Extravasation of immune and tumor cells from an endothelial perspective Free

Crossing the vascular wall is essential for immune cells to reach the site of inflammation. Leukocyte passage through the endothelium, known as transmigration, is a multistep process. Immune cells are first captured by and adhere to the vascular wall, after which they migrate and crawl along the endothelial cells (ECs) to detect the best exit site (Muller, 2011). Leukocytes, including monocytes and neutrophils, cross the endothelium either through a single EC, known as the transcellular route, or more commonly between ECs via the so called paracellular route, also known as diapedesis (Mamdouh et al., 2009; Schulte et al., 2011). Similarly, circulating tumor cells (CTCs), which originate from primary tumors or other metastatic sites and disseminate through the blood circulation, must pass through the endothelium to seed in distant organs and form metastases. CTCs are commonly described as ‘hijackers’ of the leukocyte transmigration process, as they mimic leukocyte tethering, arrest and extravasation mechanisms (Leong et al., 2014; Miles et al., 2008; Paku et al., 2000).

Under homeostatic conditions, the endothelium constitutes a natural barrier against immune or tumor cell transmigration. ECs are activated by vasoactive compounds (cytokines and chemokines) that are secreted by infected or injured tissues during inflammation or by the primary tumor during metastatic progression (García-Román and Zentella-Dehesa, 2013; Mantovani et al., 1992; Mendoza et al., 2004). In addition, when subjected to inflammatory or tumoral stimuli, vascular cells undergo significant alterations in gene expression and function, including the overexpression of adhesion receptors (such as E-selectin) and chemoattractants (Hiratsuka et al., 2011; van Buul et al., 2007b). Chemokines and adhesion receptors presented at the surface of ECs promote the adhesion of leukocytes and CTCs to the vascular wall, guiding them to the best site for extravasation (Gassmann et al., 2009; Vestweber, 2015). Cytokines also modulate the permeability of the endothelium by promoting endothelial cell–cell junction disassembly (Broermann et al., 2011; Holopainen et al., 2012) and EC retraction (Garcia et al., 1998; Honn et al., 1989), facilitating the opening of gaps in the endothelium through which leukocytes and tumor cells can pass to reach the perivascular environment.

Although endothelial permeability is restored after leukocyte transmigration, the mechanisms of tumor cell extravasation can irreversibly damage the endothelium and the underlying basement membrane (Paku et al., 2017; Strilic et al., 2016). This highlights important discrepancies between leukocyte and cancer cell extravasation. On the one hand, endothelial involvement supports the delivery of leukocytes to injured tissues; on the other hand, it facilitates cancer cells homing to distant organs and the development of metastasis. Extravasation represents a bottleneck in metastatic dissemination, with ∼1–5% of CTCs successfully exiting the vasculature (Lambert et al., 2017). This specific step in the metastatic cascade is a promising druggable target for preventing metastatic colonization (Follain et al., 2021; Nozaki et al., 2010). Understanding the differing mechanisms of endothelial involvement during inflammation versus metastasis could greatly benefit the development of antimetastatic therapies.

In this Review, we describe how the endothelium actively remodels to favor the extravasation of immune and tumor cells. We particularly focus on the similarities and differences between endothelial functions during immune cell and CTC adhesion and diapedesis. Finally, we discuss how the transmigratory sites and modalities during inflammation and metastasis are regulated by the biochemical and biophysical characteristics of the endothelium.

While circulating in the bloodstream, immune cells must detect inflammation sites to extravasate precisely where tissues are damaged. During the initiation phase of inflammation, the endothelium is activated to guide immune cells to the injury site. Injured tissues secrete inflammatory mediators, which elicit endothelial activation through regulated gene expression and morphological modifications (Fig. 1). Inflammatory stimuli include cytokines, such as TNF, IFNγ and interleukins (e.g. IL-1β). Other stimuli include pathogen-associated molecular patterns (PAMPs), damage-associated molecular patterns (DAMPs) and bacterial endotoxins (Theofilis et al., 2021). PAMPs are molecules associated with groups of pathogens, whereas DAMPs are molecules released from damaged or dying cells. The binding of cytokines to their endothelial receptors induces the activation of the nuclear factor (NF)-κB pathway, which upregulates the expression of endothelial adhesion receptors, including E-selectin, P-selectin, vascular cell adhesion-1 (VCAM-1) and intracellular cell adhesion-1 (ICAM-1) (Anwar et al., 2004; Bochner et al., 1991; Collins et al., 1995; Rahman et al., 1998; Wertheimer et al., 1992; Wyble et al., 1997). Inflammatory cytokines also promote the expression of the monocyte chemoattractant protein-1 (MCP-1; also known as CCL2) and drive monocyte recruitment (Rollins et al., 1990). The upregulation of adhesion receptors and chemoattractants creates local adhesion hotspots, enabling circulating leukocytes to engage with the endothelium at the site of inflammation. For example, E-selectin and P-selectin facilitate leukocyte capture and rolling, whereas ICAM-1 and VCAM-1 mediate stronger binding to the endothelium (Lawrence and Springer, 1991). Selectins are a family of transmembrane glycoproteins comprising L-selectin, which is expressed by most leukocytes, as well as E-selectin and P-selectin, which are expressed by inflamed ECs. Selectins bind specifically to glycoproteins capped with the sialyl Lewis x (sLe^x), a carbohydrate structure that determines selectin binding specificity. sLe^x typically decorates endothelial CD34, glycosylation-dependent cell adhesion molecule-1 (GlyCAM-1) and mucosal vascular addressin cell adhesion molecule-1 (MadCAM-1) (Baumheter et al., 1993; Berg et al., 1993). E-selectins bind to E-selectin ligand-1 (ESL-1; also known as GLG1), P-selectin glycoprotein ligand-1 (PSGL-1; also known as SELPLG), glycosylated CD44 and sLe^x-decorated L-selectin expressed on the surface of leukocytes (Zarbock et al., 2011). Meanwhile, L-selectin engagement with endothelial sLe^x residues captures leukocytes at the endothelium (Ivetic et al., 2019). Leukocytes express PSGL-1 (which also binds to endothelial P-selectin; McEver and Cummings, 1997), CD44 (a receptor for hyaluronan that also binds selectins; Clark et al., 1996; Katayama et al., 2005) and fibronectin (which is found on the luminal side of the endothelial wall; Jalkanen and Jalkanen, 1992; Barbazán et al., 2017; Osmani et al., 2019). Engagement of selectins with their ligands is reversible and transient, leading to leukocyte rolling on the surface of the endothelium monolayer (Kaplanski et al., 1993; Ley et al., 1995; Mayadas et al., 1993). Interestingly, the shear stress typically found in vasculature of organs stabilizes selectin bonds and enhances leukocyte selectin-mediated adhesion to the vascular wall (Lawrence et al., 1997; Yago et al., 2004). Furthermore, rolling neutrophils, as well as effector and regulatory T cells, extrude membrane tethers at the rear of the cell and slings at the front. These membrane protrusions are enriched with PSGL-1, which further stabilizes leukocyte adhesion to the endothelium (Abadier et al., 2017; Ramachandran et al., 2004; Sundd et al., 2012). Here, engagement of PSGL-1 by endothelial E- and P-selectin, and of CD44 by E-selectin, triggers the activation of signaling pathways (involving Src kinase, GTPases and Ca²⁺ signaling), which eventually stimulate the activation of integrins (Alon and Ley, 2008; Luo et al., 2007; Zarbock et al., 2011).

