In situations where migrating cells first encounter each other to form junctions, changes in cellular morphology have been observed and correlated with the spatiotemporal distribution of cadherins and catenins (alpha-catenin, beta-catenin, delta-catenin). Cell culture experiments show the projection of exploratory filopodial processes from apposing cells as they reach out to make contact with each other. These contacts are short-lived and extend and retract multiple times before finally forming a stable contact. Once this contact stabilizes it extends in a zippering fashion through the protrusion of lamellipodial processes. Cadherins and catenins begin to cluster at these contact sites . E-cadherin is known to bind β-catenin immediately upon export of these proteins from the endoplamisc reticulum, with α-catenin joining this complex only after it is recruited to the plasma membrane . All three of these proteins continue to accumulate at and strengthen the nascent junction 
These findings from cell culture work are further supported through studies in model organisms, such as the fruit fly (Drosophila melanogaster)  and nematode (Caenorhabditis elegans)  (as reviewed in ). In Drosophila embryos the highly regulated process of dorsal closure of the epidermis involves the projection of filopodia and lamellipodia from cells at the leading edge of the epithelial sheets, in much the same way as previously described in cell culture studies .
In C.elegans embryos, sealing of the epidermal sheets around the entirety of the embryo in a process termed ventral enclosure, also utilizes filopodia. Moreover, α-catenin was shown to rapidly accumulate at filopodial contact points in a cadherin-dependent manner and in doing so promote junction formation. In cells at the leading edge made mutant by removing cadherin expression, filopodial protrusions were still extended, however catenin accumulation and subsequent junction formation failed. The term ‘filopodial priming’ was coined to describe the formation of nascent junctions via the recruitment of α-catenin to filopodial tips engaged in cell-cell contact via their cell surface cadherins. This process leads to further recruitment of cadherin-catenin complexes in order to strengthen the nascent junction .
Another family of proteins that have been detected at nascent junctions are the nectins and nectin-like (Necl) proteins (see Figure below). These proteins have been suggested to initiate AJ formation prior to the involvement of cadherins in some circumstances, as evidenced predominantly from mammalian cell culture studies (as reviewed in ). Nectins comprise a single extracellular domain of three Ig-like loops, a transmembrane domain and a cytoplasmic domain capable of binding the actin-binding protein afadin (as reviewed in ). Nectins, like cadherins, can interact with each other in trans between apposing cells and in cis on the same cell. Nectins utilize their first and outermost Ig-like loop to engage in trans interactions and their second Ig-like loop to engage in cis interactions, which is a pre-requisite to the subsequent formation of trans interactions . In contrast to cadherins, these interactions do not require calcium.
The formation of nectin-nectin interactions at sites of cell-cell contact have been shown to promote further nectin-nectin interactions and increase the clustering of nectin-afadin interactions , with afadin suggested to provide a direct link to the cytoskeleton. The resulting nectin-based adhesion sites are thought to serve as focal points for the subsequent recruitment of cadherins and eventual formation of adherens junctions. This forms the basis of the ‘fork initiation and zipper’ model for cell-cell adhesive junction formation, as proposed by Irie et al  and supported through mechanical force measurements .
At the single molecule level, Tsukasaki et al demonstrated that trans interactions between nectin domains were uncooperative, i.e. they showed ‘zipper-like’ unbinding. In contrast trans interactions between cadherin domains were cooperative, i.e. they showed ‘parallel-like’ unbinding (see Figure below). These dynamic properties of nectin and cadherin binding were suggested to fit to a ‘fork initiation and zipper’ model of junction formation . In this model, initial cell-cell contacts are mediated by the shorter-lived, uncooperative nectin-nectin domain interactions, which being more rapid in binding and unbinding are considered better suited to exploratory cell behavior. Initial nectin-nectin trans interactions stimulate downstream signaling resulting in increased cell-cell contact sites, for example through promoting increased filopodial protrusions. This process is described as ‘fork initiation’. Following the ‘fork initiation’, ‘zippering’ occurs to stabilize junction formation. Longer-lived, cooperative cadherin-cadherin domain interactions are suggested to mediate this phase, due to their more robust nature when compared to uncooperative, nectin-nectin domain interactions.
