Spatial regulation of endosomes in growing dendrites

Abstract

Many membrane proteins are highly enriched in either dendrites or axons. This non-uniform distribution is a critical feature of neuronal polarity and underlies neuronal function. The molecular mechanisms responsible for polarized distribution of membrane proteins has been studied for some time and many answers have emerged. A less well studied feature of neurons is that organelles are also frequently non-uniformly distributed. For instance, EEA1-positive early endosomes are somatodendritic whereas synaptic vesicles are axonal. In addition, some organelles are present in both axons and dendrites, but not distributed uniformly along the processes. One well known example are lysosomes which are abundant in the soma and proximal dendrite, but sparse in the distal dendrite and the distal axon. The mechanisms that determine the spatial distribution of organelles along dendrites are only starting to be studied. In this review, we will discuss the cell biological mechanisms of how the distribution of diverse sets of endosomes along the proximal-distal axis of dendrites might be regulated. In particular, we will focus on the regulation of bulk homeostatic mechanisms as opposed to local regulation. We posit that immature dendrites regulate organelle motility differently from mature dendrites in order to spatially organize dendrite growth, branching and sculpting.

1. Neuronal polarity and the challenge of maintaining distinct axon and dendrite domains

Neurons are highly specialized cells with a very complex morphology. First, neurons extend long processes and thus require long distance coordination of transport, maintenance and turnover of their proteins and organelles. Second, neurons establish and maintain distinct domains which differ greatly in their composition. At the basic level, this involves differentiating distinct axons and dendrites. The process of growing an axon and several dendrites is often referred to as establishing neuronal polarity (Dotti et al., 1988; Takano et al., 2015; Schelski and Bradke, 2017). Axons and dendrites share some organelles (such as mitochondria, endosomes, lysosomes, and autophagosomes), but also have unique organelles, most notably synaptic vesicles which are only found in axons (Ferguson, 2018; Rajgor et al., 2021; Funahashi et al., 2020). Even within any given neuronal domain, there is spatial regulation of different compartments. In dendrites, the soma and proximal portions often differ from distal portions in the density of the compartments found within them (Yap et al., 2018; Maday and Holzbaur, 2016; Cheng et al., 2018). In axons, distal portions differ from proximal portions as well. In addition, there are local differences of organelles at synapses. Some compartments are mostly found in the soma, such as the Golgi apparatus, and only sparse Golgi elements are found as outposts in a subset of dendrites (Ye et al., 2007; Lewis and Polleux, 2012; Mikhaylova et al., 2016; Delandre et al., 2016; Bowen et al., 2017). Golgi elements are largely absent in the axon, but TGN markers can be detected in axons (Sytnyk et al., 2002; Lie et al., 2021; González et al., 2018). They likely consist of TGN-derived carriers delivering newly made proteins.

