Regulation of Dynactin through the Differential Expression of p150^Glued Isoforms^*,S⃞

EXPERIMENTAL PROCEDURES

Library Screening and RT-PCR—Multiple isoforms of p150^Glued were identified in a screen of a human fetal brain cDNA library (Stratagene). RT-PCR was performed using an Enhanced Avian RT-PCR kit (Sigma), poly(A)+ RNA from adult human brain, skeletal muscle and kidney (Clontech), and primers corresponding to sequences from the 5′-end of exon 3 and the 3′-end of exon 8 of the human DCTN1 gene. The PCR products were fractionated on 4% NuSieve-GTG (BMA, Rockland, ME) gel, and purified with QIAEX II Gel Extraction Kit (Qiagen) for TA cloning followed by sequence analysis.

Cell Cultures, Transfections, and Immunocytochemistry—COS7, HeLa-M and MN1 cells were maintained as described (17). Transient transfection assays were performed using Fugene-6 (Roche Applied Science) and plasmids encoding Myc-tagged or GFP-tagged forms of p150^Glued. For immunofluorescence assays, COS7 and HeLa-M cells were fixed in -20 °C methanol for 10 min and MN1 cells were fixed in 4% paraformaldehyde for 20 min and then permeabilized with 0.25% Triton X-100. Immunostaining was performed using antibodies to tubulin (DM1A, Sigma or YL1/2, Serotech), p150^Glued (BD Biosciences and polyclonal antibody UP502(17)), Myc (Invitrogen), Golgi protein GM130 (BD Biosciences), and TGN46 (Serotech).

Live Cell Imaging—For time-lapse imaging using OpenLab software (Improvision), cells were seeded on glass bottom dishes (World Precision Instruments) and imaged in phenol red-free Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 25 mm HEPES, 1% OxyFluor (Oxyrase), and sealed with mineral oil (Sigma). The microscope stage was maintained at 37 °C. Images were acquired at 5-s intervals for 5 min.

RNA Interference—HeLa-M cells were transfected using Lipofectamine RNAiMAX (Invitrogen) with 10 nm RNA duplexes (Dharmacon) targeted to the 3′-untranslated region of human DCTN1 (GenBank™ accession number NM_004082). Two independent RNA duplexes were used: UTR1, 5′-ccaccaccaaagguuaaguuu-3′ and UTR2, 5′-gacuucaccccuugauuaauu-3′, both of which gave similar results. Maximal p150^Glued knockdown was achieved 72 h after transfection. Rescue experiments were performed by transfecting RNAi-treated cells 72 h later with the indicated DNA constructs using Fugene (Roche Diagnostics), in comparison to mock-treated controls. Overall levels of p150^Glued expression were determined by quantitative Western blot of whole cell lysates; we observed no significant differences between the mean expression levels of mock-treated HeLa-M cells and cells rescued with either wild-type p150^Glued or any of the specifically mutated constructs (p = 0.6 by one-way analysis of variance). We also compared the relative signal intensity at the cellular level via immunofluorescence measurements analyzed with NIH ImageJ and consistently found that expression of wild-type p150 to about 50% of the control level was sufficient to rescue Golgi morphology. We used this expression level as a standard to compare the effects of wild-type and mutant proteins. The average fluorescence intensity of p150 for each condition was: 113.5 ± 17.8 (n = 84 cells) for mock-treated cells, 56.2 ± 16.5 (n = 100 cells) for cells rescued with wild-type p150^Glued, 55.5 ± 14.6 (n = 43 cells) for cells rescued with the M571T construct, 51.2 ± 15.7 (n = 60 cells) for cells rescued with the R785W construct, 48.0 ± 9.3 (n = 42 cells) for cells rescued with the R1101K construct, and 49.5 ± 13.1 (n = 44 cells) for cells rescued with the T1249I construct.

