

Pathway enrichment: Prioritizes top associated biological pathways

   Previous studies indicated that recapitulating epigenetic modifications
   of 2-cell embryos, such as chromatin accessibility landscape
   ([105]Hendrickson et al., 2017) and broad H3K4me3 ([106]Yang et al.,
   2022), can facilitate totipotency stabilization in vitro. This
   underscores the importance of characterizing activated biological
   pathways under the establishment, self-renewal, or exit of totipotency.
   Therefore, we developed the “Pathway Enrichment” sub-module
   ([107]Figure 2C), providing users with an intuitive and interactive
   chart, along with a sortable and downloadable table that prioritizes
   key relevant biological pathways based on gene ontology (GO)
   ([108]Ashburner et al., 2000) and Kyoto Encyclopedia of Genes and
   Genomes ([109]Kanehisa and Goto, 2000). This sub-module enables users
   to gain insights into biological context-dependent epigenetic
   modifications in in vivo, in vitro, or genome-edited totipotent,
   pluripotent, and differentiated cells. Each resulting table can be
   sorted and downloaded. More information, such as genes involved in each
   pathway, could be accessible in the downloaded CSV file.

scRNA-seq analysis: Dissects transcriptome nuances at single-cell resolution

   Using scRNA-seq data sequenced at different time points of Dux
   induction, we previously confirmed the existence of an intermediate
   state during the 2C-like transition ([110]Fu et al., 2019), emphasizing
   that scrutiny of transcriptome at single-cell resolution could promote
   a better understanding of molecular mechanisms underlying the
   establishment and exit of totipotency. Here, we present an interactive
   “scRNA-seq analysis” interface to facilitate the intuitive comparison
   of gene expression or co-expression across cells under different
   developmental stages and TSC-like cells with/without genome-editing
   ([111]Figure 2C). To our knowledge, this is the first online scRNA-seq
   analysis tool tailored to investigation in transcriptome nuances of
   totipotency.

A case study: A systematic integrated analysis of epigenetic and
transcriptomic architectures in totipotency

   Here, we present a case study to illustrate how the functional modules
   in Toti, such as “Featured Gene,” “Genome Browse,” “Motif Enrichment,”
   “Pathway Enrichment,” “scRNA-seq analysis,” and “Download,” can be
   effectively utilized to explore the epigenetic and transcriptomic
   nuances of totipotency. In this case study, we mainly chose five mouse
   cell types for comparison, including pluripotent mESCs, embryonic TSCs
   (2-cell embryos), and three in vitro TSCs models (2CLCs, TPSs, and
   ciTotiSCs). 2CLCs are naturally occurring, transient totipotent-like
   cell populations within mESC cultures, whereas TPSs and ciTotiSCs are
   totipotent-like cell lines generated through distinct chemical
   reprogramming protocols. All these three in vitro TSC models were
   previously found to show similar upregulated expression of totipotency
   factors as in totipotent 2-cell (2C) embryos but were distinguished
   from mESCs ([112]Hu et al., 2023). Our case study aims to delineate
   epigenetic and transcriptomic architectures recapitulated by these in
   vitro TSCs through systematic integrated analysis of multi-omics data
   available in Toti. This case study not only sheds light on how to
   leverage these modules in totipotency exploration but also highlights
   their potential applications in advancing our understanding of
   totipotent states.

Deciphering transcriptome discrepancies in totipotency across in vitro TSC
models

   The in vitro TSCs exhibit totipotent-like features in terms of
   transcriptome and developmental potential and thus serve as alternative
   models for totipotency study ([113]Xu and Liang, 2022). To evaluate
   totipotency-associated characteristics these in vitro TSCs
   recapitulated, we chose embryos across different developmental stages
   to compare the molecular (dis)similarity across these cellular models
   and in vivo cells using scRNA-seq analysis. In addition, mESCs were
   also included in the analysis as controls. Consistent with previous
   findings ([114]Hu et al., 2023), we observed that all in vitro TSCs
   showed higher similarity with the totipotent 2-cell embryos compared to
   mESCs ([115]Figure 3A). Interestingly, although these TSC models were
   all considered in vitro totipotent-like cells sharing core totipotent
   features, discrepancies in their gene expression patterns suggest that
   they may capture different molecular states within the totipotency
   spectrum, as reflected by their separated clustering positions in the
   UMAP ([116]Figure 3A).

