Huntington's disease is a severe progressive neurodegenerative disorder caused by a CAG expansion in the IT15 gene, which encodes huntingtin. The disease primarily affects the neostriatum and cerebral cortex and also associates with increased incidence of diabetes. Here, we show that mutant huntingtin disrupts intracellular transport and insulin secretion by direct interference with microtubular β-tubulin. We demonstrate that mutant huntingtin impairs glucose-stimulated insulin secretion in insulin-producing β-cells, without altering stored levels of insulin. Using VSVG-YFP, we show that mutant huntingtin retards post-Golgi transport. Moreover, we demonstrate that the speed of insulin vesicle trafficking is reduced. Using immunoprecipitation of mutant and wild-type huntingtin in combination with mass spectrometry, we reveal an enhanced and aberrant interaction between mutant huntingtin and β-tubulin, implying the underlying mechanism of impaired intracellular transport. Thus, our findings have revealed a novel pathogenetic process by which mutant huntingtin may disrupt hormone exocytosis from β-cells and possibly impair vesicular transport in any cell that expresses the pathogenic protein.
Huntington's disease (HD) is caused by a CAG triplet expansion in the first exon of the IT15 gene, coding for the protein huntingtin. The disease primarily affects the neostriatum and cerebral cortex. However, HD is also associated with an increased incidence of diabetes (1,2). The function of huntingtin is not yet fully understood, but a role in regulating intracellular trafficking has been suggested. Recently, huntingtin was shown to affect kinesin activity via alterations in the state of phosphorylation of huntingtin (3,4). Moreover, huntingtin has been shown to interact with dynein (5), HAP-1/p150Glued (6,7) and kinesin light chain (8), all of which are part of the transport machinery. The protein has also been reported to be associated with microtubules (9,10). A normal and highly regulated interplay between microtubules and molecular motors is crucial for an effective transport. Mutant huntingtin has been shown to exert negative effects on axonal transport in neurons (11–15), again emphasizing the importance of huntingtin in intracellular trafficking events.
In a transgenic mouse model (R6/2) of HD in which the mice express exon 1 of huntingtin with 155 CAG repeats (16), insulin secretion is deficient in late disease stages (17). β-Cell mass, insulin content of the islets and secretion of insulin are all reduced. Here, we have, in order to further elucidate the defects seen in the mouse model, studied the effects of expressing mutant huntingtin in an acute fashion in a clonal insulin-producing cell line. We found that mutant huntingtin rapidly reduces the efficiency of intracellular transport and decreases stimulated insulin secretion despite normal content of insulin and a normal number of insulin vesicles being present in the cells. We also report that exon 1 mutant huntingtin interacts more strongly with β-tubulin than its wild-type (wt) counterpart. We suggest that this interaction blocks intracellular transport, rendering it progressively more inefficient as huntingtin accumulates. This could potentially be a novel pathogenetic mechanism in HD, occurring in the pancreatic β-cells, and perhaps also in other cell types in which vesicular transport is crucial, e.g. neurons.
Altered insulin levels, localization and secretion
We first used semi-quantitative real-time PCR to study insulin mRNA expression in 12-week-old R6/2 and wt mice. As expected, we found dramatically reduced insulin mRNA levels in R6/2 mice, whereas mRNAs of kinesin 1a and huntingtin-associated protein 1 (HAP1) were unchanged (Fig. 1A). Moreover, the insulin content in R6/2 β-cells was drastically reduced as judged by immunocytochemistry (Fig. 1B). In immuno-electron microscopy micrographs, we found that the number of insulin vesicles was reduced in R6/2 β-cells as previously reported (Fig. 1C) (17). Interestingly, the remaining insulin/proinsulin in the R6/2 β-cells was mainly localized to the Golgi apparatus (Fig. 1D and E). These data indicate that insulin production, vesicular biogenesis and/or vesicle trafficking are impaired in the R6/2 mouse. The decreased insulin mRNA in the R6/2 mouse model is likely due to a transcriptional effect, possibly due to decreased PDXI or p300 as previously demonstrated in this model (18).
