Integrins are heterodimeric transmembrane receptors that play an essential role in enabling cells to sense and bind to extracellular ligands. Activation and clustering of integrins leads to the formation of focal adhesions at the plasma membrane that subsequently initiate signalling pathways to control a broad range of functional endpoints including cell migration, proliferation and survival. The α4 and α9 integrins form a small sub-family of receptors that share some specific ligands and binding partners. Although relatively poorly studied compared with other integrin family members, emerging evidence suggests that despite restricted cell and tissue expression profiles, these integrins play a key role in the regulation of signalling pathways controlling cytoskeletal remodelling and migration in both adherent and non-adherent cell types. This review summarises the known shared and specific roles for α4 and α9 integrins and highlights the importance of these receptors in controlling cell migration within both homeostatic and disease settings.
The migration of cells is a vital process involved in the maintenance of homeostatic conditions, for example, during wound healing, but which also plays an important role in the progression of pathologies such as cancer metastasis. Cell migration is governed by the interplay of numerous cell surface receptors, which integrate external signals to initiate or inhibit cell movement. One such group of cell surface receptors are the integrins. Integrins exist as transmembrane non-covalently bonded heterodimers of one α and one β subunit. To date, 24 known combinations of α and β subunits have been identified in mammals, which are formed from the combination of 18 α and 8 β subunits (Figure 1) . These αβ dimers consist of a large extracellular domain, a transmembrane domain and a short intracellular tail. Unlike many signalling receptors, such as the receptor tyrosine kinases, integrins have no innate catalytic activity. Instead, they utilise large intracellular complexes of adaptor proteins to relay signals from the extracellular matrix (ECM) to the cell interior. Integrins are therefore vital mediators of cellular responses to the ECM environment and enable rapid responses to changes in local tissue composition.
α4 and α9 integrins form a distinct subgroup of α integrin subunits because of their relative sequence homologies.
Integrins sense the ECM by binding short amino acid recognition sequences, such as the RGD motif found in fibronectin and laminin . Crystal structures have suggested that at rest, some integrins are in an inactive ‘bent’ conformation with important signalling epitopes sequestered within the integrin structure . Therefore, for receptor activation, the integrin needs to extend into an open conformation enabling the binding of a signalling molecule . This activation mechanism can be triggered through one of the two possible pathways: outside-in and inside-out signalling. Outside-in activation occurs when integrin ligand engagement at the interface between the α and β subunit heads triggers the activation of cell type-specific intracellular signalling pathways. This is achieved through the recruitment of intracellular protein adaptors such as Src and Syk family tyrosine kinases. Inside-out activation, on the other hand, is a complex process whereby other cellular receptors are activated through alternative pathways and ultimately trigger integrin signalling through the association of adaptor proteins such as talin  and kindlins  to the β-integrin subunit. Once active, integrins assemble multi-protein intracellular adhesion complexes to indirectly couple the ECM to the F-actin cytoskeleton. This provides the intrinsic force to propel the cell and control shape change during migration .
Following activation, integrins form a biochemical signalling platform, commonly termed the ‘adhesome’ , at their cytoplasmic tail to transmit intracellular signals. The formation of the adhesome is a highly organised process where maturing transient focal complexes forming behind the leading edge of lamellipodia recruit additional adaptor proteins to become increasingly stable focal adhesions (FA). Some of these FAs subsequently mature further into more stable fibrillar adhesions, thereby anchoring the cell to the ECM through the integrin adhesome [9,10]. The proteins recruited to these adhesomes depend, in part, on the integrin heterodimer that has been activated, and therefore enable the initiation of receptor-specific signalling pathways. These pathways act via the Rho family of small GTPases and F-actin dynamics to regulate cell spreading, polarity, migration and ECM remodelling as well as longer-term events such as survival, growth and proliferation. The integrin intracellular tails are short (20–70 amino acids long, except for the β4 subunit which has an intracellular tail that is 1089 amino acids long); therefore, adaptor proteins must compete for binding to this small region. This competition offers a means to fine-tune the cellular effect of integrin activation. Integrin signalling can be further focussed by post-translational modifications of the cytoplasmic tails, as well as through internalisation and trafficking , transdominant effects of other integrins , and clustering of integrin receptors at the cell surface to increase avidity.
