To nucleate or not, that is the question in neurons

Abstract

Microtubules are the structural center of neurons, stretching in overlapping arrays from the cell body to the far reaches of axons and dendrites. They also act as the tracks for long-range transport mediated by dynein and kinesin motors. Transcription and most translation take place in the cell body, and newly made cargoes must be shipped from this site of synthesis to sites of function in axons and dendrites. This constant demand for transport means that the microtubule array must be present without gaps throughout the cell over the lifetime of the animal. This task is made slightly easier in many animals by the relatively long, stable microtubules present in neurons. However, even stable neuronal microtubules have ends that are dynamic, and individual microtubules typically last on the order of hours, while the neurons around them last a lifetime. “Birth” of new microtubules is therefore required to maintain the neuronal microtubule array. In this review we discuss the nucleation of new microtubules in axons and dendrites, including how and where they are nucleated. In addition, it is becoming clear that neuronal microtubule nucleation is highly regulated, with unexpected machinery impinging on the decision of whether nucleation sites are active or inactive through space and time.

Introduction to microtubule nucleation and neuronal polarity

Microtubules are polymers of α-β tubulin heterodimers that are assembled into linear protofilaments that in turn form a hollow tube, typically with 13 protofilaments. They are intrinsically polarized with the site of rapid subunit addition designated the plus end; this end has the β-tubulin side of the heterodimer exposed. The first steps in forming a microtubule from free heterodimers can occur spontaneously in solution, but in vivo new microtubule nucleation is typically templated by the γ-tubulin ring complex (γ-TuRC) (Kollman et al., 2011; Teixido-Travesa et al., 2012; Tovey et al., 2018). In animal cells the γ-TuRC contains 14 γ-tubulin subunits organized into a cone-like structure by γ-tubulin complex proteins (GCPs) 2–6 (Kollman et al., 2011; Teixido-Travesa et al., 2012; Tovey et al., 2018). Recent structures of frog and human γ-TuRC complexes position additional subunits actin and MOZART1 in the lumen of the cone and MOZART2 at the periphery (Consolati et al., 2020; Liu et al., 2020; Wieczorek et al., 2020a; Wieczorek et al., 2020b; Zimmermann et al., 2020).

On its own the γ-TuRC is not a very efficient stimulator of nucleation (Consolati et al., 2020). Several proteins that enhance nucleation have been identified. For example, XMAP215 promotes microtubule polymerization at the plus end (Brouhard et al., 2008), but also during in vitro reconstitution assays binds to γ-tubulin and works with the γ-TuRC to nucleate microtubules (Thawani et al., 2018). CM1 domain-containing proteins including CDK5RAP2/centrosomin (cnn) also promote γ-TuRC-mediated nucleation (Choi et al., 2010). In addition to their role as stimulators of γ-TuRC-mediated nucleation these proteins can help recruit nucleation machinery to specific cellular sites. These sites include the classic centrosomal MTOC and an ever-expanding array of cytoplasmic organelle surfaces, ranging from the nuclear envelope to mitochondria (Muroyama and Lechler, 2017; Sanchez and Feldman, 2017; Tillery et al., 2018). Positioning of nucleation sites, as well as microtubule minus ends more broadly, is tightly coupled to cell shape and function so varies widely in differentiated cells (Muroyama and Lechler, 2017), even within the same animal (Tillery et al., 2018). Because neurons are the focus here, it is important to map out the basic layout of microtubules in axons and dendrites before considering neuronal sites of nucleation.

