Regulation of microtubule dynamics in neurons is critical, as defects in the microtubule-based transport of axonal organelles lead to neurodegenerative disease. The microtubule motor cytoplasmic dynein and its partner complex dynactin drive retrograde transport from the distal axon.
Regulation of microtubule dynamics in neurons is critical, as defects in the microtubule-based transport of axonal organelles lead to neurodegenerative disease. The microtubule motor cytoplasmic dynein and its partner complex dynactin drive retrograde transport from the distal axon. We have recently shown that the p150Glued subunit of dynactin promotes the initiation of dynein-driven cargo motility from the microtubule plus-end. Because plus end-localized microtubule-associated proteins like p150Glued may also modulate the dynamics of microtubules, we hypothesized that p150Glued might promote cargo initiation by stabilizing the microtubule track. Here, we demonstrate in vitro using assembly assays and TIRF microscopy, and in primary neurons using live-cell imaging, that p150Glued is a potent anti-catastrophe factor for microtubules. p150Glued alters microtubule dynamics by binding both to microtubules and to tubulin dimers; both the N-terminal CAP-Gly and basic domains of p150Glued are required in tandem for this activity. p150Glued is alternatively spliced in vivo, with the full-length isoform including these two domains expressed primarily in neurons. Accordingly, we find that RNAi of p150Glued in nonpolarized cells does not alter microtubule dynamics, while depletion of p150Glued in neurons leads to a dramatic increase in microtubule catastrophe. Strikingly, a mutation in p150Glued causal for the lethal neurodegenerative disorder Perry syndrome abrogates this anti-catastrophe activity. Thus, we find that dynactin has multiple functions in neurons, both activating dynein-mediated retrograde axonal transport and enhancing microtubule stability through a novel anti-catastrophe mechanism regulated by tissue-specific isoform expression; disruption of either or both of these functions may contribute to neurodegenerative disease.
Microtubules are polymers of tubulin that undergo successive cycles of growth and shrinkage so that the cell can maintain a stable yet adaptable cytoskeleton. In neurons, the microtubule motor protein dynein and its partner complex dynactin drive retrograde transport along microtubules from the distal axon towards the cell body. In addition to binding to dynein, the p150Glued subunit of dynactin independently binds directly to microtubules. We hypothesized that by binding to microtubules, p150Glued might also alter microtubule dynamics. We demonstrate using biochemistry and microscopy in vitro and in cells that p150Glued stabilizes microtubules by inhibiting the transition from growth to shrinkage. We show that specific domains of p150Glued encoded by neuronally enriched splice-forms are necessary for this activity. Although depletion of p150Glued in nonpolarized cells does not alter microtubule dynamics, depletion of endogenous p150Glued in neurons leads to dramatic microtubule instability. Strikingly, a mutation in p150Glued known to cause the neurodegenerative disorder Perry syndrome abolishes this activity. In summary, we identified a previously unappreciated function of dynactin in direct regulation of the microtubule cytoskeleton. This activity may enhance generic microtubule stability in the cell, but could be especially important in specific areas of the cell where dynactin and dynein are loaded onto microtubules.
Microtubules are dynamic, polarized polymers of tubulin that serve as tracks for long-distance transport in eukaryotic cells. In neurons, transport along microtubules is especially important yet particularly vulnerable to disruption, as these cells are long-lived and postmitotic with elongated axonal processes that can extend up to a meter . There is accumulating evidence that axonal transport is disrupted in multiple neurodegenerative diseases, including amyotrophic lateral sclerosis and Huntington’s disease . In neurodegeneration, defects in microtubule dynamics may precede transport defects .
The rate-limiting step in microtubule formation is nucleation from soluble tubulin, which in the canonical pathway is catalyzed by γ-TuRC enriched at the centrosome . However, in large, postmitotic cells like neurons, noncentrosomal nucleation may be particularly important –. Following nucleation, the dynamics of polymerization and depolymerization are strongly influenced by microtubule-associated proteins (MAPs). In particular, a spatially specialized group of MAPs that localize to the microtubule plus end, the plus end-tracking proteins (+TIPs), are ideally poised to modulate dynamics in cells .
