Like in any other cell, many functions of actin in neurons relate to its ability to polymerize and depolymerize in response to cellular signaling. Although not specifically studied in dendrites, numerous ABPs are known that cooperate in controlling the structure and stability of F-actin networks and their ability to shape cellular membranes. A summary can be found in Table 1, whereas in the text below we will focus on the mechanisms that may be particularly relevant in the regulation of dendritic F-actin.

As the rate-limiting step in actin polymerization, nucleation is a crucial point in regulating F-actin dynamics. Several actin nucleators, including the Arp2/3-complex, WASP-homology-2 (WH2) domain proteins and formin-homology (FH) proteins, facilitate this process. The Arp2/3-complex is the only known regulator for actin branching. It requires an existing actin filament, from which it nucleates a new filament branch (Smith et al., 2013). The Arp2/3-complex is activated by membrane-associated interactors, such as neuronal Wiskott–Aldrich Syndrome protein (N-WASP) or WASP-family verprolin-homologous protein (WAVE) (Korobova and Svitkina, 2010). Arp2/3-complex-dependent polymerization of branched actin networks generates widespread pushing forces against the plasma membrane, accounting for its prominent role in the maturation and enlargement of dendritic spines (Bosch et al., 2014; Spence et al., 2016). Another mechanism of activation involves the F-actin binding protein cortactin, which can bind and activate the Arp2/3-complex both directly and indirectly via N-WASP (Kowalski et al., 2005; Korobova and Svitkina, 2008). The Arp2/3-complex and cortactin are enriched in both axonal and dendritic growth cones of young hippocampal neurons (Strasser et al., 2004) and in dendritic spines of mature neurons (Hering and Sheng, 2003). While overexpression of Arp2/3-complex subunits or N-WASP affect both dendrite and axon development, a deficiency of those proteins induces excessive growth and branching exclusively of the axon (Strasser et al., 2004; Pinyol et al., 2007). Dendritic phenotypes seen at the later stages of development are mostly related to attenuated filopodia and spine formation (Spence et al., 2016). The precise molecular mechanisms behind such differential effects have yet to be elucidated (Sainath and Gallo, 2014). Still, it hints at a functional redundancy with other actin nucleators specific to dendrite development. Here, elaboration critically depends on the WH2-domain nucleator Cobl (Ahuja et al., 2007), which acts as a positive regulator of neurite outgrowth and branching in rat primary hippocampal neurons (Hou et al., 2015).

Formins are actin nucleators downstream of Rho-GTPases (Matusek et al., 2008; Kühn and Geyer, 2014). They nucleate unbranched actin filaments and are mainly associated with the outgrowth of filopodia (Hotulainen et al., 2009). Additionally, they play a role in coordinating MT functions, since they have a distinct MT bundling activity (Bartolini et al., 2008). Formins are involved in proper axon development (Matusek et al., 2008), and in the formation of a deep actin network within the axon, where actin filaments are nucleated from the surface of stationary endosomes in so-called “F-actin hotspots” (Ganguly et al., 2015). Whether the same mechanism is behind the formation of F-actin patches and longitudinal F-actin bundles that have been observed in dendrites (D’Este et al., 2015; Sidenstein et al., 2016) is unknown. Interestingly, another WH2-domain nucleator, Spire (Spir-1/2), directly interacts with the formins Fmn-1/2 (Pechlivanis et al., 2009). It was shown that in several non-neuronal cell types, those two proteins are recruited to recycling endosomes and cooperate in the nucleation of F-actin from the vesicle’s surface (Schuh, 2011; Pylypenko et al., 2016). Whether this mechanism is active in neurons has not been investigated so far. However, since the expression patterns of Spire1 and Fmn-2 markedly overlap in the mouse brain (Schumacher et al., 2004), the existence of such a mechanism in neurons seems plausible.

F-actin turnover is greatly accelerated by filament severing proteins, like the closely related ADF and cofilin-1 (Sarmiere and Bamburg, 2004). They increase the number of uncapped ends that may undergo polymerization and regulate the G/F-actin pool (Andrianantoandro and Pollard, 2006). Binding of ADF/cofilin to F-actin additionally induces a conformational change, which can affect binding of other ABPs (Ngo et al., 2016). The activity of ADF/cofilin is tightly regulated via several mechanisms, including phosphorylation (CaMKII, LIMK) and dephosphorylation (calcineurin, slingshot). For a detailed review on ADF/cofilin, see (Kanellos and Frame, 2016). Of note, cofilin-1 activity is instrumental for the dynamic plasticity of dendritic spines (Noguchi et al., 2016), and it is possible that activated cofilin-1 could spread out from a single activated spine to drive re-organization of F-actin in associated dendritic compartments.

There is a vast number of studies addressing the role of cell adhesion molecules (CAMs) and extracellular guidance cues in neuronal cell migration, axon pathfinding, axon-dendrite contact formation and dendritic spine plasticity (Togashi et al., 2009), whereas their role in dendritogenesis has been somewhat overlooked. Those cell surface receptors associate directly with ABPs, thereby translating environmental cues into local changes in actin dynamics (Leshchyns’ka and Sytnyk, 2016).