Neuronal microtubules are highly polarized and have different arrangements in axons and dendrites. Axonal microtubules are organized with plus ends uniformly pointed away from the cell body, while dendrites contain minus-end-out microtubules (Baas and Lin, 2011). While neuronal microtubule polarity was originally mapped in vertebrate neurons (Baas et al., 1988; Burton and Paige, 1981; Heidemann et al., 1981), subsequent studies in flies (Rolls et al., 2007; Stone et al., 2008) and worms (Goodwin et al., 2012; Harterink et al., 2018) showed that these bilaterian invertebrates also have axons and dendrites with distinct microtubule organization. In all cases axonal microtubules have plus-end-out polarity. However, dendritic microtubule arrangement is a bit more complicated. Mature vertebrate dendrites have mixed polarity with plus-end-out and minus-end-out microtubules (Baas et al., 1988; Stepanova et al., 2003; Yau et al., 2016). Invertebrate dendrites take polarity to an extreme, and mature dendrites are almost entirely minus-end-out (Goodwin et al., 2012; Harterink et al., 2018; Stone et al., 2008). Immature dendrites, however, have mixed microtubule polarity like those of vertebrates (Hill et al., 2012). This drastic difference in microtubule polarity between axons and dendrites means that minus ends and nucleation may be controlled in different ways in these two compartments.

Where does nucleation occur in neurons?

The first part of the answer to this question is: not at the centrosome. The centrosome is the classic MTOC found within most animal cells and functions during mitosis to organize the spindle. In developing mammalian neurons γ -tubulin is found at the centrosome, but as cultured cortical, hypothalamic and hippocampal neurons mature, γ -tubulin is lost from the centrosome (Leask et al., 1997; Stiess et al., 2010). Moreover, laser ablation of the centrosome does not disrupt axon extension in hippocampal neurons (Stiess et al., 2010). In Drosophila neurons, the centriole is free of γ-tubulin even very early in neuronal development and it plays no role in neuronal microtubule organization (Nguyen et al., 2011). The only neuron type in which centrosomal nucleation has been shown to persist at maturity is large ganglion cells in the sea anemone Nematostella vectensis (Stone et al., 2020). These neurons have three equivalent neurites with plus-end-out microtubules (Stone et al., 2020) and so it is possible that the continued persistence of a centrosomal MTOC is related to lack of overall polarization. The other type of exception to the centriole being inactivated as a nucleation site during neuronal development is in ciliated neurons. Rather than branched dendrites, some sensory neurons, including vertebrate hair cells and Drosophila chordotonal neurons, use a modified cilium to receive input. The centriole is repurposed as a basal body in these neurons and often is found at some distance from the cell body in the dendritic projection. The resulting microtubule arrangement is controlled by the basal body and has been nicely described in C. elegans ciliated neurons (Harterink et al., 2018). In these neurons the basal body is quite close to the dendrite tip. The region beyond the basal body is the cilium, which should have plus-end-out very stable microtubules as in other cilia. The region between the cell body and base of the cilium also relies on the basal body for its organization, and has microtubule minus ends arranged towards the basal body (Harterink et al., 2018). For the rest of the review, we will be considering neurons that have non-ciliated dendrites, and no longer have a centriolar MTOC when mature.

The second part of the answer, specifying where nucleation actually does happen, has been much more elusive. In part this is because components of the γ-TuRC are easy to visualize when clumped together at the centrosome, but much harder to pin down when distributed throughout a very long cell. Indeed, an early attempt to localize γ -tubulin in cultured sympathetic neurons resulted in the conclusion that nucleation sites were restricted to the cell body (Baas and Joshi, 1992). It has since become clear, however, that local nucleation is important in axons and dendrites (Luders, 2020; Wilkes and Moore, 2020). For example, changing levels of γ-tubulin in Drosophila sensory neurons alters dendrite branching (Ori-McKenney et al., 2012) and results in mixing of microtubule polarity in axons and dendrites (Nguyen et al., 2014). Reduction of γ-tubulin in cultured hippocampal neurons reduces microtubule number in dendrites (Yau et al., 2014) and axons (Sanchez-Huertas et al., 2016). Despite the increasing awareness that local nucleation occurs in axons and dendrites, the difficulty in defining specific sites of nucleation means that large holes in our knowledge remain. There is now information on axonal nucleation in vertebrate neurons, and dendritic nucleation in invertebrate neurons, but no complete picture of where nucleation occurs in vertebrate or invertebrate neurons.