Figure 3.

   [117]Figure 3
   [118]Open in a new tab

   Transcriptome discrepancies in totipotency across in vitro TSC models

   (A) The UMAP of in vitro mouse TSC models (2CLCs, ciTotiSCs, and TPSs),
   mESCs, and mouse embryo cells spanning from zygote to blastocyst (n =
   34,874 cells). 15 cell clusters were identified.

   (B) Expression patterns of maternal, minor ZGA, major ZGA, totipotent,
   and pluripotent genes in oocyte, zygote, 2-cell embryo, mESC, 2CLC,
   ciTotiSC, and TPS. ZGA, zygotic genome activation; mESC, mouse
   embryonic stem cell; 2CLC, 2-cell-like cell; ciTotiSC, chemically
   induced totipotent stem cell; TPS, totipotent potential stem cell.

   (C) Heatmaps showing gene expression patterns of maternal genes in
   oocytes, 2-cell embryos, 2CLCs, TPSs, ciTotiSCs, and mESCs (left).
   Functional enrichment analysis results (right). The x axis is the
   −log[10] (two-sided p value); the y axis represents the enriched
   pathways.

   (D) Heatmaps showing gene expression patterns of totipotent genes in
   2-cell embryos, 2CLCs, TPSs, ciTotiSCs, and mESCs (left). Functional
   enrichment (gene ontology, GO) analysis results (right).

   (E) Heatmaps showing gene expression patterns of minor ZGA genes in
   2-cell embryos, 2CLCs, TPSs, ciTotiSCs, and mESCs (left). Functional
   enrichment (gene ontology, GO) analysis results (right).

   (F) Heatmaps showing gene expression patterns of major ZGA genes in
   2-cell embryos, 2CLCs, TPSs, ciTotiSCs, and mESCs (left). Functional
   enrichment (gene ontology, GO) analysis results (right). The x axis is
   the −log[10] (two-sided p value); the y axis represents the enriched
   pathways. Shared, genes expressed in all three in vitro TSC models
   (2CLCs, ciTotiSCs, and TPSs); 2CLC-sig/ciTotiSC-sig/TPS-sig,
   cell-type-specific expressed genes.

   To investigate the discrepancies among these TSC models, we utilized
   the “Featured Gene” module to examine the expression patterns of
   previously reported signature genes, including maternal, minor ZGA,
   major ZGA, totipotent, and pluripotent genes ([119]Yang et al., 2022).
   In line with previous findings, these TSC models exhibited elevated
   expression of totipotent and ZGA genes and decreased expression of
   pluripotent genes compared to mESCs ([120]Figure 3B) ([121]Xu and
   Liang, 2022). Notably, we also observed differences in expression of
   maternal and totipotent genes in these in vitro TSCs, while comparing
   to that in zygotes ([122]Figure 3B). This suggests that, although all
   these three TSC models are widely used as in vitro models of
   totipotency and share core totipotent features, they also exhibit
   unique transcriptional signatures, likely reflecting their different
   origins, derivation methods, and molecular states within the totipotent
   spectrum. This diversity offers a valuable opportunity to explore
   multiple facets of totipotency that may not be fully captured in any
   single model.

