Protein phosphatase 1 regulates huntingtin exon 1 aggregation and toxicity

Abstract

Huntington’s disease is neurodegenerative disorder caused by a polyglutamine expansion in the N-terminal region of the huntingtin protein (N17). Here, we analysed the relative contribution of each phosphorylatable residue in the N17 region (T3, S13 and S16) towards huntingtin exon 1 (HTTex1) oligomerization, aggregation and toxicity in human cells and Drosophila neurons. We used bimolecular fluorescence complementation to show that expression of single phosphomimic mutations completely abolished HTTex1 aggregation in human cells. In Drosophila, mimicking phosphorylation at T3 decreased HTTex1 aggregation both in larvae and adult flies. Interestingly, pharmacological or genetic inhibition of protein phosphatase 1 (PP1) prevented HTTex1 aggregation in both human cells and Drosophila while increasing neurotoxicity in flies. Our findings suggest that PP1 modulates HTTex1 aggregation by regulating phosphorylation on T3. In summary, our study suggests that modulation of HTTex1 single phosphorylation events by PP1 could constitute an efficient and direct molecular target for therapeutic interventions in Huntington’s disease.

Introduction

Huntington’s disease (HD) is characterized by the loss of medium spiny neurons in the striatum. The main histopathological hallmark of HD is the misfolding and subsequent intracellular aggregation of a mutant form of huntingtin (HTT) (1). HTT is a very large protein (∼350 kDa), but expression of exon 1 is sufficient to produce HD-like features in various cellular and animal models (2–4). HTT exon 1 (HTTex1) contains a polyglutamine (polyQ) tract that, in normal conditions, is constituted by 6–35 glutamine residues. An expansion of the polyQ tract beyond 35 glutamines induces the misfolding and aggregation of mutant HTT and causes HD (5,6). Mutant HTT with longer polyQ expansions is more prone to aggregate, and leads to earlier onset of the disease (2,7,8).

The polyQ tract is preceded by an N-terminal sequence of 17 amino acids (N17 domain) that is highly conserved, suggesting that this domain plays an important role in the function of HTT (9–15). The N17 domain plays a key role in the aggregation pathway of HTT, where the protein associates first into alpha-helical oligomers and acts as a seed to concentrate and facilitate the formation of larger aggregates and fibrils (16–20). Deletion or posttranslational modifications such as phosphorylation, ubiquitination and SUMOylation in the N17 produce striking effects in the stability and aggregation of HTT, as well as in cell viability (21–34). The N17 domain has three phosphorylatable amino acid residues—threonine at position 3 (T3), and serine residues at positions 13 and 16 (S13 and S16). Constitutive phosphorylation of T3 enhances mutant HTT aggregation, but its role in HTT toxicity remains unclear (29). Previous studies have focused on double S13/S16 phosphorylation (30,31,35–38), despite the fact that double S13/S16 phosphorylation is less frequent than single T3 or S13 phosphorylation (39,40), and that overexpression of particular kinases is required in order to achieve double S13/S16 phosphorylation (30).

Single phosphorylation and phosphomimetic modifications in S13 or S16 modulates the formation of HTT fibrils and reduces the oligomerization rate in cell-free systems, just as the double S13/S16 phosphorylation does, both in vitro and in vivo (26,28,30,31). Casein kinase 2 (CK2) inhibitors reduce S13/S16 phosphorylation and enhance toxicity (37), while GM1 ganglioside induces S13/S16 phosphorylation and restores motor and molecular deficits in HD mice (36). IKK, a kinase involved in inflammatory responses, also regulates T3 and S13/S16 phosphorylation and modulates HTT aggregation (30,37,39,41). While several protein phosphatases control HTT dephosphorylation beyond exon 1 and modulate its toxicity (42–46), it is not known which protein phosphatases, if any, regulate phosphorylation of the N17 domain.

Here, we elucidate the contribution of single N17 phosphorylation events towards HTTex1 oligomerization, aggregation and toxicity. We screened a collection of protein phosphatase chemical inhibitors to identify modulators of HTTex1 oligomerization and aggregation in human cells. Inhibition of PP1 prevents HTTex1 aggregation but not oligomerization in human cells. In addition, downregulation of PP1 in Drosophila neurons reduces HTTex1 aggregation and increases its toxicity. In total, our findings point to a critical role of T3 phosphorylation in HTTex1 aggregation and support the targeting of PP1 for therapeutic interventions in HD.

Results

Single N17 phosphomutants modulate HTTex1 aggregation in human cells

In order to investigate the contribution of each phosphorylatable residue within the N17 region (T3, S13 and S16) towards HTTex1 aggregation, we used the bimolecular complementation assay (BiFC) system for the visualization of both oligomeric species and inclusion bodies of HTTex1 in living cells, which we have previously described (47–49). In this system, wild-type (19Q) or disease-causing (97Q) HTTex1 are fused to non-fluorescent halves of the Venus fluorescent protein (Fig. 1A). Thus, upon dimerization of HTTex1 fragments, the Venus halves are brought together and reconstitute the functional fluorophore. Therefore, fluorescence is proportional to the extent of HTTex1 dimerization/oligomerization. We introduced point mutations in each of the phosphorylatable residues within the N17, changing these amino acids to either alanine (A mutants, which cannot be phosphorylated—phosphoresistant) or to aspartic acid (D mutants, which mimic the phosphorylated state—phosphomimic). Importantly, these phosphomutants behave like phosphorylated peptides in terms of their aggregation in cell-free systems and primary neuronal cultures (26,28).

Figure 1.

