On the amniotic side, EXMC contribute to the formation of the mesodermal layer that overlays the amniotic epithelium. Cells in this amniotic mesoderm exhibit high expression of mesoderm-associated genes such as Fn1 and Foxf1, consistent with their mesenchymal identity, and reduced expression of epithelial markers including Epcam and Cldn7. In addition, amniotic mesoderm cells show robust expression of Postn, Bmp2 and Bmp4, among other factors^[21]. Functional studies indicate that BMP2 is required for normal morphogenesis of the amnion and chorion: BMP2 deficiency leads to abnormal closure of the pro-amniotic canal and subsequent defects in amniotic and chorionic development^[43].

Within the visceral yolk sac, local thickening of EXMC is observed around E7.5, coincident with the emergence of blood islands along the lateral walls. These structures comprise hemato-endothelial progenitors embedded in an EXMC-derived stromal matrix and represent the earliest site of primitive hematopoiesis. Transforming growth factor-β (TGF-β) signalling is critical for maintaining both the angiogenic and primitive hematopoietic functions of EXMC in the yolk sac^[44]. TGF-β promotes EXMC-derived angiogenesis by enhancing endothelial cell formation and maintaining vascular stability^[45,46]. Homozygous mutation of the basic helix-loop-helix transcription factor Hand1 results in embryonic lethality associated with defective EXMC development, leading to yolk sac malformations and abnormal cardiac looping^[47]. These findings position EXMC at the nexus of extraembryonic vascular, hematopoietic and cardiac development. They further demonstrate that perturbations of extraembryonic mesoderm propagate across multiple embryonic and extraembryonic organ systems.

On the placental side, EXMC together with ExE form the chorionic plate. By E8.5, the allantois—entirely derived from extraembryonic mesoderm—extends through the exocoelom towards the mural side and fuses with the chorion^[48,49] [Figure 1F]. Successful chorio-allantoic fusion permits blood vessels originating in the extraembryonic mesoderm of the allantois to invade the chorionic trophoblast. The intimate interdigitation of these fetal vessels with maternal blood sinuses establishes the labyrinthine layer of the mouse placenta, which is indispensable for effective maternal-fetal gas and nutrient exchange and therefore for embryonic survival and normal growth^[15]. Failure of chorio-allantoic fusion results in placental insufficiency and embryonic lethality.

BMP4 plays a pivotal role in allantoic development: at E6.5, Bmp4 is highly expressed at the posterior end of the primitive streak, and by E7.5 it is strongly upregulated in the emerging allantois, where it promotes mesodermal differentiation and outgrowth^[50]. Disruption of the adhesion molecule VCAM-1 or its receptor integrin α4, as well as combined deficiencies in BMP5 and BMP7, impairs or abolishes chorio-allantoic fusion^[42,51]. Moreover, Rspo3 deficiency prevents vascular invasion of the chorion^[52], while loss of Hgf or Wnt2 perturbs chorionic branching morphogenesis and labyrinth formation^[51]. Together, these studies highlight a complex signalling network in which BMP, WNT, RSPO and growth factor pathways orchestrate EXMC differentiation, allantoic elongation and placental morphogenesis.

During CS5a, continued thickening and cavitation of the epiblast generates the pro-amniotic cavity^[53,54]. By CS5c, as Exmes progressively fill the extraembryonic space, the amnion is organized as a bilayered structure comprising an epithelial sheet and an overlying Exmes-rich mesenchyme. In this configuration, Exmes provide a stromal scaffold that anchors the embryo in a relatively fixed position within the chorionic cavity. Subsequent asymmetrical growth and patterning along the A-P axis cause regional stretching of the Exmes-rich domain towards the posterior pole, where it condenses into the connecting stalk^[35]. This early stalk not only serves as a mechanical tether but also demarcates the posterior side of the bilaminar disc, coupling Exmes morphogenesis to axial specification.