   Next, based on the expression patterns of previously reported signature
   genes ([123]Yang et al., 2022) in the three in vitro TSC models, we
   categorized maternal, totipotent, and ZGA genes into both shared and
   cell-type-specific clusters using unsupervised hierarchical clustering.
   These included a shared cluster and three cell model-specific cluster,
   defined as the 2CLC-signature cluster, cTotiSC-signature cluster, and
   TSC-signature cluster ([124]Figures 3C–3F; [125]Table S3). Our analysis
   revealed both common and distinct expression patterns of these
   signature genes across the three in vitro TSC models ([126]Figures
   3C–3F). Specifically, while each model displayed unique transcriptional
   signatures, they all shared core totipotency features. As shown in
   [127]Figure 3D, the totipotent genes shared among the three in vitro
   TSC models were functionally enriched in pathways related to cell
   development and maturation, supporting a common totipotent identity.
   However, each in vitro TSC model also expressed a unique subset of
   totipotent genes, reflecting molecular distinctions among them.
   Notably, 2CLCs were uniquely enriched for totipotent genes associated
   with chromosome telomeric regions ([128]Figure 3D), including the
   well-established Zscan4 family genes, consistent with previous
   transcriptomic and epigenetic studies of 2CLCs ([129]Eckersley-Maslin
   et al., 2016). In contrast, cTotiSC-signature totipotent genes were
   enriched in pathways related to cell fate commitment, pattern
   specification, and regionalization ([130]Figure 3D). TPS-specific
   totipotent genes showed enrichment in immune response related pathways.
   These distinct features likely reflect differences in the origin and
   induction strategy of the models and suggest that each captures a
   different molecular state within the totipotent spectrum.

Key histone modifications correlated with gene expression in 2CLCs

   Our analysis showed that 2CLCs partially recapitulated the
   transcriptome features of totipotency, as evidenced by the extremely
   high expression of totipotent and minor ZGA genes in 2CLCs compared to
   other cells ([131]Figure 3B). To explore key epigenetic modifications
   modulating totipotent signatures in 2CLCs, we initially examined the
   correlation between gene expression and histone modifications, which
   were previously shown to regulate early development ([132]Zhou and
   Dean, 2015), including H3K4me3, H3K27ac, H3K27me3, and H3K79me3. When
   considering all protein-coding genes (PCGs), whose promoters regions
   overlap with histone modification peaks (±5 kb around the transcription
   start site [TSS] of each PCG), we only observed that H3K4me3
   modification showed association with their expression ([133]Figure 4A).
   Among the histone marks examined, H3K4me3 showed the strongest
   correlation with both totipotent genes and minor ZGA genes, compared to
   all PCGs in 2CLCs ([134]Figure 4A; [135]Table S4). In addition, the
   marker H3K27ac positively correlated with expression of both totipotent
   and minor ZGA genes ([136]Figures S1A and S1B). The Polycomb repression
   marker H3K27me3 ([137]Kundaje et al., 2015) showed no correlation with
   expression of all PCGs, totipotent genes, and minor ZGA genes in 2CLCs
   ([138]Figures 4B and [139]S1C), which is in line with the previous
   finding that H3K27me3 does not modulate 2C-like transition in mESCs
   ([140]Rodriguez-Terrones et al., 2018). H3K79me3 only showed a slightly
   positive association with totipotent genes but not all PCGs
   ([141]Figures 4B and [142]S1D). Furthermore, among the histone
   modifications analyzed, only H3K4me3 showed enrichment around the TSS
   of PCGs in 2CLCs ([143]Figure 4C). Collectively, these findings suggest
   that among histone modifications known to regulate embryonic gene
   expression, H3K4me3 is particularly strongly associated with the
   transcriptional activation of totipotent genes and minor ZGA genes in
   2CLCs. This observation is consistent with previous studies showing
   that H3K4me3 deposition closely correlates with the expression of
   upregulated 2C-specific genes ([144]Zhang et al., 2021), further
   supporting the robustness of the standardized data analysis pipeline
   implemented in Toti.

Figure 4.

   [145]Figure 4
   [146]Open in a new tab

   Identification of key histone modifications that modulate gene
   expression in 2CLCs

   (A) Correlation between H3K4me3 modification and expression of all
   protein-coding genes (PCGs; left), totipotent genes (middle), and minor
   ZGA genes (right).

   (B) Correlation between expression of all protein-coding genes (PCGs)
   and H3K27ac (left), H3K79me3 (middle), and H3K27me3 (right)
   modification. The x axis is normalized gene expression (log[2] (RPKM
   +1)) and the y axis is the density of reads overlapping with histone
   modified regions (±5 kb around TSS) and in 2CLCs.

