Intestinal immune cell trafficking has been identified as a central event in the pathogenesis of inflammatory bowel diseases (IBD). Intensive research on different aspects of the immune mechanisms controlling and controlled by T cell trafficking and retention has led to the approval of the anti-α4β7 antibody vedolizumab, the ongoing development of a number of further anti-trafficking agents (ATAs) such as the anti-β7 antibody etrolizumab or the anti-MAdCAM-1 antibody ontamalimab and the identification of potential future targets like G-protein coupled receptor 15. However, several aspects of the biology of immune cell trafficking and regarding the mechanism of action of ATAs are still unclear, for example, which impact these compounds have on the trafficking of non-lymphocyte populations like monocytes and how precisely these therapies differ with regard to their effect on immune cell subpopulations. This review will summarise recent advances of basic science in the field of intestinal immune cell trafficking and discuss these findings with regard to different pharmacological approaches from a translational perspective.
- Crohn’s disease
- ulcerative colitis
- cell adhesion
- inflammatory bowel disease
Statistics from Altmetric.com
The incidence and prevalence of inflammatory bowel diseases (IBD) such as Crohn’s disease (CD) and ulcerative colitis (UC) is increasing worldwide and a medical cure is still not available.1 However, pivotal advances in our understanding of IBD pathogenesis in recent years have helped to optimise the medical management of these diseases by facilitating the introduction of new targeted therapeutic agents.2 One of these advances was the discovery and translational exploration of specific gut homing and cell trafficking pathways controlling the locomotion of immune cells in the healthy and diseased intestine, which led to the development of specific antibodies interfering with such cell trafficking.
The current concept of IBD pathogenesis assumes that a complex interplay of genetic predisposition and environmental factors leads to translocation of luminal antigens over a weakened epithelial barrier resulting in undercontrolled activation of the intestinal immune system with subsequent structural alterations of the bowel wall.3–5 Most established therapeutic concepts tackle the disease by impacting on the intestinal immune system, for example, by inducing cell death of immune cells or by blocking proinflammatory signalling pathways.2 Since such mechanisms are mostly shared between the immune cells of different organs across the body, this entails the potential of systemic complications such as infections.6 Hence, the idea of specifically interfering with aspects of immune cell trafficking to or within the gut is an intriguing concept potentially combining local immunosuppression with minimisation of systemic side effects.
The following review will describe and illustrate the basic principles of intestinal immune cell trafficking in the context of IBD. Moreover, we will detail the translational progress in the field by discussing past, ongoing and future efforts to design therapeutic anti-trafficking agents (ATAs) for the treatment of IBD with strong focus on the basic science evidence supporting these approaches. Finally, we will outline major open questions remaining to be solved.
Immune cell trafficking: physiological and pathological aspects
Trafficking describes all processes controlling the localisation of immune cells. More specifically, in the case of T cells, this implies regulation of homing, recirculation and retention in the various differentiation states.7
Adhesion is a central step of several aspects of leucocyte trafficking. It is a tightly regulated process eventually leading to firm contact of previously circulating cells with the endothelium of high endothelial venules (HEVs) as a prerequisite for subsequent paracellular or transcellular transendothelial migration.8
Adhesion itself requires interaction of integrins on the surface of the immune cell with adhesion molecules expressed on endothelial cells. Integrins are heterodimers consisting of an α and a β subunit. 18 different α and eight different β subunits exist that associate to 24 distinct αβ combinations. Each monomer is a surface glycoprotein of >1600 amino acids length consisting of a short intracellular domain, a transmembrane domain and a large extracellular domain (figure 1).9 The extracellular domain acts as receptor for various ligands, while the intracellular domain has important functions for signal transduction and is linked to signalling pathways controlling cell motility, survival and proliferation.10 Many integrin heterodimers bind to components of the extracellular matrix, such as collagens, laminins, fibronectin or vitronectin. In the case of gut homing, interaction with endothelial cell adhesion molecules (CAMs) is important.7 9 10 Endothelial CAMs belong to the immunoglobulin superfamily of receptors and their extracellular part contains a combination of immunoglobulin and mucin domains with a variety of alternative splice variants.11 12 Intercellular adhesion molecule (ICAM)-1, vascular cell adhesion molecule (VCAM)-1 and mucosal vascular addressin cell adhesion molecule (MAdCAM)-1 have been shown to control integrin-dependent homing to the gut.13–15 While ICAM-1 and VCAM-1 are expressed in venules throughout the body, MAdCAM-1 expression is considered to be largely gut-specific, although expression in some extra-intestinal tissues like bone marrow or placenta has also been shown.16 Importantly, binding of ligands to integrins requires the presence of bivalent cations and a metal-ion-dependent adhesion site is part of the I-like domain of the β chain and the I domain, which is part of some α subunits (eg, αL and αM, but not α4).10
However, integrins cannot readily interact with endothelial CAMs, since they are in an inactive state on quiescent circulating immune cells17 and, moreover, cells need to be previously recruited to the vessel wall and to be slowed down. Thus, the adhesion cascade (figure 2) commences with so called ‘rolling’, which is mediated by selectins on the lymphocytes and selectin-ligands on the endothelium. These interactions establish a rather loose contact between the cells, which is repetitiously detached and re-established resulting in a slow rolling behaviour of leucocytes along the vessel wall.18 This exposes the cells to mediators released from the tissue, particularly chemokines. Such chemokines activate the cells by G-protein coupled chemokine receptor signalling (‘outside-in signalling’) and consecutively induce a conformational change of surface integrins (‘inside-out signalling’). During this conformational change the previously bent position of the heterodimers is elongated and, thus, the ligand binding pocket is fully accessible and the affinity increased (‘affinity regulation’). Moreover, such activated integrins have increased mobility on the cell surface and are therefore more likely to establish multivalent contact with their ligands (‘valency regulation’). Once such contact occurs, inside-out signalling also supports the stabilisation of the integrin-CAM interaction eventually resulting in firm adhesion of immune cells to endothelial cells.17
Firm adhesion, vice versa, triggers integrin-dependent outside-in signalling affecting the cytoskeleton and resulting in flattening and crawling of the cells along the endothelium.10 Under the influence of other tissue stimuli and endothelial surface molecules, these cells may finally transmigrate to the tissue and thereby complete homing.