To further investigate this phenomenon, we used 832/13 β-cells (INS-1 cells) with adenovirally mediated expression of exon 1 huntingtin containing either 17 (wt) or 69 (mutant) glutamine residues. First, we analyzed insulin secretion in response to glucose. Cells expressing 69Q huntingtin exhibited normal basal release of insulin, while the response to glucose was impaired (Fig. 2A). The fold-change after stimulation was significantly higher in 17Q cells than in cells expressing pathogenic 69Q huntingtin (Fig. 2B). Total insulin/proinsulin levels were comparable as determined by radioimmunoassay (Fig. 2C), indicating that the molecular manipulations did not affect insulin content. Next, using immunofluorescence labeling of insulin/proinsulin and phogrin, an insulin vesicle marker, we assessed the subcellular localization of the hormone. 17Q- and 69Q-expressing cells displayed similar patterns of insulin/proinsulin and phogrin immunoreactivity, a punctuate staining in the periphery and a stronger proinsulin signal from the Golgi apparatus (Fig. 2D). Although we found a slight increase in early apoptosis in cells expressing 69Q huntingtin when compared with 17Q cells (Fig. 2E), the defect in insulin secretion could not be attributed to cell death. When we triple-stained cells with endoplasmatic reticulum (ER), cis-Golgi and trans-Golgi network markers, we found no alterations in ER or Golgi morphology upon huntingtin expression (data not shown).
Impaired intracellular transport
Huntingtin has previously been reported to affect intracellular trafficking in neurons (11,12,14). Therefore, we used vesicular stomatitis viral glycoprotein coupled to a yellow fluorescent protein (VSVG-YFP) to study the transport of protein from the ER to the plasma membrane. The VSVG protein reversibly misfolds at 39°C and therefore remains in the ER. When the temperature is lowered to 33°C, the protein refolds correctly and translocates to the plasma membrane via the Golgi apparatus (19). In both 17Q- and 69Q-expressing cells, the VSVG reached the Golgi apparatus after 15 min at 33°C. In contrast, the transport from the Golgi to the plasma membrane was significantly slower in the cells expressing 69Q huntingtin (Fig. 3A–C).
To verify the transport defects, we studied insulin vesicle transport to and from the trans-Golgi network in two fluorescence recovery after photobleaching (FRAP) experiments. We visualized insulin vesicles by transduction with an islet amyloid polypeptide (IAPP)-EGFP virus. First, the Golgi apparatus was photobleached. The recovery of fluorescence was assessed, monitoring the refilling of the Golgi from the ER. We did not observe any changes in the rate of recovery in 69Q compared with 17Q cells (Fig. 3D), indicating normal protein trafficking between ER and Golgi apparatus. In the second experiment, we performed an inverse FRAP to determine transport from the Golgi apparatus by monitoring fading of the network. We found that the rate of fluorescence fading was significantly slower in the cells expressing 69Q huntingtin (Fig. 3E, P < 0.05), indicating impaired post-Golgi transport.
Insulin vesicle transport and release
The initial phase of insulin secretion is contributed by exocytosis of a ‘release competent’ pool of insulin-containing vesicles situated close to the plasma membrane (20). Sustained insulin release depends on the recruitment of vesicles from a vesicular pool in the interior of the cell, a process dependent on microtubule-associated transport. In order to further dissect the time-dependency of insulin release, we transfected cells with a CD4-phogrin chimeric protein (21). The protein localizes to the insulin vesicles with the CD4 epitope exposed on the luminal surface of the granule. Vesicles, endocytosed after fusing with the plasma membrane, can thus be detected with CD4 antibodies present in the stimulation buffer. We found the number of fused insulin vesicles to be normal in 69Q cells at the early time points, corresponding to the first-phase insulin secretion. However, the expected increase in insulin secretion at the later time points, corresponding to the second or sustained phase of insulin, was blunted in cells expressing mutant huntingtin (Fig. 4A).
To estimate the number and speed of vesicles moving along microtubules, we acquired high-velocity scans of cells transduced with IAPP-EGFP and the different types of huntingtin. We tracked the vesicles in 17Q- and 69Q-expressing cells and analyzed the peak velocities, which reflect directed transport in contrast to the slower random vesicle movements. The fast, directional type of movement is dependent on a normal integrity of the microtubules (22). Indeed, we observed a dramatic reduction in the number of vesicles moving in a directional fashion in the 69Q-expressing cells (Fig. 4B and C; representative vesicle trajectories in a 17Q- and a 69Q-expressing cell are depicted in Fig. 4D and E, respectively). However, the total number of vesicles was not altered by expression of mutant huntingtin (Fig. 4F). These data indicate that microtubule-based transport of insulin vesicles is perturbed.