One of these +TIPs is dynactin , a large complex that binds and activates cytoplasmic dynein , and also associates with microtubules through its dimeric p150Glued subunit . p150Glued has two alternatively spliced microtubule-binding domains at its N-terminus: a cytoskeleton associated protein glycine-rich (CAP-Gly) domain , followed by a serine-rich basic domain –. The p150Glued microtubule-binding N-terminus is dispensable for most dynein-mediated organelle transport –. However, we and others have recently shown that it is specifically required for efficient transport initiation from the distal axon in neurons ,.
Because p150Glued is specifically enriched at the microtubule plus end, we hypothesized that it might modify microtubule dynamics. Here, using solution assays and direct visualization of microtubule dynamics using TIRF microscopy, we show that p150Glued promotes microtubule formation by binding both to microtubules and to soluble tubulin. Both the CAP-Gly and basic domains are required for tubulin-binding in vitro. The full-length isoform encoding both these domains in tandem is primarily expressed in neurons, so we hypothesized that this pro-polymerization activity might be a neuron-specific function of p150Glued. Accordingly, we find that in epithelial cells depleted of p150Glued there is no effect on microtubule dynamics, while in primary neurons, we observed a significant increase in catastrophe upon depletion of p150Glued that was specifically rescued by expression of the neuronal isoform. Finally, we find that a mutation in p150Glued causative for Perry syndrome, a lethal Parkinson’s syndrome, inhibits the anti-catastrophe activity. Thus, the novel neuron-specific anti-catastrophe activity described here may facilitate microtubule stabilization in neurons. We speculate that disruption of this function may contribute to neurodegeneration.
To investigate the dynein-independent effects of dynactin on microtubule dynamics, we designed N-terminal recombinant polypeptides truncated before the dynein binding site within the first p150Glued coiled-coil domain (CC1, Figure 1A) . Since CC1 is also required for endogenous dimerization of p150Glued, we replaced this domain with a short, well-characterized GCN4 coiled-coil . In order to assay the effects of dimerization on microtubule dynamics, we also generated a corresponding construct lacking the dimerization domain, but including both the N-terminal CAP-Gly and basic microtubule-binding domains.
Figure 1. p150Glued promotes microtubule formation.
(A) Schematic depicting endogenous full-length p150Glued and dimeric and monomeric constructs. (B) Light scattering traces for recombinant polypeptides or buffer control incubated with tubulin show that the dimeric p150 Nt-GCN4 and neuronal MAPs tau23 and DCX-GFP robustly promote microtubule assembly, while monomeric p150 Nt and rKin430-GFP do not. (C) TIRF elongation assay. As shown in the schematic at top, rhodamine-labeled tubulin is polymerized from biotinylated Alexa488-labeled GMPCPP microtubule seeds. Montage below shows free tubulin assembly (red) from a stabilized seed (green) in the absence of p150. Arrowhead identifies the microtubule plus-end. (D) Kymograph plots showing representative examples with buffer, 200 nM p150 Nt, or p150 Nt-GCN4. (E) Polymerization rates and (F) catastrophe frequencies from seeded assembly shows that p150 Nt-GCN4 promotes polymerization and inhibits catastrophe. All error bars represent SEM of three or more independent experiments. Statistical testing was performed with a two-tailed t test. * p<0.05; ** p<0.01. For (E), p<0.01 for all conditions compared to control.
To characterize our recombinant polypeptides, we performed glutaraldehyde cross-linking and hydrodynamic analysis, and confirmed that p150 Nt-GCN4 dimerized as expected (Figure S1A). Both the monomeric and dimeric constructs were soluble and monodisperse following purification (Figure S1B). Hydrodynamic analysis also indicated that both polypeptides are highly elongated (Rs/Rmin>1.9, Figure S1B), consistent with previous electron microscopy images of dynactin  that indicate that the p150Glued dimer projects outward from the Arp1 filament that forms the base of the dynactin complex. We measured the relative affinities of both the dimeric and monomeric constructs for microtubules, and found that both bound to paclitaxel-stabilized microtubules with moderate affinities (Kd of 340 nM and 890 nM, respectively, Figure S1C), within the typical range for CAP-Gly proteins .
Previous work from our lab suggested that the N-terminus of p150Glued promotes bulk microtubule formation . To confirm and extend this result, we first compared the activities of the dimeric p150 Nt-GCN4 and the monomeric p150 Nt constructs using in vitro assembly assays. Only the dimeric construct induced a large increase in light scattering, a measure of increased microtubule polymerization (Figure 1B).