Rolling slows leukocyte velocity, allowing leukocyte integrins to interact with luminal endothelial receptors, which results in stronger adhesion interactions with the endothelium (Fig. 2). Integrin activation occurs via outside-in signaling, the activation of integrins following receptor binding to their ligand, downstream of selectin binding and in response to endothelium-displayed chemokines (Alon and Ley, 2008; Shamri et al., 2005). Leukocytes express α4β1 integrin (also known as very late antigen-4, VLA-4) and β2 integrins [αLβ2 and αMβ2; also known as lymphocyte function-associated antigen 1 (LFA-1) and macrophage-1 antigen (Mac-1), respectively] (Wen et al., 2022; Yago et al., 2010). Specifically, α4β1 binds to endothelial VCAM-1 (Alon et al., 1995), whereas LFA-1 and Mac-1 bind to ICAM-1 and ICAM-2 on the EC surface (Kuwano et al., 2010; Sundd et al., 2012). Shear stress promotes ICAM-1 nanoclustering to favor β2-integrin-dependent leukocyte arrest (Piechocka et al., 2021). It has recently been suggested that ICAM-1 binding favors a high-affinity state for β2 integrins (Fan et al., 2019). Finally, activation of β2 integrins following engagement to ICAM-1 is finetuned by E-selectin and P-selectin adhesion kinetics (Zhou et al., 2021).

ICAM-1 and VCAM-1 are members of the immunoglobulin superfamily with an extracellular domain composed of several Ig domains, and intracellular domains that support outside-in signaling required for the stabilization of leukocyte adhesion and further transmigration (Greenwood et al., 2003; Lyck et al., 2003; Staunton et al., 1988; van Buul et al., 2007a; Yang et al., 2006). Upon binding to leukocyte LFA-1 and Mac-1, ICAM-1 forms homophilic dimers that cluster into plaques surrounding adhered leukocytes. ICAM-1 clustering and VCAM-1 ligation to α4β1 induce the recruitment of VCAM-1 to the adhesive plaque, demonstrating that outside-in signaling also occurs in the endothelium upon leukocyte binding (Barreiro et al., 2002). In addition, early binding of neutrophils to the endothelium induces Ca²⁺ entry mediated by PIEZO1, a mechanogated Ca²⁺ channel, in an ICAM-dependent manner. This process leads to remodeling of the actin cytoskeleton and an increase in myosin contractility of ECs (Wang et al., 2022). These changes might contribute to a localized alteration in endothelial cortical stiffness, which has been suggested to be necessary for efficient exploration of the endothelial surface by leukocytes to identify extravasation hotspots (Schaefer and Hordijk, 2015). Meanwhile, actin filaments have been observed to polymerize into microvilli-like protrusions extending towards adhered leukocytes. The role of these microvilli-like structures is uncertain, reflected in the various names attributed to them, including filopodia, surface or junctional membrane protrusion and ‘endothelial docking structures’ (Fig. 1), demonstrating their necessity for stable leukocyte adhesion (Barreiro et al., 2002). After leukocyte adhesion, ICAM-1, VCAM-1 and their actin-binding protein (ABP) partners cluster into these actin-rich structures. ICAM-1 clustering relies on the actin nucleation-promoting factor cortactin, as its depletion compromises leukocyte transition from rolling to firm adhesion (Schnoor et al., 2011). Thus, the docking structure acts as an adhesive plaque that facilitates the transition of leukocytes from rolling to crawling, although it does not support diapedesis (Kroon et al., 2018). Conversely, several reports suggest that large endothelial protrusions form an apical cup (discussed in detail later) that promotes leukocyte transmigration but not adhesion (Carman and Springer, 2004; van Buul et al., 2007a). Nevertheless, there is consensus that endothelial microvilli enhance tight adhesive interactions with arrested leukocytes. Interestingly, microvilli-like structures also stimulate the exchange of chemokines between the endothelium and immune cells. For instance, IL-8 has been observed by electron microscopy to localize at the tip of endothelial microvilli (Middleton et al., 1997). Cytokines, such as IL-8 (also known as CXCL8) or CCL25, induce the activation of β2 integrins in leukocytes, thereby strengthening their adhesion to the endothelium (Detmers et al., 1990; Li et al., 2024). Furthermore, endothelial CD44 colocalizes with ICAM-1 in filopodia and additionally recruits hyaluronan-bound chemokines to the docking structure. Chemokines recruited through CD44–hyaluronan interaction support leukocyte adhesion and extravasation (van Steen et al., 2023).