Aberrant interaction of mutant huntingtin with β-tubulin
HAP1 has been shown to interact with kinesin light chain (8). Furthermore, the poly-Q expansion in the androgen receptor affects kinesin activity negatively via the cJun N-terminal kinase (JNK) pathway (23). Finally, sustained insulin release is dependent on the normal activity of kinesin motors (24). We therefore studied the interaction between huntingtin and the kinesins. Using proximity ligation, we found that there is a close co-localization of both kinesin heavy and light chain and huntingtin (Fig. 5A). However, we found no direct interactions, as monitored by co-immunoprecipitation (Fig. 5B). Next, we studied the effects of inhibiting the JNK pathway by incubating cells in 10 µm of JNK inhibitor SP600125. Despite the beneficial effects of such inhibition reported in neurons (23), we did not observe any positive effects of this drug in 832/13 β-cells expressing 69Q (Fig. 5C).
To elucidate the mechanism underlying impaired post-Golgi transport, we immunoprecipitated mutant and wt huntingtin. We separated the immunoprecipitates by SDS–PAGE and identified interacting proteins by MALDI-TOF mass spectrometry. Apart from previously reported interactions with histones and DNA-binding proteins (15), we identified β5- and α1b-tubulin as direct interactors with mutant huntingtin (>95% certainty; z-values = 1.98. and 2.09, respectively). Using immunoprecipitation followed by immunoblotting for β-tubulin, we verified that mutant exon 1 huntingtin interacts more strongly with β-tubulin than does the wt protein. We estimated the levels of pulled-down β-tubulin in relation to the non-bound β-tubulin in the flow through. Of the total β-tubulin, 0.03% (median value; range 0–0.18%) was pulled down in 17Q cells, whereas 0.32% (median value; range 0.04–1.79%) was pulled down in 69Q huntingtin-expressing cells, indicating a stronger interaction of β-tubulin with the mutant form of huntingtin (Fig. 6A and B). We could not detect any interaction with α-tubulin (data not shown).
To elucidate whether the presence of mutant or wt huntingtin would affect microtubular polymerization or stability, we analyzed the repolymerization of microtubules after nocodazole treatment. We spun cell lysates at 9200g to separate soluble β-tubulin from polymerized tubulin. We found no differences between 17Q and 69Q cells before treatment with nocodazole [median pellet/supernatant ratios; 17Q 1.39 (range 1.26–2.28), 69Q 1.31 (range 1.22–1.71). Mann–Whitney U test P = 0.49]. After nocodazole treatment, we allowed the cells to polymerize microtubules for 15–30 min before analysis. We found that expressing mutant huntingtin did not alter the microtubular dynamics compared with cells expressing wt huntingtin (Fig. 6C, two-way ANOVA genotype P = 0.23, time P = 0.041, no genotype–time interaction).
To further address the interaction between huntingtin and β-tubulin, we performed double-immunofluorescence stainings in 17Q and 69Q cells with antibodies against huntingtin and β-tubulin. In both wt and mutant huntingtin-expressing cells, we observed partial co-localization of huntingtin and β-tubulin (Fig. 7A and B). Tubular structures appeared as typical β-tubulin tubulin (Fig. 7, green panels) in INS-1 cells and fibroblasts as a control (3T3). wt huntingtin appeared more robust in the tips of processes of 17Q cells (arrowheads, in Fig. 7A). Mutant huntingtin frequently formed cytoplasmic aggregates, which were in close contact with β-tubulin (arrows, in Fig. 7B). The data strengthen the notion that huntingtin is associated with β-tubulin.
Huntingtin has previously been suggested to be associated with microtubules (9,10). However, the strength of the interactions has not previously been scrutinized. Here, we have shown that mutant huntingtin binds with a higher affinity to β-tubulin than does the wt protein. This raises the possibility that microtubule-dependent transport is disrupted by mutant huntingtin through a physical block of transport by huntingtin monomers, microaggregates and larger aggregates that stick to the microtubules, making microtubule-dependent transport less effective (Fig. 8). Notably, huntingtin is present in cells at much lower levels than β-tubulin; however, an accumulation over time of mutant huntingtin monomers and aggregates on microtubules may result in an increasing impairment with longer disease duration.