We also compared the activities of these constructs to that of the well-characterized neuronal MAPs tau (Tau23) and doublecortin (DCX-GFP) (Figures 1B and S1D), both of which have been shown to promote the formation of microtubules, a function that appears defective in the setting of disease ,. The increase in light scattering induced by 1 µM p150 Nt-GCN4 was similar in magnitude to the increase induced by addition of 10 µM DCX-GFP, and approximately 4-fold more than the increase induced by 2 µM Tau23 (Figure 1B). Thus, p150, like these classical neuronal MAPs, enhances the polymerization of microtubules from soluble tubulin. In contrast, the microtubule-binding motor domain of kinesin-1 (rKin430-GFP) did not promote the formation of microtubules (Figure 1B), indicating that this effect is not a nonspecific one characteristic of all microtubule-binding proteins.
Since microtubule bundling, as well as microtubule polymerization, can lead to increases in light scattering signal, we pelleted the reaction mixtures and analyzed the resulting microtubule pellets by SDS-PAGE (Figure S1E). This analysis confirmed that p150 Nt-GCN4 induces a large increase in tubulin polymerization. Additionally, we tested to ensure that increased turbidity from protein aggregation was not responsible for the light scattering signal; when the purified p150Nt-GCN4 is incubated in the absence of tubulin, light scattering remained low and constant (Figure S1F). As a final control, because the fusion of a His-affinity tag has been reported to disrupt the activity of another plus-tip protein, EB1 , we compared our His-tagged p150 constructs to equivalent Strep-tagged constructs, as well as to the activity of an untagged construct purified by ion exchange chromatography (Figure S2A). The untagged construct was identical to the His-tagged construct on size exclusion chromatography (Figure S2B), and the activities we measured were identical (Figure S2C, S2D), indicating that the promotion of microtubule polymerization by dimeric p150 is not affected by the nature of the N-terminal purification tag.
Though previous work concluded that monomeric p150 constructs could influence microtubule polymerization at high concentrations ,, in our assay, monomeric p150 Nt only marginally increased light scattering as compared to dimeric p150 Nt-GCN4 (Figure 1B); nor did we see increased tubulin in polymer in sedimentation assays as compared to buffer controls (Figure S1E). Thus, we find that dimerization of p150Glued is required for robust pro-polymerization activity in vitro.
We hypothesized that p150Glued, which localizes to microtubule plus-ends in the cell ,, might, like other +TIPS , influence the parameters of dynamic instability. In the absence of MAPs, microtubule dynamic instability is characterized by periods of slow polymer growth ending in catastrophe and rapid depolymerization; following rescue, microtubule growth resumes by addition of new tubulin subunits at the plus end. To test the hypothesis that p150Glued alters these parameters, we used total internal reflection fluorescence (TIRF) microscopy to directly observe microtubule polymerization from pre-formed GMPCPP-stabilized microtubule seeds (Figure 1C) ,.
Addition of tubulin alone to pre-formed, stabilized seeds induced characteristic periods of slow growth, with transitions to rapid shrinkage following a catastrophe (Figure 1D). In contrast, addition of a low (100–200 nM), physiological concentration of p150 Nt-GCN4  to the assay led to more rapid microtubule growth (Figure 1D). The polymerization rate increased more than 2-fold at 200 nM p150 Nt-GCN4 (p<0.01, Figure 1E). We also observed that growth was more persistent, with the catastrophe frequency reduced 2-fold at 100 nM and 4-fold at 200 nM (p<0.05 and <0.01, Figure 1F).
In agreement with the results from the light scattering assay described above, here by direct observation we also see that dimerization of p150Glued is necessary to influence microtubules. The monomeric p150 Nt construct had only a modest effect on microtubule polymerization (Figure 1E), and did not affect the catastrophe frequency (p>0.8, Figure 1F), while dimeric p150Glued had a robust effect on microtubule dynamics, acting to both enhance the polymerization rate and suppress the catastrophe frequency in vitro.
