The DOCK protein family in vascular development and disease

Rho GTPases function as molecular switches, cycling between their active guanosine-5′-triphosphate (GTP)-bound and inactive guanosine diphosphate (GDP)-bound states. The best characterised of these are RHOA, ras-related C3 botulinum toxin substrate 1 (RAC1) and cell division cycle 42 (CDC42) [12]. Rho GTPase activity is controlled by the opposing actions of guanine nucleotide exchange factors (GEFs), GTPase-activating proteins (GAPs) and guanine nucleotide dissociation inhibitors (GDIs). GEF regulators activate GTPases by promoting the exchange of GDP for GTP. The Rho GEF family comprises approximately 80 members, divided into typical (Dbl) and atypical (DOCK) categories, and has been most commonly implicated in the development of neurological disorders and cancers [11, 13, 14]. Conversely, the GAP family, consisting of around 60 members, catalyse the intrinsic GTPase activity of small Rho GTPases, leading to their inactivation [15, 16]. In a similar fashion, GDIs bind to and stabilise GDP-bound small GTPases, sequestering them in the cytoplasm and therefore preventing the spontaneous GEF-catalysed guanine nucleotide exchange reaction [15].

GEF-mediated activation of Rho GTPases contributes to dynamic actin cytoskeletal assembly and rearrangement, forming the basis of cell-to-cell adhesion and migration [17, 18]. CDC42 is a small Rho GTPase present in most eukaryotes and is known for its role in the regulation of actin-based morphogenesis and cell polarity [19]. By directing filopodia formation, CDC42 modulates cell adhesion, migration and invasion in many cell types, including neurons, endothelial cells and VSMCs [17, 18]. CDC42-driven filopodia formation at the leading edge of endothelial tip cells has been shown to be required for angiogenesis and vessel patterning through the development of membrane protrusions, micropinocytosis, endothelial cell barriers, adherens junctions and tube formation [17‐21]. In vascular development, filopodia emerge from endothelial tip cells in response to CDC42 activation, to regulate cell shape and migration, vascular branching and sensing of the microenvironment [22]. Filopodia are produced by the polymerisation of unbranched actin filaments, which are then arranged in parallel bundles at the filopodium tip [23]. This CDC42-induced localised actin polymerisation at the cell membrane promotes filament elongation via numerous pathways; CDC42 directly activates formin family proteins to nucleate and advance actin filament extension, whereas activation of Wiskott-Aldrich syndrome proteins (WASPs) up-regulates the actin-related protein 2/3 (ARP2/3) complex to produce branched actin filament networks [24]. Actin elongation is stimulated by the activation of uncapping proteins, for example, vasodilator-stimulated phosphoprotein (VASP), delivered to the growing filopodium tip by molecules such as myosin X [21]. CDC42 may also activate the IRSp53 adapter protein, which links actin to the membrane and promotes clustering of the uncapping protein (Fig. 1) [25, 26]. Filopodia formation is a dynamic process owing to the continuous assembly and disassembly of the actin filaments. Actin disassembly at the filopodium tip is typically cofilin- or gelsolin-mediated, but may also be driven by actin depolymerisation and retraction forces applied to the filopodium actin filaments by non-muscle myosin II. Interestingly, several molecules are common to both neural and vascular development, for example, neuropilin-1 (NRP1) is required for tip cell guidance in the developing central nervous system (CNS) as well as for CDC42-mediated filopodia formation in endothelial tips cells [27, 28]. To facilitate cell migration, filopodial protrusions must be stabilised by adherence of the filopodia to the ECM via focal adhesions (FAs). In both neuronal growth cones and endothelial tip cells, this mechanism physically links the actin cytoskeleton to the ECM via integrins to mediate cell advance.