Microtubule Binding Assay—His-tagged alternatively spliced isoforms of p150^Glued, spanning residues 1-333, were expressed in E. coli and purified by Ni²⁺-affinity chromatography. Alternatively, cDNA constructs encoding these isoforms or wild type and mutant forms of p150^Glued were expressed in vitro using the TnTT7 Quick system (Promega) and clarified by centrifugation at 39,000 × g for 30 min. Either the purified recombinant or in vitro-expressed proteins were then incubated with increasing concentrations of microtubules polymerized in vitro from purified bovine tubulin (Cytoskeleton) in the presence of 0.1 mm GTP and 20 μm Taxol (Cytoskeleton) for 20 min at 37 °C. Microtubule-bound and unbound proteins were separated by centrifugation at 39,000 × g for 20 min and analyzed by SDS-PAGE and Western blot analysis. Results were quantified by densitometry using NIH ImageJ.

Dynein Intermediate Chain Pull-down Assay—His-tagged full-length recombinant dynein intermediate chain was expressed in Escherichia coli, purified by Ni²⁺ affinity chromatography, and then coupled to activated CH-Sepharose 4B beads (8). Myc-tagged p150^Glued proteins were expressed using the TnTT7 Quick system (Promega) and clarified by centrifugation at 39,000 × g for 30 min. A fixed volume of DIC-Sepharose beads (equivalent to ∼15 μg DIC) were incubated with increasing amounts of p150^Glued proteins for 1 h at 25 °C on a platform shaker at 200 rpm; bound and unbound proteins were separated by centrifugation at 1,300 × g for 2 min and analyzed by SDS-PAGE and Western blot. Results were quantified by densitometry using NIH ImageJ.

Sucrose Density Gradient Fractionation—COS7 cells transiently expressing Myc-tagged wild-type or mutant p150^Glued proteins were lysed using 100 mm Na-PIPES (pH 6.8), 1 mm MgSO₄, 1 mm EGTA, 25 mm NaCl, 0.5 mm dithiothreitol, 1% IGEPAL CA-630, and protease inhibitors (leupeptin, pepstatin A, N-p-tosyl-l-arginine methyl ester, and phenylmethylsulfonyl fluoride). The cell lysate was clarified by centrifugation at 38,000 × g for 10 min at 4 °C, and the resulting supernatant was layered over a 5-25% linear sucrose density gradient and centrifuged at 120,000 × g for 18 h at 4 °C. 1-ml gradient fractions were collected and analyzed by SDS-PAGE and Western blot analysis using monoclonal antibodies to Myc (Invitrogen), DIC (MAB 1618, Chemicon), and dynamitin (BD Biosciences). Results were quantified by densitometry using NIH ImageJ.

RESULTS

Differential Expression of Alternatively Spliced Isoforms of p150^Glued—The human gene encoding p150^Glued is composed of 32 exons (20). We have previously identified an alternatively spliced form of p150^Glued that lacks the N-terminal CAP-Gly and basic microtubule-binding domains (Fig. 1A). This p135 isoform is expressed primarily in neurons (15). We screened a fetal human cDNA library to probe for additional alternatively spliced isoforms of p150^Glued and found multiple alternatively spliced cDNAs, which differ in an N-terminal domain encoded by exons 5-7; three small exons that encode 7, 6, and 7 amino acid residues, respectively (Fig. 1A). Specifically, we identified isoforms lacking exons 5 and 6 (Δ5,6) and exons 5 through 7 (Δ5,6,7).

We performed RT-PCR to compare the expression of these alternatively spliced transcripts in several human tissues, including brain, kidney, and muscle (Fig. 1B). The identity of the bands observed by gel electrophoresis was confirmed by direct sequencing of the PCR products. We noted the expression of multiple isoforms of p150^Glued in adult brain, including a major band corresponding to the full-length polypeptide that includes all three exons (FL), as well as two more minor bands corresponding to isoforms lacking either exons 5 and 6 (Δ5,6) or lacking exons 5 through 7 (Δ5,6,7). In skeletal muscle we also observed the expression of multiple p150^Glued isoforms, including a major band corresponding to Δ5,6 and minor bands corresponding to an isoform lacking only exon 5 (Δ5) as well as Δ5,6,7. In contrast, the predominant isoform of p150^Glued expressed in kidney was Δ5,6,7, which lacks the basic domain. Together, these observations suggest tissue-specific regulation of expression of p150^Glued isoforms.

