Neutrophil transendothelial migration hotspots – mechanisms and implications Free

To evoke an immune response, immune cells are recruited towards the site of inflammation. For this to happen, immune cells must first breech the vascular wall, which constitutes a barrier and is composed of an endothelial monolayer with pericytes wrapped around it. The endothelium lines the inner layer of the blood vessels and is therefore the first barrier immune cells have to breech. It plays a crucial role in guiding leukocytes through a process referred to as transendothelial migration (TEM) or leukocyte extravasation. This process comprises a complex cascade with multiple steps of interaction between leukocytes and endothelial cells. Currently, the classic textbook model of leukocyte extravasation, originally proposed by Butcher (Butcher, 1991) and further refined by Springer (Springer, 1994), consists of leukocyte capture, rolling, firm adhesion, diapedesis and abluminal crawling (Fig. 1). Leukocytes can extravasate at junction regions between two endothelial cells (paracellular), or through the cell body of a single endothelial cell (transcellular). Box 1 gives a concise summary of leukocyte TEM and its main players. For a more detailed overview of this complex process, we refer to several excellent review papers (Butcher, 1991; Springer, 1994; Ley et al., 2007; Vestweber, 2015; Muller, 2016; van Steen et al., 2020).

It is easy to picture TEM as a straightforward cascade of several steps occurring in a sequential manner, starting with the capture of a leukocyte on the endothelium and ending with the leukocyte reaching the source of infection in the interstitium. However, two observations show that leukocyte extravasation is not as simple as this and is not solely a sequential process. First, none of the steps in the model are irreversible. Not all leukocytes start rolling, not all rolling leukocytes will eventually adhere firmly, many adhering leukocytes will return to the bloodstream and some leukocytes even undergo reverse TEM (Bradfield et al., 2007; Woodfin et al., 2011; Muller, 2015). Secondly, if one compares the diameter of a leukocyte [7–10 μm (Downey et al., 1990), and when fully spread up to 20 μm (Heemskerk et al., 2016)] with the thickness of the endothelium at cell-to-cell junctions (0.2–0.8 μm; Cahill and Redmond, 2016), one quickly realizes that, for most of the extravasation process, different steps of the cascade occur in overlapping timeframes. This is visualized in transmission electron microscopy images produced decades ago that show how leukocytes simultaneously adhere to the apical side of the endothelium while at the same time having initiated diapedesis and being partly present underneath the endothelium (McDonald, 1988).

Although many proteins and processes we describe are involved in leukocyte extravasation in general, in this Review, we will focus on neutrophils, because they act as a first line of defense when the immune system is challenged. As it has been observed that neutrophils leave the bloodstream at non-random locations (Proebstl et al., 2012; Rigby et al., 2015; Hyun et al., 2019), we wish to discuss the concept of the ‘transmigration hotspot’, a region in the endothelium and in the pericyte layer associated with increased neutrophil TEM events. We will elaborate on the biological relevance of such TEM hotspots and compile several factors that may contribute to the existence of hotspots, both in the endothelial and the pericyte layer.

When studying TEM in vivo or in vitro, at first glance, leukocytes appear to cross the vascular barrier in a random manner. However, upon careful observation, it seems that there are spots at the luminal side of the vessel wall where leukocytes preferentially exit. What is the cause of this preference? Perhaps neutrophils actively search for an ideal diapedesis site to exit the endothelium and the pericyte layer (Hyun et al., 2019). Such sites may serve as gateways that allow for the serial passage of neutrophils, meaning that neutrophils apparently ‘follow each other’, although it is not clear how they do so (Feng et al., 1998). These observations suggest that there may be predefined exit sites on the endothelium that neutrophils use to extravasate; these sites can be classified as transmigration hotspots.

When considering vessel homeostasis and integrity, it could make sense that spatiotemporal mechanisms exist to regulate where neutrophils exit the endothelium. A possible hypothesis, which was, among others, raised by Hyun and colleagues, is that when the majority of neutrophils extravasate at the same spot, disruption of the vessel wall as a whole is minimized (Hyun et al., 2019). Conversely, one could argue that a large simultaneous TEM of leukocytes across a specific site in the monolayer may locally disrupt the integrity of the vessel wall in a way that could become detrimental for endothelial barrier function. In any case, as vessel integrity and neutrophil trafficking are impaired in common inflammatory diseases, such as rheumatoid arthritis (Szekanecz and Koch, 2008), inflammatory bowel disease (Panés and Granger, 1998) and atherosclerosis (Chistiakov et al., 2015), it will be critical to unravel the regulation of TEM hotspots. So far, there have actually been many hypotheses postulated regarding factors that could play a role in determining hotspots. Here, we will highlight several of these determinants, and in what way they may steer neutrophil exit through local TEM hotspots.