Altogether, the engagement of leukocytes with the endothelial wall represents a central event preceding extravasation, encoded in a two-step mechanism dependent on specific adhesion receptors.

The dissemination of cancer cells throughout the body and the formation of metastases often exhibit a non-random distribution pattern where CTCs specifically settle in particular organs, a phenomenon known as organotropism (Gao et al., 2019). Several factors contribute to the specific arrest and adhesion of cancer cells to distinct vascular beds. Initially, the ‘seed and soil’ hypothesis proposes that this specificity arises from compatible molecular interactions between CTCs (‘seed’) and the local microenvironment (‘soil’) (Fidler, 2003). Although originally stating that tumor cells would only form metastases in distant organ niches that fit their specific requirements, the seed and soil hypothesis can now be extended to suggest that CTCs will arrest in vascular beds where the surface adhesion molecules on the blood vessel endothelium match those in their own adhesome (Osmani et al., 2019; Offeddu et al., 2021). Similar to the preparatory phase during leukocyte extravasation, endothelial pre-activation supports the arrest and adhesion of CTCs, which are crucial steps in the extravasation cascade (Fig. 1). Tumor cells secrete a variety of soluble factors, such as growth factors, inflammatory cytokines and chemokines, as well as extracellular vesicles (EVs). These factors circulate in the bloodstream and promote the establishment of a conducive microenvironment known as the premetastatic niche (García-Román and Zentella-Dehesa, 2013; Peinado et al., 2017) (Box 1). The premetastatic niche is characterized by the presence of immunosuppressive cells [myeloid-derived suppressor cells, macrophages and regulatory T cells (Treg)], inflammation markers, increased angiogenesis and enhanced vascular permeability (Liu and Cao, 2016). Additionally, the secreted factors are directly taken up by ECs in the capillary beds of distant organs, facilitating their priming (Ghoroghi et al., 2021; Mary et al., 2023). Interestingly, the adhesion repertoire expressed by ECs is influenced by pre-activating signals secreted by tumor cells. Specifically, E-selectin expression is upregulated prior to CTC arrest and adhesion. In vitro studies have shown that PCI-24 pancreatic cells secret IL-1α and IL-6, which induce E-selectin expression in human umbilical endothelial cells (HUVECs) and activates in synergy with monocytes the expression of E-selectin in HUVECs through IL-1β (Narita et al., 1995). Colon cancer cells stimulate hepatic Kuppfer cells to produce IL-β1, TNF and IL-6 (Auguste et al., 2007; Gangopadhyay et al., 1998), whereas breast cancer cells induce macrophage-derived TNF secretion (Eichbaum et al., 2011). Altogether this upregulates ICAM-1, VCAM-1 and E-selectin expression in ECs (Auguste et al., 2007; Eichbaum et al., 2011; Gangopadhyay et al., 1998).

The seed and soil hypothesis is further elaborated by integrating mechanical cues that CTCs encounter during their vascular dissemination (Azevedo et al., 2015; Follain et al., 2020). CTC arrest can occur via vascular occlusion or cellular adhesion, depending on whether hemodynamic forces exceed the adhesive forces between the ECs and CTCs (Follain et al., 2018). Adhesion of CTCs to the endothelium involves several adhesion receptors (Fig. 2), similar to leukocytes, starting with weak but rapid engagement followed by stronger but slower adhesion (Osmani et al., 2019). E-selectin is among the initial endothelial receptors involved in CTC adhesion, and CTCs preferentially arrest in vessels where E-selectin presentation is upregulated (Auguste et al., 2007; Barthel et al., 2013; Shea et al., 2017). Treatment of mice with anti-E-selectin antibodies has been shown to reduce metastasis of colorectal and lung cancer to the liver (Brodt et al., 1997), whereas overexpression of E-selectin in the liver in vivo redirects metastasis to this organ (Biancone et al., 1996). Cancer cells express various ligands decorated with sLe^x or sLe^a including CD44, PSGL-1, CD24, mucin 1 (MUC1), galectin-3 and death receptor 3 (DR-3, also known as TNFRSF25) (Aigner et al., 1998; Buffone and Weaver, 2020; Burdick et al., 2006; Gout et al., 2006; Miles et al., 2008; Osmani et al., 2019; Shirure et al., 2012; Strell and Entschladen, 2008).