Two important advances in understanding axonal nucleation have emerged recently: identification of a role for the HAUS/augmin complex and connecting nucleation to presynaptic sites (Figure 1). The HAUS complex binds to the side of a microtubule and recruits the γTuRC, resulting in branched microtubule nucleation (Petry et al., 2013). This type of nucleation is stimulated by RanGTP and TPX2 (Petry et al., 2013). The new microtubules tend to align in parallel with the existing one they branch from; in fact in Xenopus extracts over 40% of new microtubules have a 0 degree angle with their parent microtubule, meaning they are perfectly aligned in parallel (Petry et al., 2013). Thus, although this type of nucleation is termed branched, it is well-suited to the geometric constraints of axons, and could help reinforce polarity. Indeed, when the HAUS complex is depleted from neurons, axonal microtubule polarity becomes more mixed (Cunha-Ferreira et al., 2018; Sanchez-Huertas et al., 2016). Reduction of HAUS subunits also lowers microtubule density in axons (Cunha-Ferreira et al., 2018; Sanchez-Huertas et al., 2016) and dendrites (Cunha-Ferreira et al., 2018). Spots of endogenous γ-tubulin and HAUS7 are scattered through axons and tend to colocalize with excitatory synapses (Qu et al., 2019). A role for these spots is suggested by emergence of new microtubule plus ends at presynaptic sites, and concomitant reduction when γ-tubulin levels are low (Qu et al., 2019). Pre-synaptic localization of nucleation sites may occur broadly across neuron types and species as endogenous γ-tubulin is also seen at large synaptic boutons of Drosophila motor neurons (Nguyen et al., 2014).

Figure 1. Dendritic and axonal microtubule nucleation.

A. An overview of a neuron with key nucleation mechanisms indicated in axons and dendrites. Pink lines represent minus-end-out microtubules and green lines plus-end-out ones. Circles in dendrites show sites of endosomal nucleation and pink barbs indicate branched nucleation in axons.

B. Summary of endosomal nucleation. A diagram of a dendrite branch point is shown containing an early endosome with Wnt signaling machinery to control microtubule nucleation. This diagram is based on nucleation in Drosophila sensory neuron dendrites.

C. A region of an axon is shown containing a microtubule with the HAUS/augmin complex nucleating a new microtubules. This type of nucleation has been shown to occur in mammalian axons. The different pieces of the diagram linked by arrows indicate a time sequence.

D. Developing dendrites in some invertebrate neurons have a hotspot of nucleation near the growth cone. This nucleates microtubules in both directions, but its position near the tip means that most of the dendrite is minus-end-out.

A role for HAUS complex-mediated local nucleation in axons is supported by multiple lines of evidence. Scattered nucleation of new microtubules along existing ones, with hotspots at presynaptic sites, seems likely to contribute to maintenance of axonal microtubule polarity and density. HAUS complex function in dendrites is not as clear. Although changes in dendrite outgrowth and microtubule density were seen in cultured neurons with reduced HAUS subunits (Cunha-Ferreira et al., 2018) the complex has not been localized within dendrites. In Drosophila, where dendrites are uniformly minus-end out, a deletion mutant of one of the augmin (HAUS complex in Drosophila) subunits has no detectable effect on microtubule organization or on dendrite shape, although it can suppress phenotypes due to centrosomin (cnn) loss (Yalgin et al., 2015). Additional studies are needed to determine whether/how the HAUS complex contributes to nucleation in dendrites.

