Geon The Targeted Transport of PSD-95 MT


As mentioned in Chapter 10, most plasticity-related proteins (PRPs) or their mRNAs do not need specific microtubule tracks (MTTs) for their delivery to any particular spines. The most important PRP which requires targeted transport is PSD-95, since it can recruit other PRPs for synaptogenesis (Mogha et al., 2012; Boersma et al., 2011). Therefore, a spine will be active if it can get PSD-95 via a specific MTT, and inactive if no such MTT exists.

The targeted transport can be achieved by three post-translational modifications: palmitoylation, acetylation and polyglutamylation. Palmitoylation directs PSD-95 to a specific dendritic branch while acetylation and polyglutamylation may guide motor proteins to targeted synapses.

Regulation of Vesicular Transport by Palmitoylation


Figure A-1. Chemical structure of palmitic acid. [Source: Wikipedia]

Translocation of proteins from the soma to dendrites is typically mediated by vesicles. Insertion of proteins into a vesicle requires palmitoylation which adds 16C palmitic acid to the protein (Conibear and Davis, 2010). The enzymes that catalyze palmitoylation are called palmitoyl acyltransferases (PATs), localized mainly to endoplasmic reticulum and Golgi apparatus (Ohno et al., 2006). The Golgi apparatus located outside of the soma is known as Golgi outpost. A neuron contains many Golgi outposts, each is located at the branching point of a dendrite (Jan and Jan, 2010). This strategic location allows the Golgi outpost to serve as a gate for the associated dendrite. A protein may enter a dendrite via vesicles only if it is palmitoylated at the outpost associated with the dendrite.

Over 20 PATs have been identified, regulating different proteins. AMPAR is regulated by ZDHHC8 and ZDHHC5 (Thomas et al., 2012) while PSD-95 by ZDHHC 8, 2, 3, 7, 15 and 17 (Yoshii et al., 2011). ZDHHC8 is involved in both AMPAR and PSD-95. This may explain why ZDHHC8 deficiency reduces the density of dendritic spines (Mukai et al., 2008).

Regulation of MTTs by Acetylation and Polyglutamylation

Acetylation is a process that adds an acetyl group to a protein. Microtubules can be acetylated on lysine 40 of the α-tubulin, which makes the structure more stable (Sadoul et al., 2011). It has been demonstrated that kinesin-1 (KIF5) walks preferentially along acetylated microtubules (Cai et al., 2009).

Two acetyltransferases able to acetylate microtubules have been discovered: Elp3 and Mec-17. Elp3 controls the migration and differentiation of cortical neurons (Creppe et al., 2009). MEC-17 deficiency leads to impaired migration of cortical neurons (Li et al., 2012).

HDAC6 is a major deacetylase for microtubules. Although HDAC6 is a member of the histone deacetylase (HDAC) family, it targets non-histone proteins such as α-tubulin, Hsp90, and cortactin. The enzyme plays a critical role in stress response (Kwon et al., 2007;Lee et al., 2012). It is also implicated in emotional behavior (Fukada et al., 2012).

While acetylation provides stable tracks for kinesin-1, polyglutamylation can modify tracks for most motors, either reduce or enhance their mobility depending on the adaptor protein. Thus, polyglutamylation is a fine regulator of MTTs.

Polyglutamylation adds one or more glutamyl units to a glutamate residue on a protein. In microtubules, the modified site is also the binding site for many microtubule associated proteins (MAPs) and motor proteins (Janke et al., 2008). Therefore, not only can polyglutamylation affect the binding of MAPs, it can also regulate the mobility of motor proteins.

The effects of polyglutamylation on motor mobility depend on the adaptor protein, not the motor itself. In a study using cultured hippocampal neurons, polyglutamylation induced by neuronal activity interfered with trafficking of glycine receptor (GlyR) by KIF5, without affecting AMPAR transport by the same motor KIF5. Their difference is that GlyR interacts with KIF5 via the adaptor gephyrin whereas AMPAR uses GRIP1 as the adaptor (Maas et al., 2009; Dumoulin et al., 2010). In another study, loss of polyglutamylation resulted in abnormal targeting of KIF1A without affecting KIF3A and KIF5 (Ikegami et al., 2007). In zebrafish, the adaptor CSAP is required for normal brain development and proper left-right asymmetry (Backer et al., 2012).

Huntingtin, the protein mutated in the Huntington’s disease, is also an adaptor protein, but with unique functions. It coordinates vesicular transport by dynein and/or kinesin. When huntingtin is phosphorylated at serine 421, the complex moves in anterograde direction (outward from the soma). Remarkably, in the unphosphorylated state, the complex reverses its moving direction from anterograde to retrograde (Caviston and Holzbaur 2009).

The Targeted Transport


Figure A-2. The targeted transport. Spine 1 is inactive because the track that may deliver PSD-95 to this site is blocked by polyglutamylation. Spine 2 is inactive because the associated Golgi does not have sufficient PATs to catalyze palmitoylation of PSD-95 for the vesicular transport. Spine 3 is active as it can receive adequate supply of PSD-95 via efficient MTTs.

Cargos are typically transported from the soma to destination in a saltatory manner, hopping from one microtubule to another. Some microtubules are stable, connected by MAPs. Others are dynamically changing, which may represent the population undergoing memory consolidation. In dendrites, the polarities of microtubules are mixed, allowing both kinesin and dynein to move in either anterograde or retrograde direction. The unique function of huntingtin can even reverse the moving direction simply by changing its phosphorylation state. There are several advantages for the bidirectional transport. One of them is to get around roadblocks (Jolly and Gelfand, 2011). The actual MTTs are much more complex than those illustrated in Figure 11-2.

A microtubule does not extend into the spine unless the spine is stimulated (Dent et al., 2011). Therefore, PSD-95 will be unloaded at the dendritic shaft. Inevitably, some of them may diffuse to neighboring areas. This explains why a cluster of spines often form a functional unit (Govindarajan et al., 2011; Winnubst and Lohmann, 2012; Fu et al., 2012).


Author: Frank Lee
First Published: November, 2012
Last Updated: May, 2013