Unlike leukocytes, which roll over the endothelium after tethering to the vascular wall, the rolling mechanism of CTCs remains debated or poorly described. Studies have reported in vitro rolling of cancer cells over activated ECs through PSGL-1 interactions with P-selectin and N-cadherin in colon and breast cancer, respectively (Burdick et al., 2003; Giavazzi et al., 1993; Hazan et al., 2000; Strell and Entschladen, 2008). In vivo and ex vivo, observations in zebrafish embryos have shown breast cancer cells rolling and forming microvilli, reminiscent of leukocyte slings and tethers (Glinsky et al., 2003; Liu et al., 2018). Intravital imaging studies demonstrate that CTCs are capable of firmly adhering to the endothelium (Borriello et al., 2022; Glinskii et al., 2003; Osmani et al., 2019). However, we and others have suggested that CTCs might not rely on a rolling mechanism to stably adhere to the vasculature. Rather, they can stop instantly by engaging weak bonds with the luminal side of the endothelium that further evolve in stable long-lasting adhesions (Borriello et al., 2022; Haier et al., 2003; Osmani et al., 2019; Shen et al., 2010). This variability suggests that the rolling properties of CTCs might depend on their adhesion repertoire (discussed in detail later). These inconsistencies underscore that CTC rolling is not universally observed and might be contingent upon the adhesion potential of CTCs. Although further research is needed to clarify the existence of CTC rolling, multiple studies support a consensus on a two-step mechanism for CTC arrest. This mechanism involves initial engagement with weak but rapidly engaging adhesion receptors for CTC margination, followed by stronger but slower interactions to fully stabilize CTCs (Glinskii et al., 2003; Osmani et al., 2019). For instance, metastable selectin-mediated adhesions are transient, providing time for CTCs to engage in stronger adhesion to the endothelium (Dimitroff et al., 2004; Myung et al., 2011; Shea et al., 2017). Similarly, CD44 and integrin αvβ3 have been shown to facilitate transient adhesion to the endothelial wall (Osmani et al., 2019). The adhesion of CTCs to the vascular wall allows for the exchange of cytokines between tumor and ECs. For example, growth factors from the EGF and HGF families are secreted by the endothelium and stimulate the expression of β1 integrins in breast and hepatocellular cancer cells, enhancing their strong adhesion to the endothelium (Kawakami-Kimura et al., 1997; Narita et al., 2009). In vivo, intravascular injections of CTCs in mice result in the upregulation of adhesion molecules on brain ECs, including activated leukocyte cell adhesion molecule (ALCAM), VCAM-1, ICAM-1 and α4β1 (Soto et al., 2014). Highly metastatic breast cancer cells express α4β1 and ALCAM on their surface, which respectively bind to endothelial VCAM-1 and ALCAM in the brain. Neutralizing tumor cell adhesion with anti-VCAM or anti-ALCAM antibodies has been shown to impair the formation of metastatic tumors in the brain (Soto et al., 2014) and the adhesion of melanoma cells to endothelial cells in vitro (Klemke et al., 2007). Meanwhile, homophilic ICAM-1–ICAM-1 interactions, as well as interactions between tumoral MUC-1 and ICAM-1, drive clustering of CTCs and their adhesion to the endothelium, thereby facilitating transendothelial migration and metastasis (Finzel et al., 2004; Rahn et al., 2005; Regimbald et al., 1996; Sundar Rajan et al., 2017). Decreasing ICAM-1 expression in ECs and CTCs synergistically prevents CTC transmigration. Treatment with anti-ICAM-1 antibodies prior to injection of triple-negative breast cancer cells has been shown to reduce CTC metastasis and colonization in the lung (Taftaf et al., 2021). However, whether ICAM-1 and MUC-1 mediate transient or stable adhesion remains unclear. Similarly, CD44 has been implicated in promoting CTC clustering, in addition to its role in transient adhesion to the endothelium (Liu et al., 2019; Osmani et al., 2019). It is also uncertain whether the endothelial outside-in signaling observed during the leukocyte transmigration cascade, which involves ICAM-1 and VCAM-1, is similarly involved in cancer cell adhesion and transendothelial migration. Recent studies support the local activation of the endothelium within minutes of cancer cell adhesion. In 3D engineered vascular networks, ECs locally extend finger-like protrusions towards the adhered breast cancer cells. Mechanistically, the adhesion of breast cancer cells is proposed to elevate the activity of myosin II regulatory light chain (RLC; in humans, encoded by MYL9, MYL12A and MYL12B), downstream of Ca²⁺-calmodulin-dependent myosin light chain kinase activation. RLC recruitment subsequently modulates endothelial cytoskeleton contraction (Khuon et al., 2010). In the zebrafish embryo, ECs have also been observed to extend intraluminal finger-like protrusions specifically interacting with those formed by adhered breast cancer cells (Follain et al., 2018). Although the role of endothelial protrusions remains elusive, insights can be drawn from the leukocyte transmigration cascade. Knowing that breast cancer cells extend protrusions enriched with adhesion molecules such as CD44 and integrin-β1 (Shibue et al., 2012; Wolf et al., 2020), a hypothesis could be that interactions between endothelium and tumor protrusions enhance adhesion receptor binding and potentially facilitate chemokine exchange. This tight interrelation between ECs and CTCs could reinforce cancer cell adhesion to the vascular wall and/or provide activating signals to prepare for cancer cell extravasation, similar to the endothelial ‘docking structure’ or ‘transmigratory cup’ formed during leukocyte adhesion and transmigration (Fig. 1). Further experiments are required to precisely identify the role of endothelial finger-like protrusions during CTC extravasation and the activation pathways leading to their extension.

Leukocyte transmigration occurs preferentially in post-capillary venules, where the endothelium is more reactive to inflammation, exhibits an increased molecular adhesome repertoire and has endothelial junctions that are more prone to opening (Claesson-Welsh et al., 2021). Unlike veins and arteries, which display a continuous endothelium, ECs in post-capillary venules have loosened junctions upon inflammation and might even have gaps between neighboring cells. Besides, endothelial activation could favor the formation of reversible transcellular pores, altogether facilitating leukocyte transmigration (Aird, 2007). Notably, in the lymph nodes, high endothelial venules (HEV) express a specific adhesion repertoire that supports constitutive leukocyte trafficking (Miyasaka and Tanaka, 2004).