So far, little is known about how γ-tubulin is localized within vertebrate dendrites. Instead, recent advances in understanding sites of dendritic nucleation have come from studies in Drosophila and C. elegans. Dendritic branch points were first proposed to be hotspots of microtubule nucleation based on analysis of emergence of microtubule plus ends (Ori-McKenney et al., 2012). γ-tubulin was later shown to concentrate at branch points in several types of Drosophila sensory neurons (Nguyen et al., 2014). It was proposed that Golgi outposts act as mini-MTOCs at these branch points (Ori-McKenney et al., 2012; Yalgin et al., 2015). While the Golgi may be involved in nucleation in the cell bodies of Drosophila neurons (Mukherjee et al., 2020), removing Golgi from dendrites did not reduce γ-tubulin localization at branch points and so it seemed unlikely Golgi outposts house dendritic nucleation sites (Nguyen et al., 2014). Additional studies have continued to support the idea that dendritic Golgi outposts do not contribute to microtubule organization (Mukherjee et al., 2020; Yang and Wildonger, 2020). Instead, studies in Drosophila sensory neurons and developing C. elegans neurons have converged on endosomes as a site of dendritic microtubule nucleation (Figure 1).

During development of the PVD sensory neuron in C. elegans, minus-end-out polarity of the anterior dendrite is established by positioning a MTOC within the growth cone (Liang et al., 2020). Tagged γ-TuRC components localize to this region and colocalize with recycling endosome marker Rab11 (Liang et al., 2020). Moreover, dominant negative or mutant Rab11 impairs γ-TuRC localization in some neurons (Liang et al., 2020). This tip-localized MTOC only seems to be active during development, as mature PVD neurons do not have a similar hotspot of microtubule initiation (Harterink et al., 2018). The involvement of Rab11 in nucleation echoes an earlier finding that Rab11 endosomes help recruit nucleation sites to mitotic spindle poles in mammalian cells (Hehnly and Doxsey, 2014). In mature Drosophila dendrites, early endosomes marked with Rab5 localize to dendrite branch points and are important for localization of nucleation sites there (Weiner, 2020). Reduction of Rab5, but not other Rabs, reduces γ-tubulin concentration at branch points and new microtubules emerge from Rab5-labeled endosomes (Weiner, 2020). Thus, early endosomes are likely the MTOC at dendrite branch points in these cells. The earlier identification of Golgi markers at branch points may have been due to leakage of overexpressed Golgi markers into endosomes (Weiner, 2020).

While work in C. elegans and Drosophila neurons independently identified endosomes as sources of microtubule nucleation in dendrites, the story in Drosophila includes a complex signaling pathway. This pathway was identified serendipitously as a regulator of nucleation. A previous study showed Wnt signaling proteins control localization of a different microtubule regulator, Apc2, to dendrite branch points (Weiner et al., 2018). Wnt signaling proteins were therefore included as candidates to test for control of γ-tubulin localization to the same site (Weiner et al., 2020). Because the output of this Wnt pathway is microtubule nucleation rather than transcription regulation through β-catenin, it was termed apocryphal Wnt signaling (Weiner et al., 2020). Aside from β -catenin, it involves all the canonical Wnt signaling proteins from receptors Frizzled (fz), and fz2, co-receptors arrow (LRP5/6) and Ror, and the major scaffolds dishevelled (dsh) and Axin (Nye et al., 2020; Weiner et al., 2020). Most of the players, including dsh and Axin, localize to Rab5 endosomes (Weiner, 2020) Although most canonical Wnt signaling proteins are involved, it is not yet known whether a Wnt ligand is required. The key player upstream of γ-tubulin seems to be Axin, as targeting Axin to mitochondria is sufficient to recruit γ-tubulin to this ectopic site (Weiner, 2020). It will be interesting to determine whether this endosomal Wnt signaling pathway is used to control microtubule nucleation in other cell types and organisms. A role of Axin in localizing γ-tubulin to centrosomes of dividing cultured cells and oocytes suggests that it might be (Fumoto et al., 2009; He et al., 2016).

With the identification of a role for the HAUS complex in localizing axonal nucleation sites to the sides of existing microtubules, and for endosomes in housing dendritic nucleation sites, some initial spotlights on neuronal microtubule nucleation shine. However, much more remains to be illuminated. For example, we do not know whether the role of endosomes is specific to invertebrate neurons or whether all axonal microtubules are nucleated from the side of existing microtubules.