Single N17 phosphomimic mutations modulate 97QHTTex1 aggregation but not oligomerization in human H4 glioma cells. (A) HTTex1 (grey bars) was fused to two non-fluorescent halves of the Venus protein. When HTTex1 dimerizes, Venus recovers fluorescence. The three N17 phosphorylatable residues were mutated to mimic phosphorylation or dephosphorylation. (B) Cells transfected with phosphoresistant BiFC pairs resemble the non-mutated 97QHTTex1 phenotype in terms of aggregation, while phosphomimic pairs showed a total absence of aggregates. (C) Quantitative analyses of microscopy pictures. *Significant versus 97Q/97Q, P < 0.05. (D) Filter trap assays (FT) were consistent with microscopy results. Phosphomimic pairs did not produce insoluble aggregates (FT), but showed similar levels of oligomeric species (Native-PAGE) comparing to non-mutated 97QHTTex1. Expression levels of each pair were evaluated by SDS-PAGE. Scale bar, 20 µm.

We had previously shown that 97QHTTex1-Venus BiFC pairs oligomerize and aggregate more readily than wild-type 19QHTTex1-Venus BiFC pairs (49). Strikingly, all single phosphomimic mutations of 97QHTTex1-Venus BiFC constructs (T3D, S13D, S16D) completely abolished the formation of inclusion bodies in human cells (Fig. 1B and C) while the phosphoresistant mutants (T3A, S13A, S16A) behaved similarly to non-mutated 97QHTTex1-Venus BiFC pairs (Fig. 1B and C and Supplementary Material, Fig. S1). These phenotypes were further confirmed by filter trap assays, where non-mutated 97QHTTex1 and phosphoresistant pairs appear as large SDS-insoluble aggregates, as opposed to wild-type 19QHTTex1 and phosphomimic pairs [Fig. 1D, filter trap (FT)]. No differences in oligomerization/fluorescence levels were observed between the phosphomutants, and non-mutated 97QHTTex1, as determined by flow cytometry (Supplementary Material, Fig. S2). Native-PAGE analyses confirmed that all phosphomutants formed oligomeric species (Fig. 1D). These results are consistent with a recent report where expanded HTTex1 was shown to exist as tetramers but not monomers (50).

Owing to its dynamic nature, phosphorylation usually affects only a fraction of the total pool of any given protein. Thus, we hypothesized that subpopulations or pools of 97QHTTex1 with different extents of phosphorylation may co-exist in cells. In order to determine whether 97QHTTex1 fragments with different phosphorylation status interacted in living cells, we took advantage of the unique features of our BiFC system to screen all possible pairwise combinations of phosphomutants and non-mutated 97QHTTex1 control (Table 1). Combinations of phosphomimic with non-mutated 97QHTTex1 oligomerized to the same extent as the non-mutated 97QHTTex1 pair, as determined by flow cytometry (Supplementary Material, Fig. S3). However, phosphomimic pairs did not form inclusions, independently of the mutated residue (Table 1). Combinations of phosphomimics with non-mutated 97QHTTex1 (Table 1 and Fig. 2A and B) or phosphoresistant mutants (Table 1) resulted in intermediate aggregation phenotypes, depending on which residue was mutated. Combinations including T3D, S13D or S16D resulted in 0%, 14.1–22.0% or 19.8–29.1% of cells with inclusions, respectively (Table 1), which is significantly less than the percentage of cells with inclusions observed with the non-mutated 97QHTTex1 pair (36.1%). In contrast, phosphoresistant pairs generally resulted in an increased percentage of cells displaying inclusions (43.5–48.5% cells with inclusions), with the exception of the S13A/S16A combination (35.3%) (Table 1).

Table 1.

Percentage of cells containing HTT inclusions (average ± SEM) when co-transfected with different combinations of HTTex1-, α-syn- or Tau-Venus BiFC plasmids

		Venus 2
		T3D	S13D	S16D	T3A	S13A	S16A	97Q	α-syn	Tau
Venus 1	T3D	0	0	0	0	0	0	0	0	15.5±2.2
	S13D		0	0	22.0±3.4	17.1±2.3	14.5±1.2	14.1±1.3	5.6±1.7	15.7±2.2
	S16D			0	27.5±1.5	29.1±3.6	19.8±3.0	22.9±1.0	10.0±4.8	16.5±1.7
	T3A				46.4±1.3	43.5±3.6	45.9±2.5	34.5±2.2	N/D	N/D
	S13A					48.5±1.3	35.3±2.4	37.8±1.9	N/D	N/D
	S16A						48.2±2.6	36.6±6.1	N/D	N/D
	97Q							36.1±2.2	39.3±3.9	16.2±1.3

N/D, not determined.

Figure 2.

Effect of phosphomimic mutants on 97QHTTex1 aggregation is protein-specific. (A) Combinations of 97QHTTex1-Venus 1 BiFC constructs with synuclein- or tau-Venus 2 BiFC constructs revealed different co-aggregation responses to phosphomimic constructs. 97QHTTex1 (B) and synuclein (C) showed similar patterns of aggregation in the presence of single phosphomimic mutations, with T3D being the most restrictive for decreasing aggregation. (D) Tau co-aggregated with 97QHTTex1 regardless of the N17 mutation. *Significant versus 97Q/97Q (B) or 97Q/Syn (C), P < 0.05. Scale bar, 20 µm.

Importantly, the effect of phosphomimic mutants on 97QHTTex1 aggregation appears to be protein-specific (Fig. 2). We and others have recently shown that HTTex1 co-aggregates with alpha-synuclein (α-syn) and Tau, and that these interactions change the aggregation profile of HTTex1 (47,48,51,52). Combinations of phosphomimic 97QHTTex1 with α-syn BiFC constructs resulted in a residue-dependent reduction in the percentage of cells with inclusions, with the T3D/Syn showing no inclusions (Fig. 2A and C). On the other hand, combinations of Tau-Venus plasmids showed the same aggregation pattern independently of the mutated residue (Fig. 2A and D). These observations further support that single N17 phosphomutants modulate HTTex1 aggregation, and that the T3D is the most restrictive modification in preventing HTTex1 aggregation.