Chemokines, or chemoattractant cytokines, play an indispensable role in TEM, as they activate and instruct leukocytes on where to go (Thelen and Stein, 2008). During the crawling and diapedesis stages of TEM, neutrophils extensively form protrusions as a sensing mechanism to detect chemotactic cues on the endothelium. They can even poke through cell–cell junctions into the abluminal space, most likely to screen the local environment for migration cues (Nourshargh and Alon, 2014). Several chemokines that are used by neutrophils in a sequential manner during TEM have been identified (Middleton et al., 2002). Endothelial- and pericyte-derived chemokine ligand 1 (CXCL1) is presented by the endothelium and used by neutrophils during luminal and abluminal crawling (Girbl et al., 2018). In contrast, neutrophils themselves secrete chemokine ligand 2 (CXCL2), which is then presented by the endothelium via atypical chemokine receptor 1 (ACKR1; also known as DARC) at junction regions (Girbl et al., 2018). In mice, it has been shown that ACKR1 is only expressed in post-capillary venules and small collecting venules, and is enriched at cell–cell junction regions of the endothelium (Thiriot et al., 2017). Even more strikingly, only venules that were positive for ACKR1 showed leukocyte–endothelium interactions (Thiriot et al., 2017). Likewise, the ACKR1-mediated chemokine cue is required for efficient diapedesis in mice (Girbl et al., 2018). As ACKR1 is known to non-specifically bind chemokines without initiating intracellular signaling pathways, this molecule may be a major player in presenting chemokines to leukocytes in order to facilitate TEM, and may thus be part of local TEM hotspots (Fig. 2).

Another way of presenting chemokines to leukocytes on the endothelium is by non-specific binding of chemokines to glycosaminoglycans (GAGs) (Proudfoot, 2006). This is exemplified by the GAG heparan sulfate, which is highly anionic, that is, negatively charged (Kreuger and Kjellén, 2012), and thereby capable of binding chemokines, which often have a positively charged domain (Lortat-Jacob et al., 2002; Lortat-Jacob, 2009). These chemokine–GAG complexes may in turn bind adhesion molecules such as L-selectin on leukocytes to strengthen the interactions between the circulating leukocyte and the endothelium during the rolling, adhesion and crawling stages on the endothelium and in subendothelial spaces (Kumar et al., 2015). Interleukin-8 (IL-8; also known as CXCL8), a well-studied neutrophil chemoattractant, has been shown to be presented by the endothelium on GAGs (Middleton et al., 1997). IL-8 binds C-X-C chemokine receptor type 1 and 2 (CXCR1 and CXCR2) on neutrophils, which in turn leads to integrin β2 activation (Rajarathnam et al., 2019). In this way, IL-8 not only activates the neutrophil, it also provides directionality for efficient TEM (Middleton et al., 1997). In an experiment that ‘scavenged’ endothelial-bound IL-8 using soluble GAGs to disturb the chemokine presentation on the endothelium there was a reduction in neutrophil TEM (Gschwandtner et al., 2017). IL-8 binds GAGs through a positively charged C-terminal region. By adding a peptide to the endothelium that is based on this GAG-binding domain, IL-8 can be excluded from the endothelium, resulting in reduced neutrophil adhesion (Martínez-Burgo et al., 2019). Like the study by Gschwandtner and colleagues, this demonstrates that the IL-8 chemokine gradient on the endothelium is pivotal for neutrophil TEM.

An alternative model argues that chemokines are in fact not bound to GAGs when they are presented to leukocytes. This ‘chemokine cloud’ model states that chemokines exist in solution, and that their diffusion is not prevented by their binding to GAGs, but because they are sequestered within the hydrated glycocalyx, a carbohydrate-rich layer on the luminal side of the endothelium (Graham et al., 2019). Data supporting this model include studies showing that chemokine interactions with GAGs and leukocyte receptors require the same residues within chemokines, meaning GAGs and leukocyte receptors are likely to compete for chemokine binding (Lau et al., 2004; Sepuru and Rajarathnam, 2016; Sepuru et al., 2016). Additionally, antibody treatments against solvent chemokines result in a much more effective reduction of leukocyte adhesion compared to antibody treatments targeting GAG-bound chemokines (Bonvin et al., 2017).