In a similar fashion, RAC1 mediates lamellipodia formation, which is required for blood vessel development and has been shown to promote migration, adhesion, angiogenic sprouting and the permeability responses of endothelial cells to VEGF both in vivo and in vitro [17, 18, 29, 30]. Like CDC42, RAC1-mediated lamellipodia formation may be driven by the activation of the WASP-related WAVE regulatory complex, which promotes actin filament nucleation [31]. The ARP2/3 complex is stimulated to advance branched actin nucleation, whilst filament length is controlled by elongation factors, such as VASP and formin family members [31]. The serine/threonine-protein kinase (PAK) family contributes towards actin polymerisation in both filopodia and lamellipodia formation, and becomes active only when bound to either RAC1 or CDC42 [32]. PAKs have many downstream effectors, including LIM kinase (LIMK), to promote actin polymerisation and bundling [32]. As with filopodia, lamellipodial actin depolymerisation is propelled by severing proteins, such as cofilin (Fig. 1).

Further support for CDC42 involvement in mammalian vascular development is demonstrated by its conserved function in zebrafish, where Cdc42 is activated by pro-angiogenic Arfgef9b and has been found to regulate filopodia and endothelial cell guidance mechanisms during angiogenesis [33]. In mice, compromised blood vessel formation during development is seen in Cdc42 knockout and endothelial-specific deletion models [18]. Moreover, in the mouse retinal vasculature, CDC42 activation has been associated with the pro-angiogenic factor NRP1, demonstrating inhibition of cellular protrusion upon Nrp1 and Cdc42 knockdown, a finding that has been replicated in zebrafish models [28]. Both CDC42 and RAC1 have also been shown to be essential for vascular development in vitro. Whilst CDC42 is required for protrusion formation and angiogenic sprouting, RAC1 knockout endothelial cells are unable to contribute towards important angiogenic processes, such as tip cell migration, owing to their inability to form lamellipodia and FAs [29]. This is exemplified by Rac1-deficient mice, which are embryonic lethal by embryonic day 9.5 (E9.5), most likely due to defective development of major blood vessels and lack of small vessels observed in the embryo [29]. Therefore, through their respective functions in controlling actin cytoskeleton reorganisation, both CDC42 and RAC1 demonstrate roles in key cellular processes required for vasculogenesis and angiogenesis.

The DOCK proteins are a family of Rho GEFs known to regulate the small Rho GTPases CDC42 and RAC1. The family is subdivided into four groups, depending on their GTPase specificity and functional domains (Fig. 2). The DOCK-A (DOCK1, -2, and -5) and DOCK-B (DOCK3 and -4) subfamilies are generally specific for RAC1 activation and share the highest protein structure similarity of all DOCK subfamilies, including an N-terminal Src homology 3 (SH3) domain, two DOCK homology regions (DHR-1 and DHR-2), an armadillo repeat motif and a C-terminal proline-rich domain [34‐38]. In addition, DOCK-A proteins, but not DOCK-B, contain a helical domain towards their N-terminus. The DOCK-C subgroup (DOCK6, -7 and -8) regulate the activity of both RAC1 and CDC42 [39, 40]. The only recognisable functional domains of this subset are the DHR-1 and -2 domains and armadillo repeat. Finally, the DOCK-D subfamily (DOCK9, -10 and -11) all contain an N-terminal pleckstrin homology (PH) domain, which binds phosphoinositides to facilitate DOCK-D protein translocation to the membrane. The DOCK-D proteins generally display CDC42 specificity, although both DOCK9 and -10 have also been implicated in RAC1 activation [35, 38, 41‐43].

Unlike other Rho GEFs, DOCK proteins are not reliant upon PH and Dbl homology (DH) domains for GEF activity and membrane localisation [36, 39]. Rather, DOCK proteins are structurally related by their conserved functional DHR-1 and DHR-2 domains. DHR-1 mediates the localisation of the protein to the membrane via lipid, namely phosphatidylinositol (3,4,5)-trisphosphate (PIP₃), interaction. PIP₃ is a product of phosphatidylinositol 3-kinase (PI3K) activation, a key pathway in the regulation of proliferation, migration and adhesion, with a well-documented role in angiogenic processes through its phosphorylation of protein kinase B (AKT) [44]. At the plasma membrane, the DHR-2 domain binds the target small Rho GTPase to catalyse GTP-GDP exchange, leading to its activation and thus initiating the cascade of Rho GTPase effector proteins resulting in actin polymerisation and, in turn, lamellipodia and filopodia formation (Fig. 1) [36, 37]. The DHR-2 region comprises three lobes, known as A, B and C [45]. Through extensive contact with lobe B, lobe A stabilises the domain, whereas lobes B and C contain the GTPase binding site and catalytic centre to mediate guanine exchange of the target small Rho GTPase [37, 46]. Lobe B is comprised of two anti-parallel sheets and lobes A and C are near-identical anti-parallel alpha helical bundles, each consisting of four alpha-helices [45]. The SH3 domain, shared by the DOCK-A and -B subfamilies, binds to the PH domain of engulfment and cell motility (ELMO) proteins, required for activation of DOCK GEF activity [47‐50]. The DOCK-A and -B C-terminal proline-rich region has been found to bind phosphatidic acid (PA) as well as SH3-containing adaptor proteins, such as CRK, which contribute towards the functional complex required for DOCK-mediated Rho GTPase activity [48, 51].