While the β subunits are promiscuous in their associations with the α subunits, each heterodimer is thought to have a specific role, as demonstrated by their tissue-specific profiles of expression and discrete phenotypes in knockout mouse models [13,14]. Thus, the α subunit largely determines the specificity of ligand binding and dictates downstream signalling.
α4 and α9 integrins
18 α subunits have been characterised in the mammalian genome, many of which have been characterised in detail. The α4 and α9 integrins are equally as divergent from all other human integrin subunits (Figure 1A), suggesting they are evolutionarily related. This sequence conservation means that α4 and α9 integrins share several structural and functional properties and therefore have similar expression profiles (Table 1). As such, they have been classified into a pseudo-subgroup within the integrin family . Structurally, the ectodomains of α4 and α9 integrins are distinct from other integrins as neither receptor contains I-domains or post-translational cleavage sites that are common features of other integrin subgroups [15,16]. In addition, this sub-family presents similar ligand-binding profiles . For example, both bind the SVVYGLR motif in osteopontin [18,19] and cryptic epitopes within the EDA region of fibronectin (EDA-FN) . α4 and α9 integrin cytoplasmic domains also share 44% amino acid sequence identity (Figure 1B), including the highly conserved α-integrin juxtamembrane motif which is known to interact with several integrin modulators , and Yxxφ motif which interacts with AP2 to facilitate integrin trafficking  (Figure 2). Taken together, this suggests that as well as co-ordinating signals from the same ECM stimuli, α4 and α9 integrins may also be trafficked and modulated in a similar manner. However, to date, there have been no studies investigating the co-trafficking of these integrins. New tools currently in development could provide increased understanding of spatio-temporal α4 and α9 integrin trafficking in response to physiological stimulations which may provide new insight into specific and shared roles for these receptor subunits .
α4 and α9 integrin cytoplasmic binding partners.
|α4 integrin||α9 integrin|
|β partner(s)||β1 ||β1 |
|β7 — leukocyte-specific|
|Knockout mice||Embryonic lethal, E11–E14 [13,14]|
Defects in placenta and heart 
|Death, <12 days after birth|
(K14-Cre) reduced re-epithelialisation 
|Cell expression||Haemopoietic stem cells|
|Haemopoietic stem cells|
α4β7 in T cells
|α3β1 in epidermis |
|α4 integrin||α9 integrin|
|β partner(s)||β1 ||β1 |
|β7 — leukocyte-specific|
|Knockout mice||Embryonic lethal, E11–E14 [13,14]|
Defects in placenta and heart 
|Death, <12 days after birth|
(K14-Cre) reduced re-epithelialisation 
|Cell expression||Haemopoietic stem cells|
|Haemopoietic stem cells|
α4β7 in T cells
|α3β1 in epidermis |
Functionally, both integrins have been shown to co-ordinate cellular migration in homeostatic and disease states; α4 integrin facilitates T-cell migration, and both α4 and α9 integrins are associated with cancer progression and metastasis in several cell types. In vitro experiments have demonstrated the importance of α integrin cytoplasmic domain specificity for this biological function. CHO cells expressing α2 and α5 cytoplasmic regions fused to α4 ectodomains changed the localisation of the αβ1 heterodimer from puncta to FA-like complexes , suggesting that the α4 cytoplasmic tail may selectively restrain the αβ1 heterodimer from entering FAs. Studies in rhabdomyosarcoma cells also showed that expression of a fused α2 integrin ectodomain and α4 integrin tail led to reduced collagen gel contraction . In both studies, expression of α4 integrin tail with any of the integrin ectodomains led to reduced cell spreading and increased migration compared with wild-type integrins [23,24]. This infers that α4 integrin facilitates transient FA formation, which is associated with rapidly migrating cells.
It was assumed until recently that α4 and α9 integrins share similar biological roles because of the similarities previously discussed. However, recent discoveries have highlighted important differences. For example, only α9β1 can bind to Tenascin-C , a protein in the early wound provisional matrix . Furthermore, α9β1 expression is up-regulated during cutaneous wound healing . This highlights a potential role for α9β1 in modulating adhesion and migration in response to injury, a function currently not assigned to α4 integrin.