Regulation of Tip Tracking Behavior by Differential Expression of p150^Glued Isoforms—The region encoded by exons 5-7 spans a basic domain of the p150^Glued polypeptide (Fig. 1A) that has recently been identified as possessing a weak affinity for microtubules (11). Truncation constructs of p150^Glued that lack the N-terminal CAP-Gly domain but include this basic domain have been shown to exhibit one-dimensional diffusion along the microtubule, termed “skating” behavior (11). In contrast, constructs that include the N-terminal CAP-Gly domain bind more stably to the microtubule. Therefore, the additional splice forms that we identified may encode functionally distinct isoforms of the polypeptide, which differ in their relative association with the microtubule.

We tested whether these isoforms differ in their affinity for microtubules. As shown in Fig. 2, a comparison of the binding of the p150-FL and p150-Δ5,6,7 isoforms in a microtubule sedimentation assay indicates that the loss of the basic domain significantly alters the overall affinity for microtubules. p150-FL bound to microtubules with a K_d of 0.40 ± 0.07 μm. p150-Δ5,6,7 bound to microtubules much more weakly; we did not reach saturation (the extrapolated B_max was 0.7 ± 0.1) potentially due to weak binding, but we estimated a K_d of 5.3 ± 1.5 μm. Consistent with this observation, we have found that a recombinant construct of p150^Glued spanning the CAP-Gly domain (residues 1-107) binds to microtubules with a significantly lower affinity than a longer construct (1-333) that also includes the basic domain (data not shown).

Cellular assays have demonstrated that GFP-labeled constructs of p150^Glued can dynamically track growing microtubule plus ends (12, 13), but we have previously observed that the dynamic plus end localization of p150^Glued is cell type-specific (21). Given the cell type-specific expression of p150^Glued isoforms, we hypothesized that the difference in tip tracking behavior might be due to the differential expression of alternatively spliced isoforms of the polypeptide.

To test this hypothesis, we tagged both the p150-FL construct and the Δ5,6,7 isoform with GFP and expressed these polypeptides in HeLa cells. As shown in Fig. 3A, we saw a clear difference in the cellular localization of these constructs. P150-FL fully decorates cellular microtubules. Microtubule bundling was observed at higher expression levels, as previously noted (10). In contrast, the Δ5,6,7 isoform shows marked tip decoration at a similar expression level. Live cell assays demonstrate that the tip decoration seen in fixed cells is due to dynamic tip tracking of the alternatively spliced isoform (Fig. 3B for stills and see supplemental Movie S2). In contrast, tip tracking is not seen in dynamic assays with the p150-FL isoform (Fig. 3B and supplemental Movie S1). Similar results were obtained upon expression of these constructs in COS-7 cells (data not shown).

Is the Microtubule Binding Domain of p150^Glued Required for Dynactin Function?—The expression of distinct isoforms of p150^Glued that vary in their affinity for microtubules both in vitro and in cellular assays suggests that modulation of microtubule binding may have a significant effect on dynactin function. While cytoplasmic dynein alone is a robust microtubule motor in vitro, dynactin is required for most of dynein functions in the cell. Dynactin is not required for processivity of the dynein motor, but single molecule studies have shown that dynactin enhances dynein run length along the microtubules, potentially due to the ATP-independent binding of dynactin to the microtubule (6, 7). However, in vitro studies have shown that when multiple dynein motors are bound to the same bead cargo the effects of dynactin on overall run length are negligible (22). Therefore, it has been suggested that the microtubule-binding domain of dynactin may not be required for intracellular organelle transport; studies in Drosophila S2 cells support this hypothesis (23).

We asked whether the microtubule-binding domain of p150^Glued is required to maintain Golgi organization in mammalian cells. Dynactin is required to maintain a compact perinuclear Golgi in HeLa cells (17); either depletion or transient overexpression of p150^Glued causes Golgi disruption. Transfection of HeLa cells with either the p150-FL or the p150-Δ5,6,7 isoforms leads to vesiculation of the Golgi (Fig. 3C). This disruption is expression-level dependent; lower levels of GFP-p150 expression do not measurably disrupt the Golgi (see cell marked with an arrowhead in Fig. 3C), but higher expression levels lead to complete dispersal (see cell marked with a double arrowhead in Fig. 3C).