   (C) Density distribution of reads overlapping with histone-modified
   regions (±5 kb around TSS) in 2CLCs. The x axis is the distance from
   the transcription start site. And the y axis is the read density.

The potential role of H3K4me3-Zscan4 axis in 2CLCs

   Zscan4 family genes are typically totipotent signature genes in 2CLCs
   ([147]Falco et al., 2007). Next, we investigated gene expression and
   epigenetic regulatory patterns around Zscan4a, Zscan4c, Zscan4d, and
   Zscan4f genes in 2-cell embryos, selected in vitro TSCs models (2CLCs,
   TPSs, ciTotiSCs, and TLSCs), and pluripotent mESCs. Zscan4c only
   expressed in 2-cell embryos and in vitro TSCs models but not in mESCs
   ([148]Figure 5A). Open accessible regions around Zscan4c also show a
   similar totipotent-specific pattern with accessible peaks existing only
   in 2-cell embryos and 2CLCs, when compared to mESCs. Importantly,
   H3K4me3 peaks around Zscan4c were specifically presented in 2CLCs and
   2-cell embryos but not in mESCs as well ([149]Figure 5A), suggesting
   the potential totipotency-relevant role of H3K4me3-Zscan4c axis.
   Similar expression and epigenetic modification patterns were also
   observed around Zscan4a, Zscan4d, and Zscan4f ([150]Figure S2). These
   findings are consistent with previous findings ([151]Genet and
   Torres-Padilla, 2020) showing totipotent-specific patterns in both
   expression and H3K4me3 modification surrounding Zscan4 family genes.

Figure 5.

   [152]Figure 5
   [153]Open in a new tab

   The potential role of H3K4me3-Zscan4c axis in 2CLCs

   (A) Read coverage represents gene expression, chromatin accessibility,
   and H3K4me3 modifications (from left to right) around gene Zscan4c in
   2C embryo, 2CLC, TPS, ciTotiSC, TLSC, and mESC. 2CLC, 2-cell-like cell;
   TPS, totipotent potential stem cell; ciTotiSC, chemically induced
   totipotent stem cell; TLSC, totipotent-like stem cell; mESC, mouse
   embryonic stem cell.

   (B) Functional annotation (gene ontology) of H3K4me3-Zscan4c axis
   (H3K4me3 modified regions located within ±2 kb around TSS of Zscan4c)
   in 2CLCs. The x axis is the −log[10] (two-sided p value), and the y
   axis represents the enriched GO terms.

   (C) Read coverage represents H3K4me3 modification and gene expression
   around Ctc1 in 2CLC and mESC.

   (D) Read coverage represents H3K4me3 modification and gene expression
   around Tinf2 in 2CLC and mESC.

   (E) Motif enrichment results for Zscan4c in Zscan4c-over-expressed
   mESCs.

   (F) Schematic illustrating the putative role of Zscan4c in maintaining
   genomic stability and telomere elongation.

   Next, we investigated the function role of H3K4me3-Zscan4c axis in
   2CLCs using the “Pathway Enrichment” module. Intriguingly, we found
   that H3K4me3-Zscan4c axis suggested enrichment in telomere maintenance-
   and chromosome telomeric region-relevant pathways ([154]Figure 5B). In
   addition to Zscan4c, we also observed that H3K4me3 modification and
   expression of other genes involved in telomere regulation and
   maintenance, such as Ctc1 ([155]Wang and Chai, 2018) and Tinf2
   ([156]Schmutz et al., 2020), showed similar patterns of enrichment in
   2CLCs compared to mESCs ([157]Figures 5C and 5D). Altogether, these
   results indicate the potential role of H3K4me3-Zscan4 axis in
   regulating the telomere in 2CLCs ([158]Figures 5B–5D; [159]Table S5).