The effects of the huntingtin mutation on vesicular transport of brain-derived neurotrophic factor (BDNF) have been studied previously (11,12). Proposed mechanisms for the observed defects are altered interaction patterns between huntingtin and HAP1/p150glued, and alterations in phosphorylation of huntingtin serine 421 (3,4). However, these proposed mechanisms depend on parts of the huntingtin protein other than that encoded by exon 1. We now show that the huntingtin protein encoded by exon 1, containing the polyglutamine expansion, is sufficient to cause a defect in intracellular transport and insulin release. Clearly, the present and previous mechanisms are not mutually exclusive in full-length models of HD. Trushina et al. (15) report that intracellular transport is defective in neurons of a full-length model of HD (HD72). Mitochondria in this model move slower and stop more frequently (15). More recently, vesicles in Drosophila motor axons have been shown to move at a normal velocity, but to pause more often and for longer intervals in the presence of mutant huntingtin (25). These findings are in line with our hypothesis of a physical block to microtubular transport by mutant huntingtin. Moreover, Trushina et al. (15) show that motor proteins and tubulins interact and associate with mutant huntingtin aggregates in the HD patient brain.
Here, we demonstrate that expression of mutant exon 1 huntingtin in 832/13 β-cells slows intracellular trafficking, reduces the number of rapidly moving vesicles and decreases stimulated insulin release. The defect occurs as early as 48 h following transfection/transduction, and despite normal cellular levels of insulin, suggesting that it is not caused by transcriptional alterations or defects in vesicle biogenesis. We show that the second phase of insulin release is blunted, indicative of an insufficient replenishment of vesicles from the interior of the cell, a process highly dependent on functional microtubular transport in the β-cell (22). Using mass spectrometry, we identified β-tubulin as an interacting partner of mutant huntingtin. Immunostaining showed that β-tubulin partially overlaps with huntingtin and even with the aggregates. By immunoblotting, we show that the β-tubulin interaction is stronger with mutant than with wt huntingtin. However, mutant huntingtin does not affect microtubular dynamics. Therefore, we suggest that mutant huntingtin, by binding to microtubules, becomes a physical block to microtubule-dependent motors, thereby rendering intracellular transport less efficient. We cannot rule out interactions of huntingtin with other proteins being responsible for part of this defect. However, we did not find any evidence of interactions with other microtubular- or exocytosis-related proteins in our mass spectrometry experiment.
We propose an intracellular transport defect mechanism that is dependent on exon 1 huntingtin only. This could be a general phenomenon in cells containing mutant huntingtin, not being restricted to vesicles of specific content, such as insulin granules or BDNF-containing vesicles. Deficient intracellular trafficking could thus explain the propensity of animal models of HD and HD patients for developing diabetes. Although insulin secretion is a functional parameter that may be more sensitive to microtubular dysfunction than neurotransmitter release, neurons are also heavily dependent on intracellular and axonal transport for normal function. In view of this, our results may provide a mechanism for why neurons are prone to be impacted by mutant huntingtin.
MATERIALS AND METHODS
We used the R6/2 mouse strain generated by Mangiarini et al. (16). Genotyping was performed as described (16). The animals were housed with food and water ad libitum under a 12 h light–dark cycle. We conducted all the work involving animals according to rules set by the Ethical Committee for the Use of Laboratory Animals at Lund University, Sweden. In our colony, these mice develop subtle motor deficits at ∼4 weeks of age and reach an end stage of the disease at ∼12 weeks. We studied the mice at 7 weeks or at 12 weeks of age.
Clonal insulin-producing cells and tissue culture
We cultured clonal human insulin-expressing 832/13 INS-1 cells as described (26), at 37°C in a humidified atmosphere containing 95% air and 5% CO2 in RPMI-1640. We supplemented the medium with 10% fetal bovine serum, 100 U/ml penicillin, 100 µg/ml streptomycin, 10 mm HEPES, 1 mm sodium pyruvate and 50 µm β-mercaptoethanol. We transduced cells with an adenovirus containing an HA-tagged exon 1 huntingtin with either 17 or 69 glutamine residues (17Q and 69Q) under the control of the CMV promoter or transiently transfected cells with plasmids with the same huntingtin constructs. We cultured the cells for an additional 48 h prior to the experiments, to assure a robust expression of the transgene.
Insulin mRNA levels
We isolated islets by standard collagenase digestion, and handpicking under a stereomicroscope. We allowed the isolated islets to recover for 2–3 h in the RPMI-1640 medium, containing 10% fetal calf serum and 11.1 mm glucose, in an incubator at 37°C with humidified atmosphere and 5% CO2. We then purified RNA from the islets using an RNeazy MINI Kit (Qiagen) and reverse-transcribed the RNA with the Advantage™ RT-for-PCR Kit (BD Biosciences). Using a semi-quantitative PCR-based low-density array from Applied Biosystems, on the ABI PRISM® 7900 HT Sequence Detection System (Applied Biosystems), we determined mRNA levels. Assays used were insulin (Mm00731595_gH), HAP1 (Mm00468825_m1), kinesin family member 1a (Mm00492863_m1); 18S RNA was used as an internal control (4342379-18S). n = 5 for each genotype.