In addition to the endothelium, other cell types, such as perivascular macrophages, can produce neutrophil-attracting chemokines that are subsequently presented on the luminal side of the endothelium (Abtin et al., 2014). An example of a protein transporting chemokines to the apical side is ACKR1, which can bind chemokines at the basolateral side of endothelial cells and transport them to the apical side via transcytosis (Pruenster et al., 2009). In this way, the endothelium can present cues that are produced in the underlying tissue. Evidently, several mechanisms exist by which chemotactic cues are presented to neutrophils by the endothelium. It is challenging, however, to monitor chemokine presence on the surface of the endothelium, though this could potentially be overcome by super-resolution microscopy. This may also reveal how chemokine gradients are organized and help to establish TEM hotspots.

Recently, it was shown that von Willebrand factor (VWF), a protein restricted to the endothelium, is expressed in a mosaic pattern within a monolayer, with one cell expressing much more protein than its neighbor (Yuan et al., 2016). The authors show that this heterogeneity is generated by noise-induced changes in DNA methylation of the VWF promotor (Yuan et al., 2016). Interestingly, they also show that this heterogenic behavior is not restricted to VWF, but that at least two other proteins, endothelial-specific molecule 1 (ESM1) and ephrin-B2, also follow similar mosaic patterns. However, this is not a generalized phenomenon in endothelial cells, as roundabout homolog 4 (ROBO4) and endothelial-specific transcription factor ETS-related gene (ERG) show a homogenous pattern (Yuan et al., 2016). VWF has historically been associated with hemostasis (Schillemans et al., 2019), but there have also been studies linking the protein with TEM; in an experiment using VWF-blocking antibodies, a 50% decrease in neutrophil TEM in the mouse cremaster muscle was observed (Petri et al., 2010). VWF does not affect rolling or adhesion, but influences the later diapedesis stages of TEM (Petri et al., 2010). Recently, it was shown that VWF supports local adhesion of platelets by releasing angiopoietin 1 to activate the endothelial angiopoietin-1 receptor (TEK, also known as Tie2; a receptor tyrosine kinase), which limits plasma leakage upon leukocyte transendothelial migration (Braun et al., 2020). Taken together, these observations make for an attractive hypothesis – heterogenous expression of proteins related to TEM could possibly determine transmigration hotspots.

For many TEM-related proteins, their localization and the effects of their decreased expression have been studied (Table 1). Additionally, differences in the expression of these proteins in vessel types and organs are known to be the reason why leukocytes, for example, mainly transmigrate in postcapillary venules (Aird, 2003; Kalucka et al., 2020). However, whether mosaic expression of these proteins within a vascular bed could be important for hotspot formation is yet to be described.

In the context of heterogeneity, an interesting candidate to study could be the leukocyte integrin-binding proteins intercellular adhesion molecule 1 and 2 (ICAM-1 and ICAM-2). The importance of these proteins in the neutrophil TEM cascade has already been well established (Box 1). Furthermore, significant changes in TEM have been described upon increased or decreased expression of these proteins. For instance, blocking of ICAM-1 and ICAM-2 results in fewer monocytes reaching endothelial junctions (Schenkel et al., 2004). Additional work has demonstrated that ICAM-1 is important for T-cell TEM across the blood–brain barrier, where high ICAM-1 expression promotes transcellular TEM, whereas intermediate ICAM-1 levels results in more paracellular TEM (Abadier et al., 2015). Another study showed the same effect for neutrophils (Yang et al., 2005). Expression of exogenous ICAM-1–GFP in endothelial cells results in more transcellular TEM through tumor necrosis factor (TNF)-stimulated human umbilical vein endothelial cells (HUVECs), although the absolute number of neutrophils that transmigrates does not differ so much (Yang et al., 2005). Of note, it has been shown both in vitro and in vivo that neutrophils have a tendency to transmigrate at tricellular junctions, spots in the monolayer where three endothelial cells share a junction (Wang et al., 2006; Gorina et al., 2014). Interestingly, another group linked this type of junction preference with an enrichment of ICAM-1 near tricellular junctions (Sumagin and Sarelius, 2010). They postulated that ICAM-1 at these tricellular junctions acts as a recognition site for neutrophils to transmigrate; thus, such junctions may also qualify as a local TEM hotspot.