To date, DOCK3, -8, -10 and -11 have not been associated with vascular development and remain primarily implicated in the pathogenesis of neurological disease and/or immune-related disorders (Table 1). By contrast, mounting evidence supports an argument for the importance of the other DOCK family members in critical vascular processes, potentially implicating the wider DOCK protein family in the development of vascular disease.

DOCK1, formerly named DOCK180 due to its 180 kDa protein mass, was the first of the DOCK family proteins to be characterised [52]. Mammalian DOCK1 protein expression is distributed throughout all tissues, and enriched DOCK1 expression has been identified in pulmonary artery endothelial cells, human umbilical vein endothelial cells (HUVECs) and human microvascular endothelial cells (HMVECs) [16, 53, 54]. Despite exhibiting relatively low tissue specificity, DOCK1 plays a specific role in the formation of skeletal muscle, as demonstrated in zebrafish, fruit fly and mouse models [55, 56]. This activity is regulated on a cellular level by the formation of several regulatory complexes between DOCK1 and different binding partners (Fig. 3a). Of particular interest, DOCK1, as well as DOCK2, -4 and -5, interacts with ELMO1 or -2, whereby the binding of ELMO1/2 induces a conformational change in DOCK1, releasing an auto-inhibitory mechanism and enabling DOCK1 to act as a RAC1-specific GEF [35, 36, 55]. Once associated with DOCK1, ELMO1 can interact with RHOG, promoting translocation of DOCK1 to the plasma membrane, where the DOCK1/ELMO1/RHOG complex activates RAC1 [9, 57]. Alternatively, DOCK1 translocation from the cytosol to the cell membrane can be mediated by interactions between the DOCK1 DHR-1 domain and membranous PIP₃ [36, 58]. Upon membrane localisation, DOCK1 activates RAC1 to facilitate RAC1-dependent actin reorganisation at the leading edge of the cell, which is required for directional cell movement [36, 59].

Conversely, ELMO2 has been found to recruit and complex with DOCK1 at initial cell-cell contacts in Madin-Darby canine kidney cells [50]. The DOCK1/ELMO2 complex is essential for rapid recruitment and spreading of E-cadherin, actin reorganisation, localised RAC1 activity and strong cell-cell adhesion. In a vascular context, loss-of-function mutations in ELMO2 can cause autosomal recessive intraosseous vascular malformation (VMOS), which results in malformed blood vessels lacking a mature vascular smooth muscle layer [60]. Analysis of VMOS fibroblasts identified a correlation of ELMO2 deficiency with the down-regulation of DOCK1, resulting in impaired RAC1-dependent cell migration [60].