   Furthermore, we examined the target sequences of Zscan4c in
   Zscan4c-over-expressed mESCs (ChIP-seq data in Toti) using the “Motif
   Enrichment” module. The top 2 enriched binding motifs displayed a
   repeated CA/TG pattern, suggesting the microsatellite signatures
   ([160]Figure 5E). The (CA)n/(TG)n microsatellites are susceptible to
   chromosomal recombination, which affects the genome integrity
   ([161]Gendrel et al., 2000). Zscan4 was previously found to recognize
   and bind to (CA)n/(TG)n microsatellites in nucleosomes, preventing them
   from forming an energetically unfavorable structure known as Z-DNA and
   disassembling under torsional stress ([162]Srinivasan et al., 2020).
   This mechanism potentially functions as a developmentally regulated
   safeguard that maintains genomic integrity in totipotent embryos, which
   is in line with the observed enrichment in “regulation of DNA
   recombination” pathway for H3K4me3-Zscan4c axis illustrated in
   [163]Figure 5B.

   Collectively, our primary analysis suggests the hypothesized role of
   Zscan4 in preserving genomic integrity by maintaining telomeres and
   safeguarding microsatellite regions in 2CLCs. As illustrated by this
   case study, we show that Toti could facilitate users to understand the
   potential function of H3K4me3 in regulating genomic stability in TSCs
   through activating Zscan4 ([164]Figure 5F). Our case study exemplified
   how to apply Toti to explore discrepancies of transcriptome and
   epigenome nuances underlying totipotency across mESCs, 2-cell embryos,
   and in vitro TSC models.

Discussion

   To the best of our knowledge, Toti is the first and unique multi-omics
   database exclusively developed for investigation of transcriptional and
   epigenetic factors shaping totipotency and can be accessed through a
   user-friendly Web interface ([165]http://toti.zju.edu.cn/). Toti
   unprecedentedly provides a comprehensive repository of transcriptomic
   and epigenomic data from 4,680 human and 3,604 mouse samples, covering
   in vivo, in vitro, and genome-edited embryonic TSCs, TSC-like cells,
   PSCs, and embryos spanning preimplantation stages, setting it apart
   from other embryonic and stem cell resources, such as dbEmbryo,
   DevOmics, and StemCellDB. The “Search” module enables users to easily
   and effectively focus on their interested studies/datasets by searching
   for genes, sequencing types, or other relevant keywords. This “Browse”
   module provides insights into the (dis)similarity across in vivo and in
   vitro totipotent cells, PSCs, and other embryonic cells in human and
   mouse by allowing intuitive and comparative visualization of epigenetic
   and transcriptomic features. With three sub-modules designed in the
   “[166]analysis” section, Toti also facilitates users to prioritize key
   motifs and biological pathways that might be able to capture gene
   regulatory logic underlying totipotency. Furthermore, Toti allows an
   online scRNA-seq analysis module tailored to investigation on cellular
   transcriptome underpinning of the establishment and exit of totipotency
   at single-cell resolution. We conducted a case study to provide a
   walkthrough of various modules, such as “Featured Gene,” “Genome
   Browse,” “Motif Enrichment,” “Pathway Enrichment,” “scRNA-seq
   analysis,” and “Download,” demonstrating how they can be effectively
   used to explore the potential applications of Toti for totipotency
   research. Our case study successfully replicated previously reported
   findings, testifying to the robustness, consistency, and reliability of
   our standardized data analysis pipeline. While we acknowledge that the
   analyses presented in this case study may not meet the stringent
   standards typically expected for standalone biological discoveries,
   they sufficiently demonstrate how researchers can utilize the
   integrated multi-omics datasets and multifaceted analytical modules in
   Toti for totipotency-related research. We believe Toti constitutes a
   comprehensive resource and opens new avenues for understanding the
   intricacies of totipotency.

   In the future, we will continuously maintain and update Toti by
   curating and incorporating the latest findings in TSC studies. As TSCs
   are the progenitors of all cell types in mammals and have great
   potential for regenerative medicine, we will continue to expand our
   collection of publicly available multi-omics data of TSCs to more
   species, such as pigs and monkeys. In addition, we will continue to
   integrate other data types, such as scATAC-seq and Hi-C data, to
   further improve our understanding of totipotency-associated regulatory
   networks. These additions are anticipated to provide insights into the
   molecular basis of totipotency and facilitate researchers to stabilize
   totipotency in vitro.