Beyond position: regulation of nucleation activity in neurons.

While position of nucleation sites is critical for maintaining microtubule density and polarity throughout the neuron, there are hints that activity of nucleation sites is also regulated. It is tempting to think of microtubule nucleation only as a housekeeping function that acts constantly in the background to maintain microtubule number. However, it now seems likely that the baseline rate of nucleation uses only a subset of the nucleation capacity present in the neuron, meaning many nucleation sites are inactive. The best evidence for this interpretation is the increase in number of growing microtubule plus ends initiating at presynaptic sites within a minute of neuronal activation by bicuculline or BDNF (Qu et al., 2019). As this increase depends on γ-tubulin it almost certainly represents new microtubule nucleation, and the rapidity of the response indicates that the machinery is present and waiting to be activated (Qu et al., 2019). Inactive γ-TuRC conformations are consistent with recent structural studies (Consolati et al., 2020; Zimmermann et al., 2020). Axon injury and stress also increase microtubule plus ends in neurons in flies (Chen et al., 2012; Stone et al., 2010), mammals (Kleele et al., 2014), and in flies this depends on γ-tubulin (Chen et al., 2012). This increase is part of a neuroprotective response that occurs before axon regeneration and dampens regenerative outgrowth if not properly regulated (Chen et al., 2016; Chen et al., 2012). Although neuroprotection-associated nucleation also hints that some nucleation sites are unused at baseline, it requires hours and the transcription factor fos to turn on (Chen et al., 2016; Chen et al., 2012). While it seems more likely that a nucleation regulator is the transcriptional target of fos, it is possible that synthesis of the entire γ-TuRC is upregulated. Another hint that activity of γ-TuRCs is dampened at baseline in neurons, is a mutant version of γ-tubulin (the bmps1 allele) that can increase plus end number in neurons (Nguyen et al., 2014). This mutant has a single amino acid change (Mahoney et al., 2006) and so seems more likely to represent an activated form than to increase levels of the γ-TuRC.

If γ-TuRCs are only partially active at baseline in neurons, what might regulate them? Although numerous modifications including phosphorylation of γ-TuRC subunit have been identified (Tovey and Conduit, 2018), how these relate to activity in cells is not clear. The best candidate for a nucleation regulator is CDK5RAP2/cnn (Choi et al., 2010). Overexpression of a small region of CDK5RAP2, called the γ-TuNA, is sufficient to increase the number of plus ends in axons (Sanchez-Huertas et al., 2016). It is important to note that it can be difficult to separate regulation of activity and localization in cells, and CDK5RAP2 is a good example of this as it also helps localize γ-tubulin to centrosomes (Fong et al., 2008) and the Drosophila ortholog, cnn, is required to position γ-tubulin at dendrite branch points (Weiner, 2020). Like overexpression of the CDK5RAP2 γ-TuNA, extra Nmnat, an NAD biosynthesis enzyme, is sufficient to increase microtubule nucleation in neurons (Chen et al., 2016). In this case, however, it is likely the mechanism is indirect.