Further involvement of DOCK1 in cell migration may be partially regulated in a VEGF-independent manner, as telomerase-immortalised HMVECs showed increased motility following polypyrimidine tract binding protein 1 (PTPB1) inhibition, through activation of the DOCK1/RAC1 pathway [61]. PTPB1 is known to impede VEGF receptor (VEGFR) signalling in endothelial cells, resulting in decreased cell motility [62]. Upon integrin stimulation at the membrane, FAK, epidermal growth factor receptor (EGFR) and platelet-derived growth factor receptor (PDGFR) induce DOCK1 phosphorylation by Src family kinases, promoting the association of DOCK1 with the P130CAS (BCAR1) and CRK scaffold proteins (Fig. 3a) [54]. Interestingly, FAK, P130CAS and DOCK1 have all been implicated in axonal outgrowth, driven by DCC receptor interaction in netrin-1 (NTN1) signalling, an established angiogenic pathway [5, 63]. Under PTPB1 inhibition, DOCK1-P130CAS-CRK complex formation is promoted, leading to increased RAC1 activity. Alternatively, the DOCK1-P130CAS-CRK complex can be recruited to FAs by further association with FA-essential paxillin, wherein RAC1 activation specifically drives increased migration of human glioma cells [54]. In human microvascular cells, this paxillin-CRK-DOCK1 complex was found to be disassembled by TIMP metallopeptidase inhibitor 2 (TIMP2) in favour of paxillin-CRK-C3G complex formation, switching GTPase activation from RAC1 to RAP1 [64]. As TIMP2 is a suppressor of migration and impedes DOCK1-mediated cell motility, these data imply a potentially global role for DOCK1 in cell migration, further substantiated by siRNA-mediated DOCK1 knockdown studies in HUVECs and intestinal epithelial cells [65]. In this study, DOCK1 depletion reduced cell spreading, cell migration and lamellipodial extensions at the migratory edge of the cell, indicating that reduced RAC1 activation caused by DOCK1 knockdown is the primary cause for these phenotypes. These findings were mimicked by DOCK5 depletion in the same study [65]. The high amino acid conservation within DHR-1 domains across the DOCK family may point towards a common PIP₃-associated mechanism for mediating cell migration in DOCK-expressing cell types. Interestingly, the DOCK1/ELMO1 complex has also been found to associate with the Dbl-RhoGEFs Intersectin-1 and -2 (ITSN1/2), known to contribute to intercellular endothelial adhesions [66]. DOCK1 is the only DOCK protein known so far to interact with typical RhoGEFs and, when complexed, DOCK1/ELMO1-ITSN1/2 directly interacts with G-protein coupled receptor 124 (GPR124) to aid cell adhesion [66]. As GPR124 is associated with brain angiogenesis and blood-brain-barrier integrity, and has been shown to regulate CDC42-mediated filopodia formation in human brain vascular pericytes [67], these observations provide further evidence to implicate both typical and atypical Rho GTPase regulators in vascular development.

A relationship between DOCK1 and the cardiovascular system in vivo was first observed in 2010, when Dock1-deficient mice were observed to have developed large edema at E14.5 [51]. Upon dissection of Dock1-deficient embryos, defects in cardiovascular development were identified, namely sub-membranous ventricular septal defect and double outlet right ventricle [51]. These cardiac abnormalities were noted to resemble the phenotype observed with knockdown of chemokine receptor type 4 (Cxcr4), a gene encoding the receptor for CXCL12 and known to be critical for vascular development and endothelial cell morphology and branching [68‐70]. Co-expression of DOCK1 and CXCR4 in cardiac and vascular endothelial cell lineages suggested a possible role for DOCK1 in endothelial cellular processes, such as cell invasion of the ECM, although it is currently unclear whether DOCK proteins are activated during endothelial ECM cell invasion [51]. Of interest, CDC42 is a key player in the formation of invasive podosomes, through interaction with its downstream effectors WASP and ARP2/3. However, the role of DOCK proteins in this process remains to be delineated (Fig. 1) [71]. Subsequently, investigation of DOCK1 function in vascular endothelial cells using chemotaxis assays identified a role for DOCK1 downstream of CXCR4 in cardiac endothelial cell migration and membrane ruffling in response to CXCL12. As CXCL12-induced activation of CXCR4 leads to association with ELMO1 to promote recruitment and activation of DOCK1 to the plasma membrane [72], these findings point towards a mechanism of action whereby CXCR4 stimulates DOCK1-mediated RAC1 activation to control endothelial cell migration and invasion via actin cytoskeleton reorganisation during cardiovascular development [51].