While these steps are roughly equal with regard to homing to different tissues, the integrins, CAMs, chemokines and chemokine receptors differ. Thus, the receptor equipment of the cells can be considered as a kind of ‘zip code’ guiding subsequent tissue homing.
The most important integrins controlling adhesion to HEVs of the intestine are αLβ2, αMβ2, α4β1 and α4β7. β2 integrin heterodimers adhere to ICAM-1, whereas α4β1 interacts with VCAM-1 and α4β7 binds to MAdCAM-1 and with reduced affinity also to VCAM-1.19
From naïve to intestinal memory T cells: recirculation and retention
The nature and tasks of adaptive immune cells require further trafficking potential in addition to the above described process of homing to peripheral tissues. In particular, T cell recirculation and retention occurs, which is controlled by differential endowment of T cells with integrins and chemokine receptors depending on their differentiation status. Naïve T cells released from the thymus do not home to peripheral tissues, but need to access secondary lymphoid organs (SLOs) patrolling for encounter of their cognate antigen. Therefore, they are endowed with expression of CD62L (L-selectin), C-C-chemokine receptor (CCR)7 and also low levels of α4β7 integrin20 enabling them to home to the gut-associated lymphoid tissue (GALT), which expresses MAdCAM-1 on its HEVs.21 Two exceptions from the above detailed general principle have to be noted in this context: MAdCAM-1 and the α4β7 integrin may mediate both rolling and adhesion in the GALT. On the one hand, CD62L-dependent rolling occurs through interaction with MAdCAM-1 post-translationally modified with carbohydrate groups.22 On the other hand, low-affinity ligation of α4β7 (in a so called ‘extended-closed headpiece’ conformation) with MAdCAM-1 also mediates rolling.23 C-C-chemokine ligand (CCL)19 and CCL21 released from the GALT are ligands of CCR7 regulating the activation step and facilitating subsequent conformational change of α4β7 and firm α4β7-dependent adhesion to MAdCAM-1.20
If a naïve T cell encounters its cognate antigen presented by antigen-presenting cells in the GALT and appropriate co-stimulation occurs, recirculation is delayed and T cells undergo profound changes giving rise to a clonal pool of effector T cells.24 25 In the further course of the immune response these T cells bequeath different entities of memory T lymphocytes.26 27 If no antigen encounter happens (which is by far more frequent), naïve T cells recirculate to be able to sample the material presented in other SLOs. Such SLO egress is also actively regulated by sphingosine-1 phosphate (S1P) in the peripheral blood, attracting naïve T cells back to the circulation via ligation of S1P receptors (S1PRs).28
Two different levels of T cell differentiation have to be considered (figure 3). On the one hand, on the—so to speak—longitudinal axis, effector and memory T cells develop following antigen recognition in the course of an immune response. Effector T cells are terminally differentiated cells specialised to fight the intruding antigen. Most of them die following elimination of the antigen. To the contrary, memory T cells survive the acute immune response to constitute a long-term immunological memory. As such, they facilitate the establishment of a new adaptive immune response including the generation of new effector T cells in case of later re-exposure with the antigen.26 27 Different models exist to explain the relationship between effector and memory T cells. While some authors favour a model, in which selected effector T cells receive cues to acquire memory function, increasing evidence suggests that memory T cells develop independently and in parallel from activated naïve T cells.29 On the rather transverse axis, different forms of effector T (TEff) cells develop depending on the cytokine environment present during T cell priming. Specifically, interleukin (IL) 12 and transcriptional regulation by the transcription factors T-bet and STAT4 lead to a T helper cell 1 (TH1) phenotype characterised by secretion of IFN-γ,30 31 while impact of IL-4 together with or without TGF-β results in a TH2 or TH9 phenotype producing IL-4 or IL-9 and characterised by transcriptional control by GATA3 or PU.1, respectively.32 33 In contrast, the combination of IL-23 with IL-6 and TGF-β leads to the formation of TH17 cells via the transcription factor RORγt.34 Alternatively, regulatory T cells with anti-inflammatory properties and secreting TGF-β and IL-10 may be induced by TGF-β and retinoic acid and are guided by transcription factor Foxp3.35 36 Both levels of T cell differentiation are important with regard to T cell trafficking since qualitative and quantitative differences in the expression of homing molecules exist as described below.