The adhesion and transmigration sites of leukocytes occur at different spots. Once adhered, immune cells crawl or migrate over the endothelial layer toward the extravasation site. Monocyte crawling relies on the interaction between LFA-1 and endothelial ICAM-1 (Schenkel et al., 2004), whereas neutrophils crawl by engaging the Mac-1 receptor with endothelial ICAM-2 (Halai et al., 2014; Phillipson et al., 2006). Paracellular diapedesis occurs at the cell junction between neighboring ECs, mostly at tricellular junctions (Castro Dias et al., 2021). ICAM-1 and VCAM-1 further mediate leukocyte transmigration, alongside junctional adhesion molecules, such as junctional adhesion molecule A (JAM-A; also known as F11R), platelet EC adhesion molecule 1 (PECAM-1), ICAM-2, CD99 and CD99L2 (Lyck et al., 2003; Mamdouh et al., 2009; Schenkel et al., 2007; Seelige et al., 2013; Su et al., 2002; Woodfin et al., 2007). Engagement with these receptors supports endothelial signaling events that induce inter-EC junction loosening, the recruitment of leukocyte adhesion receptors at the cell junction and endothelial actin contraction. For example, ICAM-1 and PECAM-1 engagements are responsible for the destabilization of vascular endothelial (VE)-cadherin homophilic interactions between ECs. VE-cadherin forms adherens junctions, and is stabilized and linked to the cytoskeleton by associating with catenins (p120-catenin, β-catenin and γ-catenin). Upon binding to leukocyte ligands, ICAM-1 modifies the phosphorylation state of VE-cadherin and its associated catenins, disassembling the junctional complex and thus opening adherens junctions. In addition, ICAM-1 clustering, coupled with shear stress, increases endothelial membrane tension, leading to the opening of PIEZO1 (Wang et al., 2022). Ca²⁺ influx following PIEZO-1 activation has been shown to disrupt junctional VE-cadherin through Src and Pyk2 (also known as PTK2B)-mediated phosphorylation of the intracellular domain of VE-cadherin (Allingham et al., 2007; Etienne-Manneville et al., 2000). Meanwhile, PECAM-1 promotes VE-cadherin dephosphorylation at Tyr731 through the recruitment of the tyrosine phosphatase SHP2 (also known as PTPN11). Tyr731 is normally masked by the catenin complex, but increased membrane tension following leukocyte adhesion reveals the amino acid to SHP2 (Arif et al., 2021). VE-cadherin dephosphorylation triggers its endocytosis, which supports leukocyte transmigration (Wessel et al., 2014). Notably, the dissociation of the tyrosine phosphatase VE-PTP in complex with VE-cadherin is also mediated by VEGF, the main vascular growth factor, and the inflammatory lipopolysaccharide (LPS), a major component of bacterial envelope (Broermann et al., 2011; Nottebaum et al., 2008). As endothelial junctions destabilize, leukocytes squeeze between ECs and gain new adhesions with the endothelium. Homophilic PECAM-1–PECAM-1 interactions between transmigrating leukocytes and ECs induces the recruitment to the junction of the endothelial lateral border recycling compartment (LBRC), a membrane compartment connected to the endothelial cell surface at the cell borders (Mamdouh et al., 2009). LBRC fusing to the plasma membrane increases the endothelial membrane surface, and the concentration of adhesion receptors, including PECAM-1, CD99 and JAM-A. Homophilic CD99 engagement triggers a second wave of LBRC recruitment to the endothelial borders, further promoting leukocyte transmigration (Watson et al., 2015). LBRC is also delivered at the site of transcellular diapedesis to promote leukocyte adhesion and transcytosis (Carman et al., 2007; Shaw et al., 2004). Following successful leukocyte transmigration, the endothelial barrier must reseal, and both tight and adherens junctions must be restored to prevent vascular leakage. Interestingly, the heterogeneous expression of adhesion receptors, particularly ICAM-1, limits the vascular leakage during and after leukocyte transendothelial migration (Grönloh et al., 2023). Transmigratory pore closure relies on endothelial actin polymerization into ventral lamellipodia (Martinelli et al., 2013) and acto-myosin contractility (Heemskerk et al., 2016).

Thus, it has been suggested that the endothelium is not merely a passive barrier crossed by leukocytes, but a surface that actively remodels itself by forming endothelial cups that foster leukocyte transmigration (Carman and Springer, 2004; van Buul et al., 2007a). However, the relative contribution of these cups to diapedesis remains unclear (Fig. 1). Interestingly, ultrastructural studies of neutrophil transmigration reported that the endothelium completely envelops extravasating neutrophils, thus preserving vascular permeability (Petri et al., 2011; Phillipson et al., 2008). This suggests that the EC docking structure, or cups, might further evolve into a dome during leukocyte extravasation.

Although the extravasation of cancer cells is much less understood than leukocyte diapedesis, many mechanisms are thought to be shared between leukocyte and cancer cell transmigration. For instance, similarly to leukocytes, integrin β1 drives stable cancer cell adhesion to ECs and facilitates extravasation (Osmani et al., 2019; Reymond et al., 2012; Shibue et al., 2013; Stoletov et al., 2010). One possible mechanism is that cancer cells extend filopodia-like structures through the endothelial barrier and bind to the underlying endothelial matrix. The adhesion of the laminin receptors α3 and α6β1 integrins to subendothelial laminin is required for cancer cell transmigration and subsequent metastatic colonization (Chen et al., 2016). Although the adhesion repertoire of leukocytes is relatively well-defined in physiological conditions, cancer cells are a highly heterogenous population, even within a single tumor. This heterogeneity results from a highly plastic phenotype, allowing cells to transition along a spectrum of epithelial–mesenchymal states (Nieto et al., 2016), and between different mechanical phenotypes (Gensbittel et al., 2021). For instance, α3 and α6β1 integrins are typically expressed by cancer cells with at least a partial epithelial phenotype. Thus, the requirement for specific integrins for extravasation might depend on the phenotype of CTCs along the epithelial–mesenchymal spectrum. However, the contribution of different cancer phenotypes to extravasation efficiency has yet to be fully characterized.