Increased dynamics of 97QHTTex1 inclusions containing S13 or S16 phosphomimic mutants

We next analysed the dynamics of 97QHTTex1 inclusions using fluorescence recovery after photobleaching (FRAP). We observed a faster fluorescence recovery in inclusions formed by pairs containing phosphomimic mutants (S13D or S16D) and non-mutated 97QHTTex1 in comparison to inclusions formed exclusively by non-mutated 97QHTTex1 (Fig. 3A and B;Supplementary Material, Videos S1–S3). The phosphoresistant (T3A or S13A)/97QHTTex1 pairs, but not the S16A/97QHTTex1, presented slower fluorescence recovery in inclusions (Fig. 3C). Phosphomimic-containing inclusions often had less defined boundaries, appearing rather continuous with the cytosolic fluorescence, consistent with areas of more dynamic exchange of HTT protein. Inclusions formed by the S13D/97QHTTex1 combination (Fig. 3D and Supplementary Material, Video S4) showed a fluid-like behaviour as opposed to the more rigid behaviour of the inclusions formed by combinations containing phosphoresistant mutants (Supplementary Material, Video S5) or non-mutated 97QHTTex1 BiFC pairs (Supplementary Material, Video S6).

Figure 3.

Phosphomimic-containing inclusions of 97QHTTex1 are more dynamic. (A) Time lapse of FRAP experiments on aggregates containing non-mutated 97QHTTex1 alone or in combination with phosphomimic mutant in H4 glioma cells. T3D mutants produced no aggregates under any circumstance. (B) Aggregates containing S13D or S16D mutants recovered significantly faster than the non-mutated 97QHTTex1. (C) Aggregates combining non-mutated 97QHTTex1 with T3A or S13A recovered more slowly than aggregates made only of non-mutated 97QHTTex1, with the exception of the S16A/97Q combination. (D) Time lapse of aggregates formed by non-mutated 97QHTTex1 and S13D (Supplementary Material, Video S4). Aggregates containing phosphomimic versions are highly dynamic and flexible, and move (arrow indicates moving aggregate in each frame) through the cell with more freedom than non-mutated 97QHTTex1 aggregates (Supplementary Material, Video S1). Scale bars, 20 µm.

Overall, the results suggest that single phosphorylation events within N17 domain prevent HTTex1 aggregation but not its oligomerization, and that phosphorylation at the T3 residue might play a critical role in modulating HTTex1 aggregation. In addition, our observations support the idea that inclusions might be composed of non-phosphorylated HTT or mixtures of non-phosphorylated and phosphorylated pools of molecules, consistent with data indicating that disease-causing HTT is hypophosphorylated in the N17 region (29,37).

Protein phosphatases regulate HTTex1 aggregation in human cells

The results above indicate that single N17 phosphorylation can modulate HTTex1 aggregation, which is consistent with the idea that N17 phosphorylation might be enhanced by the activation of kinases (30,37,39) or, alternatively, by the inhibition of phosphatases. In order to identify protein phosphatases that might mediate HTTex1 dephosphorylation, we screened a library of 33 phosphatase chemical inhibitors for their effect on HTTex1 aggregation and oligomerization, using our 97QHTTex1-Venus BiFC system (Fig. 4 and Supplementary Material, Table S1). We found that inhibitors of PP1/PP2A, CD45, and Cdc25 prevented 97QHTTex1 aggregation as determined by filter trap assay (Fig. 4A). These results were confirmed by fluorescence microscopy, where we observed that inhibitors for those phosphatases significantly decreased the percentage of cells with inclusions (Fig. 4B). Interestingly, the decrease in oligomerization upon treatment with these phosphatase inhibitors was much less dramatic than the reduction observed in aggregation levels (B01, B02, B07, B08 and C04). Although a slight reduction in fluorescence/oligomerization was observed for 28 out of the 33 inhibitors assayed, the few exceptions where a striking decrease was observed (C06, D01 and D03) were owing to cytotoxicity (Supplementary Material, Table S1). Importantly, cells treated with CD45 inhibitors showed reduced levels of HTTex1 expression and increased toxicity (Supplementary Material, Table S1), and were therefore excluded from the study.

Figure 4.

Protein phosphatases regulate HTTex1 aggregation in mammalian cells. (A) Representative filter trap of protein extracts from cells transfected with non-mutated 97QHTTex1 BiFC constructs and treated with chemical inhibitors of protein phosphatases. The graph shows oligomerization levels as determined by flow cytometry (black bars) and aggregation levels as determined by optic density of filter trap dots (white bars). The names of phosphatase inhibitors (Enzo Life Sciences) and the concentrations used are described in Supplementary Material, Table S1. (B) Selected inhibitors PP1/PP2A (B01 and B02) and Cdc25 (C04 and C05) also reduced the percentage of H4 cells with large HTTex1 inclusions as determined by quantitative analyses of microscopy pictures. *Significant versus DMSO, P < 0.05.

Single N17 phosphomutants modulate HTTex1 aggregation in Drosophila

To further investigate the role of N17 phosphorylation on HTTex1 aggregation, we expressed single phosphoresistant or phosphomimic versions of 97QHTTex1 in Drosophila and assessed for the formation of inclusions. We generated flies expressing different phosphomutant versions of 97QHTTex1 fused to mCherry in adult dopaminergic neurons, under the control of tyrosine hydroxylase (TH)-GAL4 (Fig. 5, columns 1–3), or in larval imaginal discs using the eye-specific glass multiple reporter (GMR)-GAL4 driver (Supplementary Material, Fig. S4). Both T3A (Fig. 5B) and T3D (Fig. 5E) mutants formed fewer inclusions than non-mutated 97QHTTex1 (Fig. 5A), while S13D (Fig. 5F) and S16D (Fig. 5G) mutants showed a significantly larger number of aggregates (quantification in Fig. 5O). S13A (Fig. 5C) and S16A (Fig. 5D) showed no difference in the number of inclusions in comparison with non-mutated 97QHTTex1 (Fig. 5O). In the larval eye imaginal discs, all phosphomutants showed decreased aggregation, with the exception of the S13D mutant that formed more inclusions than the non-mutated 97QHTTex1 (Supplementary Material, Fig. S4).