Importantly, such T cell priming in the GALT is accompanied by the induction of gut homing markers due to specific features of intestinal CD103+ dendritic cells (DCs). These cells are endowed with the potential to synthetise retinoic acid, which drives the expression α4 integrin and CCR9 and downregulates CD62L and CCR7.25 37
Memory T cells are categorised into central memory T (TCM) cells, effector memory T (TEM) cells and resident memory T (TRM) cells depending on the compartment they supervise (figure 4).38 39 TCM cells recirculate between SLOs and the peripheral blood. Accordingly, they constitute an exception from the above described switch in homing markers and their homing marker profile is similar to naïve T cells. To the contrary, gut-homing TEM cells and also TEff cells primed in the GALT are induced to express α4 and CCR9 enabling them to access the gut tissue via α4β7 and α4β1. Thus, after recirculation, these cells may roll on intestinal HEVs mediated by low-affinity interaction of α4 integrins with MAdCAM-1 and VCAM-1.40 In the small intestine, CCL25 may specifically activate gut homing cells via CCR9,41 while no such specific combination has been identified in the large intestine. Regarding the reported importance of the G-protein coupled receptor (GPR)15 for colon homing42–44 and the apparent similarities of its recently described ligand GPR15L with chemokines and demonstrated chemotactic features,45 46 one might speculate that GPR15L could have this role. Finally, α4β7 establishes firm contact to MAdCAM-1 and VCAM-1 and α4β1 to VCAM-1.47 Recirculation of these cells occurs directly to the blood, a process that is also mediated by the gradient of S1P between the blood and the tissue via S1PRs,48 49 or through afferent lymph vessels, which requires CCR7 expression50 (figure 5).
TRM cells differ from the other T cells in that it is their explicit function not to recycle, but to remain within the gut tissue as a resident cell in close association with the epithelium or in the lamina propria. Their maintenance is controlled by tissue factors like TGF-β or IL-15 and they express specific transcription factors and surface molecules. For example, the expression of KLRG1, KLF2 and TCF1 is downregulated, while the expression of Blimp-1 is increased.39 51–53 Moreover, the transcription factor Hobit is specifically expressed in mouse TRM cells, whereas its expression is not similarly restricted in human T cells.54 Such transcriptional regulation leads to upregulation of CD69 and sometimes also CD103 (αE integrin). CD69 counteracts the function of S1PR1, the main S1PR responsible for tissue egress from the gut.55 CD103 is the ligand of E-cadherin expressed on epithelial cells.56 Thus, by inhibiting egress and promoting retention, together, these factors, lead to tissue residence of TRM cells. Interestingly, such TRM cells have recently been identified as potent mediators of chronic intestinal inflammation and flaring episodes by controlling the recruitment and differentiation of other immune cells including innate immune cells and thereby governing a proinflammatory immune response57 (figure 6).
Taken together, T cell trafficking is a sophisticated process with complex steps that are individually regulated. Mechanistic insights generated over recent years have revealed a multitude of potential targets for therapy.
Trafficking of other immune cells
Although most evidence on immune cell trafficking has been generated with T cells, evidence on the trafficking properties of other immune cells is also available.
In general, homing of B lymphocytes seems to be controlled in a similar way as in T cells. Naïve and central memory B cells are recruited to the GALT through interaction of CD62L and α4β7 integrin with their ligands on HEVs as well as by activation of CCR7 by CCL21. However, the B cell lectin CD22 also interacts with MAdCAM-1, which is a pathway apparently unique for naïve and central memory B lymphocytes, since knockout of CD22 led to dramatic impairment of IgD+ B cell recruitment to Peyer’s pateches (PPs), whereas T cell trafficking was not affected.58 Recirculation of B cells from PP is controlled by CXCR4, CXCR5 and S1PR1.59 In contrast to memory B cells that are characterised by expression of CCR6, activated plasmablasts upregulate CCR9, CCR10 and α4β7 integrin.60 α4β7 integrin seems to be critical for gut homing of plasmablasts since β7-/- mice are unable to mount an IgA response.61 In analogy to T cells, the upregulation of gut homing markers like CCR9 and α4β7 integrin in B lymphocytes is mediated by DCs secreting retinoic acid.62
Regarding innate immune cells like monocytes and neutrophils, homing to peripheral organs is mostly regulated by selectins (L-selectin, E-selectin, P-selectin) and interaction of the integrins αLβ2 and αMβ2 with ICAM-1.63 64 In eosinophils and monocytes, rolling and firm adhesion mediated by interaction of α4β1 integrin with VCAM-1 and fibronectin also seem to play a role.65 66 Most likely due to the absence of organ-specific imprinting, gut-specific mechanisms of homing have rarely been described. However, there is evidence that eosinophils make use of α4β7 integrin to home to the gut.67 Moreover, we recently also observed a role of α4β7-mediated gut homing in non-classical monocytes with functional consequences for the balance of macrophage subsets in the gut and intestinal wound healing.68 In addition, innate lymphoid cells (ILCs) type 1 and 3 may undergo gut-specific imprinting mediated by retinoic acid to upregulate α4β7 integrin. To the contrary, α4β7 expression of ILCs type 2 is acquired in the bone marrow.69 Moreover, it seems that the previous view that innate cell trafficking is an exclusively unidirectional process from the site of their generation to peripheral organs needs to be corrected, since recirculation of monocytes has recently been demonstrated.70 However, trafficking of innate immune cell subsets in the context of IBD remains poorly elucidated and further efforts will be necessary to fully understand the complexity of trafficking of these cells.
