|Home | About | Journals | Submit | Contact Us | Français|
Huntington’s disease (HD) is caused by polyglutamine expansion (exp) in huntingtin (Htt). The type 1 inositol (1,4,5)-triphosphate receptor (InsP3R1) is an intracellular calcium (Ca2+) release channel that plays an important role in neuronal function. In a yeast two-hybrid screen with the InsP3R1 carboxy terminus, we isolated Htt-associated protein-1A (HAP1A). We show that an InsP3R1-HAP1A-Htt ternary complex is formed in vitro and in vivo. In planar lipid bilayer reconstitution experiments, InsP3R1 activation by InsP3 is sensitized by Httexp, but not by normal Htt. Transfection of full-length Httexp or caspase-resistant Httexp, but not normal Htt, into medium spiny striatal neurons faciliates Ca2+ release in response to threshold concentrations of the selective mGluR1/5 agonist 3,5-DHPG. Our findings identify a novel molecular link between Htt and InsP3R1-mediated neuronal Ca2+ signaling and provide an explanation for the derangement of cytosolic Ca2+ signaling in HD patients and mouse models.
Huntington’s disease (HD) has onset usually between 35 and 50 years with chorea and psychiatric disturbances and gradual but inexorable intellectual decline to death after 15–20 years (Vonsattel and DiFiglia, 1998). Neuropathological analysis reveals selective and progressive neuronal loss in the striatum (Vonsattel et al., 1985), particularly affecting the GABAergic medium spiny striatal neurons (MSNs). At the molecular level, the cause of HD is a polyglutamine (polyQ) expansion (exp) in the amino terminus of huntingtin (Htt), a 350 kDa ubiquitously expressed cytoplasmic protein (HDCRG, 1993; Nasir et al., 1996). A number of transgenic HD mouse models have been generated, which reproduce many HD-like features (Menalled and Chesselet, 2002; Rubinsztein, 2002). Despite significant progress, cellular mechanisms that link the mutation with the disease remain controversial (Tobin and Signer, 2000).
A number of Htt binding partners have been identified in yeast two-hybrid (Y2H) screens with an Htt amino-terminal fragment (Gusella and MacDonald, 1998; Kalchman et al., 1997; Singaraja et al., 2002). Htt-associated protein-1 (HAP1) was the first identified Htt binding partner (Li et al., 1995; Gutekunst et al, 1998; Page at al, 1998). Importantly, the HD-causing polyQ expansion of Htt (Httexp) promotes Htt-HAP1 association (Li et al., 1995, 1998b). In rodents, two HAP1 protein isoforms differing in their carboxy termini are expressed via alternative splicing —HAP1A and HAP1B—both of which bind Htt (Li et al., 1995; Nasir et al., 1998). Only one HAP1 isoform has been identified in humans, and this is most similar to rodent HAP1A (Li et al., 1998b). Association of HAP1 with hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) (Li et al., 2002), the p150Glued subunit of dynactin (Engelender et al., 1997; Li et al., 1998a), and the Rac1 guanine nucleotide exchange factor Kalirin-7/Duo (Colomer et al., 1997) has been discovered in Y2H screens. Targeted disruption of the HAP1 gene in mice results in postnatal death and depressed feeding behavior, suggesting an important role of HAP1 in hypothalamic function (Chan et al., 2002). Despite all these data, the role of HAP1 in neuronal signaling and the pathogenesis of HD remain unclear (Bertaux et al., 1998).
The inositol (1,4,5)-triphosphate receptor (InsP3R) is an intracellular calcium (Ca2+) release channel that plays an important role in neuronal Ca2+ signaling (Berridge, 1998). Three isoforms of InsP3R have been identified (Furuichi et al., 1994). The type 1 receptor (InsP3R1) is the predominant neuronal isoform. Mice lacking InsP3R1 display severe ataxic behavior (Matsumoto et al., 1996), and mice with a spontaneous mutation in the InsP3R1 gene experience convulsions and ataxia (Street et al., 1997), suggesting a major role of InsP3R1 in neuronal function. To identify novel InsP3R1 neuronal binding partners, we performed a Y2H screen of a rat brain cDNA library and isolated neuronal cytoskeleton 4.1N protein (Maximov et al., 2003) and HAP1A. In the present manuscript, we describe biochemical and functional interactions of InsP3R1 with HAP1A and Htt. The discovered association of InsP3R1 with HAP1A and Htt provides additional insight into HAP1 function in the brain and to our knowledge for the first time links neuronal InsP3R1 function with Htt and HD.