Previous studies with a monomeric p150Glued CAP-Gly polypeptide lacking the basic domain suggested that EB1 was required for p150Glued to modify microtubule dynamics ,. However, native p150Glued is dimeric, and our dimeric p150 Nt-GCN4 construct has potent effects on microtubule dynamics in the absence of EB1. However, we wondered if the addition of EB1 to our assays might further modulate the effects we observed.
Using analytical size exclusion chromatography, we confirmed that, as expected ,,, p150 Nt-GCN4 forms a stable complex with recombinant full-length EB1 (Figure S3A–B). However, the ability of p150Glued to promote microtubule formation appears to be independent of this interaction; the addition of EB1 caused a minor decrease on the pro-polymerization activity of p150Glued (Figure S3C, S3E) and did not affect its ability to inhibit microtubule catastrophe (p>0.5, Figure S3D, S3F).
These data confirm that p150Glued has an intrinsic ability to modify microtubule dynamics and, further, that p150Glued inhibits the inherent pro-catastrophe effects of EB1 in vitro.
We have shown that p150 Nt-GCN4 acts as a pro-polymerization and anti-catastrophe factor in vitro, acting on preformed microtubules. However, we suspected that p150 Nt-GCN4 might also promote microtubule nucleation because we observed a decrease in the lag time for microtubule formation that was inversely dependent on the p150 Nt-GCN4 concentration (Figure 2A, 2B).
Figure 2. p150 catalyzes microtubules nucleation.
(A) Light scattering traces for increasing concentrations of p150 Nt-GCN4. At right, the scale was magnified to illustrate the initial increase in light scattering observed with increasing concentrations of p150 Nt-GCN4. Error bars represent SEM of three or more independent experiments and are omitted for clarity at right. (B) p150 Nt-GCN4 decreases nucleation time, defined as initial appearance of signal over background. Fit is to a double exponential decay. Note that error bars are smaller than symbols. (C) Schematic of the TIRF nucleation assay. p150 constructs are immobilized and the chamber is then perfused with rhodamine-labeled tubulin. (D) Montage from TIRF nucleation assay shows that dimeric p150 Nt-GCN4 catalyzes microtubule nucleation in contrast to monomeric p150 Nt. Scale bar, 12.5 µm. (E) Montage of representative microtubules nucleated in the presence of a low concentration of p150 dimer. Yellow arrows identify the site of nucleation. Red arrows identify the growing plus- and minus-ends. Scale bar, 5 µm.
We designed a TIRF assay to directly test this hypothesis. We specifically oriented the p150 N-terminal microtubule-binding domains toward the microscopy chamber by immobilizing 250 nM p150 polypeptides on the coverslip via an antibody to the C-terminal His-tag (Figure 2C). We then washed the chamber to remove unbound p150, and perfused in 3.5 µM soluble tubulin, which is below the critical concentration for spontaneous microtubule nucleation.
Consistent with our hypothesis, in the presence of dimeric p150 Nt-GCN4, we observed the appearance of many short microtubules within 2 min (Figure 2D). We did not observe nucleation when monomeric p150 Nt was immobilized in the chamber (Figure 2D). When we reduced the amount of p150 Nt-GCN4 on the chamber surface so that we could observe the nucleation of individual filaments, we observed growth from both the microtubule minus- and plus-ends (Figure 2E). This suggested to us that p150 was not promoting nucleation via a template-and-capping mode like γ-TuRC , but instead might stabilize an oligomeric tubulin species, similar to the action of doublecortin , or by binding directly to soluble tubulin dimers.
p150Glued is well-characterized as a MAP, but we hypothesized that it might affect microtubule dynamics by also binding to soluble tubulin through its CAP-Gly domain. The CAP-Gly domains of other structurally related proteins, such as CLIP-170 and tubulin binding cofactors B and E, have been shown to bind soluble tubulin . We used analytical size exclusion chromatography to assay complex formation. To preclude microtubule nucleation, we performed all experiments at 4 C.