While the p150-Δ5,6,7 isoform binds to microtubules more weakly than the p150-FL polypeptide, this isoform still has an intact CAP-Gly domain. To more fully test the effects of the microtubule-binding domain, we compared the expression of p150-FL with p135, a neuronally expressed isoform that lacks both the CAP-Gly motif and the basic residues that mediate microtubule binding (Fig. 1A). As shown in Fig. 4A, overexpression of p135 does not disrupt the Golgi to the same extent as expression of p150-FL. In most cells expressing p135, the Golgi appears normal. In ∼40% of cells overexpressing p135, the Golgi appears tubulated (Fig. 4A), indicating partial rather than full disruption (24).

To test whether the microtubule-binding domain is required to maintain Golgi organization, we used an RNAi approach to determine if the p135 isoform can functionally substitute for full-length p150^Glued in HeLa cells. Endogenous p150^Glued was depleted using either of two independent siRNA oligos to the 3′-UTR (supplemental Fig. S1), leading to 70-80% knockdown of expression 72 h after transfection. Complete rescue was observed upon transfection with a full-length construct of human p150^Glued that lacks the 3′-UTR and is thus resistant to the targeted siRNA (Fig. 4B).

Quantitative analysis indicates that the p135 isoform is indistinguishable from wild-type p150^Glued in terms of rescuing Golgi localization following knockdown of endogenous p150^Glued protein in HeLa cells (Fig. 4B). These results suggest that the microtubule-binding domain of p150^Glued is not important for dynactin's role in maintaining a compact perinuclear Golgi in mammalian cells, consistent with the observation that this domain is dispensable for peroxisome transport in Drosophila (23).

Functional Analysis of Sequence Variants of p150^Glued Identified in Patients with ALS—While the experiments shown in Fig. 4 indicate that the microtubule-binding domain of p150^Glued is not required for normal Golgi organization in nonneuronal cells, a mutation in this domain (G59S; Fig. 1A) has a significant effect on dynactin function leading to motor neuron degeneration (5). This mutation exerts its effects through mechanisms involving both loss-of-function, and dominant gain of a toxic function (2, 17, 25). Additional allelic variants of the p150^Glued polypeptide have been identified in ALS patients but not in controls (18, 19). These allelic variants, M571T, R785W, R1101K, and T1249I, affect residues that are highly conserved among the vertebrate p150^Glued proteins and are therefore potential candidates as causal or risk factors for ALS. However, there is no conclusive genetic evidence linking these mutations to ALS. To test whether these new p150^Glued mutations lead to functional disruption of dynein-dynactin activity, we tested the effects of these mutations in both in vitro and cellular assays.

We expressed Myc-tagged constructs corresponding to each variant in both COS-7 cells (Fig. 5A) and the motor neuron-like cell line MN1 (supplemental Fig. S2), and compared the localization of these variants to wild-type full-length p150^Glued and to p150^Glued with the G59S mutation. The M571T, R785W, R1101K, and T1249I constructs all bound robustly to cellular microtubules, indistinguishable from wild-type p150^Glued; only the G59S mutation significantly disrupts microtubule binding in this assay. Each of the Myc-tagged constructs, including the G59S construct, disrupted the Golgi in transfected cells. The dispersed Golgi staining seen in transfected cells differs significantly from the perinuclear localization in untransfected cells on the same coverslip (Fig. 5A). To quantitate these results we measured the distribution of Golgi staining as a function of distance from the nucleus in both COS7 and MN1 cells expressing Myc-tagged wild-type p150^Glued, or the M571T, R785W, R1101K, and T1249I constructs. No significant differences were observed as compared with expression of Myc-tagged wild-type p150^Glued (Fig. 5B and supplemental Fig. S2). These experiments suggest that these four amino acid substitutions do not perturb microtubule binding. We also verified this in microtubule sedimentation assays; as shown in Fig. 6A, no significant differences were observed in the association of these constructs with microtubules in vitro.