Prior to analysis, we washed the cells in HEPES balanced salt solution (HBSS; 114 mm NaCl; 4.7 mm KCl; 1.2 mm KH2PO4; 1.16 mm MgSO4; 20 mm HEPES; 2.5 mm CaCl2; 25.5 mm NaHCO3; 0.2% BSA, pH 7.2) supplemented with 2.8 mm glucose for 2 h at 37° C. We then measured insulin secretion by static incubation of the cells for 1 h in 0.8 ml HBSS containing 2.8 or 16.7 mm glucose. We assessed insulin release using the Coat-a-Count Kit (DPC), which recognizes human insulin and cross-reacts ∼20% with rat insulin. We normalized released insulin to the amount of protein in the well as measured with the BCA Protein Assay Kit (Pierce).
JNK inhibition experiments
We cultured INS cells as described earlier and performed insulin secretion in the presence of the JNK inhibitor SP600125 (Calbiochem) or vehicle (DMSO, Sigma-Aldrich) at indicated concentrations in the HEPES buffer during preincubation and stimulation.
We dispersed primary β-cells from isolated islets mechanically into a single-cell suspension and cultured the cells in RPMI-1640 supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 µg/ml streptomycin for 24 h. We then spun the cells gently at 1000g for 10 min and resuspended them in PBS. We placed drops of cell suspension on gelatin-coated slides, allowed them to air-dry before fixation with 4% paraformaldehyde (PFA) in 0.1 m phosphate buffer, pH 7.4, for 10 min. We fixed INS-1 cells in 4% PFA for 30 min. After brief rinsing, we blocked cells in 5% normal donkey serum in 0.1 m PBS containing 0.3% Triton X-100, followed by primary antibody incubation overnight at room temperature (RT) in PBS containing 2% serum, 0.3% Triton X-100. Primary antibodies used were mouse-GM130 (1:250; BD 610822), rabbit-calnexin (1:400; Nordic BioSite, SPA-865), sheep-TGN38 (1:100; Nordic BioSite, T3660), mouse insulin (1:2000; Abcam, ab8304), guinea pig insulin (1:2500; Euro-Diagnostica, B 65-1), rabbit-phogrin (kind gift from Dr T. Takeuchi and Dr S. Torii), mouse-anti-polyglutamine 1C2 (1:1000; Chemicon, MAB1574) and rabbit-β-tubulin (1:500; Cell Signaling, 2146). We used FITC, Cy-3- or -5-labeled secondary antibodies (Jackson Lab; 1:200) for detection in a Leica confocal microscope. For Golgi/insulin co-localization studies, we overlaid the images in Adobe PhotoShop CS, and we toggled the Golgi using the Magic Wand Tool (tolerance 35). The toggled area was cut out of the insulin layer and the image stored (Golgi). The toggled area was then inverted and a new image stored (cell except Golgi). We analyzed intensities using NIH ImageJ.
Immuno electron microscopy
We anesthetized and transcardially perfused animals with 0.5% glutaraldehyde and 1.5% PFA. We then dissected out pieces of the pancreases that were post-fixed for 1 h in 1.5% PFA. We rinsed the tissue pieces in 0.1 m Sörensen's phosphate buffer (pH 7.4) and low-temperature-embedded the tissue pieces in Lowicryl to preserve immunogenicity (27). Using an LKB supernova ultramicrotome, we sectioned the embedded tissue blocks and mounted sections on gold grids. We blocked the sections in 0.1 m PBS, pH 7.4, containing 0.5% BSA before incubation overnight at 4°C with a murine insulin antibody (Abcam, ab8304, 1:200). We incubated the sections in secondary antibodies conjugated to 10 nm colloidal gold (BB International, 1:20) for 1 h at RT. After rinsing in PBS and distilled water, we contrasted the sections in 4% uranyl acetate. We acquired images in a Philips CM-10 electron microscope.