Consistent with the vascular phenotype in Dock1^−/− mice, morpholino-directed dock1 knockdown in zebrafish induced notable vascular defects, but only at higher oligo concentrations [55]. It is unclear whether this observation was a secondary effect derived from the known Dock1 function in myoblast fusion, leading to trunk defects upon dock1 knockdown [55]. Nonetheless, the shift in zebrafish dock1 expression from a ubiquitous pattern during early somitogenesis, to a more specific vascular expression in the dorsal aorta and posterior cardinal vein later in development, would support a potential role during vascular development [55]. This temporal and spatial expression pattern is mimicked by elmo1, which demonstrates non-specific expression throughout the embryo during early somitogenesis and adopts a vasculature- and CNS-specific pattern by 24 hours post fertilisation [55]. Investigation of elmo1 activity in early zebrafish development confirmed its important role in the developing vasculature, with morpholino knockdown inducing multiple vascular-related phenotypes, including heart edema, accumulation of blood in the yolk sac, deformation of the intersomitic vessels and agenesis of the dorsal longitudinal anastomotic vessel [55]. Of note, this study identified Ntn1 as an upstream activator of the Elmo1/Dock1/Rac1 signalling cascade in zebrafish, confirming the Ntn1/Dock1 interaction in a vascular context as well as in axonal outgrowth, as discussed above [5, 55]. Additional zebrafish studies have demonstrated the effects of Dock1 and Elmo1 overexpression in reducing endothelial cell apoptosis and inducing angiogenesis, further supported by HUVEC studies confirming the same results under DOCK1 and ELMO1 overexpression and identifying activation of the pro-survival PI3K/AKT signalling cascade by DOCK1 and ELMO1 [63]. As DOCK1-mediated GEF activity is contingent on formation of an ELMO1/DOCK1 complex in both zebrafish and mammalian systems, these observations indicate an essential mechanism of ELMO1/DOCK1-mediated RAC1 activation during vascular development.

Although not yet implicated in human vascular disease, these combined data imply a conserved role for DOCK1 in the control of mammalian cardiac development and vascularisation, under the regulation of multiple, vascular cell-specific genes.

In stark contrast to DOCK1, which is expressed throughout all tissue types aside from peripheral blood leukocytes, DOCK2 expression is confined predominantly to haematopoietic cells, but is confirmed to also have low-level expression in HMVECs [53, 59]. DOCK2 shares 63.2% identity with DOCK1, excluding the carboxyl terminal variable regions [45]. Unlike DOCK1, induction of DOCK2 expression in normal rat kidney cells led to a rounding and detachment phenotype in vitro, thought to be caused by RAC1 activation [59]. Despite both proteins driving RAC1 activation, it is thought that the different morphological effects observed with DOCK1 and DOCK2 over-expression may stem from alternative intracellular localisation of these proteins [59]. However, this is yet to be fully elucidated, and different, cell-specific mechanisms of DOCK1 and -2 activation may also contribute to the disparate cellular responses.

Known for its involvement in immune regulation, DOCK2 regulates lymphocyte activation and migration [34, 73]. This has been demonstrated in a Dock2^−/− mouse model, where T and B lymphocyte migration was inhibited, and RAC1 activation and actin polymerisation were virtually abolished [34]. Similar to DOCK1, DOCK2 is regulated by ELMO1 which drives binding of DOCK2 to PIP₃. PIP₃ association translocates DOCK2 to the membrane in a PI3K-dependent manner, leading to RAC1 activation which in turn serves as a positive feedback loop, stabilising PIP₃ at the cell’s leading edge.

Interestingly, DOCK2 has also been implicated in vascular remodelling [74], wherein VSMCs de-differentiate from a contractile to a synthetic phenotype, triggered by damage to the blood vessel walls and followed by VSMC migration and proliferation. This can lead to the narrowing of the blood vessel lumen. Low endogenous DOCK2 expression is maintained in VSMCs, but is drastically increased by the addition of platelet-derived growth factor subunit B (PDGF-BB), a molecule known to stimulate VSMC phenotypic modulation [74]. Stimulation of DOCK2 expression correlates with a decrease in the expression of VSMC marker proteins, such as serum response factor (SRF) and Mycod. Conversely, DOCK2 knockout was shown to increase VSMC marker protein expression, supporting an argument for the requirement of low level DOCK2 expression in VSMCs to maintain contractile protein homeostasis [74]. A further study has implicated a role for DOCK2 in neointima formation, which is the generation of scar tissue in the blood vessel lumen after vein graft [75]. This was identified when unusually high levels of DOCK2 expression were detected in vein grafts of male Sprague-Dawley rats that underwent jugular vein-carotid artery bypass grafting [75]. Adding to existing knowledge of the role of DOCK2 in VSMC differentiation, DOCK2 knockdown was noted to inhibit rat VSMC migration and proliferation, whereas over-expression significantly increased these processes [75]. As VSMC migration and proliferation are key processes in angiogenesis, these findings point towards a role for DOCK2 in the maintenance of healthy mammalian blood vessels. To further validate the involvement of DOCK2 in vessel maintenance, the knockdown of endogenous DOCK2 in grafted veins was shown to reduce neointimal formation and lead to improved hemodynamics in vein grafts [75]. Despite this clear association with vessel maintenance, DOCK2 has not yet been implicated in human vascular disease, but has been reported to cause an inherited immunodeficiency disorder [76].