Both the M571T and R785W substitutions occur in or near the dynein-binding domain of p150^Glued. Therefore we performed pull-down assays with recombinant DIC covalently bound to Sepharose beads. We noted no significant differences in the association of any of the four Myc-tagged constructs with DIC in this assay (Fig. 6B). Thus, none of the mutations have a measurable effect on the affinity of the dynein-dynactin interaction.

As noted above, the G59S mutation in p150^Glued induces both a loss-of-function, and a gain-of-function, the later due to an increased tendency for the mutant protein to aggregate (17). To test whether any of ALS-associated p150^Glued variants induce a similar aggregation, or whether they are less effectively incorporated into the dynactin complex, we expressed the Myc-tagged M571T, R785W, R1101K, and T1249I constructs in COS-7 cells, and then fractionated the resulting cellular lysates over sucrose density gradients. Comparisons of these gradients (Fig. 6C and data not shown) indicate comparable levels of incorporation of wild type and Myc-tagged M571T, R785W, R1101K, and T1249I constructs of human p150^Glued into the endogenous dynein-dynactin complex. Additionally, both sucrose density gradient analysis and immunofluorescence analysis of transfected cells indicates that none of these mutations result in aberrant p150^Glued aggregation. We also examined the effects of overexpression of wild type and Myc-tagged M571T, R785W, R1101K, and T1249I constructs of human p150^Glued in the MN1 neuronal cell line (supplemental Fig. S2). Again, we observed no aggregation induced by any of the constructs, in contrast to previous observations on the formation of protein aggregates induced by the expression of the G59S mutation in this cell type (17).

Together, these results suggest that the M571T, R785W, R1101K, and T1249I substitutions in the p150^Glued polypeptide do not result in significant perturbation of dynactin function. In order to detect more subtle defects in p150^Glued function, we used RNAi to specifically knockdown endogenous p150^Glued in HeLa cells (supplemental Fig. S1) and examined the ability of the Myc-tagged M571T, R785W, R1101K, and T1249I constructs to rescue cellular function, using a rescue of Golgi disruption as a readout for wild-type dynein-dynactin activity in these experiments (Fig. 7). Expression levels of each of the constructs were monitored relative to expression in mock-treated controls by quantitative Western blot as well as quantitative analysis at the cellular level to ensure cells with similar expression levels were compared (as described under “Experimental Procedures”). While there were no clear-cut differences among the constructs tested, we found that the G59S construct was least effective at restoring normal Golgi morphology, consistent with a mild loss-of-function effect for this polypeptide (17). None of the ALS-associated variants were significantly different from wild type in this assay.

DISCUSSION

Cytoplasmic dynein and dynactin are essential for a wide range of intracellular functions in higher eukaryotes, including interphase roles of intracellular trafficking and mitotic roles in spindle assembly and checkpoint. Given these varied and important cellular functions, it is not surprising that loss of either dynein or dynactin is lethal early in development. However, it is interesting that subtle defects in dynein or dynactin caused by missense mutations in subunits of either complex result in a neuron-specific degeneration in both mice and humans.

The observation that a G59S mutation in the p150^Glued subunit of dynactin results in slowly progressive neurodegenerative disease had lead to multiple studies looking for additional mutations in the polypeptide in patients with motor neuron disease. Munch et al. (18, 19) have identified three alterations in the sequence of p150^Glued in patients with ALS, as well as one in a patient with ALS and FTD. Insufficient genetic data were available to causally associate the sequence variants with disease.