Similar to leukocyte diapedesis, cancer cells can transmigrate through the paracellular route, which requires EC junction disruption, notably VE-cadherin destabilization (Reymond et al., 2013). Further facilitating their own dissemination, CTCs can increase vascular permeability to promote extravasation either through direct or indirect secretion such as EV secretion from the primary tumor or cytokines secreted by blood cells or directly by tumor cells (Padua et al., 2008; Schumacher et al., 2013; Ghoroghi et al., 2021; Qian et al., 2011). VEGF-A, a ligand for VEGFR1 and VEGFR2, plays an important role in promoting angiogenesis and vascular remodeling, including the loosening of EC junctions (Gerhardt et al., 2003; Simons et al., 2016). The aggressiveness of ovarian (Jean et al., 2014; Masoumi Moghaddam et al., 2012), breast (Arias-Pulido et al., 2012) and melanoma (Palmer et al., 2011) cancers correlates with VEGF-A expression. In ovarian cancer cells, VEGF-A activates focal adhesion kinase (FAK; also known as PTK1) downstream of VEGFR2 and Src, leading to VE-cadherin phosphorylation at Y658. This phosphorylation disrupts VE-cadherin interaction with β-catenin causing its internalization and endothelial junction destabilization. Consequently, FAK activation increases vascular permeability and downstream VEGFR activation aiding CTC extravasation (Weis et al., 2004; Jean et al., 2014). Similarly, VEGFR2 activation through Tyr949 phosphorylation on the endothelium has been shown to promote liver colonization by pancreatic cancer cells (Li et al., 2016). In breast cancer cells, VEGF-A expression sustains extravasation in vivo in a model of experimental metastasis in zebrafish embryos (Stoletov et al., 2010) by inducing von Willebrand factor (VWF) expression in ECs. VWF enhances vascular permeability and facilitates extravasation in vitro (Dhami et al., 2022). Moreover, VWF is directly expressed by osteosarcoma cells in patients, supporting tumor cell extravasation ex ovo (Mojiri et al., 2017).

Vascular permeability is also regulated by Ca²⁺ signaling in ECs (Curry, 1992). Fibrosarcoma cells increase intracellular Ca²⁺ levels in ECs by secreting ATP, which binds to P2Y receptors and mobilizes Ca²⁺ through downstream signaling involving phospholipase C and its second messenger, inositol trisphosphate (IP₃) (Nejime et al., 2008). ATP can also be released by tumor cell-associated platelets or by ECs following cancer cell-induced death (necroptosis), further increasing endothelial permeability (Schumacher et al., 2013; Strilic et al., 2016). Ca²⁺ also regulates VEGF-mediated VE-cadherin destabilization and cancer cell extravasation, as this process can be inhibited by treating ECs with the Ca²⁺ chelator BAPTA (Lee et al., 2003).

Another mechanism utilized by leukocytes that is also employed by CTCs during extravasation is the use of the adhesion receptors ICAM-1 and VCAM-1. For example, tumoral ICAM-1 binds to endothelial ICAM-1, to facilitate transmigration (Taftaf et al., 2021). In vitro, inhibiting ICAM-1 expression by siRNA in breast cancer cells prevents them from crossing the endothelial layer in a Boyden chamber assay (Taftaf et al., 2021). In vivo, anti-ICAM-1 neutralizing antibodies efficiently reduce breast cancer cell extravasation and metastatic outgrowth in the lungs (Taftaf et al., 2021).

Early ultrastructural analysis of cancer cell extravasation has demonstrated that the endothelial intercellular junctions are not only destabilized, but that ECs might also partially retract near adhered CTCs in vitro (Honn et al., 1994). This retraction helps contact between the cancer cells and the subendothelial matrix, aiding subsequent extravasation. Subendothelial matrix has also been found to be accessible to CTC in mouse lungs (Wang et al., 2004). Furthermore, similar to the endothelial dome wrapping of neutrophils, ECs can also cover the CTC to reform vessels in a process, called endothelial remodeling (also known as endothelialization or angiopellosis), further illustrating the active role of ECs in cancer cell extravasation (Allen et al., 2019; Follain et al., 2018; Lapis et al., 1988; Paku et al., 2000). This early observation is strikingly reminiscent of leukocyte transmigratory cups (Fig. 1). ECs have been shown to form filopodia-like intraluminal protrusions in vitro in 3D vascular networks (Khuon et al., 2010), in vivo around breast cancer cells arrested in the caudal plexus of zebrafish embryo (Follain et al., 2018) and in vivo in the mouse brain near adhered melanoma and breast cancer cells (Haskó et al., 2019; Herman et al., 2019; Karreman et al., 2023). These studies suggest that endothelial intraluminal protrusions might support cancer cell extravasation, although the exact cellular and molecular mechanisms remain elusive.

Mechanical forces in the vasculature also influence extravasation. For example, hemodynamic forces sustain the activation of ECs near attached cancer cells (Follain et al., 2018). This activation might lead to EC retraction and reorganization due to the dynamic mechanical forces at play. Gaps spontaneously form at vertices between three ECs, where CTCs also tend to extravasate (Escribano et al., 2019). In the zebrafish embryo, endothelial remodeling requires flow sensing through endothelial VEGFR2, a receptor sensitive to both VEGF-A and shear stress (Baeyens and Schwartz, 2016; Follain et al., 2021; Simons et al., 2016). Interestingly, VEGF-A is not required for endothelial remodeling in mouse brain metastasis (Karreman et al., 2023), suggesting that in this context VEGFR2 might be VEGF-A-independent and rely on its flow-sensing properties. Endothelial pocketing is observed in specific regions of the vasculature where flow forces are strong enough to activate ECs but still permissive for cancer cell arrest. Indeed, inhibiting VEGFR2 signaling with anti-angiogenic drugs, such as sunitinib or cediranib, decreased endothelial remodeling and prevented cancer cell transmigration (Follain et al., 2021). In the mouse brain, the endothelial remodeling is activated by the secretion of tumoral matrix metalloproteinase-9 (MMP9) (Karreman et al., 2023), which might degrade the endothelial matrix and release endothelial-activating chemokines.

Once extravasation is complete, the lasting effects on the structure and integrity of the endothelium also differ between leukocytes and cancer cells. Unlike leukocyte diapedesis, where the endothelium remains intact, ECs can be highly damaged by tumor cells. For example, an alternate mechanism involves adhered cancer cells releasing amyloid precursor protein, which binds to endothelial death receptor (DR) 6 (also known as TNFRSF21), inducing programmed necroptosis. This endothelial necroptosis fosters tumor cell extravasation by physically destroying the vascular barrier, promoting metastasis in vivo (Strilic et al., 2016).