When both p150 Nt-GCN4 and p150 Nt were incubated with tubulin, we observed a pronounced shift in the elution peak, indicating complex formation (Figure 3A, Figure S4A–B). We varied the molar ratio of p150∶tubulin until we no longer observed uncomplexed species . In this way, we found that p150 Nt-GCN4 formed a stable complex with tubulin in a 1∶1 ratio (Figure S4C,E–F). However, we observed a tail trailing the elution peak of the 1∶1 p150 Nt-GCN4∶tubulin complex (Figure 3A), which could indicate an exchange process between free p150 Nt-GCN4 and tubulin during the elution. In agreement with this possibility, we found that when we incubated excess tubulin with p150 Nt-GCN4 to encourage saturated binding, the elution peak shifted further to the left (Figure 3A, Figure S4H–I), indicating that a larger complex had formed, and suggesting that p150 Nt-GCN4 might promote microtubule nucleation by transiently binding multiple tubulin subunits and encouraging the formation of a stable seed.
Figure 3. p150Glued forms a complex with soluble tubulin.
(A) Size exclusion chromatograms for p150 Nt-GCN4 run alone or pre-incubated with tubulin reveal that p150 forms a stable complex with tubulin. For the red trace, maximum absorbance is scaled to the complex peak. (B) Size exclusion chromatograms for subtilisin-treated tubulin incubated with p150 Nt-GCN4 suggest that the p150Glued binds tubulin C-termini in solution, as complex formation is not observed after cleavage. (C) Size exclusion chromatograms for p150 Nt-GCN4 incubated with tubulin and eluted at increasing ionic strength indicate that electrostatic interactions between tubulin and the p150 basic domain stabilize the p150-tubulin complex because progressively reduced complex formation is observed.
We know that the acidic C-terminal domain of tubulin is important in the binding of p150Glued to microtubules ,. To determine if p150Glued also binds to soluble tubulin dimers via their C-terminal tails, we cleaved these tails from tubulin using the protease subtilisin (Figure S4J–K) , and again assayed for complex formation via size exclusion chromatography. After cleavage, we could not detect a tubulin-p150 complex (Figure 3B), confirming that the p150 CAP-Gly domain interacts with the C-terminus of tubulin dimers in solution.
This orientation might leave the p150 basic domain free to stabilize the complex through electrostatic interactions. In support of this model, we found that when we eluted the tubulin-p150 Nt-GCN4 complex under increasing ionic strength, which should disrupt such interactions, the relative abundance of the complex was decreased at the expense of free tubulin (Figure 3C). This suggests that while either the CAP-Gly or the basic domains are sufficient to bind microtubules –, both domains may be necessary for p150Glued to bind tubulin dimers and promote microtubule assembly.
p150Glued is alternatively spliced in a tissue-specific manner with expression of full-length p150 containing both the CAP-Gly and basic domains restricted to the nervous system (Figure 4A) ,. The two other predominant p150 spliceforms expressed in vivo lack either the CAP-Gly domain or the basic domain. p135 is an isoform that, like p150, is enriched in the nervous system; it arises from an alternative start site and thus lacks the CAP-Gly domain and exon 5 . In nonneuronal tissues, all or part of the basic domain is spliced out ,. Thus we were able to use physiologically relevant splice forms to test our model.
Figure 4. Tissue-specific p150Glued isoforms differentially modify microtubule assembly dynamics.
(A) Schematic depicting p150Glued spliceforms (and corresponding recombinant proteins): the neuronally-enriched p150 and p135 (p135 Nt-GCN4), and the ubiquitous isoform lacking the basic region (exons 5–7, Δ5–7 Nt-GCN4). (B, D) Size exclusion chromatography with (B) p135 Nt-GCN4 or (D) Δ5–7 Nt-GCN4 indicates that they cannot form stable complexes with tubulin. (C) Microtubule affinity of 400 nM dimeric polypeptides for increasing concentrations of Taxol-stabilized microtubules demonstrates that these constructs bind microtubules. (E) Light scattering traces for 1 µM constructs suggest that only the brain-specific full-length p150 isoform can promote microtubule formation. Error bars (± SEM) are smaller than the symbols. (F) Single frames from 6 min posttubulin perfusion contrasting the inability of Δ5–7 Nt-GCN4 or p135 Nt-GCN4 to nucleate microtubules with the full-length p150 Nt-GCN4. Scale bar, 12.5 µm.