One of the most well-known endothelial cell–cell junction proteins is vascular endothelial cadherin (VE-cadherin). VE-cadherin is present at endothelial junctions and is extremely important for the barrier integrity of a vessel (Vestweber, 2008). In addition to its function in barrier integrity, the role of VE-cadherin in leukocyte TEM is well established. Leukocyte transmigration can be enhanced in vivo by adding VE-cadherin-destabilizing antibodies (Gotsch et al., 1997), whereas stabilizing the endothelial junctions by increasing the adhesive function of VE-cadherin decreases TEM (Schulte et al., 2011). The role of VE-cadherin in leukocyte TEM is distinct from its role in vascular permeability, however, as it has been shown that both these processes are regulated by different phosphorylation sites on the cytosolic side of VE-cadherin (Wessel et al., 2014). Specifically, leukocyte TEM requires dephosphorylation of VE-cadherin Tyr731, which can be triggered by leukocytes via tyrosine phosphatase Src homology region 2 (SH2)-containing protein tyrosine phosphatase-2 (SHP-2; also known as PTPN11), which in turn allows endocytosis of VE-cadherin. This suggests that leukocytes can create an exit site where junction stability is transiently and locally decreased because of VE-cadherin internalization. In essence, this process could then lead to a local hotspot for other leukocytes (Fig. 2).

Altogether, protein localization or activity within and heterogeneity between endothelial cells could be important determinants for neutrophil exit. The interplay between adhesion molecule abundance and junction phenotype is certainly interesting and a topic for future research. Owing to space limitations, we are not able to discuss all TEM-related proteins in detail. An overview of the potential role of these proteins is given in Table 1.

Heterogeneity of the inflamed endothelial monolayer is not only restricted to protein expression; it is also true for specific membrane structures. The endothelium displays dynamic membrane protrusions at cell–cell junction regions (Breslin et al., 2015). Moreover, under inflammatory conditions, finger-like protrusions appear on the surface of the endothelium (Oh et al., 2007; van Buul et al., 2010a,b). These structures have been characterized as filopodia (Kroon et al., 2018). Filopodia form a platform for adhesion molecules, such as ICAM-1, and it is believed that they can efficiently present their ligands to the neutrophil integrins lymphocyte function-associated antigen 1 (LFA-1; α_Lβ₂ integrin) and Macrophage antigen 1 (Mac-1; α_Mβ₂ integrin) (Kroon et al., 2018). The binding of ICAM-1 to these integrins results in firm adhesion and crawling of the leukocyte prior to diapedesis (Alon and Feigelson, 2009). Endothelial filopodia seem to have the same regulatory mechanism as filopodia of migrating cancer cells (Mattila and Lappalainen, 2008). They require myosin-X expression, its motor domain and the activity of the small Rho family GTPase Cdc42 (Kroon et al., 2018). The presence of these filopodia is highly variable within the endothelium. Whereas the apical side of some endothelial cells is completely covered with protrusions, other endothelial cells in the same monolayer hardly display any (Kroon et al., 2018).

Filopodia are believed to present ligands, and potentially also chemokines, to the leukocytes that crawl along. Next, membrane structures protrude on the luminal side of the endothelium once a leukocyte becomes firmly adhered to it (Buccione et al., 2004). This process is a result of ICAM-1 clustering upon binding to the leukocyte integrins, followed by activation of the small GTPase RhoG, leading to cortactin-mediated actin remodeling and the formation of actin-based structures (van Buul et al., 2007; Schnoor et al., 2011). Throughout the years, several different names have been proposed for these structures. Whereas Barreiro and colleagues, who first discovered them, coined the term ‘docking structures’ (Barreiro et al., 2002), the protrusions were later named transmigratory cups, due to the fact that they correlate with transmigration events (Carman and Springer, 2004). Other names proposed include actin-rich cup structures or ICAM-1-enriched contact areas (Barreiro et al., 2002; Vestweber et al., 2013; Mooren et al., 2014), in vivo endothelial dome structures (Phillipson et al., 2008; Petri et al., 2011) and apical cup structures (van Buul et al., 2007). Despite these different names, the structures described in all these reports are strongly correlated with diapedesis events and can be found during both paracellular and transcellular diapedesis (Heemskerk et al., 2016). As Barreiro and co-workers were the first to describe these structures (Barreiro et al., 2002), we will use the term docking structures in this Review. Docking structures require dynamic actin remodeling and express ICAM-1 and vascular cell adhesion molecule 1 (VCAM-1) (van Buul et al., 2010a,b). Platelet endothelial cell adhesion molecule (PECAM-1) can also be found on these structures, whereas VE-cadherin is usually absent (Carman and Springer, 2004). Whether docking structures are formed de novo during diapedesis around the neutrophil or whether they originate from filopodia is not known. However, it is an attractive hypothesis that multiple filopodia are used to build up a membrane protrusion. This mechanism is well understood for migrating tumor cells, which first extend lateral filopodia, followed by the formation of membrane protrusions or lamellipodia that arise from the existing filopodia (Ridley, 2011). To date, there is no proof that such mechanism is also at work in the inflamed endothelium for the formation of docking structures.