DOCK4 demonstrates diverse tissue expression, including in HUVECs and HMVECs, and has been observed to regulate the activity of both RAC1 and, in contrast to other DOCK family members, RAP1 [16, 53, 77]. RAC1 activation requires the DHR-2 domain and DOCK4 binding to the PH domain-containing protein ELMO1/2, similar to other DOCK-A and -B subfamily members [78]. However, the DOCK4 SH3 domain has a unique role in negatively regulating RAC1 activity [78]. The DOCK4 functional domains are also important in protein expression localisation, as exemplified in a splice isoform of DOCK4 lacking exon 49, which contains a conserved region [77]. Tissue expression of the alternatively spliced isoform was restricted to the stereocilia of the cochlea and retina, as opposed to the usual widespread expression of full-length DOCK4 in a diverse range of tissues [77].

Analogous to other DOCK family proteins, DOCK4 is known to contribute towards cell migration [8, 78‐80]. This has been demonstrated in fibroblast cells, where DOCK4 activated RAC1 at the cell membrane [78]. In this study, over-expression of wild-type DOCK4 was found to significantly increase cell migration, by comparison to constructs containing either a C-terminal or DHR-2 domain mutation.

Several reports have implicated DOCK4 involvement in vascular-type cells, noted initially when the microRNA mir-21 exhibited downstream effects on DOCK4, DOCK5 and DOCK7 expression in VSMCs [79]. mir-21 is known to play a critical function in the development of cardiac hypertrophy and fibrosis, and influences cell proliferation and migration through targeting the knockdown of downstream effectors [68]. Speculation for the role of mir-21 in vascular development is strengthened by its association with bone morphogenetic protein (BMP) and transforming growth factor-beta (TGF-β) signalling, known to promote a contractible phenotype in VSMCs by targeting the programmed cell death 4 (PDCD4) gene [79]. Indeed, upon mir-21 over-expression in pulmonary artery smooth muscle cells, DOCK proteins 4–7, 9 and 10 were found to be down-regulated, indicating a notable association between the DOCK family of proteins and this cell type [79]. Further investigation using siRNA identified a significant decrease of VSMC migration upon knockdown of DOCK4, -5 and -7, and a reduction of VSMC contraction upon DOCK4 and DOCK5 knockdown, but not DOCK7. This suggests a specific function for DOCK4 and -5 in the regulation of vascular cell contractility, substantiated by the elevated expression of contractile genes upon siRNA targeting of DOCK4 and -5 [79]. As these proteins contain multiple SH3 domains which are not present in DOCK7, it is possible that cytoskeletal remodelling may be mediated through SH3 domain interactions with other signalling molecules.