A final consideration is that unlike leukocytes, cancer cells do not disseminate only as single cells. Indeed, CTC clusters are more aggressive than their single-cell counterparts (Aceto et al., 2014). Furthermore, cancer cells with an epithelial identity, when trapped in the vasculature, might proliferate within the lumen prior to extravasation or endothelial wrapping (Al-Mehdi et al., 2000). Therefore, the endothelium plays a direct role in determining the fate of CTCs towards either intravascular growth or extravasation, mediated through Wnt signaling pathways (Jakab et al., 2024). The active process of endothelium-mediated extravasation might explain how large CTC clusters can successfully extravasate through endothelial remodeling processes (Follain et al., 2018; Allen et al., 2019).

An important distinction between immune cells and cancer cells lies in their mechano-morphological properties; CTCs are inherently larger and less deformable than leukocytes (Alibert et al., 2017; Bagnall et al., 2015; Barzilai et al., 2017). Thus, while leukocytes have evolved to navigate through the entire vasculature, the arrest sites of CTCs are predominantly determined by vascular topology and the deformability of CTCs themselves (Gensbittel et al., 2021; Perea Paizal et al., 2021).

Immune cell recruitment and extravasation are influenced by the adhesion and chemokine repertoire expressed specifically on the luminal side of ECs, which varies among different tissues (Chi et al., 2003). For example, ECs in HEVs constitutively express peripheral node addressin (PNAd; encoded by CD34 and GLYCAM1) and MadCAM-1, which bind to L-selectin and facilitate T-cell homing to lymph nodes. Inflammation signals locally induce the expression of selectins and ICAM-1 on ECs to recruit immune cells to sites of tissue damage. Furthermore, ECs actively support peripheral diapedesis by extending intraluminal protrusions that guide leukocyte transmigration through junctional protrusions (Arts et al., 2021).

Similarly, the intravascular arrest potential of CTCs depends on their adhesion repertoire. For instance, studies using intravital imaging combined with experimental metastasis models in zebrafish embryos suggest that high levels of integrin β1 expression on breast cancer CTCs favor their stable arrest and extravasation in the caudal capillary plexus compared to the brain microvasculature (Osmani et al., 2019; Paul et al., 2019).

Circulating platelets bound to CTCs promote their arrest by enabling CTCs to engage with endothelial platelet-specific adhesion receptors (Garcia-Leon et al., 2024; Labelle et al., 2011; Mammadova-Bach et al., 2020). Similarly, platelets have recently been shown to facilitate neutrophil recruitment (Burkard et al., 2023). These common mechanisms are likely enhanced by the ability of platelets to readily adhere to primed ECs (Frenette et al., 1995).

Altogether, this underscores the role of the adhesion repertoire expressed on the luminal side of specific vascular subregions in dictating particular vascular dissemination patterns. Recent findings suggest that ICAM-1 is heterogeneously expressed on endothelial lumens, with higher expression levels determining extravasation hotspots for leukocytes (Grönloh et al., 2023). It is speculated that specific vascular adhesion patterns might also influence metastatic organotropism. This idea is supported by observations that tumor EVs with distinct adhesion repertoires target different organs; for example, α6β4 and α6β1 direct EVs to the lungs whereas αvβ5 targets EVs to the liver, indicating differential endothelial counter-receptor expression in these two organs (Hoshino et al., 2015). Whether a similar mechanism operates for circulating CTCs remains to be formally demonstrated.

Following their stable adhesion, leukocytes often crawl toward endothelial cell–cell junctions (CCJs) in an ICAM-1 and/or VCAM-1 dependent manner (Shulman et al., 2006). Given that diapedesis occurs through EC junction, CCJs emerge as obvious hotspots for leukocyte extravasation. Recent reports indicate that these junctions are highly dynamic at the cytoskeletal level and are enriched in adhesion receptors that mean cells are prone to extravasation (Arts et al., 2021; Kroon et al., 2018). For instance, ICAM-1 localized at tri-cellular junctions promotes paracellular transmigration of leukocytes (Sumagin and Sarelius, 2010; Castro Dias et al., 2021). EC junctions appear essential, suggesting that transcellular diapedesis might be a dominant mechanism for leukocytes, at least in vivo (Woodfin et al., 2007, 2011; Schulte et al., 2011).

Similarly to leukocytes, endothelial junctions, particularly tri-cellular junctions, are hotspots for cancer cell transmigration. Tri-cellular junctions are weaker, and dynamic endothelial forces drive gap formation between ECs, facilitating CTC extravasation through diapedesis (Escribano et al., 2019). For efficient extravasation, leukocytes tightly engage the endothelial adhesion machinery (Vestweber, 2015). This suggests that CTCs using diapedesis might have an adapted adhesion repertoire allowing them to hijack leukocyte diapedesis mechanisms. This is further supported by a recent study using a microfluidic vessel-on-a-chip model, which demonstrated that arrested cancer cells can crawl in a similar manner to leukocytes, although randomly, on ECs before proceeding to extravasation (Javanmardi et al., 2023). Alternatively, from an endothelial perspective, active diapedesis might occur in regions where the endothelial tightness is lower, whereas endothelial remodeling may occur in areas where the endothelial barrier is highly preserved, such as in the brain. This remodeling would facilitate removal of metastatic emboli from circulation and reperfusion of blood vessels (Follain et al., 2018; Karreman et al., 2023).

The strength of intercellular adhesion between ECs has been shown to depend on the stiffness of the underlying subendothelial matrix in vitro (Javanmardi et al., 2023; Krishnan et al., 2011).