It should however be noted that genetic studies in model organisms have brought into question the relative importance of nectins compared to cadherins in the formation of AJs (as reviewed in cite]20571587[/cite]). Studies in knockout mice lacking different nectins showed that the loss of nectin was not embryonic lethal, as one would expect if nectins were crucial to the formation of all AJs . Knockout mice lacking the nectin-binding protein afadin did show AJ defects during an early stage of embryonic development called gastrulation . However, these defects were less severe and occurred later than those observed with mice lacking the cadherin-catenin complex. Similar results were obtained in fruit fly embryos with cells mutant for Echinoid (equivalent to nectin) or Canoe (equivalent to afadin), AJs were still able to form . The involvement of nectins in AJ formation may be therefore be cell type and context specific, with the cadherin-catenin system being a more globally implemented system.
Nectin-nectin cis interactions enable the formation of nectin-nectin trans interactions on apposing cells. These interactions, being uncooperative and short-lived in nature, are suited to the exploratory cell behavior initiated between cells coming into close contact .
Nectin-nectin trans interactions stimulate a signaling cascade (as reviewed in ) starting with the activation of the tyrosine kinase cellular-Src (c-Src) and leading to the activation of the Rho family GTPase Cdc42 via the Cdc42 GEF (guanine nucleotide exchange factor) FRG . Activation of FRG requires the convergence of two phosphorylation pathways; one mediated by c-Src and one mediated by the small G protein Rap1, whose activation is itself a result of a c-Src pathway involving the Crk–C3G complex (an adaptor protein and a guanine nucleotide releasing protein, respectively) .
Cdc42 activation leads to increased filopodia formation and activation of the Rho family GTPase Rac that leads to increased lamellipodia formation. It should be noted that in this context Rac activation requires both Cdc42 and c-Src acting via the Rac GEF Vav2 . The increase in filopodia enhances cell-cell contact, whilst the increase in lamellipodia promotes the closure of gaps between contact sites through increased cell surface receptor interactions . These morphological events can therefore be described as ‘fork initiation’ (multiple filopodia extending outwards) and ‘zippering’ (lamellipodial protrusions sealing the gaps). This signaling cascade ultimately leads to the recruitment of cadherins to the nectin-based cell-cell contact in order to stabilize the formation of the adhesive junction (as reviewed in ).
Type I cadherins, also known as classical cadherins, are commonly found in adherens junctions (AJs) and desmosomes. They are produced as a precursor polypeptide that is non-adhesive. This pro-form must be proteolytically cleaved at its N terminus to produce a functional cadherin, after which the cadherin is trafficked to the plasma membrane for use . Cadherins comprise five extracellular cadherin (EC) repeats in their ectodomain. These highly conserved repeats are a hallmark of the cadherin superfamily and are roughly 110 amino acids in length. Each EC domain comprises seven β-strands (labeled A to G) arranged into two β-sheets, with opposite orientations to facilitate tandem repeating of these subdomains . The desmosomal cadherins, desmoglein and desmocollin, also follow this pattern and contain five EC repeats 
It is currently unknown precisely how the cadherin-catenin complex is recruited to sites of cell-cell contact, with several different models being posited including, but not limited to; nectin-based recruitment and spontaneous recruitment.
Evidence from epithelial cell culture suggests that the establishment of early nectin-nectin based adhesions leads to the recruitment of cadherins, although the requirement for nectin-based interactions in AJ formation in vivo is far less clear (as reviewed in ). In cell culture scenarios various proteins are suggested to bridge the gap between the nectin-afadin complex and the cadherin-catenin complex.
During early junction formation vinculin and ponsin are suggested to link the nectins to the cadherins. Vinculin is known to bind α-catenin that can bind β-catenin, which directly interacts with cadherin. In vitro experiments show ponsin is able to bind both afadin and vinculin, however these interactions are competitive, suggesting additional proteins are needed in vivo to allow ponsin and vinculin to recruit cadherins to nectin-based adhesion sites .
Other protein complexes connecting nectin-afadin to cadherin-catenin include LMO7–α-actinin  and ADIP-α-actinin . However it is unclear at what point during junction formation these proteins come into play, with evidence suggesting LMO7 is only involved at a more mature stage of junction development .