DOCK4 involvement in vascular development was further consolidated in a more physiologically relevant study, using siRNA-based screens in an organotypic angiogenesis system which recapitulates endothelial cell association, vessel sprouting and tubule establishment [80]. Here, the role of DOCK4 was investigated in 3D culture of endothelial cells seeded onto a confluent layer of fibroblasts. Knockdown with DOCK4-specific siRNA led to a marked decrease in the number of vessel branches that developed, suggesting that DOCK4 promotes vessel sprouting. The remaining developing tubules were thinner and with fewer lateral cell-to-cell contacts, implicating DOCK4 in tubule development and endothelial cell adhesion, a critical process in angiogenesis [80]. To further underline the importance of DOCK4 in cell adhesion, siRNA-mediated Dock4 depletion in normal mouse osteoblasts elicited a marked suppression of adherens junctions [81]. When examined in vivo, homozygous Dock4 knockout in the mouse caused early embryonic lethality, however heterozygous Dock4^+/− mice exhibited a variety of angiogenic-associated defects, including a decrease in the blood vessel lumen size, seen in the brain parenchyma of E13.5 embryos [80]. This decrease in lumen size was not associated with pericyte coverage of the vessel, suggesting that another mechanism may be affected, for example cell contractility or motility [79, 80]. Combined, these results provide evidence indicating a role for DOCK4 in endothelial cell protrusive activity, tubule remodelling, organisation of lateral contacts and lumen formation.

In the context of vascular disease, DOCK4 was recently implicated in a mechanism for atherogenesis, the formation of arterial fatty deposits, and in neovascularisation following an ischaemic event [82, 83]. Specifically, DOCK4 was noted as a critical interacting partner of scavenger receptor class B type 1 (SR-B1), a receptor observed in mice to bind LDL and mediate its delivery into the arteries, instigating the cascade towards the development of atherosclerosis [82]. Co-expression of SR-B1 and DOCK4 in vivo was observed in the aortic endothelium of mouse models, and increased levels of both proteins detected in the aorta prior to atherosclerotic lesion formation [82]. DOCK4 knockdown decreased the internalisation of SR-B1 in endothelial cells, impeding the delivery of circulating LDL into the sub-endothelial space by SR-B1 and therefore reducing atherosclerotic lesion formation [82].

Additional support for the potential role of DOCK5 in vascular development lies in the shared regulatory mechanisms acting on both DOCK1 and DOCK5. Like DOCK1, DOCK5 localises to FAs via a DOCK5-P130CAS-CRK complex, implying a similar role for DOCK5 in cell adhesion. Of interest, Dock5 knockout mice display increased trabecular bone mass due to defective osteoclasts impeding microtubule-dependent podosome assembly [84]. However, DOCK5 interactions with ARF GTPase-activating protein 2 (GIT2) result in reduced GEF activities, in contrast to the positive regulation of DOCK1 by GIT1 and -2 for recruitment to FAs [85]. Although compelling, both DOCK5 and DOCK7 have yet to be thoroughly investigated in a vascular context, hence additional work is required to fully elucidate their respective regulatory mechanisms.

DOCK6, alternatively named ZIR1, has been strongly associated with a role in neuronal cell migration [7, 39]. Nonetheless, Dock6 demonstrates localised expression to the developing heart and growing limb buds in the E10.5- and E11.5-stage mouse embryo and is shown to be highly expressed in both HUVECs and HMVECs [16, 53, 86]. In a neuronal context, DOCK6 activity is known to be regulated by its phosphorylation state; DOCK6 GEF activity is inhibited by AKT-mediated phosphorylation at Ser1194, whereas Ser1194 dephosphorylation by phosphatase PP2A interaction promotes DOCK6-mediated RAC1 and CDC42 activation [7]. However, regulation of DOCK6-mediated GEF activity in a vascular context remains to be explored.

DOCK6 was identified as the second causal gene of Adams-Oliver syndrome (AOS), a rare developmental disorder characterised primarily by the congenital absence of skin on the scalp, known as aplasia cutis congenita (ACC), and terminal transverse limb defects (TTLD), both of which present with a wide range of severity. These primary features may be associated with a wide array of additional clinical features, including cardiovascular, neurological and ocular defects. Importantly, congenital heart defects and cutis marmorata telangiectatica congenita (CMTC), a rare cutaneous vascular anomaly, are each observed in ~ 20 % of AOS cases [87]. DOCK6 mutations underlie an autosomal recessive form of AOS and the vast majority are predicted to lead to a disruption of DOCK6 function, due to either premature protein truncation or degradation of mutant transcripts by nonsense-mediated mRNA decay. Consistent with a loss of GEF activity, patient-derived fibroblasts revealed actin cytoskeletal defects resembling CDC42 and/or RAC1 inactivation [86]. To date, ~ 20 independent DOCK6 mutations have been described, accounting for 7 % of all reported AOS cases [88]. Predicted loss-of-function variants cover all mutation types including missense and nonsense, and are located across the length of the gene [10, 86, 89‐91].