Following primary yolk sac formation, a mesenchymal cell population emerges overlying the outer surface of the yolk sac endoderm, recognized as yolk sac Exmes, which provide structural and signaling support for subsequent primitive haematopoiesis. From around CS6b, focal aggregations of Exmes become evident along the yolk sac, corresponding to the earliest rudiments of blood islands, marking the onset of primitive wave of yolk sac hematopoiesis. By CS7, endothelial cell-lined blood channels emerge, and by CS8, primitive hematopoietic cells are observed within these nascent vessels^[4], marking the onset of the pro-definitive wave of yolk sac hematopoiesis. Recent human yolk sac cell-atlas studies have further shown that the yolk sac endoderm provides key hematopoietic growth factors that support the hemato-endothelial differentiation of Exmes/EXMC-derived progenitors^[55], underscoring a functional crosstalk between endodermal and mesodermal compartments.

Around 13-15 days post-fertilization^[34], Exmes/EXMC invade the solid primary villi formed by pCTB and pSTB, generating the mesenchymal core of the nascent placental villi. This Exmes/EXMC -derived villous stroma progressively differentiates into vascular endothelium, perivascular mesenchyme and placental macrophages-Hofbauer cells (HBCs). HBCs, in turn, contribute to vascular remodelling and angiogenic signalling. Notably, angiogenic activity and the emergence of embryonic macrophages precede the establishment of a fully functional fetoplacental circulation, raising the possibility that the placenta itself may act as an early extraembryonic haematopoietic niche. Consistent with prevailing models, HBCs may arise either from yolk sac-derived progenitors that migrate to the placenta or from hemogenic endothelium within placental vasculature. However, the precise developmental origin and relative contribution of these sources to primitive haematopoietic lineages in the placenta remain unresolved and will be discussed in subsequent sections.

In summary, the brief but highly dynamic window (CS3-CS9 in primates; E4.5-E8.5 in rodents) from implantation to gastrulation lays the groundwork for healthy embryogenesis. Proper early patterning and maturation of extraembryonic tissues-anchored structurally and functionally by Exmes/EXMC-are essential for sustained intrauterine growth and a physiologically normal pregnancy. The amnion safeguards the embryonic milieu, the yolk sac supports primary hematopoiesis, the allantois enables gas and nutrient transport, and the placenta orchestrates maternal-fetal exchange and immune regulation^[3,15,56-58]. These processes must unfold rapidly and with exquisite spatiotemporal precision, with Exmes/EXMC undergoing parallel maturation to build the stromal, vascular and immunological infrastructure of extraembryonic organs. Yet our understanding of these events remains fragmentary.

Ethical constraints^[59,60] and the scarcity of very early human and non-human primate samples present major obstacles, and species-specific variation further complicates comparisons. In rodents, extraembryonic mesoderm can be clearly traced to mesoderm emerging through the primitive streak during gastrulation. By contrast, in primates, as in other eutherian mammals, mesenchymal-like cell types in the extraembryonic compartment are observed before the onset of gastrulation^[61,62]. As a result, three principal lineage candidates have been proposed for the earliest Exmes: trophoblast, epiblast or hypoblast. At present, there is no definitive consensus on the developmental origin, lineage relationships or differentiation trajectories of Exmes in very early primate pregnancy, making this a central open question that next-generation in vivo and in vitro models must address.

An epiblast origin of primate Exmes has long been considered plausible because rodent EXMC are unequivocally derived from epiblast cells ingressing through the primitive streak during gastrulation. Morphological evidence supports this view. Re-analysis of human and rhesus monkey Carnegie embryos by Luckett^[3] revealed that primate Exmes development is homologous to that of other eutherian mammals, contradicting Hertig’s hypothesis that trophoblast cells are pluripotent and generate Exmes. Earlier, Florian had inferred from a 12-13-day human embryo that at least a portion of Exmes originate from the caudal epiblast margin^[68]. Luckett expanded this idea by proposing that this caudal epiblast represents a precociously forming primitive streak at CS5c-CS6 and may serve as the source of all Exmes in primates^[3]. Additionally, persistent expression of pluripotency-associated genes (OCT4, NANOG) in placental mesenchyme^[67] supports a potential epiblast lineage relationship.