Establishing and maintaining distinct axonal and dendritic membrane protein composition (i.e. neuronal polarity) depends on several mechanisms. First, membrane proteins destined for axons or dendrites are sorted from the somatic Golgi/trans-Golgi network to either axons or dendrites (Guardia et al., 2018). The machinery underlying this biosynthetic polarized membrane traffic was recently reviewed (Radler et al., 2020). Cytoplasmic tail signals are recognized by adaptor complexes that sort cargos into distinct vesicles which carry cargos to axons or dendrites (Winckler, 2004; Radler et al., 2020; Al-Bassam et al., 2012). The molecular details of how the vesicle finds the correct process and avoids the wrong one are still under active investigation. There is good evidence that mechanisms exist to restrict entry into the incorrect domain, i.e. exclusion zones/barriers. For example, the axon initial segment constitutes a diffusion barrier to membrane proteins on the surface (Winckler et al., 1999; Albrecht et al., 2016). This diffusion barrier is actin/ankyrin/spectrin-based. Specifically, regularly spaced circumferential actin rings limit lateral diffusion through the initial segment and prevent intermixing of axonal and somatodendritic membrane proteins (Song et al., 2009; Albrecht et al., 2016; Hedstrom et al., 2008; Huang and Rasband, 2018). In addition, dendritic vesicles are prevented from traveling far into the axon by a myosinVI/actin-based capture mechanism that stalls their motility (Lewis et al., 2011; Watanabe et al., 2012). Dynein motors then eject dendritic vesicles from the axon initial segment (Guo et al., 2016; Klinman et al., 2017). Septin proteins also selectively restrict entry of motors into dendrites (Spiliotis and Kesisova, 2021). Lastly, maintenance of differential membrane protein composition requires functioning endocytosis and endosomal trafficking to remove wrongly inserted membrane proteins from the inappropriate domain (Winckler and Yap, 2011; Bel et al., 2009; Lasiecka and Winckler, 2011; Leterrier et al., 2006). For example, inhibition of endocytosis leads to accumulation of axonal membrane proteins on the somatodendritic surface (Yap et al. 2008, 2012; Wisco et al., 2003; Bel et al., 2009; Ribeiro et al., 2019). Subsequent to endocytosis, receptors can be degraded or, in some cases, the endocytosed receptor can be transported in endosomes to a new domain and exocytosed there. This second process is called transcytosis and can occur in either the axon-to-dendrite (Suo et al., 2014) or the dendrite-to-axon direction (Wisco et al., 2003; Yap et al., 2008; Yamashita et al., 2017; Scott-Solomon and Kuruvilla, 2018). Transcytosis together with degradation contribute to correcting sorting mistakes. In addition, transcytosis appears to be a regulated pathway for certain receptors to carry out additional functions, such as redirecting naïve TrkA receptors to the axon in an NGF-dependent manner (Yamashita et al., 2017) or diversifying endosomal population carrying activated TrkA (Suo et al., 2014; Barford et al., 2018). These mechanisms in concert thus not only ensure proper localization of membrane proteins and are essential for maintaining neuronal polarity (Lasiecka and Winckler, 2011), but also regulate signaling and duration downstream of ligand-receptor activation (Barford et al., 2017; von Zastrow and Sorkin, 2021). The machinery for transcytosis is poorly understood, but the sortilin family member SorCS1 has been implicated in sorting of Neurexin 1α and L1/NgCAM in somatodendritic early endosomes towards recycling endosomes, thus biasing trafficking towards transcytosis over local recycling or degradation (Ribeiro et al., 2019). Caspr2 which is axonally polarized by removal and degradation of somatodendritic receptors rather than by transcytosis (Bel et al., 2009) is not affected by SorCS1 depletion. Layered upon these domain-based processes is local regulation of membrane composition, notably at synapses (Cognet et al., 2006). These are not considered in this review. Rather, we will focus on the mechanisms of endosome and lysosome distribution in the soma and along dendrites. Readers interested in axonal endosome pathways and their regulation are directed to the following recent reviews (Zahavi and Hoogenraad, 2021; Aiken and Holzbaur, 2021; Ferguson, 2018; Maday, 2016; Lie and Nixon, 2019; Roney et al., 2022). Our review also focuses on maturing neurons as they grow and elaborate dendrites and establish synaptic connections, rather than on mature neurons which have completed dendritogenesis and synapse formation, or on aging neurons. Lastly, we will use examples of cargos which we have studied ourselves (NSG1/2 and TrkA) to illustrate concepts. Many other studies exist which illuminate the regulation of some of the same pathways. We apologize for not being able to exhaustively cite all the excellent papers.

2. The endosomal-lysosomal pathway – a maturational system consuming and regenerating organelles

Membrane-bound organelles are abundant in all cell types and necessary for delivery of newly synthesized membrane proteins and secreted proteins to their final destinations (biosynthetic arm) as well as for trafficking of internalized cargos to many cellular places (endosomal arm). On the biosynthetic arm, secreted proteins or transmembrane proteins mature via the ER and Golgi and are packaged into diverse carrier vesicles in the trans-Golgi network (TGN) from where they associate with microtubule motors to travel to their final destinations for fusion. On the endosomal arm, extracellular proteins or plasma membrane-resident membrane proteins can be internalized via several endocytosis pathways, all of which converge in the early endosome. Markers commonly used to identify early endosomes are Rab5 and EEA1. The early endosome is often also called the “sorting endosome” because endocytosed cargos can be sorted towards different destinations from there. There are three major routes that cargos can enter from the early endosome (Naslavsky and Caplan, 2018): recycling pathways (marked by Rab11) back to the plasma membrane, sorting to the late endosome (marked by Rab7) for fusion with lysosomes for degradation, or retrograde transport back to the TGN (marked by TGN38/46 or M6PR) (Rozés-Salvador et al., 2020). We would like to emphasize that no single marker will definitively mark a unitary singular kind of endosome.

One striking feature of the endosomal pathways is that in addition to small vesicular carriers that traffic cargo from an earlier to a later compartment, there is also wholesale maturation of earlier into later compartments. Most well described is the maturation of early endosomes to late endosomes. This was shown to occur by switching Rab isoform from Rab5 to Rab7, a process referred to as “Rab conversion” (Wandinger-Ness and Zerial, 2014; Rink et al., 2005). As Rab conversion occurs, there is also regeneration and fission to regenerate earlier compartments (Skjeldal et al., 2021; Kalaidzidis et al., 2015). Rab conversion to active Rab7 initiates the recruitment of Rab7 effectors to enable new regulatory cascades on the maturing endosome (Guerra and Bucci, 2016). This effector switch leads to profound changes in the behavior of the endosome. Rab7 effectors are required for allowing retromer-mediated escape from the maturing late endosome for certain receptors, progressive acidification, motility along microtubules, and fusion with other late endosomes, with autophagosomes, or with lysosomes. Rab7 can be replaced from lysosomes by Arl8b recruitment, thereby initiating another conversion event on late endosomes (Jongsma et al., 2020). Rab7 is thus a pivotal master regulator of the degradative arm of the endolysosomal system (Guerra and Bucci, 2016). Not surprisingly, Rab7 is linked to human diseases (i.e. Charcot-Marie Tooth disease 2B), as are many of its effectors (Liu and Wu, 2017; Gulsuner et al., 2011; Xing et al., 2021; Cui et al., 2020; Cogli et al., 2009), highlighting the critical roles for endosomal trafficking in neuronal development and function.