Experiments in epithelial cells expressing cadherin tagged to a photoconvertible fluropohore for easy tracking, have shown AJs to be highly dynamic structures that continuously turnover cadherin molecules. Further investigation using this system resulted in a model whereby cadherins are constantly and spontaneously recruited to the plasma membrane, form lateral catenin-dependent associations and are subsequently released from these clusters in an active, ATP-dependent manner. In this model the formation of a stable AJ is suggested to occur through the inhibition of cadherin release from clusters in the plasma membrane, allowing the accumulation of cadherins at the site of cell-cell contact .
Cadherins arrive at the plasma membrane through exocytosis. In epithelial cells, soon after E-cadherin and β-catenin are synthesized and exit the endoplasmic reticulum, these two proteins bind to each other and are trafficked together to the plasma membrane . When two cadherins on apposing cells first meet, binding occurs via their ectodomain (extracellular domain), specifically their outermost EC1 domain (as reviewed in ). This process is calcium dependent, requiring three calcium ions (Ca2+) to bind between each EC repeat. Each calcium binding site is formed from the end of one EC repeat, the subsequent linker region and the start of the next EC repeat. Ca2+ binding results in the regions between each EC repeat adopting a more rigid orientation that causes the length of the ectodomain to adopt a defined curvature . It is this mature, curved form of cadherin that engages in trans interactions.
Trans cadherin interactions result in an adhesive dimer interface between the interacting EC1 domains. Structural analyses show this interface is formed by the ‘swapping’ of β-strand A of one EC1 domain with β-strand A of another EC1 domain, specifically involving the insertion of the conserved tryptophan-2 side chain of one EC1 domain into a hydrophobic pocket of the opposing EC1 domain  (see Figure below). The force driving behind this interface is that of conformational strain, which results from β-strand A of each EC1 domain being strongly anchored by tryptophan-2 at one end and Ca2+-bound- glutamine-11 at the other end. This strain provides the necessary force to flip out the β-strand for ‘swapping’ . Sequence analyses suggest a similar scenario is likely to occur between the EC1 domains of desmosomal cadherins .
The role of cis interactions between cadherins is less established, with no clear consensus on precisely which domains are engaged in these interactions (as reviewed in ). Recent evidence however suggests that these interactions are not required to facilitate trans interactions , whereas trans interactions are required in order for cis interactions to occur . Computational modeling has been carried out to determine how these interactions, both trans and cis, result in the formation of a stable junction .
These in silico experiments showed that cadherins freely floating in the plasma membrane are not able to successfully form trans dimers unless their diffusion is slowed down in some manner. The introduction of a diffusion trap permitted the formation of trans dimers, however without subsequent cis interactions these trans dimers were unable to coalesce to form an ordered structure. Though weak, cis interactions were nonetheless shown to be vital for junction formation. Moreover, the ordered formation of multiple trans dimers into a junction was shown to be polarized such that each trans dimer had a specific orientation, resulting in the production of a directional lattice .
The transition from freely diffusing cadherin monomers and trans dimers to a stable, ordered, 2-dimensional junction comprised solely of cadherin trans dimers is therefore a co-operative process requiring first trans and then cis cadherin interactions. The combination of multiple cis and trans interactions provides strength and structure to adhesion sites.
In the absence of the formation of a stable adhesion site, the cadherins at the plasma membrane are removed. This can occur through endocytosis, which has been observed in established AJs upon the release of cell-cell contact (see Figure below and ) and is known to regulate E-cadherin distribution in mature AJs . Endocytosis has been proposed as a means to prevent cadherin accumulation at the plasma membrane in the absence of cell-cell contact, which would otherwise occur due to the natural and spontaneous accumulation of cadherins 
The accumulation of cadherins, in the presence of cell-cell contact stimulates events at the cytoplasmic side of the adhesion site, including reorganisation of the cortical actin cytoskeleton through effectors downstream of Rho GTPases and through linker proteins in complex with nectins and cadherins. Collectively these processes establish the gross architecture of the adherens junction.
One of the primary roles of cell-cell adhesion complexes is to connect the actomyosin network of one cell to that of neighboring cells and facilitate the generation and transduction of mechanical forces at their interface 
Different types of adhesion complexes exist, each with their own composition. For example, while the classical cadherins in adherens junctions (AJs) are bound to catenins such as β-catenin and p120 catenin which subsequently bind to the F-actin cytoskeletal network, desmosomal cadherins, bind to distinct cytoplasmic proteins such as plakoglobin (also known as γ-catenin) which in turn link to intermediate filaments. This variation exists despite desmosmal cadherins possessing a similar domain organization to the classical cadherins . In this case the adhesion complex, known as a desmosome, is found in tissues subjected to higher levels of mechanical stress .