Molecular evidences from in vitro cultured cynomolgus embryos provide further support. Single-cell analyses identified epiblast-derived Exmes-like cells expressing MIXL1, MESP1, ETS1, PODXL and EPHA4 by day 10. By day 14, cells expressing TBXT, MIXL1 and MESP1 emerged, marking gastrulation onset. A parallel population of Exmes (expressing PODXL, ETS1, PITX1, LUM, NID2) was detected at day 12; by day 14, these cells expressed PITX1, FOXF1, PDGFRA and HAND1 but not TBXT, MIXL1 or MESP1^[69]. These findings raise the possibility that epiblast cells may contribute to Exmes even before overt primitive streak formation. Therefore, an epiblast-derived contribution to Exmes prior to gastrulation cannot yet be excluded. Resolution of this question will require future synthetic embryology platforms that leverage lineage-restricted gene editors and high-fidelity barcoding.

An alternative model posits that Exmes might arise from before gastrulation. In early primate embryos, Exmes often appear morphologically continuous with extraembryonic endodermal cells lining the primary yolk sac. This morphological continuity supports a hypoblast origin. Historical anatomical studies provide initial evidence for this model. Gerard (1932)^[70], studying pre-primitive-streak chimpanzee embryos, proposed that the reticular extraembryonic mesenchyme derives from hypoblast. Enders later reported that visceral and yolk sac endoderm can delaminate and insert into the space between the primary yolk sac and the primitive pCTB, potentially giving rise to Exmes. Ultrastructural studies further revealed free or loosely attached transitional cells in this region that contain dense endoplasmic reticulum bodies characteristic of later Exmes maturation^[58]. More compelling molecular data come from early post-implantation cynomolgus embryos. In these specimens, early Exmes expressed hypoblast markers GATA4 and GATA6, whereas primitive streak markers (TBXT, MIXL1, CDX2) were diminished. These expression patterns suggest a hypoblast identity rather than epiblast or streak origin^[71]. In vitro embryo models reinforce this interpretation. In human ESC-derived and cynomolgus embryo-like structures, PrE cells and Exmes share markers including PDGFRA, GATA4, GATA6 and NID2^[72,73]. Moreover, BMP4-based induction of PrE produces Exmes-like cells (LUM+, ANXA1+, LIX1−, TBXT−)^[61], consistent with a hypoblast-origin pathway.

In early post-implantation cynomolgus embryos (CS5), Exmes are already transcriptionally distinct from visceral and yolk sac endoderm (VE/YE). Exmes upregulate COL6A1 while downregulating FOXA1, whereas VE/YE exhibit the opposite pattern^[71]. These findings support the view that Exmes are hypoblast-related but molecularly divergent lineage, rather than a simple extension of VE/YE identity.

Stem cell-based models have provided important mechanistic insights into the specification and maintenance of Exmes. Directed differentiation studies demonstrate that modulation of BMP, WNT and Nodal signalling is sufficient to induce Exmes-like populations from human pluripotent stem cells (hPSCs). For example, primed hESCs can be differentiated into early Exmes-like cells through coordinated activation of BMP and WNT pathways alongside modulation of Nodal signalling^[10]. Similarly, naïve human pluripotent stem cells can be converted into self-renewing Exmes populations through inhibition of Nodal signalling and GSK3β, with sustained proliferation requiring activation of mTOR and BMP pathways. Consistent with this, inhibition of BMP signalling reduces SMAD1/5/9 phosphorylation and markedly impairs Exmes expansion^[74].