As all four of the mutations identified in the DCTN1 gene map to domains of interest within the p150^Glued polypeptide, we tested the effects of these amino acid substitutions using a range of biochemical and cellular assays, in comparison to both wild-type and G59S forms of the polypeptide. None of these mutations appear to induce significant functional disruption of transport or trafficking in either neuronal or nonneuronal cells. Further, none of these mutations induce the formation of prominent cellular aggregates similar to those seen previously upon expression of the G59S mutation. Thus, we conclude that despite the initial identification of these variants in patients with ALS, these changes are not sufficient to cause disease. We cannot rule out the possibility that these variants may act as risk rather than causative factors, with the potential to compromise dynactin function in a subtle manner. However, in the above assays, none of these four constructs was as effective in disrupting dynactin function as the previously identified G59S mutation. One striking aspect of these studies is that expression levels of p150^Glued are likely to be tightly regulated in the cell. Both depletion and overexpression of p150^Glued results in a similar phenotype of Golgi disruption, and may therefore be predicted to aversely affect axonal transport (17). Mutations that alter expression levels of this protein may therefore also be considered as potential risk factors for motor neuron disease. While the ALS-associated variants tested here did not noticeably affect either protein stability or protein aggregation, these aspects may be as or more important as direct perturbations in binding domains within dynactin (25).

Using RT-PCR we identified alternatively spliced isoforms of p150^Glued that do affect dynactin activity directly. Specifically, we noted that multiple isoforms of the polypeptide are expressed that differ in their microtubule binding affinity. This in turn leads to alterations in the dynamic localization of the polypeptides in live cell assays. The full-length form of p150^Glued that is expressed in human brain binds along the length of cellular microtubules, and is not specifically associated with dynamically growing and shrinking microtubule plus ends. In contrast, the shorter isoform of p150^Glued that lacks exons 5, 6, and 7 and is highly expressed in kidney has a significantly weaker affinity for microtubules in vitro. In the cell, this isoform dynamically tracks growing but not shrinking microtubule plus ends.

Based on these observations, we hypothesize that the differential affinity for binding to microtubules seen in vitro means that the shorter isoform will bind less strongly along the lengths of microtubules in the cell. However, the Δ5,6,7 isoform is still able to actively track with growing microtubule ends, most likely through a direct association with the plus end-binding protein EB1 (21, 26). In contrast, the full-length isoform binds more robustly along the length of the microtubule and therefore does not exhibit plus end-binding specificity. This full-length construct is also capable of direct binding to EB1 (21), but remains associated with the microtubule during both growth and catastrophe. Thus the isoform-specific alterations in microtubule binding affinity we have noted lead to a striking modulation in +Tip-tracking behavior in live cell assays. As we have previously observed using in vitro assays, dynamic localization of CAP-Gly proteins to the microtubule plus end involves a balance in relative affinities, a higher affinity for either the microtubule end itself and/or other end-binding proteins such as EB1, in association with a relatively lower affinity for lateral binding to the microtubule (26). Here, we see a clear modulation of tip-tracking activity in the cell based on isoform expression.

Despite these clear functional differences both in vitro and in live cell assays, it is unclear if these isoforms have differential functions in vesicle trafficking in nonneuronal cells. Kim et al. (23) have shown in Drosophila S2 cells, and we confirm their result in HeLa cells, that the loss of the N-terminal microtubule-binding domain of p150^Glued including both the CAP-Gly domain and the adjacent basic domain, does not significantly affect intracellular trafficking of organelles including peroxisomes and the Golgi. However, Kim et al. (23) noted that loss of the microtubule-binding domain of p150^Glued did affect microtubule organization during cell division. The tissue-specific regulation of isoform expression that we note here may potentially correlate with overall mitotic activity, as the more stably binding full-length isoform is expressed at a high level in brain, potentially due to higher expression levels in post-mitotic neurons. Given the neuron-specific effects of the G59S mutation in p150^Glued, this domain may play a critical role in some aspect of neuronal function. Recent evidence suggests a role for dynein and dynactin in maintaining microtubule polarity in the axon (27, 28); the independent microtubule-binding domain of dynactin may be critical for this function.

Given the wide range of cellular functions that depend on dynactin, it will be important to continue to characterize the regulation of this complex within the cell, including the expression of multiple functionally distinct isoforms as described here. Also, given the identification of a mutation in dynactin as causative for motor neuron disease, additional mutations in subunits of either dynein or dynactin may be found to be causative for neurodegenerative diseases such as ALS. However, it will be essential to thoroughly investigate the effects of these mutations in biochemical and cellular assays to fully understand the effects they may have on dynactin function in the cell.

Supplementary Material