3. Endosome maturation in dendrites

Dendritic membrane receptors undergo endocytosis at many sites along dendrites and rapidly enter early endosomes which are found all along the lengths of dendrites. Our own work has studied the small transmembrane receptors NSG1/2 which are found in dendritic endosomes, but not in axons (Yap et al., 2017; Steiner et al., 2002). They play incompletely understood roles in regulating recycling and transport of a variety of important neuronal receptors, including L1-CAM, AMPA receptors, and βAPP (Hoogenraad et al., 2010; Norstrom et al., 2010; Yap et al., 2008; Alberi et al., 2005; Chander et al., 2019; Steiner et al., 2005). We found that NSG1/2 receptors themselves are endocytosed into EEA1-positive early endosomes in a Rab5-dependent manner (Yap et al., 2017). They then rapidly traffic through early and late endosomes and fuse with lysosomes within a few hours of endocytosis. The half-life of NSG1/2 in cultured embryonic hippocampal neurons is on the order of only 2 hours (Yap et al., 2018). We thus use these short-lived dendritic membrane proteins to visualize transport of dendritic degradative cargos from endocytosis all the way to degradation.

What are the spatial characteristics of dendritic endosomes? Early endosomes (EEA1-positive) are present throughout the extent of the dendritic arbor and rapidly fill with endocytosed tracers (in minutes) (Yap et al., 2018). They are not very motile (Fig. 1A) and can be observed to exchange EEA1 in place (Lasiecka et al., 2014). This is likely an indication that Rab5/7 conversion can take place on stationary EEA1-positive early endosomes all along dendrites. Live imaging of the early endosomal Rab5 in dendrites also shows a large immobile population (Boecker et al., 2020; Yap et al., 2017). Rab7-positive late endosomes, in contrast, are much more motile (Yap et al., 2018; van Bommel et al., 2019; Schwenk et al., 2014; Boecker et al., 2020). About half of them will move in a 20-min live imaging experiment, but there is also a substantial population of stationary Rab7-positive late endosomes. Even though Rab7-positive late endosomes move bi-directionally in dendrites, their net displacement is biased over time towards the soma (net retrograde motility), but this bias is small (Schwenk et al., 2014; Boecker et al., 2020). In contrast to early and late endosomes, recycling endosomes (marked by Rab11 or the recycling cargo transferrin Tfn) move rapidly and processively in both directions (anterograde towards the dendrite tip and retrograde towards the soma) (Lasiecka et al., 2010). All of these endosomal subpopulations, early endosomes (EEA1--positive), recycling endosomes (Rab11-positive), and late endosomes (Rab7-positive), are thus found everywhere along dendrites but display very different patterns of motility (Fig. 1). Late endosomes ultimately fuse with lysosomes to create degradatively active endolysosomes (Luzio et al., 2009).

Fig. 1. Different endosomal subtypes show distinct patterns of overall motility in dendrites of DIV8/9 cultured rat hippocampal neurons.

Dual live imaging of NSG1-cherry with the early endosome marker EEA1 (A), the recycling cargo Transferrin (Tfn) (B), and the late endosome marker Rab7 (C) in dendrites. Live imaging was carried out in DIV8/9 rat hippocampal neurons. Soma is oriented towards the right for all panels. Movies are displayed as kymographs in which a line scan along the dendrites for each movie frame is displayed to show changes in position (x-axis) over time (y-axis). Still frames of the first movie frame are shown above the kymographs. Capture rates were 2 fps (frames per second) for A and 1 fps for B and C. Early endosomes (EEs) are largely stationary (A) whereas recycling endosomes move rapidly and processively in both directions (B). Late endosomes (LEs) move intermittently with frequent reversals of direction (C). Some of the vesicle trajectories are traced in the right panels. Not all trajectories can be easily traced for Tfn because some of the motility is so fast that trajectories are often discontinuous.

In thinking about the distinct motility patterns for different endosome types, it is useful to consider the function of these organelles. Being transported with strong directional bias and processivity might not be the purpose of many organelles in dendrites. Some organelles are transport carriers and motility over long distances might be part of their mission. Organelles in this category include recycling endosomes which can recycle locally but also undergo processive long-range motion, but do not display directionality. This would result in distributing recycling receptors more equally along dendrites to maintain or increase receptor levels (Sotelo et al., 2014; Choy et al., 2014; Lazo et al., 2013; Jullié et al., 2014). It is likely that fusion sites of recycling endosomes are regulated, and specificity might arise from regulating fusion rather than directional transport. Other compartments might work locally and in place and not move much over time. Organelles in this category might include mitochondria, lysosomes and early endosomes (Faits et al., 2016; Lasiecka et al., 2014). A third category are organelles that are moving receptors directionally from one place to another. Organelles in this category include late endosomes (moving degradative cargos to somatic lysosomes) (Yap et al., 2018) and BDNF-signaling endosomes (moving signaling receptors from the periphery to the soma) (González et al., 2020).