Additional studies further indicate that activation of WNT signalling can generate Exmes populations with hematopoietic potential, characterized by expression of MEF2C, CD34, PECAM1 and CDH5^[75]. Notably, multiple induction systems applied to naïve hESCs (AIC-N), naïve pluripotent stem cells and hiPS cells all yielded Exmes-like cells with transcriptional signatures closely resembling hypoblast-derived Exmes [Table 1].

Among these models, the model proposed by Pham is particularly informative regarding dual contributions from epiblast- and hypoblast-related routes to extraembryonic mesoderm. In this system, naïve hPSCs are first directed through an epiblast-like state (expressing POU5F1/OCT4 and NANOG) and then induced, using ASECRiAV conditions, to differentiate into extraembryonic mesoderm cells expressing POSTN, VIM, NID2, LUM, FOXF1, ANXA1 and DCN. In parallel, conversion of naïve hPSCs to PrE using RACL generates extraembryonic mesoderm cells expressing NID2, FOXF1, VIM and ANXA1. Notably, this hypoblast-biased route also appears to transit through an intermediate epiblast-like phase, as indicated by transient expression of POU5F1, NANOG, SOX2, DPPA2, GDF3, ZNF208 and ZNF729^[74]. Together, these data suggest the notion that Exmes can arise through multiple developmental trajectories, potentially reflecting the temporally distinct contributions inferred in vivo.

Beyond lineage origin, Exmes exhibit a distinct molecular identity consistent with their structural and regulatory roles in early development.

Exmes form an extracellular matrix (ECM)-rich mesenchymal stroma, characterized by high expression of ECM components^{[61,71,73,75]}. They produce abundant fibronectin (FN1), collagens (e.g. COL1A1, COL6A1) and laminin-111, which accumulate prominently around the embryonic-extraembryonic interface, particularly in the connecting stalk region^[71]. In trilaminar cynomolgus monkey embryos derived from naïve ESCs, Exmes show upregulation of EMT-related ECM genes such as POSTN and COL6A1^[76].

Moreover, Exmes display a mesenchymal phenotype marked by expression of PDGFRA and activation of EMT-related transcriptional programme^[7,61,71,73]. For instance, in early post-implantation cynomolgus embryos (CS5), Exmes show high expression of SNAI2 and HGF^[7], consistent with activated EMT and invasive behaviour during early development.

In addition to their structural role, Exmes function as a signalling hub that interfaces between embryonic and extraembryonic tissues via ligand-receptor networks. They express key components of TGF-β signalling which regulates EMT, including TGFB1, ZEB2, SMAD3, SNAI2 and VIM^[73], as well as ligands associated with WNT and BMP pathways^[7]. This suggests that Exmes act as a signalling hub to modulate epiblast, hypoblast and trophoblast compartments.

Taken together, these features support a model in which early Exmes are primarily generated from hypoblast-related cells via EMT and robust ECM production. As development proceeds and gastrulation begins, the bilaminar disc transitions into a trilaminar structure. The newly formed mesoderm layer expands extraembryonically, and EXMC derived from the primitive streak infiltrate regions previously occupied by Exmes. At this stage, a mixed Exmes/EXMC population builds the stromal, vascular and immunological architecture of extraembryonic tissues. It is this composite Exmes/EXMC compartment that ultimately underpins the diverse structural and functional roles traditionally attributed to the extraembryonic mesoderm.

In rodents, the allantois is not merely an excretory and respiratory appendage but also a source of hematopoietic and endothelial progenitors^[92]. Pre-fusion allantoic explants can generate erythroid and myeloid cells in vitro^[93], demonstrating an intrinsic hematopoietic potential before vascular connection with the yolk sac. Allantoic hematopoiesis is Runx1-dependent: Runx1-deficient allantois fails to produce hematopoietic cells^[94]. Vascularization within the allantois exhibits a distal-proximal polarity, with a branched vascular plexus distally and straight umbilical arteries proximally, ensuring appropriate docking with the chorion and correct vascular patterning^[48].