In conclusion, rearrangements of the endothelial actin cytoskeleton, both within and between individual cells, are highly dynamic and could very well play a role in determining the location of a TEM hotspot. As actin-based protrusions contain molecules that are not uniformly present in the monolayer, like ICAM-1, it would be very interesting to find out whether the regulation of these two types of heterogeneity could be related.

In addition to many of the chemical cues mentioned, physical properties of the vessel and the underlying basement membrane are of considerable importance in leukocyte (transendothelial) migration. The mechanism by which leukocytes sense the stiffness of the endothelium is called durotaxis and has been investigated by neutrophil migration studies on non-biological surfaces (Fig. 2). For transendothelial migration, it was shown that more neutrophils cross the endothelium when endothelial cells are grown on stiffer substrates (Stroka and Aranda-Espinoza, 2009). Recently, it was shown that perturbation of the endothelial Rho GTPase-activating protein (RhoGAP) deleted in liver cancer-1 (DLC-1) leads to decreased endothelial cell stiffness (Schimmel et al., 2018). Moreover, the authors showed that neutrophils had a prolonged rolling phase and did not fully spread, but diapedesis per se was not affected. Rescuing DLC-1 expression in endothelial cells restores the initiation of the ICAM-1 adhesome upon ICAM-1 clustering, meaning that actin adapter proteins, such as α-actinin-4 and filamin A and B, are recruited. Consequently, a local actin network is assembled that supports neutrophil crawling and efficient TEM (Schimmel et al., 2018). Such local actin networks may also represent local exit sites and may function as TEM hotspots.

Increased substrate stiffness is strongly associated with an increase in traction forces generated by endothelial cells. Although their transcriptome is left almost completely unaffected, endothelial cells exert higher traction forces when cultured on stiff hydrogels compared to soft ones (Bastounis et al., 2019). This study also found that traction forces are tremendously heterogeneous across the endothelium. Based on these findings, it is tempting to postulate that neutrophil transmigration hotspots are directly correlated to regions with higher traction forces. Strikingly, several studies have shown that, in fact, neutrophils increase local traction forces upon diapedesis (Rabodzey et al., 2008; Yeh et al., 2018). Therefore, extensive live traction force microscopy studies could reveal whether neutrophils would indeed prefer certain high traction force sites in the endothelium and use those as TEM hotspots. The fact that neutrophils themselves generate traction forces on the endothelium, however, may make the analysis more difficult. Nevertheless, it is an interesting hypothesis that needs testing. A limitation of these studies is that they have all been performed in vitro. Although substrate stiffness can be mimicked in vitro, the complex in vivo 3D microenvironment with multiple factors contributing to substrate stiffness is hard to reconstruct.

Once leukocytes have crossed the endothelial monolayer, they reach the abluminal layer, where pericytes are discontinuously wrapped around the vessel (Nourshargh and Alon, 2014). Pericytes and endothelial cells interact with each other both in a chemical and physical manner (Nourshargh and Alon, 2014). Many sorts of pericyte-endothelial cell interactions can be observed: invaginations of the pericyte into the lumen of the vessel, adherens junction-based interactions, gap junction-based interactions and adhesion plaques between the two cell types. More details can be found in a recent review paper by Dessalles and colleagues (Dessalles et al., 2021). These physical interactions consequently influence the ability of neutrophils to migrate into the abluminal layer. Apart from their role in vascular development, pericytes were recently found to be associated with regulating immune responses, including neutrophil extravasation (Rudziak et al., 2019). Upon stimulation with TNF, IL-1β or lipopolysaccharide (LPS), pericytes release IL-8, leading to increased neutrophilic phagocytic activity (Proebstl et al., 2012; Pieper et al., 2013). The number of pericytes per endothelial cell wrapped around the vessel differs between vessel types and can be as high as 1:1 in the retina to as low as ∼1:100 in striated muscle (Shepro and Morel, 1993). Therefore, the role of pericytes involved in regulating TEM may be tissue specific (Fig. 2).