We produced recombinant dimeric polypeptides recapitulating p135 (p135 Nt-GCN4) and p150 lacking exons 5, 6, and 7 (the majority of the basic region, Δ5–7 Nt-GCN4) (Figure 4A and Figure S5A–D). We found that when we co-incubated p135 Nt-GCN4 (which lacks the CAP-Gly domain) with tubulin, we did not observe complex formation by size exclusion chromatography (Figure 4B). This was unsurprising given the accepted role of CAP-Gly domains in binding to microtubules , and the role we establish here for p150-binding to tubulin dimers in solution. We next assayed Δ5–7 Nt-GCN4, which lacks only the 23 amino acid residue basic domain; deletion of this domain only marginally impacts microtubule-binding affinity as measured by co-pelleting (Kd = 470 nM versus 290 nM for p150 Nt-GCN4, Figure 4C) ,, and does not affect EB1 binding as assessed by size exclusion chromatography (Figure S5E–F) . Strikingly, we find that this isoform cannot bind tubulin (Figure 4D). Further, neither p135 Nt-GCN4 nor Δ5–7 Nt-GCN4 were able to promote the formation of microtubules as assessed both by light scattering and direct TIRF imaging (Figure 4E, Figure 4F). These results demonstrate that both the tandem p150Glued CAP-Gly and basic domains are necessary to stably complex tubulin and to promote assembly dynamics.
To determine how the cellular role of p150Glued corresponds to the biochemical activities we measured, we first depleted p150Glued from COS7 cells (Figure 5A), a primate cell line derived from kidney that should not express significant levels of the neuronal isoform . We transfected the cells with low levels of GFP-EB3 to visualize the growing microtubule plus end, and examined microtubule dynamics in an unbiased manner using the PlusTipTracker software package . Under conditions in which >95% of endogenous p150Glued was depleted (Figure 5A), we did not observe an appreciable effect on either the average displacement of EB3-GFP comets before microtubule catastrophe or on microtubule polymerization velocities (Figure 5B–C, p>0.1, and unpublished data). We obtained similar results in HeLa cells, another epithelial cell line.
Figure 5. p150Glued stabilizes microtubules in neurons.
(A) Depletion of p150Glued in COS7 cells by siRNA (KD) relative to control cells treated with scrambled (Scram) oligonucleotides; upper panel is a Western blot probed with a monoclonal antibody to p150Glued, and lower panel is Coomassie staining to show equal protein loading. (B) Representative rainbow-coded maximum intensity projections show that the overall microtubule architecture and dynamics are not perturbed in COS7 cells depleted of p150Glued relative to control cells. (C) PlusTipTracker quantitation of EB3-GFP comet velocities shows that knockdown does not alter parameters of microtubule dynamics, including distance to catastrophe, a measure of the catastrophe frequency (p>0.1). (D) Western blot showing knockdown of p150Glued in DRG neurons (KD), relative to control neurons or COS7 cells treated with scrambled oligonucleotide. Note the differential splice forms of p150Glued expressed in neurons relative to COS7 cells. (E) Space-filling fluorescence in representative neurites used to investigate microtubule dynamics in primary DRG neurons. (F) Rainbow-coded maximum intensity projections of selected EB3-GFP comets demonstrate microtubule dynamics in DRG neurons. (G) Kymographs of GFP-EB3 comets in cultured DRG neurons treated with either scrambled or p150Glued RNAi reveal that in neurons, p150Glued inhibits catastrophe. (H) Camera lucida tracing of the kymographs in panel G to indicate polymerization events persisting greater than 5 µm (black) and less than 2 µm (magenta). (I) Kymographs of GFP-EB3 comets in neurons depleted of endogenous p150Glued by RNAi and rescued with either full-length p150, Δ5–7, or the neuron-specific alternative splice form p135. (J) Analysis of microtubule dynamics in scrambled control neurons, and neurons depleted of endogenous p150Glued with or without rescue with resistant constructs of p150 or p135 and Δ5–7. Bars represent mean of comet parameters from multiple cells on multiple days ± SEM. Statistical testing was performed via t test with correction for multiple comparisons. *** p<0.001. (K) In vitro analysis of catastrophe rates demonstrates that the Perry syndrome-associated mutation Q74P p150 Nt-GCN4 does not inhibit microtubule catastrophe (p>0.5), as compared to the wild-type p150Nt-GCN4 construct. (L, M) Kymographs and quantitation of GFP-EB3 comets in cultured DRG neurons depleted of endogenous p150Glued and rescued with either wild-type or Q74P p150Glued reveal that mutant p150 is defective in inhibiting microtubule catastrophe in neurons.