The VSVG-YFP construct was a kind gift from Dr Kay Simons. We double-transfected INS-1 β-cells with VSVG-YFP and either 17Q or 69Q huntingtin plasmids, using Lipofectamine 2000 (Invitrogen). Twenty-four hours after transfection, we transferred the cells to 39°C to accumulate the viral protein in the ER (19). After another 24 h at 39°C, we moved the cells to 33°C for 0, 15 or 120 min before fixation in 4% PFA. We immunostained the cells as described earlier, using a rabbit-HA antibody (1:4000; Abcam, ab 9110) or GM130. We used either Cy-5- or AMCA-conjugated secondary antibodies to avoid any bleed through from the YFP. We captured images in a Leica confocal microscope and further analyzed the images using the NIH ImageJ software. We determined the ratio of yellow fluorescence at the plasma membrane in relation to the intensity in the cytoplasm to get a measure of the effectiveness of the transport from the ER via the Golgi to the plasma membrane.
Fluorescence recovery after photobleaching
We cultured the 832/13 β-cells (INS-1 cells) on poly-l-lysine-coated 35 mm dishes with a coverslip glued to the bottom (P35G-0-10-C, MatTek) before transduction with either 17Q or 69Q huntingtin adenoviruses. Twenty-four hours after huntingtin transduction, we transduced the cells with an adenovirus-expressing islet amyloid polypeptide fused to a GFP tag (IAPP-GFP). We allowed the cells to recover for yet another 24 h before the experiment, ensuring robust expression of both transgenes. We then transferred the cells to an incubation chamber on an inverted Zeiss axiomat confocal microscope, where the cells were maintained at 37°C with 5% CO2 throughout the experiment. We acquired images with a ×63 oil immersion objective, zoom factor ×2. The pinhole was set to 3 airy units. We bleached cells with a maximum of 488 nm laser power for 40 cycles. Images were captured every 500 ms during a period of 200 s. We normalized the results as described previously (11,28).
For examination of the transport from the Golgi complex to the cytoplasm, we used inversed FRAP for the analysis of Golgi destaining. In this experiment, we kept IAPP-GFP-transduced cells at RT for 2 h prior to the experiment, the last hour in the presence of 0.1 mg/ml of cycloheximide. These manipulations aimed at blocking the transport in the Golgi apparatus and at blocking protein translation in the ER. We then imaged the cells in the Zeiss confocal microscope. In this experiment, the whole cell, except for the Golgi apparatus, was bleached with 60 cycles with the 488 nm laser at full intensity, after which we captured images every 6 s for 24 min, using the settings mentioned above. We analyzed and normalized the Golgi intensity according to the procedure of Dundr et al. (29). We analyzed the data until the wt cells reached a background level. We repeated both FRAP and iFRAP experiments in three independent experiments with similar results.
Tracking of vesicles and estimation of the velocity
We acquired images of INS-1 β-cells transduced with IAPP-GFP in a Zeiss confocal microscope. The settings are described earlier, except that the zoom factor was set to ×12. Images (256 × 256 pixels, voxel size 0.047 µm/pixel) were captured every 200 ms for 30 s. Vesicles were tracked manually using NIH ImageJ software. Briefly, we recorded vesicle centroid coordinates from frame to frame and linked them. We calculated the traveled distance, the average speed and the peak velocity taking into account the x and y voxel sizes and the number of frames per second. We recorded trajectories and projected as maximum projection over the movie frames to draw the trajectory over one 2D frame. We then added the outline of the cell as an overlay in Adobe Illustrator. We split vesicle trajectories into spinning vesicles and vesicles moving in a linear trajectory. We analyzed all the vesicles moving linearly in a comprehensive way. We thereafter transformed velocity data by visual banding using SSPS software, ranking peak velocities and binning into speed intervals that were plotted in a histogram (n = 9/genotype).
We cultured INS-1 cells for 24 h in 24-well plates before adenoviral transduction with either 17 or 69Q huntingtin. Twenty-four hours later, we transfected the cells with CD4-phogrin using Lipofectamine 2000 (Invitrogen). Forty-eight hours following transfection, we stimulated with 16.7 mm glucose and 35 mm KCl in HBSS for 0, 1, 5 or 15 min with 0.5 µg/ml mouse anti-human CD4 antibody (Chemicon, CBL127) in the medium. After stimulation, we rinsed the cells in PBS, prior to fixation in 4% PFA. We then permeabilized cells with 0.3% Triton X-100, followed by a Cy-2-conjugated donkey-anti-mouse secondary antibody for the detection of internalized CD4/phogrin. One well of unstimulated cells per condition was kept as a transfection control. These wells were permeabilized and incubated with CD4 antibody to label all CD4-positive vesicles.