Neural CAM1 (NCAM1) has been extensively studied for its role in neuronal development (Li et al., 2013; Leshchyns’ka and Sytnyk, 2016). It was shown recently that the isoform NCAM180 is highly enriched at dendritic growth cones in rat primary hippocampal neurons during dendritogenesis (Frese et al., 2017). Knockdown of NCAM1 led to reduced dendrite lengths, most likely due to absence of NCAM1-mediated actin stabilization, since many different ABPs were found to be associated with its intracellular domain (Pollerberg et al., 2013; Frese et al., 2017). These results suggest a novel role of NCAM180 in dendritic arborization. Of note, other NCAM family proteins have also been reported to be involved in dendritic branching and morphology in C. elegans (Dong et al., 2013).

Integrins are another type of surface receptors in direct contact with the actin cytoskeleton. They interact with components of the extracellular matrix (ECM) and affect actin dynamics through associated Abl-family tyrosine kinases, of which Arg is particularly abundant in the nervous system and localizes to dendritic spines (Lin et al., 2013). Arg controls both dendritic spine and dendrite arbor stability through distinct pathways: it promotes binding of cortactin to F-actin to stabilize spines (MacGrath and Koleske, 2012), and attenuates Rho activity to stabilize dendrite arbors (Moresco, 2005; Lin et al., 2013).

An important feature of mature neurons is dendrite compartmentalization, for example the distinction between apical and basal dendrites, or between proximal and distal regions of apical dendrites. Those compartments are characterized by the expression of specific sets of ion channels (Ginger et al., 2013). Little is known about the mechanisms behind this distinction, however, several studies demonstrated that the large secreted matrix glycoprotein Reelin influences positioning of the Golgi apparatus toward the future apical dendrite (Leemhuis and Bock, 2011; Meseke et al., 2013), and that it is required for establishing and maintaining the molecular identity of the distal dendritic compartment of pyramidal neurons (Kupferman et al., 2014). Reelin signals through lipoprotein-receptors, activating both the GSK3β- and PI3K-Rho-GTPase-pathways, which influence the MT and actin cytoskeleton, respectively (González-Billault et al., 2005; Leemhuis and Bock, 2011). Additionally, Reelin-signaling was found to inactivate ADF/cofilin via LIMK, thereby stabilizing F-actin (Chai et al., 2009, 2016).

Apart from direct contact with the ECM and neighboring cells, another important factor for dendrite survival and stabilization is synaptic input (Niell et al., 2004). Increased calcium influx via glutamate receptors and L-type Ca²⁺-channels at excitatory synapses stabilizes dendritic branches (Lohmann et al., 2002). In addition, neurotrophic signaling via brain-derived neurotrophic factor (BDNF) modulates calcium signaling. Activation of the TrkB receptor by BDNF triggers multiple downstream pathways, which promote synaptic potentiation but also dendrite growth and stabilization (Horch and Katz, 2002; Wang et al., 2015). The downstream signaling is mediated by the activation of Rac1-GTPase and MAP-kinases, which influence both the actin and microtubule cytoskeleton, and of PLC-γ and PI3-kinase, which trigger the release of calcium from the endoplasmic reticulum (ER) (reviewed in Huang and Reichardt, 2003).

Although most of the Ca²⁺-dependent effects have been described in spines, Ca²⁺ diffuses from activated spines and thus can activate dendritic targets (Figure 1C). Cytoplasmic Ca²⁺-signaling is largely transduced via the ubiquitous Ca²⁺-sensor calmodulin (CaM), which rapidly activates CaM-kinases and calcineurin (Ca²⁺/CaM-dependent phosphatase). CaMKII, at the center of many signaling cascades, regulates formation, growth, and branching of dendrites locally via Rho-GTPases, which modulate cytoskeleton turnover, and globally via activation of transcription factors (reviewed in Redmond and Ghosh, 2005). Apart from this, CaMKIIβ possesses an F-actin binding ability, enabling the dodecameric holoenzyme to cross-link and stabilize actin networks (Lin and Redmond, 2008; Na et al., 2016). Activated CaMKII is then released from F-actin, which constitutes one of the many ways to link Ca²⁺-signaling to the regulation of the actin cytoskeleton. Several other ABPs are known to be directly influenced by CaM/Ca²⁺, including spectrins and actinin, ADF/cofilin and gelsolin (Oertner and Matus, 2005) and Cobl (Hou et al., 2015). However, there may still be additional, so far unidentified calcium sensors that directly couple Ca²⁺-signaling to actin dynamics.

Numerous ABPs are further indirectly activated downstream of CaM/Ca²⁺ and CaMKII via Rho-GTPases (Boekhoorn and Hoogenraad, 2013). Three well-studied Rho-GTPases that drive cytoskeleton-mediated dendrite morphogenesis are RhoA, Rac1 and Cdc42 (Negishi and Katoh, 2002). It was shown that during potentiation of synaptic spines, Rho-GTPases get activated and can then diffuse along the membrane into the dendrite and neighboring spines (Murakoshi et al., 2011). While this kind of “spillover” has been suggested to play a role in clustering of activated synaptic inputs (discussed in van Bommel and Mikhaylova, 2016), continued signaling within the dendritic shaft might as well be involved in activity-dependent stabilization of the whole dendrite (Figure 1C).

Rho-GTPases activate a myriad of both intertwining and antagonistic pathways that signal to the actin and microtubule cytoskeleton, their effectors including kinases, formins, MAPs, WASP-family proteins and other ABPs. For a detailed review on Rho-GTPases and their role in organizing the actin cytoskeleton, see (Sit and Manser, 2011).