α4 and α9 integrins in immune cell migration
During immune surveillance, leukocytes flow freely in the blood patrolling for signs of invading pathogens or sites of injury. Once an insult in the body is sensed (through the presence of chemoattractants or chemokines), leukocytes adhere to the endothelium, before flattening to reduce the impact of shear stress on the adhesive process. Adherent cells then become polarised and form an F-actin-rich lamellipodia at the leading edge and a uropod at the trailing edge, the latter of which is rich in actomyosin capable of contracting the rear of the cell. This change in morphology enables the cell to move along the endothelium and undergo a process of transendothelial migration (TEM) and uropod elongation , which has recently been shown to promote the secretion of chemokines capable of recruiting other immune cells . Given the importance of cell polarity and migration in immune cell invasion, it is perhaps unsurprising that integrins have been identified as integral mediators of each of the steps within this highly co-ordinated process; cell attachment, tethering, slow rolling and subsequently TEM.
Haematopoietic stem cells (HSC) are progenitors of all immune cells, and it has been suggested that α4 integrins are the dominant integrins within the HSC niched . Furthermore, they have been shown to be key for HSC homing to and from the bone marrow, a process required for the production of immune cells . However, recent evidence has shown that α9β1 is also expressed in human and murine HSCs [30,31]. Moreover, HSCs are also now thought to change their integrin profile as they differentiate, giving rise to predominantly α9-expressing neutrophils and α4-expressing lymphocytes , although the reasons for this remain unclear.
Roles in immune cell adhesion
During the first exploratory phase of migration, leukocytes extend long membrane projections (termed ‘tethers’), which strongly adhere to the endothelium to prevent detachment. This facilitates leukocyte attachment and arrest within the blood vessel as well as enabling the leukocytes to withstand shear stress. Selectins were initially thought to mediate tethering alone; however, α4 integrins have also been implicated in the control of this process . α4 integrins are primed to form the initial adhesions for tethers as they primarily localise to pre-existing microvilli on the leukocytes  meaning that they are in close proximity to the endothelium. Additionally, VCAM-1, a cell adhesion molecule expressed on endothelium, is a ligand for α4 integrin and can mediate this strong adhesion. Once α4 integrin binds to VCAM-1, it forms a complex with CD44  and, under flow conditions, recruits the FA protein paxillin to the cytoplasmic tail . Integrin-dependent tension is then relayed along F-actin tethers, promoting activation of the small GTPase Rac through an unknown mechanism, leading to further strengthening of the adhesion site . Following this, the leukocyte microvilli undergo extension by rapidly separating the cytoskeleton from the plasma membrane . More recently, α9β1 has also been shown to stabilise neutrophil adhesion to endothelial cells under shear flow , although the molecular mechanism for this has yet to be investigated. This suggests that despite differences in integrin expression profiles, α4 and α9 integrins may have overlapping and redundant roles within cell type-specific contexts.
α4 integrins have been studied primarily in the context of firm leukocyte adhesion and migration, with a focus on T-lymphocytes in inflammatory disease. In support of the importance of this function of α4 integrin, a humanised monoclonal antibody directed against α4 (Natalizumab) has been shown to reduce migration of T cells in inflammatory disease . Even in neutrophils, which express α4 integrins at much lower levels than α9, α4 plays an important biological role as α4-specific antibodies partially inhibit neutrophil migration . α4 integrin binding to VCAM-1 has been shown to reduce cell spreading, stress fibre and FA formation in a paxillin-dependent manner, resulting in the formation of small nascent adhesions which rapidly turn over to increase cell migration [41,42]. Unlike classical integrin signalling discussed previously, α4 integrin can trigger FA formation without requiring adaptor proteins such as talin or vinculin to link the integrin to cytoskeleton . Instead, α4 integrin specifically binds directly to the adhesion protein paxillin which triggers phosphorylation of focal adhesion kinase (FAK) leading to enhanced migration . In vivo experiments have shown that pharmacologically inhibiting α4 integrin binding to paxillin prevents leukocyte accumulation at sites of inflammation . However, whether this is due to disruption of initial α4 integrin tethering or defects in cell polarity remains unclear.