Patients with DOCK6 mutations display a broad range of clinical presentations in addition to ACC and TTLD. Of significance, a large proportion of patients exhibit anomalies associated with the heart and/or vasculature, implicating AOS-related DOCK6 dysregulation in aberrant vascular development [88]. Loss-of-function DOCK6 mutations are also strongly associated with CNS and structural eye abnormalities [10]. Furthermore, developmental delay and seizures have been identified in > 45 % of patients with microcephaly and, even in the absence of microcephaly, demonstrate strong correlation with the presence of impaired intracranial vascularisation [88]. Notably, abnormal pericyte recruitment, perhaps due to impaired pericyte migration, has been identified in post-mortem examination of AOS patients [92]. Ectasia and tortuosity of large veins, irregular medial thickness due to uneven mural cell deposition, and abnormal retinal vascularisation have all been observed [92]. Whilst the molecular causes of AOS in these patients remains unresolved, these characteristics may explain the observed clinical features in DOCK6-related AOS and suggest a role for the causal gene in the recruitment and migration of pericytes during angiogenesis [10, 86, 89, 90]. However, the functional impact of AOS-specific mutations in vascular cell types remains to be examined, thus providing much scope for future investigations.

Interestingly, DOCK6 has been implicated in the TGF-β signalling pathway, which is critical for many cellular processes including vascular development and homeostasis [93]. Importantly, TGF-β and BMP have roles in promoting VSMC differentiation and maintaining the contractile phenotype of VSMCs [94]. TGF-β signalling induces the haematopoietic cell-specific microRNA mir-142-3p, which has been shown to down-regulate DOCK6 expression and simultaneously decrease cell migration in VSMCs, providing preliminary evidence for a role for DOCK6 dysregulation in disrupted vascularisation [93]. Transient DOCK6 knockdown has been reported to exhibit a much more severe phenotype in HeLa cells by comparison to AOS patient-derived cells harbouring biallelic DOCK6 mutations [86, 95]. This phenomenon may be partly explained by overlapping functions between DOCK family members, leading to genetic compensation upon prolonged gene loss and reducing the severity of the phenotype. While one study identified an increase in RHOA activity in DOCK6-negative patient-derived cells, potentially highlighting a compensation response mechanism, alterations in DOCK family gene expression upon stable DOCK6 knockdown remain unexplored [95]. Therefore, further investigation into the functional effects of solely DOCK6 knockdown on VSMCs is required. Nonetheless, the existing findings support an important role for DOCK6-mediated Rho GTPase dysregulation in the development of AOS-related cardiovascular and neurological defects.

Similar to DOCK6, the DOCK-D subfamily member DOCK9 (also known as ZIZIMIN1) has been demonstrated to act as a GEF to both RAC1 and CDC42 [42, 43, 96]. In addition to strong associations with the nervous system, DOCK9 is implicated in vascular development and is expressed in HUVECs [16]. In human epithelial cells, DOCK9 has been documented to interact with SMAD2 and -3 in response to TGF-β1 stimulation, indicating a potential role as a downstream effector in the TGF-β-mediated regulation of cell motility [97]. VEGF promotes DOCK9 association with DOCK4 to influence cellular actin dynamics [80]. Through knockdown studies in 3D endothelial cell culture, DOCK9 has been shown to contribute towards the development of lateral and, to a lesser extent, tip filopodia [80]. More specifically, DOCK9 depletion resulted in the formation of tubules with a more linear morphology than their wild-type counterparts. In an attempt to further delineate the role of DOCK9 in cell morphology, inducible expression of DOCK9 was assessed in HeLa cells [43]. This resulted in reduced cell elongation and stress fibres, while inducing the formation of filopodia and membrane ruffles [43]. Taken together, these data indicate a requirement for DOCK9 in the remodelling of lateral organisation and the actin cytoskeleton in various cell types, including endothelial cells, and in vessel lumen morphogenesis [42, 43, 80].