Key messages on immune cell trafficking
Immune cell trafficking includes all aspects controlling the localisation of cells including adhesion and homing, retention and recirculation.
Adhesion as a central and clinically relevant step of gut homing depends on the interaction of endothelial cell adhesion molecules and integrins on the surface of immune cells.
Expression profiles of homing molecules on immune cells constitute a ‘zip code’ controlling the destination of cells in trafficking processes.
T cells express different homing marker profiles depending on their differentiation status.
Therapeutic approaches for interference with immune cell trafficking in IBD
Figure 7 summarises established and investigational approaches of interfering with T cell trafficking in IBD.
ICAM-1 antisense oligonucleotides (alicaforsen)/anti-αL (efalizumab)
The first efforts to target immune cell trafficking in IBD were related to ICAM-1. As mentioned above, ICAM-1 is the adhesion molecule responsible for αLβ2-dependent and αMβ2-dependent homing and these integrin heterodimers are expressed on a broad variety of immune cells.18 Preclinical studies had shown that anti-ICAM-1-treatment was effective as a prophylaxis in experimental colitis in rats13 and a single nucleotide polymorphism in ICAM-1 is associated with both CD and UC.71 Accordingly, alicaforsen, an ICAM-1 antisense oligonucleotide was developed and investigated in clinical trials. However, although beneficial effects have been reported in case series,72 several studies in CD and UC failed to meet their primary endpoints and the development was not continued beyond phase II.73 74
Efalizumab, an anti-αL antibody, preventing interaction of αLβ2 with ICAM-1 in multiple organs was investigated in an open-label study in CD following its approval for psoriasis and showed promising effects.75 However, due to cases of progressive multifocal leucoencephalopathy (PML) in psoriasis patients treated with efalizumab, the development was stopped.
Natalizumab is a monoclonal humanised IgG4 antibody directed against the α4 integrin subunit and was the first therapeutic agent in the class of adhesion molecule inhibitors approved for CD by the Food and Drug Administration (FDA) in 2004. As natalizumab binds to the α4 subunit, it blocks the interaction of both integrin heterodimers α4β1 and α4β7 with their respective interaction partners VCAM-1 and MAdCAM-1.14 15 Therefore, it is effective in blocking cell trafficking into the gut of CD patients through the α4β7-MAdCAM-1 and the α4β1-VCAM-1 axis as well as cell passage through the blood-brain barrier in multiple sclerosis (MS) patients through the α4β1-VCAM-1 axis.76 77
The precise mechanism of action of natalizumab in IBD has not been investigated in detail. However, several studies carried out in MS models suggest that blocking T cell and B cell infiltration into the inflamed tissue plays an important role, as remission was correlated with lower numbers of these cells in the cerebrospinal fluid.78 79 Of note, despite its efficacy in inducing and maintaining remission in CD, rare cases of new-onset PML under therapy were reported.80 Therefore, natalizumab is not in use in Europe and only available through a restricted prescription programme in the USA.81
Anti-α4β7 (vedolizumab, abrilumab)
Vedolizumab, a monoclonal humanised IgG1 antibody, was approved for the treatment of UC ad CD in 2014.82 83 It is directed against an epitope of the β7 chain, which is only accessible in heterodimeric combination with the α4 chain.84 The highest expression of α4β7 can be found on memory CD4+ T cells, but it is also expressed on other T cell subsets, B cells as well as some other leucocytes.85 In addition to neutralisation of α4β7, ligation of the integrin leads to internalisation of the antibody-integrin complex86 contributing to the inhibition of α4β7-MAdCAM-1 interaction. As demonstrated in a humanised in vivo model of homing to the inflamed gut, vedolizumab specifically inhibits the extravasation of T cells to the gut leading to a reduction of the intestinal T cell infiltration and explaining its efficacy in induction and maintenance therapy.44 Moreover, T helper cell subsets have been reported to express different levels of α4β7. In particular, α4β7 was expressed mainly on TH2 and TH17 cells87 suggesting that the efficacy of vedolizumab might differ depending on the present inflammatory environment.