We acquired z-stacks (0.5 µm optical increment) throughout the whole cell thickness in a Leica confocal microscope. A person blinded to the cell genotype and time of stimulation counted the number of vesicles in each cell. We normalized vesicle numbers to the non-stimulated transfection control cells. We assessed statistical significance using Student's t-tests. The experiment was repeated three times with similar results.
Cell death assay
To assess cell death induced by expression of our huntingtin constructs, we stained cell cultures with propidium iodide (PI) and annexin V-FITC (Invitrogen). In brief, we rinsed cells once in PBS before a 30 min incubation in PI (1 µg/ml) at 37°C. After the PI, we rinsed cells once in PBS, once in annexin V-binding buffer (Invitrogen) prior to a 15 min incubation in annexin V-FITC (Invitrogen) at RT. We then rinsed cells again in annexin V-binding buffer before fixation in 4% PFA for 20 min at RT. After fixation, we rinsed cells twice and stained with DAPI (Sigma) for 15 min at RT. We photographed random areas of cells, and a person blinded to the genotypes counted the annexin V and PI positive or double-positive cells in the images. We used DAPI to assess the total cell number. We performed five independent experiments with similar results.
For mass spectrometry
We cultured INS-1 cells as described, rinsed in PBS and harvested in homogenization buffer [4 mm HEPES, pH 7.4, 2 mm EDTA, 1% protease inhibitor cocktail (Sigma P8340)]. We sonicated the cells on ice and spun them at 1000g for 10 min. We determined the protein concentration using a Pierce BCA protein assay and used the supernatant for IP. For mass spectrometry, we rinsed 500 µl protein A bead slurry (Millipore 16–125) in PBS and incubated it rotating with 50 µl rabbit HA antibody for 1 h at RT. We then bound the antibodies covalently to the beads by incubation in 20 mm dimethyl-pimelimidate (Sigma) in 200 mm triethanolamine, pH 8.5, for 30 min at RT. After rinses in 200 mm monoethanolamine, pH 8.5, we rinsed the beads in IP buffer (1% Triton X-100, 0.1 m PBS, pH 7.4, 10 mm beta-glycerol phosphate, 10 mm sodium fluoride, 7 mm sodium pyrophosphate, 10 mm EDTA, 400 µm sodium orthovanadate, 20 nm okadaic acid) and transferred them to Poly-Prep chromatography columns (BioRad, 731–1550). We incubated the beads, rotating with 10 mg of INS-1 cell protein overnight at 4°C. We then rinsed the beads in IP buffer before the elution of the bound protein in PBS with increasing NaCl concentrations (0.5–2.5 m) and finally 100 mm glycine, pH 2.7. We precipitated the eluted protein in ethanol before dissolving it in ×2 Laemmli buffer (0.33 m Tris–HCl, pH 6.8, 20% sucrose, 6% SDS, 0.1% bromphenol blue and 3% beta-mecaptoethanol). We separated the protein eluates on 12.5% acrylamide gels, after which we stained the gels using the blue silver Coomassie method (30).
We prepared INS-1 cell samples as described and incubated 500 µg of cell lysate with primary antibody against HA (ab 9110), kinesin heavy chain (H1 or H2) or kinesin light chain (63–90) overnight at 4°C. After we had captured the antibodies using protein A beads (Millipore), we carefully rinsed the beads/antibodies in IP-buffer, followed by increasing concentrations of saline as described earlier. When precipitating the kinesins, we omitted the salt washes. We used beads without primary antibody as negative control. We separated the eluted proteins by SDS–PAGE, followed by blotting onto Hybond-P PVDF membranes (GE Healthcare). We incubated membranes in rabbit-HA (abcam 9110, 1:4000), rabbit β-tubulin (abcam 6046, 1:1000), mouse α-tubulin (abcam 7291, 1:1000), mouse-kinesin heavy chain (H1, 1:1000; kind gift from Dr S. Brady), mouse-kinesin heavy chain (H2, 1:1000; Dr S. Brady) or mouse-kinesin light chain (63–90, 1:1000; Dr S. Brady). We used HRP-conjugated secondary antibodies against mouse or rabbit heavy chains (NA931 or NA934, GE Healthcare, 1:20 000), or in the case of β-tubulin detection, rabbit light chain-specific secondary (211-032-171, Jackson; 1:10 000). We detected the signal using ECL Plus and Hyperfilm ECL-films (RPN2132 and RPN2103K, GE Healthcare).