Roles in immune cell polarity
Maintaining cell polarity is critical for efficient migration. α4β1 has been shown to play a key role in this process as selectively inhibiting α4β1 at the leading edge of T cells leads to reduced migration trajectory . Three phosphorylation sites in the α4 integrin cytoplasmic tail have been reported thus far, which are known to modulate α4 integrin affinity for adaptor proteins (Figure 3). These sites fine-tune α4 integrin-meditated control of cell polarity, as each site is preferentially activated in specific regions of migrating immune cells  (Figure 3). The α4–paxillin interaction is mediated by Y991 phosphorylation on the α4 tail  by an as yet unknown kinase. However, reversible phosphorylation of the α4 tail by cAMP-dependent protein kinase A (PKA) at S988 prevents binding of paxillin to α4 . α4 integrins are type I PKA-anchoring proteins  which mediate leading edge PKA activity in cells undergoing α4β1-dependent migration via association with β1 tail . Additionally, adhesion-dependent Rac activation has been shown to be restricted to the leading edge of migrating immune cells via paxillin LD4 domain recruitment of Git1 . Git1 is an ARF-GAP, and so this interaction consequently decreases ARF6 activity towards the rear of the cell, leading to decreased Rac1 activity . Arf6 is known to recycle inactive β1 integrins to protrusions in a Rab4-dependent manner , and so inhibiting Arf6 could also prevent trafficking of integrins to lateral sides of migrating cells. This would restrict integrin trafficking to lamellipodia, with phosphorylation and modulation of signalling occurring along lateral sides. Other work has shown that the CXCL12/CXCR4 axis is responsible for the redistribution of α4 from vesicles in the perinuclear region to the leading edge of migrating cells in response to chemotactic gradient . In parallel, an α4 integrin–14-3-3ζ–paxillin ternary complex has also been reported to supress Cdc42 activity along the edges of migrating cells  and accelerate cell migration. Phosphorylation of α4 integrin tail at S978, through an as yet unidentified kinase, enables association between the integrin tail and the regulatory protein 14-3-3ζ and prevents Cdc42 activation . Taken together, these data demonstrate that α4β1 guides immune cell polarity through the association with paxillin, which is modulated through the phosphorylation of the α4 integrin cytoplasmic tail.
Localisation of α4 integrins in discrete regions of a migrating cell.
In contrast with α4 integrin, very little is known about the role of α9β1 in controlling immune cell migration. As shown in Figure 2, α4 and α9 integrin cytoplasmic tails contain conserved paxillin-binding domains. For α9, this binding is facilitated by two critical residues (W999 and W1001), and has been shown to inhibit cell spreading thereby facilitating cell polarisation . α4 integrin is phosphorylated within a conserved DSW motif, which is also present in the α9 integrin cytoplasmic tail (Figure 2). This could indicate that α9 integrin is also regulated through phosphorylation, or that the aspartic acid in the ‘SWDWV’ sequence could act as a phospho-mimetic, but this has yet to be investigated.
Roles in leukocyte transendothelial migration
TEM is the process whereby leukocytes cross endothelial cells to access areas of inflammation outside the vasculature. This phenomenon can happen either paracellularly (PTEM) or transcellularly (TTEM) (reviewed in ). PTEM occurs when tight and adherens junctions (AJ) between endothelial cells are ‘unzipped’, allowing leukocytes to cross the endothelial barrier. Conversely, TTEM occurs when the leukocyte binds to a lateral border recycling compartment (LBRC) rich in molecules that are normally found at cell borders including ICAM-1, PECAM, JAM-A and CD99. The LBRC then aids in the migration of the leukocyte through the cell body, before reattaching to the ECM on the other side of the endothelial cell. While this complex intracellular change within the endothelial cells in response to neutrophil intrusion is well characterised, the molecular basis of TEM in leukocytes is poorly understood. Function blocking antibodies against either α4 or α9 integrins have been shown to inhibit neutrophil TEM [32,53]. Additionally, α9 integrin membrane expression is up-regulated by neutrophil transmigration through integrin recruitment from intracellular stores . Interestingly, stimulation of neutrophils with ADAM9, an intercellular molecule found at endothelial junctions that binds to α9 integrins , was shown to stimulate canonical migratory PI3K/Akt and ERK signalling pathways in neutrophils , suggesting migration during TEM could utilise similar mechanisms as other types of migration. In addition to this, and similarly to the α4 integrin/CXCR4 cross-talk discussed above, CXCR2 was shown to activate α9 integrin in these cells facilitating migration . However, to date, no more detailed molecular studies of the cause and reason for the change in expression of these integrins have been performed. It is tempting to speculate that TEM exposes the leukocyte to α9 integrin-specific ligands in the ECM, such as Tenascin-C, switching from α4 integrin-based migration in the vasculature to α9 integrin-based migration in the surrounding connective tissue. However, this possibility has yet to be investigated.