4. Lysosomes and sites of degradation – controversies abound

Where are lysosomes located in neurons? Surprisingly enough, this is still a debated topic. The disagreements are largely due to different labs using different markers to designate who is a lysosome and using the term “lysosome” either restrictively or expansively. Many papers explicitly indicate that they combine lysosomes and late endosomes into a single category that they refer to as “lysosomes”. Some papers thus use “lysosome” in the title, but do not distinguish between late endosomes and lysosomes. We and others use a more restrictive, operational definition of lysosomes as LAMP1-positive/cathepsin-positive compartments which are also acidified and have high degradative capacity (Yap et al., 2018; Farfel-Becker et al., 2020; Lie et al., 2021). Single markers, especially the popular LAMP1 or LAMP2 proteins, are found in many compartments other than acidified, degradative lysosomes (Yap et al., 2018; Cheng et al., 2018) (Fig. 2). Lie et al. (2021) similarly show that LAMP1-positive compartments move anterogradely into axons and are mostly negative for Lysotracker or a degradative sensor (MagicRed-cathepsin substrate), demonstrating that LAMP1-positivity includes compartments other than degradative lysosomes. It is thus best practice to indicate the marker/criteria used, instead of relying solely on the term “lysosome”. In contrast to early, recycling, and late endosomes, lysosomes (i.e. compartments that are LAMP1-positive and degradatively highly active) are not found evenly throughout dendrites, and the vast majority of the highly degradative compartments are found in the soma and in the proximal 25 μm of the major dendrite (Yap et al., 2018; Lie et al., 2021). This is true not only in cultured neurons, but also in neurons imaged in mouse cortical sections (Yap et al., 2018). There are plenty of LAMP1-positive compartments in dendrites more distally, but most are not highly degradative (Fig. 2C). Rather, LAMP1-positive compartments in dendrites are a mixture of Rab7-negative compartments of unknown function and Rab7-positive compartments which constitute a moderately degradative compartment (degradative activity ~ two orders of magnitude lower than somatic lysosomes) (Yap et al., 2018). Using the short-lived NSG1/2 as a degradative cargo, we determined that bulk degradation occurs in somatic and soma-near lysosomes and much less degradation (estimated to be ~20% of NSG2) occurs in dendritic compartments in neurons grown in culture for 8 days. These dendritic compartments likely correspond to Rab7-positive/LAMP1-positive moderately degradative compartments. Bulk degradation of dendritic membrane proteins thus proceeds via endocytosis in dendrites, convergence into largely stationary EEA1-positive early endosomes, conversion into Rab7-positive late endosomes all along dendrites, and inefficient net retrograde transport toward the soma where they fuse with somatic/soma-near lysosomes for degradation (Fig. 4). Of note: After synapses are formed and functional, LAMP1-positive compartments are able to respond to electrical activity, raising intriguing possibilities for regulated endosome positioning in response to activity in mature dendrites (Goo et al., 2017; Padamsey et al., 2017). If these compartments are highly degradative lysosomes or correspond to other LAMP1-positive organelles is not clear.

Fig. 2. Degradative capacity is not uniform along the length of dendrites.

(A) DIV8/9 cultured hippocampal neurons were stained for the lysosomal enzyme Cathepsin B (blue), active Cathepsin D (bodipy-pepstatin; green), and LAMP1. Compartments with high degradative capacity (degradative lysosomes: containing Cathepsins) are spatially restricted to occupy mostly the soma and soma-near dendritic regions. LAMP1 is found more broadly distributed into more distal regions of dendrites (red). (B,C) Quantification of degradative capacity along dendrites was determined by feeding degradation-sensitive BSA (DQ-BSA) together with degradation-insensitive BSA (Alexa647-BSA) and determining the ratio of DQ-BSA/Alexa647-BSA. 20 μg/ml of Alexa647-BSA with 5 μg/ml of DQBSA were added to cultured rat hippocampal neurons on DIV8 and incubated overnight. Neurons were fixed and stained against either LAMP1 or CatD. Each Alexa647-BSA positive compartment was scored for intensity of Alexa647-BSA, DQ-BSA and their ratio is plotted along the X-axis. The intensity of the respective marker (LAMP1 or CatD) was determined in the same BSA-containing compartments and plotted on the Y-axis. The distance from the soma was determined for each compartment and is indicated by the symbols. Each compartment was thus scored for degradative capacity (DQ-BSA/Alexa 647-BSA ratio), intensity of LAMP1 or CatD, and for its location along the dendrite. Raw data points are shown in (C) whereas binned data (as demarked by green, yellow and red boxes in C) are shown in (B). More distal regions of dendrites have fewer highly degradative compartments (red) but still have compartments with moderate degradative capacity (B). Highly degradative compartments have high levels of LAMP1 and Cathepsin D. Moderately degradative compartments have high levels of LAMP1 but lower levels of Cathepsin D. Compartments with low/no degradative capacity have low levels of Cathepsin D, but can be LAMP1-positive or -negative (C). Highly degradative compartments are overwhelmingly found in the soma (orange circles) or in the proximal dendrite (within 25 μm of the soma along the dendrite; blue circles). Compartments with low degradative compartments can be found at all distances measured along the dendrite (squares, triangles) (C).