Methods

Data processing

   We processed the collected data using constructed pipelines, as
   illustrated in [167]Figure S3. Our study was based on GRCm38 (mm10) and
   GRCh38 (hg38) reference genomes for mouse and human, respectively.

scRNA-seq data harmonization

   We performed downstream single-cell analysis using Seurat (v.5.0.1)
   ([168]Butler et al., 2018). Data were normalized with SCTransform
   ([169]Hao et al., 2024), which uses regularized negative binomial
   regression to remove technical variation while maintaining biological
   heterogeneity. Compared to traditional log-normalization, SCTransform
   better stabilizes variance across expression levels, retaining signals
   from both lowly and highly expressed genes. To reduce integration bias
   from disparate sample sizes, we randomly down-sampled 1,500 cells per
   cell type, balancing dataset diversity and statistical power. Prior to
   integration, we compared harmonization methods, including
   FindIntegrationAnchors (Seurat), Harmony, and LIGER, using pooled
   scRNA-seq data from embryos at different developmental stages, with the
   goal of identifying the optimal integration approach that effectively
   minimizes batch effects while preserving biologically meaningful
   relationships and developmental hierarchies. FindIntegrationAnchors
   performed the best, maintaining biologically plausible distances on the
   UMAP and preserving developmental trajectories ([170]Figure S4). Thus,
   we integrated scRNA-seq datasets using FindIntegrationAnchors (anchor
   features: 5–20; neighbors filtering anchors: 30–300), followed by
   IntegrateData to correct for batch effects. Cell identities were
   defined by experimental sorting, ensuring high confidence without
   relying on post hoc gene-based annotation. To validate the robustness
   and biological relevance of our results, we performed
   principal-component analysis on pseudobulk expression profiles from
   scRNA-seq data of embryos at different developmental stages. As shown
   in [171]Figure S5A, cell types clustered by developmental stages:
   early, mid, and late 2-cell embryos clustered closely with ciTotiSCs,
   2CLCs, and TPS, while mESCs, ICM, and blastocysts formed a separate
   cluster, confirming the robustness of our integration and annotations.

Featured gene selection

   Representative genesets (maternal, minor ZGA, major ZGA, totipotent,
   and pluripotent genes) were previously defined in the literature
   ([172]Yang et al., 2022). We validated their robustness using two
   independent datasets: scRNA-seq data ([173]Deng et al., 2014) and bulk
   RNA-seq ([174]Wu et al., 2016). Despite the absence of oocytes in the
   scRNA-seq data, signature genes showed expected developmental
   expression. Totipotent genes were enriched in zygotes and early, mid,
   and late 2-cell populations ([175]Figure S5B). Similarly, in the bulk
   RNA-seq dataset, expression patterns of maternal, totipotent, and
   pluripotent genes aligned with oocytes, early embryos, and ICM/mESCs,
   respectively ([176]Figure S5C). Minor and major ZGA genes also showed
   developmental stage-appropriate activation. These consistent expression
   patterns support the robustness of these gene sets.

Resource availability

Lead contact

   Further information and resource requests should be directed to Xin
   Sheng (shengxin@zju.edu.cn).

Materials availability

   This study did not generate new unique reagents.

Data and code availability

   Toti is freely accessible at [177]https://toti.zju.edu.cn. All raw
   sequencing data are publicly available from SRA
   ([178]http://www.ncbi.nlm.nih.gov/sra) and ENA
   ([179]http://www.ebi.ac.uk/ena/browser/search). Metadata for all
   datasets are provided in [180]Tables S1 and [181]S2. Gene/isoform
   expression matrices and epigenetically modified regions in embryonic
   cells can be downloaded from the “Download” page. Data processing code
   is available at [182]https://github.com/cosmoss8274/Toti.

Acknowledgments