Roles for α4 and α9 integrins in adherent collective cell migration
Collective migration requires co-ordination of cell–cell and cell–ECM adhesions to enable movement of groups of cells (sheets, clusters or strands) as a cohesive unit. Types of collective migration are used by different cell types during embryonic development, angiogenesis and wound healing and can occur in response to soluble or physical environmental cues. The current proposed mechanism for collective movement suggests that ‘leader’ cells at the free edge of the cell group sense the external environment and relay signals to attached follower cells behind the free edge to co-ordinate both the speed and direction of collective migration [56,57]. Critical to collective migration is the establishment and maintenance of cell–cell contacts to neighbouring cells, which are facilitated by tight, adherens and gap junctions. AJs are linked to the cell cytoskeleton through the cadherin family of transmembrane receptors, which homodimerise in trans with cadherin molecules on adjacent cells. Cadherins have been shown to recruit several proteins to AJs that also localise to FAs  and also share common effectors that remodel the cytoskeleton during migration, providing co-ordination between cell–cell and cell–matrix adhesion types. This cross-talk also indicates the importance of integrins in modulating both cell–cell and cell–ECM adhesion to facilitate co-ordinated and directed mass cell movement.
α4 integrins have not yet been well studied in collective cell migration contexts. However, α4 integrin knockout mice show embryonic lethality due to cell–cell adhesion defects leading to aberrant placenta formation and cardiac defects . Antisense oligonucleotides directed against α4 integrin in epicardial cells also result in reduced cell adhesion and an increased mesenchymal phenotype . However, further experiments in CHO cells have indicated that α4 integrins do not homo- or heterodimerise with other integrins . Moreover, the migration of B cells on an α4-specific ligand was not sufficient to induce cell–cell adhesion and clustering . This suggests that the formation of cell–cell adhesions cannot be directly facilitated by α4 integrins binding to themselves or across adjacent known extracellular ligands. While potential alternative contributions of α4 integrins in controlling the collective movement of cells remain unclear, the known role for these receptors in control of front–rear polarity and protrusion in other cell types is suggestive of involvement in this context and represents an interesting area for further investigation.
α9β1 integrins in epidermal cell migration
In contrast with α4 integrin, α9β1 has been implicated in several different processes relating to collective cell behaviour. During skin homeostasis, basal keratinocytes adjacent to the ECM constitutively express α9β1 integrin, which binds to shared α4 integrin ligands EDA-FN  and EMILIN-1 . However, during the first 7-day post-injury, expression of α9 integrin is elevated before returning to pre-injury homeostatic levels . Tenascin-C (an α9β1-specific ligand) is also highly expressed during tissue repair . It is perhaps because of these reasons that the best-defined role for α9β1 is in the context of the skin and collective cell migration during wound healing. During wound healing, there is a co-ordination between the re-epithelialisation of the wound bed and keratinocyte differentiation . Re-epithelialisation of the wound prevents invasion by external pathogens, while differentiation (the stratification and cornification of keratinocytes in the epidermis of the skin) produces the physiological properties of the skin that prevent mechanical injury and water loss from the body. α9β1 has been shown to orchestrate cell migration and re-epithelisation through common adhesion signalling molecules such as Src and FAK [66,67], and also to promote FAK auto-activation in keratinocytes , although these signalling pathways are dispensable. In contrast with migration, differentiation requires keratinocytes to detach from the basement membrane and exit from the cell cycle, while retaining tight desmosomal cell–cell adhesions . Conditional α9 integrin knockout mice under the control of Keratin-14 promotor showed that wounds favoured re-epithelialisation over structure and differentiation, suggesting that α9 integrin is involved in this process . Differentiation could be facilitated through α9β1 interaction with the recently identified α9β1 ligand SVEP1 (sushi, von Willebrand factor type A, EGF and pentraxin domain-containing protein 1). SVEP1 is a secreted, multidomain protein that facilitates cell adhesion to the ECM in several tissues  as well as modulating epidermal cell differentiation . Whether the epidermal defects in SVEP1 transgenic mice result from a direct interaction of α9β1 binding to SVEP1 or another cell surface receptor remains to be determined.