We aimed to identify novel proteins that bind to the carboxy-terminal cytosolic region of the InsP3R1. We used the carboxy-terminal region of rat InsP3R1 in the pLexN vector (IC bait, amino acids D2590–A2749) for Y2H screening of a rat brain cDNA library and isolated 16 positive clones. The clones were rescued and tested for their strength of association with the IC bait in a liquid Y2H assay. Three of the 16 isolated clones displayed especially strong interaction with the IC bait. Two of the strongly interacting clones (clones 7 and 8) corresponded to neuronal cytoskeleton protein 4.1N (Maximov et al., 2003). The third strongly interacting clone (clone 11) corresponded to a partial fragment of HAP1A.
To systematically map the HAP1A-interacting domain, a number of InsP3R1 carboxy-terminal fragments were cloned into pLexN vector, and the strength of interaction with clone 11 (partial HAP1A) was measured in a liquid Y2H assay (Figure 1A). From the obtained results we concluded that the minimal InsP3R1 region required for interaction with HAP1A corresponds to amino acids F2627–G2736 (Figure 1A, shaded). Thus, the HAP1A-interacting domain in the InsP3R1 carboxy-terminal region partially overlaps with the minimal 4.1N-interacting region (F2627–R2676; Figure 1A) and largely complements the minimal protein phosphatase 1α (PP1α) binding region (R2731–A2749; Figure 1A) mapped in our previous studies (Maximov et al., 2003; Tang et al., 2003).
Two carboxy-terminal splice variants of HAP1 protein are expressed in the rodent brain (Li et al., 1995). Clone 11 corresponds to a carboxy-terminal portion of the rat HAP1A isoform. To address whether both HAP1 isoforms bind InsP3R1 with similar affinity, we cloned full-length rat HAP1A and HAP1B cDNAs into the Y2H vectors and tested their ability to associate with IC10 bait. We observed strong interaction between IC10 and either full-length HAP1A or clone 11 (Figure 1B). In contrast, the HAP1B isoform did not display any detectable inter-action with IC10. Thus, unique carboxy-terminal sequences present in HAP1A protein (K578–L599) appear to robustly modulate association with the IC10 bait.
To confirm the InsP3R1-HAP1 association, we performed a series of in vitro binding experiments. In the first experiments, we expressed GST-IC8 and GST-IC10 InsP3R1 carboxy-terminal proteins in E. coli and performed GST pull-downs with extracts from COS7 cells transiently expressing hemagglutinin-tagged HAP1 proteins (HA-HAP1A and HA-HAP1B). In agreement with Y2H data, GST-IC10, but not GST-IC8, specifically associated with HA-HAP1A protein (Figure 1C). In further agreement with Y2H results, GST-IC10 association with HA-HAP1B was much weaker than that with HA-HAP1A (Figure 1C).
Do HAP1 proteins bind to full-length InsP3R1? To address this, we expressed full-length InsP3R1 (RT1) in Sf9 cells by baculovirus infection and performed in vitro binding experiments using lysates from COS7 cells transiently transfected with HA-HAP1A- or HA-HAP1B-expressing plasmids. The Sf9 and COS7 cell lysates were incubated together, precipitated with anti-InsP3R1 polyclonal anti-bodies, and blotted with anti-HA monoclonal antibodies. In agreement with Y2H and GST pull-down experiments, full-length InsP3R1 associated more strongly with HA-HAP1A than with HA-HAP1B (Figure 1D).