Figure 5.

PP1 knockdown prevents 97QHTTex1 aggregation in Drosophila. Confocal microscopy images of dopaminergic neurons in adult Drosophila brains expressing non-mutated 97QHTTex1-mCherry or the different phosphomutants (A–G) under the control of TH-GAL4. (H–N) Co-expression of PP1α-96A RNAi together with 97QHTTex1-mCherry or phosphomutants. 97HTTex1-mCherry is shown in red (columns 1 and 4), and dopaminergic cells (columns 2 and 5, anti-TH) are shown in blue. T3A (B) and T3D (E) prevented 97HTTex1-mCherry aggregation, while phosphomimic mutations at S13 or S16 produced a significant increase of 97HTTex1-mCherry inclusion bodies (F and G). PP1α-96A RNAi decreased the number of 97HTTex1-mCherry inclusions, with the exception of T3A (I), where no change was observed, and T3D (L) expressing flies, which showed a significant increase in 97HTTex1-mCherry aggregation, all in comparison to the respective no RNAi expressing genotype. (O) Quantification of average number of aggregates (±SEM) in each brain. *Significant versus TH-GAL4>97QHTTex1, #Significant versus their ‘no RNAi’ transgenic control (white bars), ^†significant versus TH-GAL4>97QHTT/PP1α-96A RNAi, P < 0.05. Scale bar, 10 µm.

These findings indicate that expression of N17 phosphomutants can modulate 97QHTTex1 aggregation in Drosophila, depending on the developmental stage and cellular context.

PP1 regulates HTTex1 aggregation and neurotoxicity in Drosophila

In order to test if protein phosphatases also regulate HTTex1 aggregation in Drosophila, we performed RNAi knockdown experiments for the homologues of PP1, PP2A and Cdc25 in flies expressing 97QHTTex1 in dopaminergic neurons (Fig. 5, columns 4–6, and Supplementary Material, Fig. S5). These three phosphatases caused the stronger decrease in HTTex1 aggregation in mammalian cells, upon chemical inhibition. PP1 knockdown flies showed a significant decrease in 97QHTTex1 aggregation (Fig. 5H), while PP2A or string (Cdc25 homologue) downregulation had no effect on 97QHTTex1 aggregation (Supplementary Material, Fig. S5A and B). PP1 downregulation prevented 97QHTTex1 aggregation in the presence of serine-phosphoresistant mutants (S13A or S16A) (Fig. 5J and K), but not T3A, where we did not observe any statistically significant change in the number of aggregates when compared with T3A in the absence of PP1 RNAi (Fig. 5O). These results indicate that the effect of PP1 RNAi on 97QHTTex1 aggregation might be mostly mediated by T3 phosphorylation. On the other hand, PP1 knockdown caused an increased number of aggregates in T3D background, and a reduction in the number of aggregates in S13D and S16D backgrounds (Fig. 5O). Since HTTex1 does not contain any phosphorylatable residue beyond N17 domain, the increased aggregation observed in the T3D background upon PP1 inhibition suggests that S13 and/or S16 might also be target for phosphorylation to modulate 97QHTTex1 aggregation, in our Drosophila model.

We next analysed the effect of PP1, PP2A or string RNAi knockdown on HTTex1 toxicity in the Drosophila eye photoreceptor neurons (Fig. 6 and Supplementary Material, Fig. S5). Importantly, genetic inhibition of PP1 did not compromise rhabdomere viability in flies expressing wild-type (19Q) HTTex1 under the control of rhodopsin 1 (Rh1)-GAL4 [Fig. 6A (column 4) and D]. However, PP1 knockdown significantly enhanced 97QHTTex1 toxicity, further reducing the number of rhabdomeres in an age dependent manner [Fig. 6A–C (column 5) and C]. PP2A or string RNAi had no effect in the number of rhabdomeres upon co-expression with 19QHTTex1 or 97QHTTex1 (Supplementary Material, Fig. S5C and D).

Figure 6.

Downregulation of PP1 potentiates HTTex1 neurotoxicity in adult fly photoreceptor neurons. (A–C) Representative pictures of photoreceptors observed by water immersion live imaging of the retinas. Wild-type flies, HD flies and flies expressing PP1α-87B RNAi together with 19QHTTex1 or 97QHTTex1 were analysed at Day 1, 8 and 15 post-eclosion. Wild-type and 19QHTTex1 expressing flies showed normal retinal morphology as the six outer photoreceptors were visible per ommatidium (columns 1 and 2). Flies expressing 97QHTTex1 exhibited age-dependent neurodegeneration with progressive loss of rhabdomeres from 8 to 15 days after eclosion (B and C, column 3). PP1α-87B downregulation caused an increase of neurotoxicity in flies expressing 97QHTTex1 (B and C, column 5) and did not affect photoreceptor integrity of 19QHTTex1 flies (column 4). Visualization of the rhabdomeres was done using Rh1-Gal4>UAS-GFP^ninaC, as described in Materials and Methods. (D) Quantification of mean rhabdomeres (±SEM) per ommatidium. *Significant versus 97Q, ^#Significant versus 97Q at days 1 and 8, P < 0.05. Scale bar, 10 µm.

Discussion

Protein misfolding and aggregation are intimately involved in the pathogenesis of HD and other neurodegenerative disorders. Despite extensive research in the field, the precise molecular mechanisms by which misfolded proteins aggregate and form toxic species remain elusive. Increasing evidence suggests that targeting phosphorylation events in the N17 domain of mutant HTT can influence the pathological function of the protein (28,30,36–38). However, the significance of single N17 phosphorylation events in HTT oligomerization, aggregation and toxicity is still poorly understood.