Interestingly, in this context, some differences seem to exist between CD and UC regarding the α4β7-dependent gut homing mechanisms and the impact of vedolizumab. Some clinical studies suggest that vedolizumab is more effective in UC compared with CD.88 One potential explanation is that UC is characterised by TH2 and TH17 signalling in contrast to CD, where a TH1 and TH17 signature predominates. In addition, an in vivo study from our group suggested that homing of T cells from patients with CD to the ileum via α4β1 and VCAM-1 can serve as a rescue mechanism in the case of α4β7 inhibition.89 Thus, from a translational perspective, further research is required to elucidate the impact of disease entity and localisation on intestinal T cell trafficking to optimise the use of ATAs.
It has early been demonstrated that α4β7 is also expressed on other immune cell subsets including innate immune cells and that vedolizumab binds to these cells.85 We could recently demonstrate substantial reduction of adhesion of CD4+ and CD8+ T cells as well as B cells and even granulocytes to MAdCAM-1 on treatment with vedolizumab in vitro.90 Further, recent studies suggested a substantial influence of vedolizumab treatment on monocytes and macrophages.91 92 Although this seems to contradict the prevailing view that lymphocytes are the main target of anti-α4β7 treatment,8 this is not necessarily the case and both observations can be well reconciled. As mentioned above, recent evidence suggests that there is not only a crosstalk of innate immune cells with adaptive immune cells, but also the other way around.57 This means that any pharmacological intervention affecting intestinal T cells will also reflect in the composition of the innate immune system. Thus, the promotion of anti-inflammatory compared with proinflammatory macrophages under vedolizumab therapy that was reported by the authors,92 is probably a secondary effect of interference with T cell homing—in particular, since no alternative mechanism of action on macrophage level could be detected so far. With regard to the supposed lack of effects on T cell abundance in the mentioned report, two recent publications promoted the notion that vedolizumab might rather lead to qualitative than quantitative alterations of the intestinal T cell pool. While Lord et al suggested that vedolizumab selectively reduces homing of proinflammatory TEff cells,93 Uzzan et al reported that vedolizumab led to a clear reduction of lymphoid aggregates and decreased the number of activated CD38+ T cells in the terminal ileum.94
Thus, collectively, it is highly probable that vedolizumab therapy has non-lymphocyte targets in addition to lymphocytes, for example, α4β7-dependent gut homing of ILCs69 or monocytes.68 However, impairment of lymphocyte homing by anti-α4β7 therapy can still be regarded as a central mechanism of action of vedolizumab.
The safety profile of vedolizumab is excellent,95 which is consistent with its gut-specific mode of action and the absence of systemic immunosuppression. This view has recently been questioned by reports suggesting an increased number of pregnancy-associated complications in vedolizumab-treated patients and their babies.96 In this context, it is important to note that expression of MAdCAM-1 has also been reported on placental vessels.16 Moreover, α4β7 is also a receptor for fibronectin, a component of the extracellular matrix, suggesting that other effects beyond inhibition of gut homing are possible. However, other investigations did not observe pregnancy-associated complications in vedolizumab-treated IBD patients97 suggesting that larger, prospective studies are needed to unequivocally characterise the effects of vedolizumab therapy during pregnancy.
In addition, several studies suggested that the rate of surgical-site infections in vedolizumab-treated patients is increased, most likely due to impaired wound healing.98 99 Other studies, however, reported no difference in surgery-related endpoints between patients treated with vedolizumab or other agents.100 Collectively, apart from special situations, the excellent safety of anti-α4β7 therapy is beyond doubt. However, potential effects beyond impairment of gut homing come into focus and further mechanistic and clinical studies are required to finally elucidate these aspects.
In addition to vedolizumab, another anti-α4β7 integrin antibody, abrilumab, is currently in development and has recently successfully completed a phase II study.101 Detailed mechanistic investigations on abrilumab are so far missing, but it is most likely that there will be no apparent differences from the mode of action of vedolizumab.
Etrolizumab is a humanised monoclonal antibody, which specifically binds the β7 subunit of both α4β7 and αEβ7 integrins, resulting in blocked interaction with the ligands MAdCAM-1 or VCAM-1 and E-cadherin, respectively.102 Etrolizumab has been shown to lead to internalisation of the α4β7 integrin-antibody complex and de novo synthesis of β7 leads to restoration of β7 integrin expression on the cell surface after removal of the antibody.103 On a cellular level, β7 integrin inhibition results in impaired α4β7-dependent gut homing as does anti-α4β7 treatment. However, as demonstrated in intravital trafficking studies, anti-β7 integrin treatment also includes an additional mechanism of action by disrupting αEβ7-mediated mucosal retention. This seems to be particularly relevant for T cell subsets with high expression of αEβ7 such as CD8+ T cells or TH9 cells.87 Two independent reports consistently demonstrated that αEβ7 is predominantly expressed on proinflammatory T cell subsets in the human gut.87 104 Interestingly, αEβ7-expressing cells in the inflamed intestine were also reported to share several features with TRM cells87 and a crucial pathogenetic role of CD69+ CD103+ intestinal TRM cells for chronic intestinal inflammation and the development of flares has recently been demonstrated.57 Therefore, it seems possible that such TRM cells are a major target of etrolizumab.