We next investigated the effect of p150Glued depletion on microtubule dynamics in mammalian neurons, which express the full-length isoform which includes both the CAP-Gly and basic domains that we have identified as necessary for productive tubulin binding. Interestingly, though the CAP-Gly domain is a conserved feature of p150Glued, the basic domain appears to be specifically evolved in organisms with complex, dynamic microtubule cytoskeletons. It is absent in yeasts, but is apparent in protists, filamentous fungi, and vertebrates, with mammalian taxa expressing particularly basic regions downstream of the CAP-Gly domain (Figure S6A–B, Dataset S1) .
Next, we visualized microtubule dynamics using GFP-EB3 expressed at low levels in mouse dorsal root ganglion neurons (DRGs; Figure 5E,F). Because of the high levels of tubulin in mammalian neurons, we quantitated microtubule dynamics manually using kymograph analysis (Figure 5G). When p150Glued was depleted by ∼80% (Figure 5D), we did not observe significant differences in either polymerization rates or the number of EB3-GFP comets compared to scrambled RNAi-treated cells (unpublished data). However, we noted that when p150Glued was depleted, there was a significant decrease in the distance GFP-EB3 comets traveled before catastrophe (Figure 5G,H), indicating a significant increase in the catastrophe frequency (Figure 5J). For example, compare the comets that traveled at least 5 µm before catastrophe (black) compared to those that travelled less than 2 µm (magenta) as shown in Figure 5H.
To confirm that this was a specific effect of p150Glued knockdown, we quantitated dynamics in RNAi-treated neurons rescued with either the full-length neuronal p150 isoform, or with p135, which lacks the N-terminal CAP-Gly domain. We observed that rescue of neurons with a plasmid encoding RNA-resistant p150 significantly reduced the catastrophe frequency as compared to RNAi-treated cells (Figure 5I,J). Wild-type dynamics were not completely restored, likely because we did not fully restore endogenous expression levels with the rescue plasmid (Figure S6C). In contrast, transfection with the p135 isoform, which lacks the N-terminal CAP-Gly domain, did not rescue defective polymerization dynamics in primary neurons (Figure 5I,J). Importantly, we found that transfection with the Δ5–7 construct, the nonneuronal splice form that lacks the basic domain, also did not rescue the defective polymerization dynamics induced by depletion of endogenous p150Glued (Figure 5I,J). Thus, in strong agreement with our in vitro results, these observations indicate that the tandem CAP-Gly and basic domains of p150 are required to generate anti-catastrophe activity in neurons.
Mutations in the CAP-Gly domain of p150Glued have been found to cause multiple neurodegenerative diseases, including Perry syndrome, a disease characterized by Parkinsonism, weight loss, hypoventilation, and depression. While human disease-associated mutations in p150Glued are expressed throughout the body, these mutations induce pathologies only in the nervous system . Thus, we hypothesized that the tissue-specific activity of p150Glued described above might be disrupted in this disease.
To test this hypothesis, we focused on the Q74P mutation (Figure S7A) . The Q74P mutation has been shown to disrupt both the microtubule and EB1-binding activities of the p150 CAP-Gly domain but to have only a modest effect on overall protein stability ,. To obviate any disease-induced aggregation, we expressed and purified recombinant Q74P Nt-GCN4 immediately before experimentation using a final gel filtration step to exclude higher order oligomers (Figure S7B, S7C), and confirmed proper dimerization by glutaraldehyde cross-linking (Figure S7D). In contrast to wild-type p150 Nt-GCN4, we found that Q74P Nt-GCN4 was defective in inhibiting microtubule catastrophe in vitro (Figure 5K). In primary neurons, we found that transfection with a plasmid expressing Q74P could not restore normal microtubule dynamics, in contrast to the rescue seen with the wild-type p150Glued construct (Figure 5L,M).
Here we show that p150Glued promotes microtubule formation in vitro by catalyzing nucleation, increasing the polymerization rate, and inhibiting catastrophe. These activities require dimerization and are dependent on the ability of p150Glued to form a stable complex with tubulin through interactions with both the N-terminal CAP-Gly and basic domains. In primary neurons, we observe that the dominant effect of p150Glued on microtubule dynamics is the suppression of catastrophe (Figure 6). Finally, we determine that a single point mutation within the CAP-Gly domain of p150Glued causative for a fatal familial form of Parkinson disease, known as Perry Syndrome, leaves p150Glued unable to promote microtubule assembly either in vitro or in neurons.