Sample preparation for MALDI-TOF MS
We excised Coomassie-stained bands on SDS–PAGE gels and washed extensively using 40% acetonitrile in 25 mm NH4HCO3, pH 7.8. After washing, we dried the gel pieces in a SpeedVac before digestion overnight at 37°C using 50–100 ng of sequencing grade trypsin (Promega) in 25 mm NH4HCO3, pH 7.8. We terminated the digestion by the addition of 2 µl 2% TFA, which also extracted the peptides out of the gel. After 1 h extraction at RT, we purified peptides from buffer using miniaturized reversed-phase tips (31). We eluted purified peptides directly onto the sample target. The matrix 2,5-dihydroxybenzoic acid was used on Anchorchip™ targets (Bruker Daltonik), increasing the sensitivity (32).
We identified samples with mass spectrometric studies using a Reflex III MALDI-TOF mass spectrometer (Bruker). The instrument was used in the positive ion mode with delayed extraction and an acceleration voltage of 26 kV. We analyzed peptide samples using accumulation of 50–150 single-shot spectra for improved signal-to-noise ratio. Trypsin autolysis peaks were used for internal calibration. We used the program ProFound (huntingtinp://prowl.rockefeller.edu/prowl-cgi/profound.exe) for identifying proteins in the NCBI database.
In situ proximity ligation assay
For methodological background, see Soderberg et al. (33). In brief, we immunostained cells as described earlier, using mouse-kinesin heavy chain (H1, 1:1000; Dr S. Brady) or mouse-kinesin light chain (63-90, 1:1000; Dr S. Brady) with rabbit-HA (ab9110, 1:4000; Abcam) and guinea pig-proinsulin (ab 9003, 1:2500; Eurodiagnostica). Instead of fluorescent secondary antibodies, to verify a close co-localization of proteins, we used an anti-rabbit, anti-mouse duolink in situ proximity ligation assay (PLA) kit (Olink Biosciences) according to the manufacturer's protocol. After the detection of interacting proteins, we counterstained the cells with Cy-2 anti-guinea pig secondary antibody (Jackson Immuno-Research, Inc.) and DAPI (Chemicon). We captured the images of cells at ×100 magnification in an Olympus epifluorescence microscope.
Microtubule polymerization assay
Forty-eight hours after viral transduction in multi-well plates, we cultured cells in media containing nocodazole (5 µg/ml, Sigma-Aldrich) for 1 h to depolymerize microtubules. We then washed cells with 37°C PBS and cultured the cells in normal medium or medium containing 100 nm paclitaxel (Sigma-Aldrich) for 15 or 30 min. After washing in warm PBS, we lysed the cells in 20 mm Tris–HCl (pH 8.6), 1 mm MgCl2, 2 mm EGTA, 1× PIC, 1 mm orthovanadate and 0.5% Nonidet P-40. We centrifuged lysates at 9168g to pellet polymerized microtubules. We then measured the amount of soluble and polymerized β-tubulin by separating supernatants and pellets, respectively, by SDS–PAGE followed by western blotting. Samples from untreated cells (pre) or cells just treated with nocodazole (0) were used as controls.
We used Mann–Whitney U tests, where group sizes were limited (n < 8) or results not normally distributed; t-tests (two-tailed) were used for larger normally distributed groups. Wilcoxon's signed rank test was used for assessing significance in the insulin secretion assay. Repeated-measures ANOVA was used to address statistical significance in samples where repeated measures were used. We considered statistical significance reached when P < 0.05.
This work was financed by the Swedish Research Council, the Swedish Society for Medicine, the Crafoord Foundation, the Hedlund Foundation, the Greta and Johan Kock Foundation and the Segerfalk Foundation.
We are very grateful for technical assistance from Britt Lindberg, Birgit Haraldsson, Lina Gefors, Ann-Kristin Holmén-Pålbrink and Dr Eric Carlemalm. We would like to thank Dr Kay Simons for supplying us with the VSVG-YFP construct, Dr Frédéric Saudou for making the huntingtin constructs available to us, Dr Seiji Torii and Dr Toshiyuki Takeuchi for providing us with the phogrin antibody and CD4-phogrin construct, Dr Scott Brady and Dr Gerardo Morfini for providing kinesin antibodies and important intellectual inputs. We are thankful for rewarding discussions within and support from NeuroFortis, NeuroNE and the Nordic Center of Excellence on Molecular Mechanisms of Neurodegeneration.
Conflict of Interest statement. None declared.