These pericyte characteristics make it very plausible that another type of hotspot, one through the pericyte layer, may also exist. Proebstl and colleagues showed that neutrophils prefer exit sites in the pericyte layer that correlate with places of high pericyte CXCL1 and ICAM-1 expression (Proebstl et al., 2012). Combining two-photon intravital imaging microscopy and electron microscopy techniques, Hyun and co-workers showed the existence of a second type of hotspot – when neutrophils cross the endothelial layer through the endothelial ‘hotspot I’, they next collectively migrate through the pericyte layer by searching for a new exit site that is termed ‘hotspot II’ (Hyun et al., 2019).

Apart from pericytes and stiffness, the composition of the basement membrane greatly contributes to the location of both hotspot I and hotspot II. Regarding the endothelial hotspot I, it has been demonstrated that the type of basement membrane protein that serves as a substrate for the endothelial cells influences the site and dynamics of transmigration. This was exemplified in vitro by a drastic decrease in neutrophil TEM when endothelial cells were grown on the substrate protein laminin 511 (α5, β1 and γ1 chains) compared to laminin 411 (α4, β1 and γ1 chains) (Song et al., 2017). Laminin 511, a basement membrane component that is expressed in a patched manner with regions of high and low expression, promotes the distribution of VE-cadherin to cell–cell junctions, leading to more stable junctions and, consequently, a decrease in neutrophil passage (Song et al., 2017). In line with these data, it has been shown that, in mice lacking laminin 511, that T cell TEM increases (Zhang et al., 2020). As for the pericyte hotspot II, Wang and colleagues showed that gaps in the pericyte layer correlate with so-called ‘low expression regions’ (LERs) of basement membrane proteins (Wang et al., 2006). Immunofluorescence of basement membrane components demonstrated that sites of low laminin 10, collagen IV and nidogen-2 expression correspond to gaps in the pericyte layer. These sites appear to function as type II hotspots (Wang et al., 2006). The existence of LERs, and their correlation with pericyte gaps, was later confirmed in another study that looked at the venular basement membranes of different tissues (Voisin et al., 2010). Smaller vessels seem to have more, but smaller-sized gaps in the pericyte layer. Interestingly, pericyte gap size is not static, and when stimulated with TNF or IL-1β, pericyte gaps increase in size, but not in number (Proebstl et al., 2012). In vitro confocal imaging partly explained this enlargement by showing that pericytes change morphology into a ‘starfish’-like shape when stimulated with either TNF or IL-1β (Proebstl et al., 2012). Whether this change of pericyte shape is dependent on neutrophils is still under debate. These enlarged gaps could potentially make it easier for neutrophils to find their way through the pericyte layer. There is evidence that neutrophil–pericyte contacts induce cytoskeletal relaxation in pericytes via inhibition of the RhoA/Rock pathway, leading to increased gap sizes (Wang et al., 2012). However, studies with neutrophil-depleted mice show that pericytes still change shape in response to TNF or IL-1β (Proebstl et al., 2012).

In conclusion, pericytes and the basement membrane likely play a role in determining the location of endothelial TEM hotspots I and abluminal pericyte-driven hotspots II. The rise of new vessel-on-a-chip techniques combined with advances in intravital imaging will likely increase our understanding of these processes, as they allow us to study multiple cell types in one model.