Fig. 4. Spatial regulation of endocytic flux in dendrites.

Endocytic cargos are endocytosed from the plasma membrane into early endosomes all along dendrites. Their transport along the endosomal pathway towards lysosomes (i.e. degradative flux) is regulated by the rate of maturation of early to late endosomes, rate of transport of late endosomes towards the soma, and fusion with degradative lysosomes. Many steps along this route are subject to regulation that is both temporally and spatially coordinated. Disruption of degradative flux can lead to many disturbances of signaling downstream of developmental cues and defects in dendrite growth. In older neurons, synapse formation and function as well as overall protein homeostasis are also affected by degradative flux. PM = plasma membrane. EE = early endosome. RE = recycling endosome. Tfn = transferrin. LE = late endosome. Lys = lysosome. Cat = cathepsin. TGN = trans-Golgi network. M6PR = mannose-6-phosphate receptor. LAMP = lysosome-associated membrane protein.

Multiple views also exist with respect to axonal lysosomes in terms of where they are located and where they are generated (Cheng et al., 2018; Farfel-Becker et al., 2019; Lie et al., 2021; Kulkarni and Maday, 2018; Jin et al., 2018; Jin et al., 2018). Recent papers report that there are few lysosomes in distal axons, but they are more frequent in proximal axons (see Roney et al., 2022 for a recent review). LAMP1-positive compartments (often referred to as lysosomes) can easily be found in distal axons, but appear to be less acidified and not highly degradative (Cheng et al., 2018; Lie et al., 2021). Similarly to our findings in dendrites, LAMP1-positive compartments in axons encompass both degradative lysosomes and other compartments which are not highly degradative. Work from some labs have led to the conclusion that acidified, degradative lysosomes (marked operationally by degradative sensors for GCase or active cathepsin D) can enter axons from the soma and move anterogradely towards the axon tip (Farfel-Becker et al., 2019). These compartments are able to fuse with autophagosomes. Others have presented evidence that LAMP1-positive compartments moving anterogradely in the axon are only mildly degradative and contain markers of the TGN (TGN38) but lack indicators for low pH (i.e. lysotracker) or high degradative activity (i.e. the sensor MagicRed-Cathepsin substrate) (Lie et al., 2021). They also often appear tubular. The field thus needs to start differentiating among distinct carriers with lower or higher degradative capacity.

5. The challenge of dendritic directional transport

In axons, directional transport has long been recognized and clear functional roles are associated with directional transport (Zahavi and Hoogenraad, 2021; Koppers and Farías, 2021; Rizalar et al., 2021; Vasudevan and Koushika, 2020). Directional anterograde transport in axons is critical for supplying axon terminals with newly synthesized proteins, including synaptic vesicle components or growth factor receptors, as well as replenishing damaged mitochondria (Cheng and Sheng, 2021; Ganesan and Cai, 2021). Directional retrograde transport in axons is critical for neurotrophic signaling via NGF-TrkA signaling endosomes (Yamashita and Kuruvilla, 2016; Pathak et al., 2020; Barford et al., 2017), and for transport of degradative cargos back to the soma, including maturation of autophagosomes (Maday et al., 2012). Directional motility in axons is conceptually simple, even if the machinery and regulation are not fully understood. This is because close to 100% of microtubules are oriented with their plus ends towards the tip of the axon. All retrograde transport is thus powered by minus-end directed dynein motor whereas anterograde transport relies on plus-end directed kinesin motors (Baas et al., 2016; Guillaud et al., 2020; Guedes-Dias and Holzbaur, 2019; Tang et al., 2013). Intriguingly, even though dynein powers axonal retrograde transport, dynein adaptors on retrogradely moving autophagosomes switch from one adaptor to a different adaptor as they get closer to the soma (Cason et al., 2021). The functional implications of adaptor switching for the same motor remain to be uncovered.

In dendrites, directional motility is not conceptually simple because microtubules are of mixed polarity with about 55% having their plus ends distally, and 45% having their minus end distally (Baas and Lin, 2011; Yau et al., 2016; Ayloo et al., 2017). This mixed polarity array of dendritic microtubules thus does not tell any given motor which direction is the soma, and which direction is the dendrite tip. As an example, Fig. 3 shows Rab7-positive late endosome motility which is strikingly different in axons and dendrites: in axons, Rab7-compartments move processively retrogradely towards the soma whereas in dendrites motility is bi-directional with many pauses and reversals of direction. Which motors regulate motility of different dendritic compartments is thus still a much-investigated question. It is not understood at all well how minus end out populations of microtubules are generated. In C. elegans and Drosophila neurons, more is known, and it was shown that minus end out microtubules are nucleated distally from different organelles, including Golgi outposts, early and/or recycling endosomes (Rolls and Jegla, 2015; Liang et al., 2020; Delandre et al., 2016). In mammalian neurons, new work is exploding to delineate how axonal and dendritic microtubule arrays are built (Cao et al., 2020; Waites et al., 2021; Aiken and Holzbaur, 2021; Monroy et al., 2020; Tas et al., 2017), and the field eagerly awaits more answers.