In primate embryos, rapid expansion of Exmes drives the emergence of the connecting stalk. Between days 13 and 16, Exmes initially cover the outer surface of the amnion and yolk sac^[3]. Because Exmes cannot keep pace with this expansion, the embryo grows predominantly towards the anterior side, and Exmes condenses into a stalk-like structure that remains tethered at the posterior pole^[35]. The connecting stalk later differentiates into extraembryonic structures including the umbilical cord and allantois, which link to the placenta and mediate maternal-fetal exchange. Functionally, the connecting stalk establishes a conduit between embryonic and extraembryonic tissues that enables the circulation of fetal blood between intraembryonic and extraembryonic compartments. In addition, the connecting stalk has been proposed to act as a signalling centre that helps pattern the embryonic A-P axis, potentially through cell migrations that occur after gastrulation. In this context, it is plausible that, following gastrulation, mesenchymal cells in the stalk region become admixed with EXMC generated from mesoderm, mirroring the mixed Exmes/EXMC state inferred in other extraembryonic territories.

Single-cell transcriptomic analyses of human CS7-CS8 embryos identified a population of cells at the posterior mesoderm that expressed markers associated with the connecting stalk, including CDX1, CDX2, CDH2 and LCN15^[79,80]. CDX2 also serves as a molecular marker for the posterior EXMC, which includes the connecting stalk. Knockout of CDX1/CDX2 leads to proliferation defects in EXMC and delayed allantoic development^[82,95]. Under normal conditions, FGF and BMP signalling at the posterior pole suppress Nodal/ACTIVIN activity, downregulate NANOG and upregulate BRA and CDX2, thereby promoting posterior fates while inhibiting anterior lateral plate mesoderm, definitive endoderm programmes, and cardiac development. In the anterior domain, NANOG antagonizes CDX2 expression^[82].

Gene perturbation studies of CDX2 and TBXT further indicate that EXMC downregulate the WNT inhibitor DCDC2, and that angiogenesis is attenuated through modulation of VEGF signalling-via downregulation of ANK3 and ANGPT1 and upregulation of TMSB10^[95]. These data link posterior EXMC identity to vascular patterning and hematopoietic competence in the connecting-stalk and allantoic region.

In the mouse, the placenta is not only a niche for hematopoietic stem cells (HSCs)^[96,97] but also an active hematopoietic organ^[97-99]. Early studies identified functional hematopoietic cells with immune capabilities in the placenta^[98]. Immature hematopoietic cells expressing CD41, Runx1, c-Kit, CD31, CD34 and Sca-1 have been detected in the labyrinthine zone^[100]. The number of HSCs in the placenta peaks at E12.5-E13.5, reaching about 15-fold the number in the AGM region^[96]. Importantly, hematopoietic cells capable of autonomous generation are present in the chorion before chorio-allantoic fusion^[93,94], indicating that chorionic hematopoiesis does not merely reflect colonizing from the yolk sac or circulation. The spatiotemporal expression of Runx1 in the chorion and at the chorion-allantois interface resembles that of integrin α4^[94], a key mediator of chorio-allantoic fusion^[42,51], suggesting that Runx1 may coordinate both fusion and the establishment of placental hematopoietic niches.

In primates, the expansion of Exmes play an inductive role in chorionic villus development. Villous morphogenesis begins at approximately days 11-13, when the chorion differentiates from trophoblast. pSTB invades the decidua, while pCTB proliferates to form primary villi, consisting of pCTB cores surrounded by multinucleated pSTB [Figure 3A and B]. Around days 13-15, Exmes infiltrate these primary villi to generate the first mesenchymal cores, filled with a reticular matrix and constituting secondary villi^[34] [Figure 3C and D].