α9β1 integrins in collective polarity signalling
As well as the canonical migratory pathways that are activated during α9β1-mediated cell migration in the skin, several alternative mechanisms have also been described. One alternative way in which α9β1 integrin facilitates migration is through modulation of inward-rectifier potassium channel 4.2 (Kir4.2). Inward rectification is the process by which a channel passes positive current more easily in the inward direction and can be used, for example, to help maintain membrane potentials in neurons. Kir4.2 can be blocked by voltage-dependent channel blockers such as intracellular organic cations (like polyamines spermidine and spermine) or Mg2+ . The cytoplasmic domain of α9 integrin, but not α4 integrins  can directly bind spermidine/spermine N1-acetyltransferase (SSAT) and is co-localised with Kir4.2 in FAs at the leading edge of migrating cells . This enhances migration of the cells through SSAT catabolism of spermidine/spermine to lower order putridine, which cannot occlude the channel pore. In a similar manner to α4–paxillin binding, this enhances the formation of a single lamellipodia by spatially restricting dynamic regulation of polyamine levels. Although the exact molecular mechanism of how this facilitates migration is unclear, polyamines are known to be important for the organisation of the cell cytoskeleton and migration . One example of this is in the gut, where polyamines have been implicated in modulating the interactions of Src, FAK and GEF Tiam1  to subsequently manipulate downstream signalling pathways and alter the collective epithelial migration. α9β1 has also been implicated in mediating enhanced cell migration through inducible nitric oxide synthase (iNOS) activity regulated by Src . This, in turn, has been shown to increase both the expression and activation of the small GTPases Cdc42 and Rac1 through protein kinase G activation , although the exact mechanisms by which α9β1 controls this pathway remain to be defined.
Transdominant signalling by α4 and α9 integrins in collective migration
Interestingly, both α4 and α9 integrins can influence collective cell migration through so-called transdominant effects. Integrins rely on cytoplasmic protein adaptors to transmit intracellular signals, and these signalling cascades are tightly regulated through limited levels of these proteins. Therefore, active receptors with the highest affinity for either the ECM ligand or the intracellular binding protein will sequester these adaptors away from other receptors. This reduces the available pool of proteins and thus suppresses the level of signalling initiated by the other integrins . Previous data have suggested that α9β1 plays a transdominant and opposing role to α3β1 during wound healing in vitro (Table 1; ), and interestingly, α3β1 has also been shown to reduce directionality and wound re-epithelialisation in skin-specific knockout mice . This transdominant effect is thought to occur downstream of FAK auto-activation, instead of impinging on α3β1-dependent Src-mediated Y861 and Y925 FAK phosphorylation . Phosphorylation of FAK at Y861 and Y925 sites has been shown to modulate Rac activation and turnover of focal contacts, respectively , providing a potential mechanism by which α9β1 could modulate α3β1-mediated migration. Interestingly, these transdominant effects are independent of α9-paxillin binding , which suggests that an as yet undefined protein could also be interacting with the α9β1 complex to mediate these responses.