Fig. 3. Late endosomes (GFP-Rab7) show strikingly different motility in axons and dendrites.

DIV8/9 hippocampal neurons were transfected with GFP-Rab7 to mark late endosomes and live imaged. Movies are displayed as kymographs in which a line scan along the dendrites for each movie frame is displayed to show changes in position (x-axis) over time (y-axis). Still frames of the first movie frame are shown above the kymographs. Capture rates were 1 fps (frames per second). Late endosomes in axons move processively retrogradely whereas late endosomes in dendrites move intermittently with frequent reversals of direction. Some of the vesicle trajectories are highlighted in the bottom panels: Red lines = stationary compartments. Blue lines = motile compartments.

An additional aspect of dendritic microtubule arrays is whether microtubules of like orientation are randomly interspersed or organized in bundles with the same orientation. Using super-resolution imaging to follow single motor movements on fixed cultured neurons, it was found that dendritic microtubules of like orientation are not randomly interspersed but tend to cluster in small domains (Tas et al., 2017). This creates areas of parallel oriented microtubules which is predicted to create more processive movements of vesicles in these areas since a motor that dissociates has a higher likelihood of reattaching to a microtubule with the same orientation and to continue moving in the same direction.

A second important discovery from the same study was that plus end-out and minus end-out microtubules have a different likelihood of carrying certain posttranslational modifications, namely acetylation or tyrosination (Tas et al., 2017). Plus end-out microtubules were more tyrosinated, and minus end-out microtubules were more acetylated. Since different motors prefer acetylated or tyrosinated microtubules, directional bias in transport direction could thus be achieved despite the fact that microtubule orientation is close to 50:50 plus end-out and minus end-out. For instance, the kinesin-1 motor KIF5 prefers acetylated microtubules and thus would tend to move retrogradely towards the soma on minus end-out microtubules in dendrites (Sirajuddin et al., 2014). In contrast, KIF1A prefers tyrosinated microtubules and would tend to move anterogradely.

Kinesin motility in dendrites is complex since there are over 40 distinct kinesin heavy chain genes in mammalian genomes (Hirokawa and Tanaka, 2015). The roles of most of these are not understood in detail. Systematic analysis of the preference of their motor domains for moving into axons or dendrites has uncovered that a handful of motor domains steer into dendrites, including kinesin-3 (KIF1A, 1B, 1C) and kinesin-4 (KIF21A, 21B). Most motor domains were found mostly steering to the axon and not accumulating in dendrites (Lipka et al., 2016). Analysis of the motility of full length kinesins also found dendritic inclusion of KIF1A and KIF1Bβ, but in addition identified KIF13A as a dendritic kinesin (Yang et al., 2019). The two methods thus came to some identical and some different conclusions. Using light-induced linking of motor domains to peroxisomes as an assay, other work found KIF21B to be able to move retrogradely in dendrites (Ghiretti et al., 2016), but its motility shows only very little directional bias. The field has its work cut out to fully understand the contributions of motor domains, accessory regulatory proteins, teams of motors, or vesicle-associated regulators for determining motility in dendrites. The quest to understand how directional vesicle movements in dendrites is achieved thus continues.

The second major microtubule motor class is dynein which walks towards the minus end and is the only retrograde motor in axons (Reck-Peterson et al., 2018; Olenick and Holzbaur, 2019). Evidence has been presented that dynein is both a retrograde and an anterograde motor in dendrites (Kapitein et al., 2010; Ayloo et al., 2017). Detailed analysis of dynein motility using a light-activated recruitment assay shows that dynein moves peroxisomes bidirectionally in dendrites, using plus end-out and minus-end out microtubule populations seemingly indiscriminately. Overall, labeled dynein motors display a net retrograde bias of 55% retrograde to 45% anterograde motility, mirroring the proportion of plus end-out and minus end-out microtubules without apparent preference for either population (Ayloo et al., 2017). Consistent with this finding, we found that inhibition of dynein leads to loss of both anterograde and retrograde movements of late endosomes in dendrites (Yap et al., in preparation). Cargos depending on dynein for net retrograde transport in dendrites might thus not be “smart” and use plus end-out and minus end-out microtubules indiscriminately. Cargos depending on dynein for net anterograde transport, on the other hand, presumably need to display some preference for minus end-out microtubules. The mechanisms by which such preferential association of dynein would be achieved remains to be further elucidated.