This is also highly consistent with findings from the phase II study evaluating etrolizumab in UC, which reported improved clinical remission in patients with high compared with low numbers of αE+ cells in baseline colonic biopsy samples.105 Similarly, clinical remission in patients with high ITGAE (αE integrin) expression in baseline biopsies, was more likely than in patients with low ITGAE expression. Additionally, in a post-hoc analysis, Tew et al could show that etrolizumab-treated UC patients with high GZMA (Granzyme A) mRNA levels were more likely to achieve clinical remission. In patients with high baseline expression, the expression decreased in the course of etrolizumab treatment and the authors concluded that GZMA mRNA levels in the colon tissue might be a useful biomarker to identify UC patients most likely benefiting from etrolizumab treatment.106
It is also important to note that αEβ7 integrin is also expressed on DCs in the intestine107 and an additional mechanism of etrolizumab targeting these cells seems possible. While αE+ DCs predominantly induce T cells with regulatory phenotype in intestinal homoeostasis, they seem to promote the generation of pro-inflammatory T cell subsets in active inflammation.108 Yet, it is completely unclear, whether and how ligation of αEβ7 on these DCs might affect their function and future studies addressing this question are highly warranted.
In view of the potential future approval of etrolizumab for use in IBD, the question where it should be positioned in the treatment and whether it is more of less effective and safe compared with other ATAs is obtruding. Since head-to-head studies are lacking so far, a clear answer to this question is not possible. Some experimental data, however, allow speculation on this issue. On a molecular level, etrolizumab was more effective in inducing α4β7 internalisation compared with vedolizumab,103 while there were no differences in inhibiting dynamic adhesion to MAdCAM-1.90 Expression studies have shown that bispecificity of etrolizumab for α4β7 and αEβ7 results in an increase of the target cells.87 In the peripheral blood, this is only marginal, but with regard to induction of αEβ7 expression on a significant portion of intestinal T cells with predominant pro-inflammatory characteristics,104 it is substantial in the gut. Accordingly, it could be shown in vivo that anti-β7 treatment led to a superior reduction of colonic T cell accumulation compared with anti-α4β7 treatment in subsets with high αEβ7 expression.87 Interestingly, the expression of αEβ7 differs in different T helper cell populations as it is the case for α4β7. αEβ7 is highly expressed on TH9 cells, on which α4β7 expression is comparably low87 and anti-β7 treatment might therefore help to specifically target this T cell population. In conclusion, with regard to T cells, these findings argue for a broader impact of etrolizumab compared with vedolizumab. However, the impact of etrolizumab on intestinal DCs and other immune cells is not clear so far as mentioned above. Therefore, predicting and comparing the composite effect in vivo is not yet possible. It has also to be taken into account that αEβ7-expressing cells are not only present in the intestine, but also in other peripheral organs such as the lungs.109 110 Although previous studies did not show particular systemic safety signals for etrolizumab, this aspect needs to be taken into account in the ongoing phase III trials.
Previously, clinical efficacy of etrolizumab has been evaluated in a phase II study including patients with moderate to severe UC and etrolizumab was clearly superior to placebo in inducing remission. Accordingly, several phase III trials are currently recruiting patients to assess the efficacy and safety of etrolizumab in CD and UC in the induction and maintenance phase and with regard to previous anti-TNF therapy.
Ontamalimab (also known as SHP647 and PF-00547659) is a monoclonal antibody against human MAdCAM-1. Due to the selective binding and high affinity of ontamalimab to MAdCAM-1 its interaction with α4β7 is blocked.111 Consistently, expression studies during a phase II trial demonstrated effects on naïve T cells, TCM and TEM cells by revealing increased numbers of circulating β7+ T cells from these subsets in ontamalimab-treated individuals.112
Ontamalimab is therefore the first antibody to target integrin-dependent homing on the endothelial side. From the mechanistic point of view, one might guess on the first view that this will result in exactly the same mode of action as anti-α4β7 integrin treatment. However, we anticipate that this will not be the case due to pleiotropic interactions of MAdCAM-1. In addition to α4β7-dependent gut homing of TEff and TEM cells, MAdCAM-1 also mediates L-selectin-mediated rolling of naïve and central memory T cells in the GALT.22 Thus, interference with the latter process might be an additional mechanism of anti-MAdCAM-1 compared with anti-α4β7 treatment. Moreover, extraintestinal MAdCAM-1 expression has been detected during active IBD,113 suggesting that anti-MAdCAM-1 treatment might affect extraintestinal manifestations of IBD. In conclusion, although α4β7-dependent and L-selectin-dependent rolling are partly redundant, anti-MAdCAM-1 therapy cannot be regarded as a simple copy of anti-α4β7 therapy and differential molecular effects leading to differential clinical effects (at least in patient subgroups) are likely. It will be important to investigate these molecular differences in parallel to the clinical studies to be able to properly interpret their results.