Here, we report that single N17 phosphorylation can prevent HTTex1 aggregation, but not oligomerization. We propose that each residue has a different ‘strength’ in modulating HTTex1 aggregation. Importantly, we found that protein phosphatase 1 can control HTTex1 aggregation and toxicity, suggesting that N17 phosphorylation might be mediated by this protein phosphatase. The fact that single N17 phosphorylation events are sufficient to abolish HTTex1 aggregation could be very important from an HD therapeutic perspective, since a single N17 phosphorylation could provide a simpler molecular target than double phosphorylation.

We show that single phosphomimic mutation in T3, S13 or S16 allow 97QHTTex1 to oligomerize but not to form large inclusions in human cells (Fig. 1 and Supplementary Material, Fig. S2). Double S13/S16 phosphorylation prevents aggregation both in vitro and in vivo, and toxicity in vivo (22,26). However, double S13/S16 phosphorylation is less abundant than single phosphorylation events (39), and may require the overexpression of specific kinases (30), which may confound therapeutic efforts based on this approach. Although some studies reported on the effect of single S13 or S16 phosphorylation events on HTTex1 intracellular localization, they did not focus on HTTex1 oligomerization, aggregation or toxicity (28,37). Results from cell-free systems indicate that single S13- or S16-phosphorylated HTTex1 behave similarly to double S13/S16-phosphorylated HTTex1 in terms of aggregation, being unable to form mature fibrils (26). Our findings in living human cells provide additional biological support for those in vitro studies.

The unique properties of our BiFC cellular model allowed us to analyse the relative contribution of each phosphorylatable residue towards HTTex1 oligomerization and aggregation. We found that T3D completely abolished 97QHTTex1 aggregation in the presence of any other phosphomutant or non-mutated 97QHTTex1 molecules, while S13D or S16D only had a partial reduction effect (Fig. 2 and Table 1). These results, together with the higher abundance of T3-phosphorylated pools (39,40), highlight the relevance of this residue in modulating the aggregation of HTTex1, and strongly support T3 as a promising target for HD intervention (29). Our results also suggest that, HTTex1 might be predominantly unphosphorylated under pathological conditions (29,37), as phosphoresistant combinations behave more similarly to the non-mutated 97QHTTex1 pair than to the phosphomimics, regarding their aggregation pattern (Fig. 1 and Table 1).

Previous in vitro studies demonstrate that N17 phosphorylation inhibits β-sheet conformation of mutant HTTex1 and suppresses its fibrillization, stabilizing the α-helical structure of the N17 domain which could led to altered aggregation dynamics (41,53). In FRAP experiments, we show that combinations of phosphomimic and non-mutated 97QHTTex1 induced the formation of inclusions that were more dynamic, diffuse and fluid than regular 97QHTTex1 aggregates, resembling an intermediate stage between oligomers and mature inclusions (Fig. 3, Supplementary Material, Videos S1–S4 and S6). Additionally, non-mutated 97QHTTex1 pairs did not aggregate to the same extent as the phosphoresistant combinations (Fig. 1C and Table 1). Thus, we propose a model where unphosphorylated HTTex1 fragments oligomerize and form inclusions that grow into mature fibrils until enough phosphorylated HTTex1 molecules are intercalated in the structure to interfere with the process, acting as a ‘brake’. This could explain the existence of aggregates of various sizes and morphologies.

Genetic and pharmacological inhibition of PP1 resulted in lower 97QHTTex1 aggregation and increased toxicity (Figs 4–6). Protein phosphatases PP2B/3 (Calcineurin), PP2C and PP1/PP2A have been also shown to regulate HTT phosphorylation at several residues beyond exon 1, as well as HTT toxicity (42–46). It is important to note that HTTex1 does not contain any phosphorylatable residue beyond the N17 domain and, therefore, any direct effect of protein phosphatases should happen on T3, S13 or S16. Our results indicate that PP1 affects 97QHTTex1 aggregation by regulating T3 phosphorylation. In Drosophila dopaminergic neurons, PP1 knockdown in serine-phosphomutant backgrounds leads to a decrease in 97QHTTex1 aggregation, regardless the phosphorylation-like state of these residues (Fig. 5O). Moreover, PP1 RNAi does not cause any further reduction of 97QHTTex1 aggregation when co-expressed with T3A mutant. Together, these data suggest that the effect of PP1 inhibition on HTTex1 aggregation is primarily mediated by T3 phosphorylation. However, an increase in 97QHTTex1 aggregation is observed when PP1 RNAi is co-expressed with T3D mutant (Fig. 5L and O), indicating that S13 and S16 might also be target for PP1 regulation. Since S13D or S16D increase 97QHTTex1 aggregation (Fig. 5F, G and O), we hypothesize that PP1 modulates S13 or S16 phosphorylation upon T3 phosphorylation. In fact, it is likely that different N17 mutations may contribute to subsequent phosphorylation events in other residues. For example, IKK-mediated S16 phosphorylation is facilitated by previous phosphorylation of S13 (30).

Interestingly, the striking decrease in 97QHTTex1 aggregation observed when PP1 RNAi was co-expressed with S13D or S16D mutants versus S13D or S16D alone (no RNAi) (Fig. 5O), suggests that T3 phosphorylation is dominant over S13 or S16 phosphorylation. Our human cell data also supports this hypothesis, since S13D or S16D BiFC combinations with non-mutated 97QHTTex1 still shows aggregates, while T3D BiFC combinations with any other mutant completely abolishes aggregation (Fig. 2 and Table 1).