Figure 6. p150Glued is a neuron-specific microtubule anti-catastrophe factor.
p150Glued inhibits microtubule catastrophe by binding both to microtubules and to soluble tubulin, leading to enhanced microtubule stability along the axon. Depletion of endogenous p150Glued by RNAi leads to more frequent microtubule catastrophe in neurons.
Dynactin was originally identified as a large protein complex that supported dynein-mediated vesicle transport ,. The N-terminus of the 150 kDa subunit binds microtubules independently of dynein , and increases the processivity of dynein in vitro ,. Recently, it has been demonstrated that the microtubule-binding N-terminus of p150Glued is dispensable for organelle localization and vesicular motility in nonneuronal cells ,,. However, the microtubule-binding CAP-Gly domain of dynactin is required for efficient transport initiation from the distal axon in neurons ,. A plus end-localized pool of p150Glued may serve to load dynein onto the microtubule ,. However, biochemical analyses and immunolocalization suggest that a large proportion of dynactin may in fact not be in complex with dynein ,,,. Enriched at the plus end, this population of dynactin would be perfectly poised to affect microtubule dynamics.
Our data suggest a mechanism whereby p150Glued could modify microtubule dynamics. Recently, it has been suggested that the kinetics of tubulin association and dissociation with the microtubule plus end may be much faster than previously appreciated . This makes it increasingly plausible that one mode whereby MAPs alter microtubule dynamics is by modulating the off-rate of tubulin subunits from microtubule plus ends. Since p150Glued can bind both to microtubules and to soluble tubulin, and because dimerization appears necessary for p150Glued to robustly modify dynamics, we speculate that p150Glued may be acting in this capacity by binding to both microtubules and tubulin at the same time, decreasing the off-rate and inhibiting catastrophe, enabling efficient initiation of dynein-mediated retrograde runs (Figure 6). Interestingly, this mechanism is distinct from the mode by which cytoplasmic dynein independently functions to inhibit catastrophe . Areas of the distal neuron where both cytoplasmic and dynactin are localized could be sites of particularly robust microtubule stabilization.
The regulation of these microtubule-modifying abilities of p150Glued may be multifactorial. We have shown that the basic region is necessary for the modification of microtubule dynamics by p150Glued, likely by ensuring a stable complex with the distributed acidic nature of tubulin. The basic region is also serine- and threonine-rich, and has been shown to be the target of phosphorylation by regulatory kinases ,, which might further modulate the p150-tubulin interaction during mitosis, or during development. p150Glued also binds to CLIP-170 ,,,, which could further modify the behavior of p150Glued in the cell.
In vivo, only the p150Glued isoform expressed in neurons includes both the full CAP-Gly and basic domains that we have shown are necessary to modify microtubule assembly dynamics. We have recently shown that young, developing neurons depleted of p150Glued are morphologically normal . In fact, profound depletion of both EB1 and EB3, which should effectively disrupt plus-end targeting, has no gross effects on neurite outgrowth . It may be that, as recent evidence suggests, only as neurons age, and their processes lengthen and elaborate, does the centrosome lose its function as a microtubule organizing center and microtubule dynamics become particularly reliant on plus-end regulation ,. It is perhaps telling that human patients with the Q74P p150Glued mutation do not show disease onset until the fifth decade of life . More broadly, microtubule dynamics may alter in aging or degenerating neurons, as suggested from studies of cells from patients with sporadic Parkinson’s and Alzheimer’s disease –.
In summary, we have identified and characterized a novel role for p150Glued in the tissue-specific stabilization of microtubules, and implicated defects in neurodegeneration. Further studies to disentangle the effects of the mutation on axonal transport and microtubule stability in neurons will be required to clarify the pathogenesis involved.
You must log in to like an article
Regulation of microtubule dynamics in neurons is critical, as defects in the microtubule-based transport of axonal organelles lead to neurodegenerative disease. The microtubule motor cytoplasmic dynein and its partner complex dynactin drive retrograde transport from the distal axon.