To branch or not to branch

Abstract

PSD-95, a synaptic scaffolding protein, plays important roles in the regulation of dendritic spine morphology and glutamate receptor signaling. We have recently shown that PSD-95 also plays an extrasynaptic role during development. PSD-95 shapes dendrite branching patterns in cultured rat hippocampal neurons by altering microtubule dynamics via an association with the microtubule end-binding protein-3 (EB3). We discovered that PSD-95 interacts directly with EB3 and that the result of this interaction decreases EB3 binding to and EB3 comet lifetime on microtubules. This decrease in lifetime also correlates to decreased dendrite branching. Here we present an additional effect of PSD-95 overexpression on microtubules. Neurons that overexpress PSD-95 show increased distance between microtubules in a manner that is not fully dependent on the interaction between PSD-95 and EB3. We discuss these new data in the context of the role of PSD-95 in shaping the dendritic arbor, and we extend our findings to include a discussion of how PSD-95 may guide neurons toward a more mature and synapse-oriented growth stage.

Postsynaptic density 95 protein (PSD-95) plays a critical role in targeting, anchoring and regulating receptors and signaling proteins at synapses (reviewed in ref. 1). It has also been reported that PSD-95 shapes the morphology of dendritic spines² and its presence is necessary for learning and memory.³ Recently, our laboratory reported a nonsynaptic function for PSD-95 during development: negative regulation of dendrite branching.⁴ This finding is of importance since PSD-95 is present in large amounts outside of the synapse.^4–6 Our recently published paper, “PSD-95 Alters Microtubule Dynamics via an Association With EB3” in The Journal of Neuroscience, details a novel interaction between PSD-95 and the microtubule end-binding protein 3 (EB3).⁷ This paper shows for the first time that PSD-95 directly influences microtubule dynamics and elucidates a mechanism by which the interaction of PSD-95 and EB3 shapes the dendritic arbor by changing EB3 comet dynamics and altering microtubule behavior.

Microtubules play crucial roles in the development and maintenance of neuronal polarity and in spine formation and development (reviewed in ref. 8–13). One area of research that has just begun to be highlighted is the differences that occur in microtubule behavior at dendrite branch points and spines. PSD-95, a molecule that is now linked to both dendritogenesis and spinogenesis, may be a key link to our understanding of the variable behavior of microtubules in different developmental processes. Insight into potential mechanisms for the action of PSD-95 and EB3 in each of these processes can be considered in two ways: (1) microtubule dynamics on the short-term scale and (2) microtubule architecture alterations that indicate long-term changes in neuronal behavior. This article will attempt to place the interaction of PSD-95 and EB3 in the context of development of both the dendritic arbor and spines and explore possible explanations that form a theory in which PSD-95 dynamically regulates microtubules based on cellular context.

Dendritic Branching

During dendrite branching, microtubules must enter a filopodium and be stabilized in order to develop into a branch.^8–13 This holds true for both the axon and dendrites, although less is known about this behavior in dendrites. We noted that an excess of PSD-95 competes EB3 off of microtubules in vitro, resulting in decreased EB3 comet lifetime in neuronal cell culture.⁷ This lack of EB3 binding to microtubules could mean that the needed stabilization of nascent branches may not occur. Interestingly, this idea is opposed by the notion that microtubule exploration is also critical for formation of new branches; however, this increased exploration might be balanced by a decrease in other proteins that interact with the microtubules and EB3 to cause more branching. Interaction with the actin cytoskeleton is important for exploring microtubules to have an effect on newly occurring branches. It is possible that PSD-95 excludes some of these proteins. Adenomatous polyposis coli (APC), for example, is important for interaction of microtubules and actin and in encouraging microtubule growth toward the membrane.^14,15 Through this interaction with the cytoskeleton, APC plays a role in shaping mammalian axon growth cones and dendrites of Drosophila neurons. Since there may be common mechanisms that underlie formation of axonal growth cones and axon branches, it is not unreasonable to draw correlations from growth cones and extrapolate them to dendrites.¹⁶ We hypothesize that if the presence of PSD-95 excludes APC from interacting with microtubules or EB proteins, then this could lead to decreased APC and EB3 binding to microtubules so that even if the microtubules were more exploratory, they would be less likely to stabilize a nascent branch point (Fig. 1B). This change in microtubule exploration could also explain the increase in microtubules crossing at the dendrite branch point in neurons that overexpress PSD-95.⁷ While the process of dendritic arborization has been highly studied in Drosophila neurons, the physiological relevance of this model system to mammalian neurons is questionable given that dendritic microtubules in a number of Drosophila neurons are unipolar.¹⁷

Figure 1.

Model of the roles of PSD-95 and associated proteins in the regulation of the microtubule cytoskeleton in dendrites. (A) Model of microtubule dynamics in dendrites of mammalian neurons. The spacing of microtubules in the dendritic shaft during branching is influenced by the interaction of microtubules with EB3, APC and actin. Microtubules enter a nascent branch and are stabilized by the interactions of these molecules, allowing the branch to extend. (B) Proposed model in which PSD-95 controls microtubule spacing and stabilization in nascent branches. PSD-95 causes these changes by disrupting the interaction of microtubules with APC and EB3. PSD-95 alters EB3 comet velocity by binding to APC and EB3 through separate domains and preventing APC from interacting with EB3. This interaction may also be responsible for the increased microtubule spacing seen in Figure 2. (C) PSD-95 regulates microtubule dynamics in a nascent branch by decreasing the interaction of EB3 with the microtubules. Decreased EB3 binding to microtubules leads to microtubules that still enter the branch but are not stabilized.