HTT N17 phosphorylation modulates neurotoxicity in several HD models by preventing HTT aggregation (31). However, this finding remains controversial (29,30,37). Currently, there is still debate on whether large, insoluble aggregates are the cause of neuropathology or a byproduct of accumulated misfolded HTT. Though these aggregates were originally thought to be neurotoxic, accumulating evidence suggests that the formation of large aggregates might be part of a protective cellular mechanism to sequester soluble toxic species (54–57). Our data here suggest that N17 phosphorylation modulation of HTTex1 neurotoxicity and the appearance of large aggregates are not necessarily related. Interestingly, polyQ-dependent toxicity varies widely depending on the cellular environment and cell-type specificity, indicating that additional factors may contribute to HTTex1 toxicity (58). Moreover, studies on other post-translational modifications in the N17 region reveal a complex cross-talk among these different modifications in modulating aggregation, and suggest that neurotoxicity may be owing to different mechanisms dependent upon the modifier site (30,33).

Altered PP1 activity has been linked to a variety of diseases, including several neurodegenerative disorders, such as Parkinson’s and Alzheimer’s diseases (59–61). In fact, PP1 is normally highly expressed in the brain, and is particularly abundant in medium spiny neurons, known to be significantly affected in HD (62). Based on these findings, it is possible that PP1 malfunction increases the susceptibility of certain types of neurons to mutant HTTex1-induced degeneration. Not surprisingly, inhibition of PP1 resulted in increased age-dependent neuronal loss in our Drosophila HD model, suggesting that PP1 activity imbalance might precede HTTex1 neurotoxicity. Despite considerable efforts to develop specific chemical activators of PP1 (63,64), further investigations will be necessary to clarify the relationship between PP1 activity impairment, aggregation and neurotoxicity in the context of HD and other polyQ misfolding disorders.

In summary, our results support a strong role for single N17 phosphorylation events on HTTex1 aggregation, dynamics and toxicity, and uncover the regulatory role of PP1 in these events. Ultimately, our study opens novel avenues for the therapeutic targeting of PP1 and N17 phosphorylation in HD.

Materials and Methods

Cell culture, plasmids and treatments

Human H4 glioma cells (ATCC HTB-148, LGC Standards, Barcelona, Spain) were maintained in OPTI-MEM I culture medium (Gibco, Invitrogen, Barcelona, Spain) supplemented with 10% (v/v) fetal bovine serum and 1% (w/v) of a penicillin/streptomycin commercial antibiotic mixture (Gibco), under controlled conditions of temperature and humidity (37 °C, 5% CO₂). Different types of cell culture dishes were used for cell seeding depending on the application. For flow cytometry and toxicity assays, cells were grown on 6-well plates (35 mm diameter, Techno Plastic Cultures AG, Switzerland). For microscopy, cells were seeded on glass-bottom 35 mm dishes (10 mm glass surface diameter, MatTek Corporation, Ashland, MA, USA). And, for protein extraction (PAGE and filter trap assays), cells were seeded on 100 mm dishes (Techno Plastic Cultures AG, Switzerland). For all experiments, cells were counted and seeded at a density of 10 000 cells/cm² regardless dish size. Generation of HTTex1- and tau-Venus BiFC constructs was described in detail elsewhere (47,49). Alpha-synuclein-Venus BiFC plasmids were a kind gift from Pamela J. McLean (Department of Neurology, Alzheimer’s Disease Research Unit, Massachusetts General Hospital, MA, USA). Phosphomimic (T3, S13 or S16 mutated to aspartic acid) and phosphoresistant (T3, S13 or S16 mutated to alanine) constructs were produced by PCR-based site-directed mutagenesis using 97QHTTex1-Venus plasmids as templates. Plasmid transfection was performed by means of the X-tremeGene 9 reagent (Roche diagnostics, Mannheim, Germany), following manufacturer’s instructions. Twenty-four hours after transfection, cells were collected and analysed for oligomerization, aggregation and toxicity as described below. Pharmacological inhibition of protein phosphatases was performed using a phosphatase inhibitor library (Enzo Life Sciences, Lausen, Switzerland). Briefly, cells were treated with 33 different phosphatase inhibitors upon transfection of 97QHTTex1 BiFC constructs. Phosphatase inhibitors were dissolved in DMSO and added to culture medium at variable concentrations, according to the IC50 described in manufacturer’s instructions (Supplementary Material, Table S1).

Flow cytometry

Cells were washed with Ca²⁺ and Mg²⁺ free phosphate buffer saline (PBS) (Gibco) and collected by trypsinization (0.05% w/v trypsin, 5 min, 37 °C) into BD Falcon Round-Bottom tubes (BD Biosciences, San Jose, CA, USA). Cell pellet was resuspended in PBS and analysed by means of a LSR Fortessa flow cytometer (Beckton Dickinson, Franklin Lakes, NJ, USA). Ten thousand cells were examined per experimental group. The FlowJo software (Tree Star, Inc., Ashland, OR, USA) was used for data analyses and representation.

Fluorescence microscopy and FRAP experiments

Images of living H4 cells were acquired using an Axiovert 200M widefield fluorescence microscope equipped with a CCD camera (Carl Zeiss MicroImaging GmbH, Germany). Pictures of a total of 100–150 cells per sample were scored for aggregate quantification using the ImageJ free software (http://rsbweb.nih.gov/ij/). FRAP experiments were performed using a META LSM 510 confocal microscope. Briefly, protein aggregates were focused at the central focal plane and adjusted to avoid pixel saturation. Experiments lasted for 70–150 s, taking one picture every second. After establishing the basal signal, aggregates were bleached using the 488 nm laser line at 100% laser transmission on a circular region of interest with a diameter of 30 pixels (1.31 µm radius) for 5 s (10 iterations). Fluorescence recovery was then monitored for 60–140 s with LSM software. Images were analysed and prepared for publication by means of the ImageJ free software.