As leukocytes adhere, they crawl toward the optimal transmigration spot dictated by the stiffness of both the endothelial and subendothelial matrix. ECs sitting on stiff surfaces deploy increased myosin-based contractile forces that weaken VE-cadherin and cell–cell junction integrity, thus promoting extravasation (Stroka and Aranda-Espinoza, 2011). This phenomenon is further supported by the observation that aged-associated ECM stiffening of the vascular wall facilitates leukocyte transmigration (Huynh et al., 2011). At the molecular level, ECM stiffening promotes the clustering of ICAM-1 around adhering leukocytes through recruitment of cortactin and α-actinin, facilitating transmigration (Schaefer et al., 2014). Cortactin recruitment at ICAM-1 clusters depends on DLC-1 (also known as ARHGAP7), which is upregulated in ECs on stiff substrates and stabilizes the ICAM-1 adhesome (Schimmel et al., 2018).

Conversely, cancer cells that firmly adhere to ECs on stiff matrices show decreased efficiency in extravasation, requiring longer times for successful extravasation (Javanmardi et al., 2023). Yet, in vitro microfluidic data demonstrates that increased matrix stiffness increases breast cancer cell migration ability following extravasation and MMP9 secretion. This highlights the role of MMP9 in mediating extravasation and post-extravasation processes, as previously reported during endothelial remodeling (Karreman et al., 2023). However, these discrepancies suggest a biphasic role of stiffness during extravasation with increased endothelial stiffness rather impeding extravasation but promoting subsequent migration.

In addition to chemokine activation, ECs are sensitive to blood shear stress, and local blood flow can promote the adhesion of circulating immune cells. Atherosclerotic plaques primarily occur in regions with disturbed flow patterns, such as vascular branching points (Zeiher et al., 1991). When ECs are exposed to atherogenic environments such as disturbed flow, the expression of ICAM-1, VCAM-1 and E-selectin increases, leading to leukocyte recruitment and inflammation (Gimbrone and García-Cardeña, 2013; Le et al., 2017). The modulation of the ECs adhesion repertoire by flow depends on the YAP/TAZ pathway, as demonstrated by studies showing that YAP (also known as YAP1) overexpression or knockdown exacerbates or retards atherosclerotic plaque formation, respectively (Wang et al., 2016a,b). Additionally, YAP mechanosignaling is required for vessel maintenance (Nakajima et al., 2017) and is involved in angiogenesis through the maturation of CCJs and VEGFR2 signaling (Kim et al., 2017; Wang et al., 2017), which responds to both VEGF-dependent and VEGF-independent flow mechanosensing (Baeyens and Schwartz, 2016).

Cancer cell dissemination depends on vascular flow patterns. In the zebrafish embryo, breast cancer cells preferentially adhere in low-flow regions of the capillary plexus, with flow profiles matching those of colonized organs. Their extravasation through endothelial remodeling requires a minimal flow threshold. This flow-dependent tuning of metastatic patterns has also been observed in individuals with brain metastasis (Follain et al., 2018). The dissemination and internalization of tumor-derived EVs depend on flow patterns, with uptake favored in vascular regions subjected to capillary-like hemodynamics, suggesting that endothelial priming for extravasation also relies on hemodynamic features (Mary et al., 2023). The balance between adhesion and hemodynamic forces directly dictates the location of cancer cell arrest. Interestingly, cancer cells have been shown to induce YAP activation in pericytes after extravasation, indicating that they are able to highjack YAP mechanosignaling in the brain perivascular niche (Er et al., 2018). However, whether endothelial YAP is directly involved in cancer cell extravasation remains to be demonstrated.

VEGFR2 mechanosignaling has also been implicated in leukocyte diapedesis, with VEGFR2 associated with the endothelial mechano-sensing complex involving PECAM1 and VE-cadherin (Fu et al., 2023; Tzima et al., 2005). Interestingly, PECAM is highly recruited at VEGFR2-enriched sites of endothelial remodeling, suggesting that the entire PECAM–VE-cadherin–VEGFR2 mechanosensing complex might play a role (Follain et al., 2018, 2021). We hypothesize that this means that the flow-dependent VEGFR2-dependent VEGF-independent endothelial driven CTC extravasation mechanism might leverage a similar mechano-signaling pathway, given that an arrested CTC engaged with the endothelium and pushed by flow will very likely induce similar traction forces on the luminal side of endothelial cells (Follain et al., 2020; Schaefer and Hordijk, 2015).

The metastatic progression of cancer cells has been shown to hijack several physiological processes to promote their dissemination. These processes include developmental mechanisms such as epithelial–mesenchymal plasticity (Brabletz et al., 2021) and wound healing (Ganesh and Massagué, 2021), as well as strategies resembling leukocyte trafficking, particularly extravasation. A significant difference between the behavior of tumors and immune cells lies in the fact that the main route of exit for leukocytes is paracellular extravasation, whereas cancer cells appear to rely not only on diapedesis but also on indoctrinating the endothelial wall, which remodels to facilitate the extravasation of emboli and restore perfusion. This mechanism has been documented in conditions such as blood clot formation during brain stroke and relies on MMP2 and/or MMP9 activity, similar to what is seen in brain metastatic cells (Karreman et al., 2023; Lam et al., 2010). Although the initial stages of cancer cell extravasation, driven by endothelial mechanisms resemble the formation of endothelial apical cups, the outcomes of extravasation differ significantly, which might be due to CTCs being less deformable than leukocytes and therefore less amenable to diapedesis. Upon arrest, leukocytes are able to crawl along the endothelial wall, a process associated with seeking regions of lower resistance, enabling them to target EC junctions in a manner reminiscent of durotaxis, which allows cell to migrate along stiffness gradients (Martinelli et al., 2014; Schaefer and Hordijk, 2015). Intravital imaging in experimental metastatic settings supports the idea that arrested CTCs do not exhibit this behavior. Furthermore, evidence indicates that rolling is not a universal mechanism for CTCs. These discrepancies might stem from differences in biomechanical properties and/or adhesive machinery between CTCs and leukocytes. Understanding these fundamental differences could undoubtedly pave the way for new therapeutic strategies targeting CTC extravasation specifically.

We thank all the members of the Goetz lab for helpful discussions.