So far, we have discussed factors that explain TEM hotspots from an endothelial point-of-view. The presentation of chemokines, expression of cell surface molecules, presence of membrane protrusions and the properties of the substrate do not depend on the presence of neutrophils. However, this may not be the only reason for neutrophil-induced successive TEM. It has been shown that neutrophils can form tethers and deposit microparticles on the endothelium (Schmidtke and Diamond, 2000; Marki et al., 2017, 2021). An interesting hypothesis could be that these deposits contain cues for other leukocytes to aid in their TEM. Neutrophils are highly polarized cells, as they are characterized by a well-defined leading-edge pseudopod and a contractile rear, also known as the uropod. These compartments are defined by different functions and protein compositions (Hind et al., 2016). Neutrophil polarization is self-organizing and stable. It has been shown that neutrophils do not require any type of chemokine gradient to become polarized and can break symmetry in conditions with uniform chemokine concentrations (Cramer, 2010). This stability has been confirmed by studies showing that neutrophils almost never reverse polarity, but rather make U-turns while preserving their pseudopods and uropods (Gerisch and Keller, 1981). While crawling over the endothelium, during diapedesis and while crawling in the abluminal space, neutrophils have been observed to elongate their uropods extensively (Hyun et al., 2012; Lim et al., 2015). The exact function of this elongation, and how it is regulated mechanistically, has remained elusive so far. Hyun and colleagues showed that this elongation, which is a common step during the later stages of diapedesis, occurs because of prolonged adhesion of the neutrophil uropod to the luminal side of the endothelium in a mechanism involving ICAM-1–LFA-1 binding (Hyun et al., 2012). Similar structures found on the rear end of neutrophil are tethers, which have also been observed during earlier stages of TEM. These tethers are not regulated by ICAM-1–LFA-1 binding, but form during selectin binding in the rolling phase of TEM (Marki et al., 2017), suggesting they are not necessarily the same structures as ICAM-1-induced uropods. A recent study discovered a new type of neutrophil membrane deposit that results from tether formation and called the deposits elongated neutrophil-derived structures (ENDS). They showed that ENDS do not contain DNA, mitochondria or endoplasmic reticulum (ER) and that they slowly degrade over time (Marki et al., 2021). Another study observed similar particles, but referred to the remnants that migrating neutrophils leave behind on the endothelium and in the interstitium of infected tracheal tissue as neutrophil ‘trails’ (Lim et al., 2015). They showed that these trails were enriched for the chemokine CXCL12 (also known as SDF-1) and are found close to transmigration sites where they facilitate subsequent diapedesis of influenza-specific CD8⁺ T lymphocytes (Lim et al., 2015). Neutrophils can also directly secrete chemotactic cues that are used by other leukocyte subsets. For example, monocyte extravasation is stimulated by the secretion of granules containing LL-37 (cathelicidin) and heparin-binding protein by transmigrating neutrophils (Soehnlein et al., 2008, 2009).

An alternative way for neutrophils to guide each other through the endothelium involves the basement membrane. Neutrophils are able to cleave the basement membrane protein laminin by locally releasing proteases and it was suggested that this release can promote neutrophil migration (Heck et al., 1990; Steadman et al., 1993). A follow-up study showed that degraded laminin fragments are chemoattractants for subsequent neutrophils (Mydel et al., 2008). In this way, neutrophils may pave the way for others to follow during extravasation. Neutrophils can even carry laminins on their cell surface in vivo, possibly to guide following neutrophils (Wang et al., 2006). We conclude that neutrophil deposits and basement membrane degradation may serve as ‘breadcrumbs’ that guide following leukocytes for successful TEM, suggesting that neutrophils themselves create a signature on the endothelium and in the abluminal space that is not intrinsically established by the endothelial cells. This illustrates the complex interplay between the neutrophils and the endothelium during transmigration.

When observing transmigrating neutrophils, one can appreciate the existence of transmigration hotspots, both in the endothelium and in the pericyte layer. However, how such hotspots function and whether they are present prior to neutrophil arrival remains an open question. Here, we have discussed several factors that may contribute to the existence of such hotspots, although this list may be far from complete. Endothelial monolayer heterogeneity could arise from heterogeneity of chemokine presentation or the heterogenous expression or localization of proteins.

Furthermore, the heterogeneous presence of actin-based membrane protrusions, such as filopodia, could be important for the identification of endothelial diapedesis sites. In addition, the subendothelial substrate, pericytes and basement membrane composition determine, at least in part, where neutrophils choose to exit the vasculature. As for establishing hotspots in the pericyte layer, the collected evidence so far points towards important roles of differential pericyte protein expression and distinct basement membrane compositions. And finally, the neutrophil itself may provide instructions to other neutrophils about where to exit the circulation via membrane-derived cues that it leaves behind in all stages of the diapedesis cascade.

Understanding TEM in detail is of critical importance to be able to intervene in case of diseases that are characterized by derailed TEM. By thoroughly understanding why neutrophil diapedesis takes place at specific sites, we start to understand the process in more detail. This will eventually help us to modulate TEM to evoke more powerful immune responses or to block TEM to temper the immune response in case of excessive immune reactions.