Immunoblotting

Proteins were extracted in native or denaturing conditions according to the requirements of each technique. Briefly, cells were washed with PBS 1× and collected by scraping. Cells were incubated with lysis buffer and sonicated for 10 s at 5 mA using a Soniprep 150 sonicator (Albra, Milano, Italy). For denaturalizing conditions, the lysis buffer was 1% Triton X-100, 150 mM NaCl, 50 mM Tris pH 7.4, supplemented with a protease inhibitor cocktail (Roche diagnostics). For native conditions, the lysis buffer was 173 mM NaCl, 50 mM Tris pH 7.4, 5 mM EDTA, also supplemented with protease inhibitor cocktail. Proteins were collected after cell lysate centrifugation at 10 000 g for 10 min at 4 °C and quantified by means of the BCA Protein Assay Reagent Kit (Thermo Fisher Scientific, Inc., Rockford, IL, USA), following manufacturer’s instructions. For SDS- or Native-PAGE immunoblotting, 15 μg of total protein extracts were prepared and separated by electrophoresis using a 12% SDS-polyacrylamide gel or a 5% SDS-free-polyacrylamide gel, respectively. For denaturing conditions, samples were boiled in standard loading buffer (200 mM Tris–HCl pH 6.8, 8% SDS, 40% glycerol, 6.3% β-mercaptoethanol, 0.4% bromophenol blue) for 5 min at 95 °C. For native conditions, extracts were mixed with SDS- and mercaptoethanol free loading buffer (200 mM Tris–HCl pH 6.8; 40% glycerol; 0.4% bromophenol blue) and the boiling step was omitted. Proteins were transferred onto PVDF membranes and blocked with 5% (w/v) non-fat dry milk in Tris–HCl buffer saline-Tween solution (TBS-T) (150 mM NaCl, 50 mM Tris pH 7.4, 0.5% Tween-20) for 1 h at room temperature. Membranes were incubated with primary antibodies against HTT (1:500, Millipore, Billerica, MA, USA) and GAPDH (1:30 000, Ambion, Austin, TX, USA) as specified. A secondary mouse IgG Horseradish Peroxidase-linked antibody (1:10 000, GE Healthcare Life Sciences, Uppsala, Sweden) was used for 1 h incubation at room temperature. Immunoblots were developed with enhanced chemiluminescence reagents (Millipore) and exposed to X-ray films.

Filter trap assays

Cells were pelleted by centrifugation at 700 g for 10 min and cell lysates collected in native conditions as described above. One hundred micrograms of native protein extracts were mixed with SDS to a final concentration of 0.4% (w/v). Samples were loaded on a dot-blotting device and filtered by vacuum through cellulose acetate membranes (0.22 μm pore; GE Water & Process Technologies, Fairfield, CT, USA), previously incubated with 1% (w/v) SDS solution in PBS. After filtration, membranes were washed two times and processed for immublotting detection of HTT, as described above. In these conditions, only large SDS-insoluble aggregates are retained in the filter and therefore HTT signal is proportional to the presence of large insoluble species. Analyses and quantification of blots signal were performed using ImageJ software. HTTex1 aggregation levels (Fig. 4) were calculated by densitometry analyses and normalized to total HTT expression levels and GAPDH loading control.

Drosophila stocks, genetics and crosses

Flies were maintained at 25 °C and raised on standard cornmeal medium in a light/dark cycle of 12 h. We generated eight constructs encoding for different versions of HTTex1 fused to mCherry: a wild-type version with a polyQ tail containing 19 glutamines, a mutant version with 97 glutamines, and 6 constructs encoding phosphomutant versions (T3A/D, S13A/D and S16A/D) with 97 glutamines. To establish transgenic UAS-HTTex1-mCherry lines, our constructs were cloned into pWalium10-roe by the Gateway cloning technology (Thermo Fisher Scientific), and then injected into y¹w¹¹¹⁸ embryos using phiC31 integrase-mediated DNA recombination (BestGene strain #9723, attP landing site at 2L-28E7). For genetic knockdown experiments in Drosophila, we employed UAS-RNAi-targeted lines of four phosphatases: string (homolog of cdc25 phosphatase, BL34831, HMS00146), PP1α-96A (alpha-1 isoform of PP1 catalytic subunit, BL42641, HMS02477), PP1α-87B (alpha-2 isoform of PP1 catalytic subunit, BL32414, HMS00409), and PP2A-29B (PP2A regulatory subunit, BL43283, HMS01921). RNAi stocks were obtained from the TRiP (Transgenic RNAi Project) library (65,66), courtesy of the Bloomington Drosophila Stock Center (Indiana University, Bloomington, IN, USA). Three different driver lines were used: TH-GAL4 (active in dopaminergic neurons, under the control of the TH promotor), GMR-GAL4 (active in the eye) and Rh1-GAL4 (active in the photoreceptors R1-R6, under the control of Rh1 promotor). To analyse the effect of RNAi downregulation of specific phosphatases upon HTTex1 aggregation and toxicity, UAS-97QHTTmCherry/CyO;THGAL4/TM6B was crossed with UAS-RNAi-targeted lines. Adult flies carrying UAS-97QHTTmCherry/+;TH-GAL4/UAS-stringRNAi, UAS-97QHTTmCherry/+;TH-GAL4/UAS-PP1α-87BRNAi, UAS-97QHTTmCherry/UAS-PP1α-96A;TH-GAL4/+ and UAS-97QHTTmCherry/+;TH-GAL4/UAS-PP2A-29BRNAi were selected and analysed. Unless specified otherwise, flies carrying UAS-lacZ targeted to the same location were used as controls. For deep pseudopupil (DPP) analysis, Rh1-GAL4;UAS-green fluorescent protein^ninaC (GFP^ninaC)/UAS-19QHTTmCherry;UAS-PP1-87BRNAi/+ and Rh1-GAL4;UAS-GFP^ninaC/UAS-97QHTTmCherry;UAS-PP1-87BRNAi/+ adult flies were selected and analysed. Rh1-GAL4;UAS-GFP^ninaC, Rh1-GAL4;UAS-GFP^ninaC/UAS-19QHTTmCherry;UAS-GFP and Rh1-GAL4;UAS-GFP^ninaC/UAS-97QHTTmCherry;UAS-GFP were used as controls.