The TURANDOT phase II trial investigated the efficacy and safety of ontamalimab in patients with moderate to severe UC. Significant effects on clinical response and remission as well as mucosal healing were identified.114 These data supported the notion that MAdCAM-1-dependent gut homing is a central pathogenetic element of IBD. However, and partly consistent with potential differences regarding the efficacy of vedolizumab in CD and UC, another phase II study investigating ontamalimab in patients with CD failed to detect statistically significant effects on the rates of clinical response and remission despite clear evidence for pharmacological activity due to a dose-related increase in circulating β7+ TCM cells.115 Phase III trials are ongoing.
S1PR modulators (ozanimod, fingolimod, etrasimod)
Modulators of S1PRs are small molecules that are orally available. Fingolimod, an agonist of the subtypes S1PR1, S1PR3, S1PR4 and S1PR5, has previously been established as a therapy for MS and currently, ozanimod, a modulator of S1PR1 and S1PR5 and etrasimod, acting as an agonist of S1PR1, S1PR4 and S1PR5, are developed for use in IBD.116 117 Due to the earlier introduction to MS therapy, most mechanistic insights come from this field. The mode of action seems to be an indirect one. S1PRs are important for recirculation of cells,20 in the context of colitis particularly for recirculation of T cells primed in the GALT. By modulating S1PR activity, these agents impede such T cells from exiting SLOs including the GALT leading to rapid sequestration.28 This reflects in substantially decreased circulating T cells in S1PR modulator-treated patients,118 in particular naïve T cells and TCM cells, early after initiation of treatment. Although numbers of TEff and TEM cells in the circulation are only marginally decreased,119 it is likely that the effects on colitis are mediated by these cells, since they are capable of gut homing and subsequent promotion of local inflammation. In the longer term, it is also likely that sequestration of naïve T cells and TCM cells leads to reduced generation of TEff cells due to a decrease in the probability of antigen recognition by the former cells. It is currently unclear, whether S1PR modulators also affect the recirculation of intestinal TEM cells.
A phase II trial of ozanimod in UC was recently reported. Despite significant clinical effects, the absolute effects on the primary outcome (clinical remission at week 8) were rather low.118 With regard to the clear effects on the abundance of circulating T cells, this seems surprising. However, it has to be taken into account, that only a minor part of the sequestrated T cells might be gut homing T cells, while the majority might consist of naïve T cells or TCM cells unrelated to intestinal immune responses. Regarding etrasimod, preliminary results from a phase II trial in UC are available. Both clinical and endoscopic endpoints were met. Unlike fingolimod, the safety profiles of ozanimod and etrasimod were favourable in so far studies, most likely due to lack of agonism at S1PR3.120 Further clinical studies will be needed to evaluate the role of S1PR modulators as a potential future therapy principle in IBD.
Others (eg, anti-IP10, AJM300, GPR15)
Several other compounds are currently under development and additional promising targets for therapy have been identified.
AJM300 is an orally available small molecule inhibiting α4. Therefore, it interferes with α4β7-dependent and α4β1-dependent gut homing and the central question for future clinical studies will be whether cases of PML will be observed. The reduced serum half-time compared with natalizumab treatment might be advantageous in this regard.121
Vercinon is an inhibitor of CCR9. Regarding the central role of CCR9 for the activation and subsequent recruitment of T cells to the small intestine, it was evaluated as a potential treatment for CD. However, study results were mixed and collectively disappointing leading to cessation of further development.122 Most likely, additional regulatory effects of CCR9 are responsible for these observations.123
A similar fate might happen to the anti-CXCL10 antibody edelumab. CXCL10 is another chemokine implicated in activation and recruitment of T cells via its receptor CXCR3. However, no significant effects were observed in phase II trials.124 Together, the results obtained with these agents suggest that targeting the chemokine-chemokine receptor axis is much more difficult than interfering with integrin-CAM interactions. Most likely, this is due to less restricted expression of chemokine receptors, which are present on a multitude of cells and make it difficult to forecast the overall clinical effect.
However, GPR15, another receptor with chemokine receptor properties might be worth to question this impression. It has been demonstrated in mouse models that GPR15 crucially controls the homing of TReg cells to the colon,42 while later reports demonstrated that it is responsible for TEff and particularly TH2 cell homing in humans.43 44 Recently, GPR15 has been deorphanised and the ligand GPR15L was demonstrated to have chemokine-like properties.45 Consistently, modulating the GPR15L-GPR15 axis might be an interesting target for the future.
Key messages on therapeutic approaches targeting immune cell trafficking:
The anti-α4β7 integrin antibody vedolizumab inhibits the interaction of α4β7 with MAdCAM-1 and reduces homing of T cells to the gut.
The investigational anti-β7 integrin antibody etrolizumab inhibits gut homing via the interaction of α4β7 with MAdCAM-1 and mucosal retention via interaction of αEβ7 with E-cadherin.
The investigational anti-MAdCAM-1 antibody ontamalimab targets α4β7-dependent gut homing on the endothelial side.