6. Regulating mobilization of late endosomes/lysosomes via different motor complexes

In non-neuronal cells, there are multiple documented mechanisms for regulating minus-end and plus-end directed trafficking of lysosomes (Bonifacino and Neefjes, 2017). Motors are recruited to LEs via a variety of adaptors, and this recruitment is regulated. Points of regulation include changes in lipid composition (phosphoinositides, cholesterol), calcium, and ER contact sites, but the field is only at the beginning of delineating all points of regulation for organelle mobilization. Below, we will give a few examples. For dynein, the Rab7-ORPL1-RILP tripartite complex regulates recruitment to LEs (Khobrekar et al., 2020; Jordens et al., 2001). Interestingly, this recruitment is regulated by cholesterol levels in non-neuronal cells. When cholesterol levels are high, ORPL1 promotes assembly of a Rab7-RILP-p150glued complex and stimulates dynein-mediated minus-end transport. Conversely, depletion of cholesterol led to interaction of ORPL1 with the ER-resident receptor VAP-A at ER contact sites. ORPL1-VAP-A binding releases the Rab7-RILP complex from dynein and enables a switch to kinesin-based transport (Wijdeven et al., 2016; Rocha et al., 2009). Contact sites between endosomes/lysosomes and the ER are commonly observed in non-neuronal cells (Wu et al., 2018) and are likely to play critical roles in organelle positioning in neurons as well (Raiborg et al., 2015; Özkan et al., 2021). Importantly, ER contacts with lysosomes could serve as dynamically regulated immobilizing scaffolds (Saric et al., 2021). A second mechanism involves calcium-dependent changes in dynein recruitment. ALG-2 responds to calcium efflux from lysosomes to recruit dynein/dynactin motors. This pathway is independent of Rab7-RILP (Vergarajauregui et al., 2009; Li et al., 2016; Wang et al., 2015). A third pathway, independent of RILP or ALG2, involves the lysosomal membrane protein TMEM55B which recruits dynein via JIP4 adaptor (Willett et al., 2017) and clusters LAMP1-positive compartments in response to cellular stress insults, such as cholesterol accumulation or reactive oxygen species. Multiple pathways that operate at least somewhat in parallel all coordinately position lysosomes in immortalized cell lines

Kinesin recruitment to late endosomes and lysosomes is also subject to regulation. Multiple mechanisms have been described in non-neuronal cells. One mechanism comprises Rab7-FYCO1-mediated recruitment of kinesin-1 via protrudin, an ER-localized protein which mediates transfer of kinesin 1 to its adaptor FYCO-1 (Raiborg et al. 2015, 2016). Repeated LE-ER contacts are required to initiate translocation of LE to the cell periphery for induction of neurite outgrowth in PC12 cells (Raiborg et al., 2016). A recent finding identified CPT1C, a neuron-specific carnitine palmitoyl transferase isoform found in the ER, as a new regulator of anterograde lysosomes transport and axon growth (Palomo-Guerrero et al., 2019). Again, ER contact sites appear to play a critical role. A third kinesin adaptor complex for lysosomes is BORC-Arl8b-SKIP. The multisubunit-BLOC-1 related complex (BORC) associates with the cytosolic side of lysosomes and functions to recruit Arl8b, thus linking lysosomes to kinesin proteins via SKIP (Farías et al., 2017; Jongsma et al., 2020). A recent study revealed that Arl8b is important for interstitial axon branching (Adnan et al., 2020) and that autophagy pathways are involved. Arl8b was shown to be required for anterograde transport of lysosomes into axons (Farfel-Becker et al., 2019) and affects maturation of locally formed autophagosomes. BORC-Arl8b also regulate axon growth cone dynamics and axonal anterograde transport of LAMP1-positive compartments (Farías et al., 2017), but a role for BORC-Arl8b in dendrites is not well established (Farías et al., 2017). Interestingly, LAMP1-positive compartments have been shown to promote neurite outgrowth by fusing and delivering membrane. These compartment thus might not be primarily degradative (Raiborg et al., 2015; Ghosh et al., 2016).

Axon growth has thus been shown to depend on regulated recruitment of motor adaptors to LEs and lysosomes. The interplay and cross-regulation of these different motor adaptors in dendrites, on the other hand, is still poorly understood, and how they regulate dendrite growth remains to be worked out. Additionally, many of the molecular pathways discussed above are activated under cellular stress conditions to change lysosome motility (Ballabio and Bonifacino, 2020; Sekine et al., 2021). This work is largely from non-neuronal cells. It is not clear to what degree these stress-responsive regulatory mechanisms are adapted during development or in mature neurons to regulate organelle mobilization and positioning. A recent paper in fact shows that inhibition of mTOR in neurons does not change motility of Lysotracker-positive (i.e. acidified) compartments (Sun et al., 2022). In addition, autophagy induction in neurons is regulated quite differently from other cell types with few new autophagosomes forming along dendrites (Maday and Holzbaur, 2016; Kulkarni et al., 2020).