Immunohistochemistry and confocal microscopy

For confocal imaging of adult brains, 10-day-old flies were dissected and brains prepared as previously described (51). Briefly, adult flies were anesthetized with CO₂ and brains isolated in PBS 1× from the head cuticles before being fixed in 4% paraformaldehyde-containing PBS. Dopaminergic neurons were stained by incubation for 48 h at 4 °C with mouse anti-TH antibody (1:100, Immunostar, Hudson, WI, USA) in PBST (1× PBS, 0.3% Triton X-100) containing 5% (v/v) normal goat serum. Samples were washed three times for 15 min in PBST. Mouse anti-Cy5 (1:200, Jackson ImmunoResearch, West Grove, PA, USA) was diluted in PBST-containing 5% (v/v) normal goat serum and used as secondary antibody by incubation for 24 h at 4 °C. For larva immunohistochemistry, eye imaginal discs of third instar larvae were dissected in PBS 1× and fixed for 1 h in 4% paraformaldehyde as described (67). Samples were then washed three times for 15 min in PBST and incubated overnight at 4 °C with rat anti-Elav antibody (1:100, Developmental Studies Hybridoma Bank, University of Iowa, USA). Imaginal discs were washed three times for 15 min in PBST and incubated for 2 h with rat anti-Cy5 secondary antibody (1:200, Jackson ImmunoResearch). Finally, all samples were submitted to a last step of washing and mounted in 80% glycerol PBS solution, followed by confocal microscopy analyses. Z-stack images were acquired using a LSM 710 Meta Zeiss confocal microscope (resolution of 1024 × 1024, slice thickness of 1 µm, frame average of 2). Z-projections were generated and merged using ImageJ free software and images were prepared on Adobe Photoshop CS6 (Adobe Systems Incorporated, San Jose, CA, USA). For quantification of aggregates, mCherry Z-stack images were analysed by means of the Fiji software (68) and aggregation was measured using 3D Objects Counter plugin (69). The minimum threshold value was defined as 0.144 µm³ to exclude signal from soluble HTT in the count. Volume (µm³) occupied per each aggregate, average number of aggregates and standard error were calculated for at least five imaginal discs or adult brains per genotype.

Live imaging of adult Drosophila eye

DPP analysis was performed in living animals as previously described (51,70). Briefly, flies at age of 1, 8 or 15 days were anaesthetized with CO₂ and then placed on a 50-mm petri dish, previously poured with 2% (w/v) agarose at 40 °C. Once the agarose was solidified, the anesthetized flies were covered with cool water to keep anaesthetic conditions. Adult compound eye integrity of at least five flies per genotype was examined by fluorescence microscopy with a water immersion objective (HC APO L40X/0.80W U-V-I, Leica Microsystems, Wetzlar, Germany). Images were obtained using a Leica DM5500 B microscope (Leica Microsystems) and an Andor Luca R DL-604M camera (Andor Technology Ltd., Belfast, UK). Images were analysed using Image J free software and number of fluorescing rhabdomeres was scored for >15 ommatidia per fly.

Statistical analyses

GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA, USA) or Sigmaplot software (Systat Software, Inc., San Jose, CA, USA) were used to perform the statistical analysis and graphical representation of data. In vitro results are shown as the average ± standard deviation (SD) of at least three independent experiments and Drosophila results as the average ± standard error (SEM), unless specified otherwise. Cell culture data were analysed by means of a one-way ANOVA followed by a post hoc Tukey test for average comparison. Aggregates quantification in Drosophila was analysed by means of a one-way ANOVA with Newman–Keuls post hoc test. DPP assays were analysed by means of a two-way ANOVA followed by a Bonferroni post hoc test. Results were in all cases considered significant only when P < 0.05.

Authors’ Contributions

J.B.S. and F.H. carried out the experiments in mammalian cells. J.B.S., G.M.P. and Y.P.A. performed the Drosophila experiments, under supervision of P.M.D. and F.G. F.H., J.B.S., P.M.D. and T.F.O. wrote the manuscript, which was edited by all authors. J.B.S. and F.H. prepared the figures. F.H. and T.F.O. had the original idea, supervised the work in mammalian cells and coordinated the different teams.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

The authors thank the Bloomington Drosophila Stock Center and the Developmental Studies Hybridoma Bank for fly stocks and antibodies, respectively. The authors also thank Bioimaging Unit from Instituto de Medicina Molecular and Advance Imaging Unit from Instituto Gulbenkian de Ciência for support with imaging.

Conflict of Interest statement. None declared.

Funding

T.F.O. and F.H. were supported by a seed grant from the European Huntington Disease Network (EHDN). T.F.O. is currently supported by the DFG Center for Nanoscale Microscopy and Molecular Physiology of the Brain. P.M.D. was supported by grants FCT-ANR/NEU-NMC/0006/2013, PTDC/NEU-NMC/2459/2014 and IF/00697/2014. F.H. and P.M.D. were also supported by Project LISBOA-01-0145-FEDER-007660 (Cellular Structural and Molecular Microbiology) funded by FEDER funds through COMPETE2020 - Programa Operacional Competitividade e Internacionalização (POCI) and by national funds through Fundação para a Ciência e Tecnologia (Refs SFRH/BPD/63530/2009, IF/00094/2013/CP1173/CT0005 and UID/CBQ/04612/2013). J.B.S. was supported by Fundação para a Ciência e a Tecnologia (Ref SFRH/BD/85275/2012) and FCT/MCTES (Ref CBQ/04612/ICL3535, to F.H. and P.M.D.). F.G. thanks the Medical Research Council (MRC) for funding that provided valuable infrastructure supporting this work.

Author notes

^†Present address: School of Biological Sciences, Monash University, Clayton, Victoria, Australia.

^‡Present address: Department of Oncology, Luxembourg Institute of Health, University of Luxembourg, Esch-Belval, Luxembourg.

^§These authors contributed equally to this work.