Roles in cancer cell migration
Tumour cell dissemination from the primary site (metastasis) is a stepwise process. During metastasis, epithelial cells change from adhesive epithelia, which interact with the basement membrane and one another, to a polarised mesenchymal phenotype in a process referred to as epithelial to mesenchymal transition (EMT). Through the stimulation of various signalling pathways including TGFβ, Sonic hedgehog and Notch, various transcription factors are activated which are associated with an increased migratory capacity . This requires the down-regulation of strong E-cadherin-based cell–cell adhesions , and up-regulation of transcription factors such as β-catenin and Snail  which enable the cells to detach from the primary tumour, before they extravasate into either blood vessels (haematogenous spread) or the lymphatic system (lymphogenous spread) . If the cell survives in circulation, it must then undergo a second extravasation step at a distant site before invading and proliferating. Given the similarities in these phenotypic changes to normal immune cell migration and wound healing process, both α4 and α9 integrins have been implicated in the control of EMT and metastasis.
Potential roles for α4 and α9 integrins in EMT
α9β1 is associated with invasive fronts of primary colorectal cancer (CRC) and gastric cancers [85,86]. It can facilitate EMT by binding to the unique EDA-FN motif which up-regulates expression of α9β1, Snail and vimentin while simultaneously down-regulating E-cadherin expression . How EDA regulates the expression of these proteins at the mRNA level remains elusive. Additionally, α9β1 can form a tri-partite protein complex with β-catenin and E-cadherin at cell–cell junctions, which dissociates following integrin ligand binding. Following complex dissociation, Src binds directly to the cytoplasmic domain of α9 integrin and promotes phosphorylation and nuclear translocation of β-catenin . Interestingly, this pro-EMT effect is transduced without a requirement for a change in α9β1 expression , unlike many other oncogenic factors. More recent work demonstrated that co-localisation of α9β1 and Kir4.2 at the leading edge of migrating cancer cells is facilitated through urokinase-type plasminogen activator receptor (uPAR) , which acts to concentrate urokinase proteolytic degradation of the ECM  and increase the expression of TGFβ . Taken together, this could suggest that the α9β1–SSAT–uPAR complex facilitates TGFβ-mediated EMT and enables cancer cell migration through ECM degradation. Interestingly, TGFβ also induces up-regulation of Wnt11, which can, in turn, up-regulate α4 integrin expression leading to increased cell–cell adhesion and reduced cell invasion of ovarian cancer cells in vitro . Additionally, highly metastatic E-cadherin deficient breast cell lines show down-regulation of α4 integrin expression , all of which indicates that conversely to the role of α9, α4 integrins may negatively regulate EMT.
α9 and α4 integrin expression and signalling in invasion
The association between integrin expression levels and invasiveness is complex. α9 integrin expression is reduced in ∼70% of hepatocarcinomas, leading to Rac1 and RhoA hyperactivity and increased FAK and Src phosphorylation , which is associated with an increased migratory capacity. Conversely, α9 integrin expression is heterogeneously altered in breast cancers, with ∼44% of tumours showing down-regulation or loss of ITGA9 and ∼45% showing normal or increased levels of expression . This is interesting as α9β1 can enhance migration though Src and iNOS signalling ; in mammary tumours, NOS signalling through guanylate cyclase culminates in Erk1/2-mediated cell invasion  and Src-mediated Y1055 phosphorylation of iNOS which stabilises iNOS half-life . The net result of this imbalance in breast cancers with altered α9β1 expression leads to an increased concentration of iNOS, which may contribute to increased α9β1-mediated migration and invasion. α9β1 can also influence the migratory capacity of melanoma cells. In its low affinity state, α9 integrin mediates transient adhesion and sparsely arranged FAs  in a Rac-dependent manner , thereby promoting migration. Conversely, following ligand engagement α9 integrin converts to a high affinity state, triggering Yes-dependent phosphorylation of cortactin at Y470 and α9 integrin localisation to FAs, which leads to activation of Src . Ultimately, α9β1 ligand engagement stimulates Rho kinase (ROCK)-dependent formation of F-actin stress fibres and FA maturation associated with stable adhesion to the ECM and slower migration speed . Cortactin phosphorylation has also been shown to reduce levels of active β1 integrins at the cell surface, potentially by limiting integrin recycling . This could contribute to the further decrease in α9β1-medited migration after ligand engagement, as active integrin trafficking is required during α9β1-mediated migration to maintain appropriate levels of integrin activity at the leading edge of migratory cells .