One set of proteins has been identified that seems to regulate dendritic nucleation without affecting γ-tubulin localization: kinetochore proteins (Hertzler et al., 2020). Until 2019, kinetochore proteins were thought to function only in mitosis and meiosis. Structural kinetochore proteins attach chromosomes to spindle microtubules, and this connection is monitored by regulatory proteins that ensure correct connection of sister chromatids to opposite spindle poles (Hinshaw and Harrison, 2018; Musacchio and Desai, 2017). The discovery that post-mitotic reduction of structural kinetochore proteins in C. elegans, Drosophila and mammalian neurons altered cell shape (Cheerambathur et al., 2019; Zhao et al., 2019) was therefore extremely surprising and intriguing. The involvement of the microtubule-binding domain of one of the kinetochore-microtubule attachment proteins, Ndc80, suggested that the neuronal function of the complex may also involve microtubules (Cheerambathur et al., 2019). However, no specific defects in microtubule organization were identified in any of the neurons examined (Cheerambathur et al., 2019; Zhao et al., 2019). In a subsequent study in Drosophila neurons, transcripts of kinetochore proteins were found to be upregulated by dendrite injury, and defects in dendrite regeneration were observed with post-mitotic reduction of kinetochore proteins (Hertzler et al., 2020). In this case, however, specific microtubule phenotypes were described. Depletion of kinetochore proteins increased the number of plus ends in dendrites, and suppression of this phenotype by introduction of one mutant allele of γ-tubulin suggested that this increase is due to nucleation (Hertzler et al., 2020). Thus, kinetochore proteins seem to act in neurons to suppress baseline microtubule nucleation. The suite of kinetochore proteins involved in suppression of microtubule dynamics includes structural proteins like Ndc80 and Nuf2 that connect microtubules to chromosomes in mitosis, centromeric proteins like the histone cid, as well as that spindle assembly checkpoint (SAC) proteins like Aurora B and Cdc20/fzy that monitor microtubule – kinetochore attachment (Hertzler et al., 2020). The involvement of SAC proteins suggests nucleation could be tuned in response to some intracellular cue. It is tempting to consider that it could be part of the activity-dependent increase in nucleation reported in mammalian neurons (Qu et al., 2019). However, the activity-dependent regulation was observed pre-synaptically in axons (Qu et al., 2019), and kinetochore proteins only affect microtubule number in dendrites (Hertzler et al., 2020). The very specific modulation of only dendritic, and not axonal, nucleation by kinetochore proteins (Hertzler et al., 2020) is perhaps the strongest evidence that separate regulatory mechanisms are used in the two compartments as many of the other studies focused only on axons or dendrites. While kinetochore proteins seem likely to respond to the intracellular environment, the involvement of transmembrane receptors fz, fz2, arrow and Ror in positioning γ-tubulin (Nye et al., 2020; Weiner, 2020) suggests that dendritic nucleation could also be influenced by extracellular events.

Do minus ends remain stuck to nucleation sites in neurons?

Once γ-TuRCs have templated new microtubules, they cap minus ends and prevent them from growing (Consolati et al., 2020; Wiese and Zheng, 2000). At centrosomal MTOCs, for example, minus ends remain anchored and the majority are capped, likely with γ-TuRCs (O’Toole et al., 2003). However, there are several different types of evidence that many neuronal microtubules do not maintain γ-TuRCs at their minus end. Some of the earliest support for short-lived association with the γ-TuRC came from studies of developing neurons, which transiently use the centrosome as a nucleation site (see centrosome discussion above). Sympathetic neurons in culture use the centrosome as a source of new microtubules during axon extension (Yu et al., 1993). However, the new microtubules do not remain attached to the centrosome (Yu et al., 1993). Within an hour almost all newly nucleated microtubules are freed from the centrosome, and they can then be transported into the axon to populate it (Ahmad and Baas, 1995). From these initial studies two scenarios could not be distinguished: release of minus ends capped with γ-TuRCs, or clipping the new microtubule away from the γ-TuRC which would remain at the centrosome to be reused. Support for release by severing was obtained by injecting developing sympathetic neurons with function blocking antibodies to katanin (Ahmad et al., 1999). Katanin represents one of the three main classes of microtubule severing proteins, with spastin and fidgetin being the others. They are all AAA-ATPase proteins and use ATP hydrolysis to remove tubulin subunits from microtubules (Kuo and Howard, 2020; McNally and Roll-Mecak, 2018; Sharp and Ross, 2012). In the antibody experiments, blocking katanin function led to accumulation of microtubules at the neuronal centrosome, suggesting they are normally freed by severing (Ahmad et al., 1999).