This strictly dynamic view of microtubules as motivators of choosing whether a dendrite should or should not branch is not the entire story. Changes in microtubule dynamics, most likely, lead to changes in the entire dendritic arbor as the neuron develops. Microtubules also serve key roles in the transport of materials to and from cellular locations and as stabilizers of structure. The difference in microtubule polarity between the axon and the dendrite is one of the key differences in cytoskeletal architecture in the mammalian neuron. A recent report has implicated specific motor proteins in the transport of cargo in dendrites that function because of the unique microtubule arrangement.¹⁸ Our previously unreported data show that PSD-95 increases spacing between microtubules in neurons and that blocking the interaction between PSD-95 and EB3 by deleting the SH3 domain of PSD-95 does not completely rescue this phenotype (Fig. 2). It is possible that other proteins, such as APC, which binds to the PDZ domains of PSD-95,¹⁹ and separately to EB3 itself,²⁰ can play a role in PSD-95-mediated microtubule spacing. In fact, we showed that changes in EB3 comet velocity are regulated by an indirect interaction between PSD-95 and EB3,⁷ and we hypothesize that APC may be the link between PSD-95 and EB3 in the regulation of comet velocity, and possibly, microtubule spacing (Fig. 1C). As a result of this spacing, our prediction is that changes in PSD-95 protein levels will have implications for motor-based transport of cellular cargo. Although we have not directly shown that the polarity of the microtubules has changed when the microtubules cross branch points, based on the spatial pattern of these microtubules, it is reasonable to assume that this is the case. We theorize that altered microtubule spacing and organization due to increased PSD-95 results in crosstalk and transport of cargo between dendrite branches instead of between the cell body and the dendrites. Our future studies will determine which microtubule motors are used for this type of transport and how this transport changes over the developmental time course of a neuron as it matures.

Figure 2.

Overexpression of PSD-95 results in increased microtubule spacing in the dendritic shaft. (A) Representative images of control hippocampal neurons that express GFP. (B) Representative images of hippocampal neurons that overexpress PSD-95-GFP. (C) Representative images of hippocampal neurons that overexpress PSD-95ΔSH3-GFP. (D) Quantification of the spacing between microtubules. Neurons that overexpress PSD-95-GFP have a significantly increased inter-microtubule distance. ***p < 0.001 compared to neurons that express GFP. Neurons that overexpress PSD-95ΔSH3-GFP have a microtubule spacing pattern that is in between that found in neurons that overexpress PSD-95-GFP and in control neurons that express GFP. *p < 0.05 compared to neurons that express GFP. All statistics analyzed by Kruskal-Wallis test followed by Dunn's multiple comparison test. n > 20 neurons for each condition. Scale bar = 0.5 µM.

Spine Regulation

Both microtubule and actin dynamics play important roles during spinogenesis and during synaptic plasticity. It has been accepted that actin plays a major role in spine dynamics; however, until very recently, microtubules were thought to not enter spines. Recent reports from three independent groups^21–23 show that microtubules are present in spines but that the invasion is very dynamic and short-lived. Furthermore, the data suggest that microtubule invasion increases spine density, size and stability.^22,24 A number of reviews that describe these studies favor a scenario in which EB3 interacts with another microtubule-interacting protein, CLIP-140, to regulate the actin cytoskeleton through other actin regulatory proteins, such as cortactin and src kinases.^12,13,24 It would be of great interest to determine whether EB3 binding to the SH3 domain of PSD-95 interferes with the association of EB3 with CLIP-140 and/or cortactin and src kinases since both EB3 and src kinases bind to the SH3 domain of PSD-95.^7,25

An open question in this field is what prevents these microtubules from continuing to polymerize once they enter the spine? Does stabilized actin prevent microtubule advancement similar to the role it plays in the growth cone? Could it be that actin filaments are less stable in the spine and that microtubules are regulated by PSD-95 since PSD-95 interacts with EB3 and limits the amount of time the microtubules spend in the spine? This could be an explanation of why microtubule invasion into the spine is transient. This explanation of microtubule behavior is consistent with the idea that reducing EB content at the growing ends of microtubules causes destabilization (as reviewed in ref. 26 and 27). Alternatively, EB3 and microtubules could be used to transport PSD-95 into a growing spine head. While this is an interesting possibility and would fit with a concept of the microtubule being able to deliver specific cargo to a spine (as suggested in ref. 13), this interpretation does not fit our data suggesting that the required region for the interaction of PSD-95 with EB3 (SH3 domain; reviewed in ref. 7) is not implicated in PSD-95 trafficking to and from the postsynaptic density but rather in the stability of PSD-95 at the PSD.²⁸ It may be that the temporary invasion of microtubules could play a role in stabilizing PSD-95 at the synapse after it has been delivered.

Other molecules that interact with PSD-95 play a role in spine dynamics include SPIN90 and IRSp53.²⁹ Besides these reported proteins, we propose that APC is important for mediating some of the effects of PSD-95 on spine morphology.^30–32

Dendrite and Spines: Yin-Yang

We propose that the acts of creating a dendrite branch and of creating a spine are yin-yang. This would mean that the neuron would choose to form a branch or spine, depending on the state of the cytoskeleton, and in specific, microtubules. Based on our data, we suggest that PSD-95 plays a role in this decision, and that as the neuron matures, increasing levels of PSD-95 push the dendrites to adopt a structure that makes them capable of forming more spines than branches. The resulting increased spacing between the microtubules, altered microtubule polarity, less stable microtubules from reduced EB3 binding, and increased microtubule exploration are all indicative of a neuron that sees itself as being ready to make smaller protrusions that require different cytoskeletal dynamics than in those forming a dendritic tree. Dendrites need microtubules to be growing and stable whereas spines need microtubules to be free and exploratory.