All cell-cell adhesion complexes are composed of intercellular and cytoplasmic components. Intercellular interactions are generally facilitated by cell-adhesion molecules (CAMs) such as cadherin or nectin, often in a calcium-dependent process. The complexity of the cytoplasmic region on the other hand has yet to be fully defined. This region, which has been visualized as a dense plaque of proteins (reviewed in ) that contains a vast array of structural and scaffolding proteins as well as regulatory proteins. These regulatory proteins mediate adhesion dynamics, including assembly and disassembly and will be discussed in greater detail in Regulation of Adhesion Dynamics.
In its most simplistic arrangement, the cytoplasmic component of AJs can be described as containing several core components. These include p120-catenin, which is bound to cadherin at the juxtamembrane position  and β-catenin or plakoglobin, which are bound to the cytoplasmic tail . α-catenin, which also binds to β-catenin, then links the complex to the actin cytoskeletal network. Although these proteins form the core components of AJs, and have been shown to be indispensable for adhesion function, a substantial body of research indicates a far greater complexity.
Interactions between cadherin and β-catenin, which occur shortly after the proteins are translated and before cadherin integrates into the membrane , facilitate further interactions with other adaptor proteins. These interactions may influence the activity of regulatory proteins such as GAPs and GEFs, tyrosine and serine/threonine kinases and phosphatases however they also facilitate the assembly of a physical link that is capable of mechano-signalling and force transduction. The structural composition of the plaque region is discussed below. It should be noted despite significant advancements in the field in recent years, the specifics of AJ composition and recruitment have yet to be fully defined.
Early biochemical studies exploring the structural link between the cytoskeleton and cadherins in adherens junctions (AJs) concluded that it comprised direct interactions in the following order; cadherin tails bind β-catenin, β-catenin binds the VH1 domain of α-catenin , the VH3 domain of α-catenin binds actin .
Indeed, the importance of α-catenin in this link was highlighted after chimeric cadherin-αE-catenin fusion proteins were introduced into fibroblasts lacking cadherin, and a rescue of cadherin function was observed. This was noted to be dependent on the inclusion of the αE-catenin C-terminus . α-catenin’s importance was again highlighted in vivo when a similar study was performed using Drosophila. In this case the DE-cadherin-α-catenin fusion protein was found to rescue adhesion defects independently of β-catenin .
Despite the clear evidence that α-catenin is essential in this link, it is also clear that a more complex arrangement of proteins localize to the AJ plaque region. This was proposed by the Weis and Nelson groups after in vitro biochemical experiments showed that α-catenin-β-catenin heterodimers have a low affinity for actin . α-catenin binds more readily to actin as a homodimer, yet due to the β-catenin binding site overlapping with heterodimer interface, α-catenin can only bind to the AJ as a monomer . Recently, this was attributed to β-catenin disrupting interactions that exist between four α-helix bundles of α-catenin and confer its asymmetric homodimer arrangement. These findings were determined through crystallization of a near-complete human α-catenin dimer . Furthermore, in an earlier study a novel assay using isolated cadherin-containing membrane patches to which β-catenin was added also showed that although α-catenin could bind the cadherin-β-catenin complex, subsequent addition of G- or F-actin failed to bind to this new complex of cadherin-β-catenin-α-catenin . Similarly, although the cadherin-β-catenin-α-catenin complex has successfully been isolated from cells, this has not yet been achieved with actin also bound. It is important to note, however, that both of these are basically negative results.
Additional biochemical and in vitro studies by the Weis and Nelson groups further indicate that a direct link between the cytoskeleton and α-catenin is unlikely to occur in vivo, and instead, is likely mediated by additional adaptor proteins. For example, observations from FRAP experiments revealed contrasting dynamics of actin and α-catenin at cell-cell contact sites. Here, α-catenin was seen to exhibit a significantly slower depletion rate when compared to actin, which underwent rapid exchange with a cytoplasmic pool of actin . The dynamics of actin and α-catenin were also explored using Drosophila epithelial cells and two separate populations of actin were identified, each with different rates of turnover. Whilst the majority of actin was highly dynamic, and was proposed to serve in preventing the lateral movement of homophilic E-cadherin clusters within the membrane, a smaller pool of actin was highly stable and this was correlated to adhesion stability. Interestingly α-catenin was found to be essential for tethering the cadherin clusters to the dynamic actin filament network but was not required for complex stability . It was concluded in this study that unidentified adaptor proteins likely ensure stability in mobile AJs.