The idea that many or all neuronal microtubules are only transiently capped by γ-TuRCs later gained support from another avenue: the discovery that Patronin/CAMSAP proteins recognize uncapped minus ends as soon as they are generated (Goodwin and Vale, 2010; Hendershott and Vale, 2014; Jiang et al., 2014; Yau et al., 2014). Visualization of these proteins can thus be used to detect minus ends not associated with the γ-TuRC. The family includes CAMSAP1, 2 and 3 in vertebrates and Patronin in invertebrates (Akhmanova and Hoogenraad, 2015). In vertebrate and invertebrate neurons CAMSAP/Patronin-decorated minus ends are found scattered throughout axons and dendrites (Feng et al., 2019; Jiang et al., 2014; Yau et al., 2014; Yogev et al., 2016). Of course, generation of uncapped minus ends could occur by severing anywhere along the microtubule. However, studies in cultured hippocampal neurons suggest a close relationship between nucleation and minus end binding by CAMSAP2. Reduction of γ-tubulin or CAMSAP2 caused a similar decrease in number of microtubule plus ends (Yau et al., 2014). Surprisingly, simultaneous knockdown of both did not further decrease plus end number, pointing to function in the same pathway rather than parallel pathways. In addition, knockdown of γ-tubulin resulted in fewer CAMSAP2 stretches in dendrites (Yau et al., 2014). Thus, a reasonable working model for generation of neuronal minus ends is that they are first created by nucleating a new microtubule, then converted to CAMSAP/Patronin-bound minus ends by a severing protein acting near the nucleation site (Figure 2). In mammalian dendrites, free minus ends extend over short distances to form minus end stretches decorated with CAMSAP2 (Yau et al., 2014). In Drosophila dendrites and zebrafish axons, free minus ends undergo long periods of growth, and in Drosophila this is facilitated by Patronin (Feng et al., 2019). Removal of new microtubules from γ-TuRCs would allow them to be used catalytically. Whether a few subunits remain behind after severing or the γ-TuRC is left completely exposed, it could rapidly be used to generate new microtubules. Reusing the same γ-TuRC to generate many microtubules, each with a growing minus end would mean that not many γ-TuRCs would be needed to maintain the microtubule population and could help explain why it has been so difficult to detect a perhaps very small amount of endogenous γ-tubulin in axons and dendrites.

Figure 2. Generation of growing minus ends after nucleation.

A speculative model of conversion of a capped minus end to a growing one is shown. Nucleation at dendrite branch points is shown as in Figure 1, and the same symbols are used. After nucleation a severing protein could clip off the new microtubule and the free minus end would be recognized by CAMSAP/Patronin family members. Depending on the particular family member the minus end could either be stabilized or could grow.

Outlook

Just over three decades ago the first maps of microtubule polarity in neurons were made (Baas and Lin, 2011). The presence of minus-end-out microtubules in dendrites (Baas et al., 1988) made it immediately clear that the mechanisms that control neuronal minus ends would be both interesting and important. The past few years have seen the identification of key machinery that controls minus ends. The discovery of the CAMSAP/Patronin (Akhmanova and Hoogenraad, 2015; Akhmanova and Steinmetz, 2015; Akhmanova and Steinmetz, 2019) family has been critical to understanding and visualizing neuronal minus ends freed from γ-TuRCs. And the machinery that positions and regulates nucleation is starting to be identified. Parallel microtubule nucleation by the HAUS complex (Cunha-Ferreira et al., 2018; Sanchez-Huertas et al., 2016) is a very logical and satisfying mechanism to reinforce axonal microtubule polarity. Dendritic microtubule nucleation is much more complex than anticipated. The involvement of dozens of kinetochore and Wnt signaling proteins to control activity and position of nucleation sites (Hertzler et al., 2020; Weiner, 2020) hints that regulation of nucleation in neurons is exquisitely controlled and sensitive to intracellular and extracellular inputs.

Highlights.

• Neuronal microtubules are nucleation locally in axons and dendrites.

• The HAUS/augmin complex is a key contributor to axonal nucleation.

• In dendrites nucleation sites are organized by endosomes.

• Neurons likely contain a substantial number of inactive nucleation sites at baseline.