Paracrine factors secreted by trophoblast cells are widely thought to induce villous mesenchymal differentiation and initiate vascular development^[101,102]. The first fetal capillaries appear around days 18-20. By approximately day 22, angiogenic cells and endothelial cords interconnect through loose junctions and elongate along the villous axis, ultimately giving rise to branched vessels^[101]. Perivascular and smooth muscle cells then encase these primitive tubes to form contractile vessels, which mature by approximately 28 days and contribute to tertiary villi. Chorionic villi near the connecting stalk further differentiate into the definitive placenta, which serves as the primary maternal-fetal interface.

Within this framework, Exmes expansion and differentiation provide both the scaffold and the cellular substrate for villous angiogenesis and early placental hematopoiesis, helping to explain how villi can generate blood vessels and blood cells before the umbilical circulation is fully established.

In mammals, the first transient blood cells are generated in the yolk sac^[103]. Early yolk sac hematopoiesis is classically divided into two waves. A primitive wave, initiating before or around CS7, produces early erythromyeloid and megakaryocytic progenitors, whereas a subsequent definitive wave at CS8-CS9 generates erythromyeloid and lymphoid-lineage cells. Blood islands on the yolk sac function as the first recognizable hematopoietic centres^[4] [Figure 8]. These blood islands arise from solid clusters of mesenchymal cells aggregating near the endoderm within the yolk sac mesoderm-regarded as the anlage of blood islands-and are subsequently induced to differentiate by adjacent endodermal cells [Figure 8]. Structurally, blood islands consist of flattened endothelial progenitors encasing primitive hematopoietic progenitors, forming a characteristic hemato-endothelial unit^[4] [Figure 8]. The flattened endothelial progenitors of blood islands differentiate into hemogenic endothelium through hemogenic specification, which subsequently gives rise to hematopoietic stem and progenitor cells via the endothelial-to-hematopoietic transition pathway^[104]. Rapid expansion and remodelling of these units give rise to an extensive vascular plexus that envelops the yolk sac in a complex arborizing network. The primitive wave of yolk sac hematopoiesis satisfies the early hematopoietic demands of the embryo, and with the onset of cardiac contractions at CS10^[91], yolk sac-derived blood cells are disseminated throughout the embryonic body.

In vitro models, Exmes induced from PrE recapitulate key features of this early extraembryonic hematopoietic niche. In addition to upregulating EMT- and ECM-associated genes, these Exmes populations show increased expression of angiogenesis- and hematopoiesis-related genes such as KDR and DPP4, as well as the placental angiogenesis-associated gene EMP2, from around day 8 of induction^[61]. In another human embryo-like model, the heX-embryoid-which contains yolk sac hematopoietic tissue and an extraembryonic niche-a distinct Exmes population has been identified. These cells express high levels of the surface marker BST2 and ECM components, while co-expressing endothelial genes (CD34, PECAM1) and hematopoietic genes (GATA1, KLF1, ZEB1, CD235)^[78]. In addition, a population of endothelial cells expressing CD34, CDH5, ERG and GATA2, but lacking CD73, has been observed, consistent with an EHT toward progenitor cells^[78]. Together, these findings demonstrate that Exmes derived from yolk sac mesoderm make a substantial contribution to the primitive wave of hematopoiesis during early development and highlight the capacity of in vitro models to capture early extraembryonic hematopoietic programmes.

The placenta is a pivotal extraembryonic organ required for gas exchange, nutrient supply and endocrine support of the developing embryo. Its function is critically dependent on the timely establishment of maternal-fetal circulation. The basic villous architecture of the human placenta is in place by approximately 4 weeks of gestation^[15], coinciding with the period during which Exmes fill the villous cores and undergo differentiation and maturation. However, the temporal uncoupling between the appearance of HBCs in human placental villi at approximately CS8^[90], well before the onset of embryonic heart beat and the establishment of a functional fetoplacental circulation at CS10^[91], has intensified interest in whether Exmes within the villous stroma support an early, in situ hematopoietic programme independent of fetal blood cellular colonization.