7. Developmental regulation of dendritic transport – when to stop and when to go?

Given the non-processive transport of many compartments in dendrites (see Fig. 1), one important question is how positioning of any given compartment is regulated. In more mature neurons, LAMP1-positive compartments and autophagosomes/autophagolysosomes have been observed to change their motility near synapses (Goo et al., 2017; van Bommel et al., 2019; Padamsey et al., 2017; Kulkarni et al., 2021). It is an intriguing finding that compartments near synapses could change their motility in an activity-dependent manner. But what about endosome positioning in maturing dendrites prior to synaptogenesis? Dendritogenesis consists of growth, branching, pruning, and initiation of synapse formation. The purpose of organelles might thus change with development: Young dendrites require supply of new membrane to growing dendrites and might favor anterograde directional transport. They need to recycle adhesion receptors and guidance receptors to balance growth and branching (adding membrane at specific points as well as distally) vs pruning (favoring endocytosis/membrane removal over fusion of more compartments distally) (Aguirre-Chen et al., 2011; Zong et al., 2018; Zhang et al., 2014). They need to regulate degradation of trophic receptors to fine-tune signaling strength and outcomes (Zhou et al., 2012; Moya-Alvarado et al., 2018). Once synapses are formed, local synapse-near regulation of organelle positioning might emerge newly in mature dendrites. Maturation of dendrites thus also requires changing the spatial regulation of organelles. Regulation of mobilization thus becomes a maturational process. Even though there is not an extensive literature on changes of organelle distribution or dynamics as neurons grow axons and mature their dendrites, multiple Rabs are critical for neuronal morphogenesis (Villarroel-Campos et al., 2016), making regulation of membrane transport a likely mechanism. More work is clearly needed to explore these ideas.

What molecular players could be changing to adjust organelle mobilization in coordination with dendrite maturation? Motor activity itself is regulated since both kinesins and dyneins exist in inactive conformations and require activation by binding to other proteins (Reck--Peterson et al., 2018; Ichinose et al., 2015; Hirokawa et al., 2010). It is important to remember that dynein subunits can be phosphorylated downstream of signaling (such as neurotrophins) and thus be dynamically recruited to endosomes carrying certain cargos (Mitchell et al., 2012). The mobility of endosomes is thus also controlled by the cargos themselves, and not all endosomes of any subclass will behave the same way. It will be important to visualize the transport of cargos themselves and not just transport of shared machinery, such as Rab7 or LAMP1.

Regulation could also be carried out at the level of the MTs or other cytoskeletal elements. For example, expression of microtubule-associated proteins (MAPs) changes as neurons mature. Immature neurons express DCX, but DCX is downregulated with development (Liu, 2011). MAPs are known to modulate MT dynamics, bundling and stability, but also to differentially affect the access of different motor proteins to the MT surface (Marx et al., 2006). DCX, for instance binds to KIF1A and will thus regulate motility of organelles in developing but not mature neurons (Liu et al., 2012). The MAP DCLK1 binds a subset of dendritic MTs and moves dense core vesicles to dendrites via regulation of kinesin-3 motors (Lipka et al., 2016). MAP6 is another example of a MAP that can regulate late endosome positioning by opposing their motility (Schwenk et al., 2014; Monroy et al., 2020). It is thus likely that any given compartment (late endosomes or lysosomes) moves differently and is immobilized and positioned differently as MAP composition changes alongside with dendrite maturation.

Non-MT cytoskeletal elements also change with maturation and could thus regulate organelle mobilization. For instance, actin patches near postsynaptic sites can capture and immobilize LAMP1-positive compartments, but these actin patches are not apparent in younger neurons (van Bommel et al., 2019). There is also a striking role for septins in regulating motors (both kinesins and dynein), and maturation could go hand in hand with changes in septins (Spiliotis and Kesisova, 2021; Karasmanis et al., 2018). Lastly, intermediate filaments have been shown to affect organelle immobilization (Styers et al., 2004). Intermediate filaments come in many flavors and are highly regulated in terms of their developmental expression in neurons (Bott and Winckler, 2020). It is intriguing to speculate that they might also contribute to developmental regulation of organelle mobilization, but nothing is yet known about such potential roles of intermediate filaments in either maturing or mature dendrites.

8. Summary

1. Different endosomal subtypes show different motility patterns in dendrites, depending on their functional roles. Some are mostly stationary, some are processively and rapidly moving in both directions, some are intermittently moving, undergoing slow net directional transport to the soma (Fig. 1). In addition, they display non-uniform distribution along growing dendrites. Fig. 4 summarizes the spatial regulation of endocytic flux in growing dendrites.

2. We propose the idea that immature and mature dendrites differ in how they regulate organelle positioning. Differentiation and maturation of dendrites thus includes switching on new regulatory components for organelle positioning, such as potential roles for synaptic activity.

3. Local environments (such as extracellular cues) can lead to spatial differences in organelle motility and dynamic changes in positioning as dendrites extend and branch or are pruned back. Regulating organelle motility is thus an underexplored developmental mechanism for dendrite growth, branching, pruning, and differentiation.