α4 integrins have also been associated with invasive tumours and metastatic spread . α4β1 binding to VCAM-1 increases transendothelial, haptotactic and chemotactic migration of metastatic melanoma cells in vitro  and drives lymphatic metastasis by promoting lymphatic migration and survival in vivo  potentially through interaction with CXCR4 . Similarly to the adhesions formed during lymphocyte adhesion and migration, CXCR4 expression in small cell lung cancer cells directs α4β1 attachment to ECM . In addition to this, α4β1 binding to midkine, a secreted plasma protein associated with tumour progression  transiently induces paxillin phosphorylation and increased migration of osteoblastic cells in vitro , which is required for lamellipodial protrusion . However, α4β1-mediated migration in cancer may be stage and cell type-specific; α4β1 inhibition blocked attachment to FN in small cell lung cancer but  not breast cancer cell lines  and in vivo work has suggested that α4 integrin may inhibit metastasis formation in a large number of organs at a stage subsequent to lymphoma cell dissemination .
Transdominant signalling by α4 and α9 integrins in cancer
α9β1 can also mediate EMT by influencing α4 integrin expression, thus acting in a transdominant fashion, which may partially explain the differences seen in cell- and stage-specific contributions to invasion by these integrins. Both in vitro and in vivo studies have shown that disassembly of stress fibres following α9β1-mediated Tenascin-C binding leads to the shuttling of the mechano-responsive transcription factor YAP from the nucleus to the cytoplasm. This inhibits YAP-mediated transcription of several FA proteins and regulators including RhoA, vinculin, zyxin and talin  as well as α4 integrin . Phosphorylated β-catenin has also been shown to promote α4 integrin transcription , which could indicate a feedback mechanism where activation of this key transcription and EMT-inducing factor controls expression profiles of anti-invasive receptors. This is particularly interesting as α4 integrin has been associated with a reduction in invasive potential in several cell types and is negatively correlated with primary tumour grade [91,107,111–113]. α4 integrin has also been shown to decrease the surface levels of α5 and αv integrins and subsequently reduce cell motility of HSC-3 cells  again indicating that transdominant action over other integrins may be a key mechanism by which this receptor regulates motile behaviour. However, α4 integrin overexpression was required to induce this phenotype, and in the context of melanoma, this transdominant relationship of α4 over α5 integrin increases cell motility , therefore, the physiological relevance of this in a disease setting remains unclear. The relative expression of these integrins in primary tumours and metastasis has not yet been studied together, but this would be an important approach to take in future to determine whether the integrin profile switches as cells undergo metastasis. Further analysis of the expression levels and shared or specific roles for these integrins in different cancer types may provide new routes for therapeutic intervention.
Conclusion and future perspectives
The wealth of evidence, to date, suggests that α4 and α9 integrins are important regulators of cytoskeletal dynamics and cell migration, particularly within pathological settings such as wound repair, inflammation and cancer. This makes these receptors potentially very interesting therapeutic targets; indeed α4 inhibitors are already in use for specific inflammatory diseases. However, it would also appear that these receptors act in both pro- and anti-migratory capacities depending upon the cell type and tissue environment. How these apparently opposing functions for this sub-family of receptors are controlled at the molecular level, and whether specific ligands dictate specific α4- and α9-dependent signalling outcomes remain to be defined. The potential impact of modulating α4 and α9 integrin function on the expression and activation of other membrane-bound integrin family members is also likely to be of significant importance when considering and interpreting how these receptors operate on a cell- and tissue-wide scale.
EDA region of fibronectin
epithelial to mesenchymal transition
focal adhesion kinase
haematopoietic stem cells
inducible nitric oxide synthase
lateral border recycling compartment
cAMP-dependent protein kinase A
sushi, von Willebrand factor type A, EGF and pentraxin domain-containing protein 1
transendothelial migration, paracellularly (PTEM) or transcellularly (TTEM)
urokinase-type plasminogen activator receptor
The authors thank the Medical Research Council [MRC; MR/N017242/1] and Tocris (Biotechne) for funding.
The authors thank Doug Sammon for critical reading of this manuscript.
The Authors declare that there are no competing interests associated with the manuscript.