The placenta is well established as a niche for the maturation of hematopoietic progenitors: throughout gestation, it harbours both hematopoietic progenitor cells and HSCs^[96,105,106]. Macrophages arise earlier than HSCs and derive from the first wave of yolk sac primitive hematopoiesis, subsequently colonising tissues as tissue-resident macrophages (TRMs)^[107]. In the placenta, HBCs function as TRMs and are essential for maintaining normal placental development and pregnancy homeostasis^[107]. A recent study reported that two populations of Exmes/EXMC contribute respectively to placental mesenchymal cells and to HBCs, erythrocytes, and endothelial cells^[108].

HBCs express CD14 and CD68, together with receptors including CD163, CD206, CD64 and FOLR2^[109]. In early pregnancy, HBCs derived from primitive hematopoiesis lack expression of the classical macrophage marker HLA-DR^[109], a phenotype linked to epigenetic silencing of CIITA (class II transactivator)^[110].

Studies have shown that fetal-derived CD34⁺, CD43⁺ and FXIII/CD68⁺ macrophage precursors are already present in villous cores before the establishment of placental circulation^[111]. Single-cell and functional assays indicate that early human HBCs share transcriptional signatures with yolk sac macrophages and express proliferation-associated genes such as Ki67. These findings suggest that early placental macrophages originate from primitive hematopoiesis and are capable of local proliferation and self-renewal within the placenta^[109].

HBCs can be classified into two major subpopulations based on their origin in early and late pregnancy. HBCs in early pregnancy, which arise either from villous mesenchymal stem cells or from monocyte precursors originating in the yolk sac, are potentially associated with tissue development and morphogenesis; HBCs in late pregnancy, which derive from definitive HSCs are likely involved in hematopoiesis and immune functions^[112].

These subtypes exhibit distinct expression patterns: CD68 is expressed throughout gestation; CD1, CD4, CD11c, CD14, CD16, CR3 and CXCR4 are enriched in early and late pregnancy; ACP and G6PD are highly expressed in early to mid- pregnancy; CD163, CD209, and HLA-DQ are specifically highly expressed in late pregnancy^[113].

The precise origin of HBCs in early pregnancy, whether primarily placental or yolk sac-derived, remains unresolved. Spatiotemporally, the very early appearance of HBCs is congruent with their putative functions: they must differentiate and mature rapidly to support tissue remodelling, repair and angiogenesis, rather than waiting for yolk sac hematopoiesis and cardiac-driven circulation to be fully established.

At day 18, HBCs have been observed within placental villi^[114,115]. Importantly, the appearance of HBCs precedes the onset of fetal circulation at CS10^[91], when blood flow to the placenta is not yet established, making large-scale migration of macrophage precursors from the yolk sac into villi mechanistically challenging.

It is also debated whether early HBCs arise from a yolk sac-independent primitive programme within the placenta or from yolk sac-derived primitive progenitors that colonize the villi. In the mouse, HBC-like placental macrophages can arise from erythro-myeloid progenitors (EMPs) of primitive hematopoiesis. Analogously, a population of CD34⁺ CD43⁺ HLF⁺ placental erythro-myeloid progenitors (PEMPs) has been identified in early human placenta, transcriptionally reminiscent of murine EMPs. PEMPs express high levels of CD45 and c-KIT and, in culture, give rise to HLA-DR⁻ HBCs, allowing reconstruction of their differentiation trajectory^[110].

Current hypotheses regarding the origin of HBCs in primate placenta therefore include: (i) differentiation from villous mesenchymal cells; (ii) migration and differentiation of yolk sac derived mononuclear precursors; and (iii) recruitment of yolk sac-derived macrophages after the onset of embryonic circulation. HBCs of distinct developmental origin are likely to perform non-overlapping functions, contributing to HBCs heterogeneity.

Disentangling the relative contributions of placental and yolk sac-derived hematopoiesis remains challenging, not only because of technical limitations in tracing early macrophage ontogeny, but also because very early-stage human specimens are rare and human and murine placentas differ substantially in structure, architecture and timing.