The difficulties in establishing a direct link between the α-catenin–β-catenin complex and the actin cytoskeleton underlies a higher degree of complexity in the intracellular structure of AJs. Much of this complexity stems from the multiple interactions that each component can engage. For example, p120-catenin controls cadherin’s endocytosis however it has also been shown to recruit additional adaptor proteins including afadin, and in turn ZO-1, as well as cortactin . Depending on the cell type and its localization, (in the nucleus or in the cytoplasm) p120-catenin has also been correlated with Rho A recruitment, AJ regulation and subsequently cancer cell invasiveness (reviewed in ).
To date numerous adaptor proteins have been identified within AJ sites. These proteins may be recruited to the plaque region where they function in the structural integrity of the AJ or in the regulation of complex dynamics. Vinculin, which is also found in integrin based cell-matrix adhesions, is found in AJs in association with α-catenin where it has been proposed to bind to a novel binding site that is thought to be exposed only when α-catenin has been stretched . Like vinculin, afadin, which is recruited to the AJ by p120-catenin , α-actinin and EPLIN can all bind α-catenin and the actin filament network . Fitting with the complex and dynamic nature of interactions within the AJ, it has been reported that vinculin, as well as afadin do not bind to the C-terminal domain of α-catenin  which was highlighted as being essential in the rescue of cadherin function in fibroblasts .
Recruitment of EPLIN, which is not expressed in all cell types but is found exclusively in AJ, has also been shown to be tension dependent . In this case EPLIN binds to the sides of F-actin where it stabilizes and/or crosslinks bundles of actin filaments to prevent Arp2/3 binding and subsequently, filament branching . Whilst serving this role, EPLIN may also interact with α-catenin in complex with cadherin and stabilize actin filament bundles that span from one AJ to the next (i.e. the adhesion belt) . In vitro experiments have shown that whilst monomers of α-catenin could not bind actin filaments, EPLIN could. Moreover, depletion of EPLIN from epithelial cells resulted in disruption of the adhesion belt, with the cell-cell contacts taking on a more immature appearance i.e. radial actin filaments terminating at puncta of E-cadherin present at contact sites . Collectively these studies  suggest EPLIN may promote AJ maturation by linking the catenin-cadherin complex to the actin cytoskeleton and promoting adhesion belt formation and stabilization. Again, in certain situations the other proteins can compensate for this adaptor protein, as shown by one study that reported normally formed actin bundles and zonula adhesions despite the introduction of an EPLIN binding site deficient mutant of α-catenin in epithelial cells that otherwise lacked functional α-catenin 
Importantly, adaptor proteins localized at AJs may also facilitate the recruitment of other proteins that function in associated mechanisms such as cytoskeletal organization. It is known for example that AJ adaptor proteins recruit the Arp2/3 complex, formins and other proteins involved in the regulation of the actin cytoskeleton . Overall, the complex and highly dynamic arrangement of the AJ may enable it to endure the transient yet strong forces produced during processes such as apical constriction, where the actin cytoskeleton pulls upon the adhesion complex.
It is unlikely that any single protein is responsible for linking the catenin-cadherin complex with the actin cytoskeleton. Instead, eplin and vinculin, as well as additional candidate proteins such as ZO-1, α-actinin, afadin  are likely to contribute in a manner that is cell type dependent. The interactions that facilitate this link are also likely to be transient which is consistent with both adaptor protein turnover rate, which is within minutes, and the dynamic nature of the cortical actin cytoskeleton as revealed by FRAP experiments . Whilst this makes detection of any link difficult using current experimental techniques it should be noted that even numerous weak, transient interactions could still promote strong linkage between cadherins and the actin cytoskeleton through sheer numbers (as reviewed in . These interactions may be cell type specific and given their importance in maintaining tissue integrity may also exhibit protein redundancy.