Investigational sphingosine-1 phosphate receptor modulators indirectly reduce T cell infiltration to the gut by inducing sequestration in secondary lymphoid organs.
Important mechanistic aspects with the potential to optimise clinical therapy remain to be investigated and additional targets might become relevant in the future.
Future avenues and unresolved questions
Dosing of anti-adhesion therapies
In a situation, in which ATAs have been solidly established as a new treatment principle in IBD, new questions related to treatment appear and associated complex aspects going beyond the ‘standard’ homing cascade come into focus. For example, the correlation of vedolizumab dosage and clinical response is not completely clear. While there seems to be a positive, approximately linear, correlation over a large range of serum concentrations, several studies reported decreased response or remission rates at higher doses.125 126 One potential aspect contributing to this observation might be that anti-α4β7 therapy does also target anti-inflammatory TReg cells44 127 and a complex balance of gut homing and local proliferation of TReg and TEff cells might be responsible for the overall clinical effect. Therefore, exploration of the mechanisms underlying the dose-response characteristics and further clinical investigations are necessary.
Biomarkers and personalised medicine
With regard to the increasing number of treatment options in IBD, which are all only effective in a portion of patients, the question, which therapy should be chosen for which patients is increasingly discussed. In an ideal world, the evaluation of a set of biomarkers prior to treatment initiation would tell the physician, which therapy is best suited for the individual patient.
This world is still far away, but some steps in this direction have already been taken. A prerequisite is to understand, why some patients respond to ATAs and others do not. One potential explanation for this observation might be the redundancy and pleiotropy of the system. As already mentioned, several redundant gut homing pathways exist meaning that blockade of a specific integrin-CAM interaction might be circumvented by the use of another pathway. In addition, integrins have pleiotropic function. For example, α4β7 binds to MAdCAM-1, VCAM-1 and fibronectin.19 Therefore, the importance of distinct pathways and interactions might be different in individual patients explaining differential response to therapy.
This idea has been supported by the observation that patients with an increase in α4β1 expression in the early course of vedolizumab treatment are less likely to respond to treatment, arguing for increased α4β1-dependent homing via VCAM-1 in these patients.128 Different observations have been made regarding the use of α4β7 itself as a biomarker. One study suggested high baseline expression of α4β7 as a prognostic indicator,129 while another reported a decline in soluble α4β7 in responders.130 To the contrary, two other studies reported that reduced α4β7 levels in the peripheral blood and high levels in the intestine at baseline were associated with good response to therapy.128 131 Interestingly, on a functional level, low α4β7 expression was also related to increased dynamic adhesion of α4β7-expressing cells to MAdCAM-1 in vitro.90 Thus, these prima facie contra-intuitive results might be explained by individual differences in the importance and functionality of the α4β7-MAdCAM-1 homing pathway resulting in extravasation of α4β7-expressing cells from the blood to the gut and—as a ‘negative image’—reduced numbers of α4β7+ cells in the peripheral blood.
Other investigations chose approaches not directly related to α4β7. Anantakrishnan and colleagues prospectively determined the gut microbiome in IBD patients treated with vedolizumab and found intriguing associations between baseline composition of the microbiome and clinical outcome of vedolizumab therapy suggesting that such assessment of the microbiome might be suitable to predict response to therapy.132 To the contrary, a comprehensive study of gene expression profiles in UC patients treated with vedolizumab, did not reveal any gene predictive for response to vedolizumab, although response to vedolizumab at week 52 led to differential regulation of almost 600 genes compared with week 0.133
In synopsis, different strategies have been considered to approach the problem from different sides, but none has led to a broadly applicable solution so far and further efforts in this direction will be necessary.
ATAs are a new pillar in the therapy of IBD providing a completely new mechanistic approach to disease control. The development of current and potential future agents from this class is also a masterpiece of translational success originating from basic studies at the bench. Substantial further efforts will be necessary to identify novel targets in the field, to develop pipeline agents and to refine existing therapies.
The research of SZ and MFN was supported by the Interdisciplinary Center for Clinical Research (IZKF), the transregional Collaborative Research Center TRR241 of the German Research Council (DFG), the research grant ZU 377/3-1 of the DFG, the Else Kröner-Fresenius Stiftung, the Fritz Bender Stiftung, the Litwin IBD pioneers programme of the CCFA, the German Crohn’s and Colitis Organization (DCCV), the ELAN programme of the University Erlangen-Nuremberg, the DFG topic programme on Microbiota, the Emerging Field Initiative and the DFG Collaborative Research Centers 643, 796 and 1181.
Contributors SZ, EB, LLS and MFN jointly wrote the manuscript and approved its final version.
Funding This study was supported by Interdisciplinary Center for Clinical Research (IZKF) (J63) and Deutsche Forschungsgemeinschaft (TRR 241/C04, ZU 377/3-1).
Competing interests MFN has served as an advisor for Pentax, Giuliani, MSD, Abbvie, Janssen, Takeda and Boehringer. SZ, MFN received research support from Takeda, Hoffmann-La Roche and Shire.
Provenance and peer review Commissioned; externally peer reviewed.
Patient consent for publication Not required.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.