Since HAP1 interacts with Htt (Li et al., 1995), we investigated whether an InsP3R1-HAP1A-Htt ternary complex might be formed. To explore this, we expressed full-length Htt-23Q or Htt-82Q proteins (Cooper et al., 1998) in HEK293 cells and performed GST pull-down experiments. We discovered that GST-IC10 protein, but not GST-IC8, efficiently precipitated Htt-82Q proteins from HEK293 cell lysates (Figure 1E). Compared to Htt-82Q, much smaller amounts of Htt-23Q were precipitated by GST-IC10 (Figure 1E). The addition of COS7 cell lysates containing overexpressed HA-HAP1A into the GST-IC10 pull-down reaction significantly increased the amount of precipitated Htt-23Q protein (Figure 1E), but had only a minor effect on the amount of precipitated Htt-82Q protein (Figure 1E). In control experiments we detected the expression of endogenous HAP1A protein in HEK293 cells (data not shown). One explanation of our findings is that Htt-82Q bound endogenous HAP1A with much higher affinity than Htt-23Q. As a result, endogenous levels of HAP1A in HEK293 cells were sufficient to support robust Htt-82Q, but not Htt-23Q, precipitation by GST-IC10. Alternatively, Htt-82Q, but not Htt-23Q, may be exhibiting strong and direct binding to the InsP3R1 carboxy terminus, even in the absence of HAP1A (see Discussion).
We performed additional GST pull-down experiments with rat cerebellar and cortical lysates. In agreement with data using cultured cells, GST-IC10, but not GST-IC8, precipitated endogenous Htt from rat brain cerebellar and cortical lysates (Figure 2A). To address whether InsP3R1 and Htt interacted in vivo, we performed coimmunoprecipitation experiments with rat brain cerebellar and cortical lysates using anti-InsP3R1 polyclonal antibodies. In agreement with the pull-down data, endogenous Htt protein is coimmunoprecipitated with the InsP3R1, even in the presence of GST-IC8 protein, which was added in the reaction (Figure 2B). In contrast to GST-IC8, inclusion of GST-IC10 protein into anti-InsP3R1 immunoprecipitation reactions significantly reduced the amounts of precipitated Htt (Figure 2B). These results are consistent with the formation of an InsP3R1-HAP1-Htt ternary complex in neurons within rat cerebellum and cortex regions. Similar results were obtained in experiments with rat striatal lysates (data not shown).
To determine the requirement of HAP1 for the InsP3R1-Htt association, we studied brain extracts from mice with targeted disruption of HAP1 (Chan et al., 2002). In these experiments, brain extracts from HAP1−/− mouse pups and wild-type control littermates were used in immunoprecipitaion experiments with anti-InsP3R1 polyclonal antibodies. The precipitated fractions were blotted with anti-Htt monoclonal antibodies. In agreement with the rat brain coimmunoprecipitation results, we found that anti-InsP3R1 polyclonal antibodies specifically precipitated Htt from wild-type mouse cortical lysates (Figure 2C, left panel, third lane). Slightly reduced amounts of Htt protein were precipitated by anti-InsP3R1 polyclonal antibodies from HAP1−/− cortical lysates (Figure 2C, right panel, third lane). To investigate preferentially strong protein-protein interactions, an additional 0.5 M KCl washing step was employed. The 0.5 M KCl wash did not appreciably decrease the amount of Htt precipitated by anti-InsP3R1 antibodies from wild-type mouse brain extract (Figure 2C, left panel, fourth lane). In contrast, the high-salt wash drastically reduced the amount of Htt precipitated from HAP1−/− brain extracts (Figure 2C, right panel, fourth lane). From these results we concluded that the presence of HAP1 proteins is required for the formation of a high-affinity salt-resistant InsP3R1-Htt association. Western blotting of anti-InsP3R1 immunoprecipitates with HAP1 monoclonal antibodies confirmed the preferential binding of InsP3R1 to the HAP1A isoform (Figure 2D, left panel) and the absence of both HAP1 isoforms in HAP1−/− samples (Figure 2D, right panel). The binding of InsP3R1 to HAP1 was not significantly affected by the 0.5 M KCl wash (Figure 2D, left panel).
The HAP1 binding site and the site of polyQ expansion are localized to the most amino-terminal region of Htt protein (Li et al., 1995). We expressed amino-terminal fragments (Htt-N, amino acids 1–158) of Htt-15Q and Htt-138Q as GST-fusion proteins and confirmed association of Htt-N-15Q and Htt-N-138Q with HA-HAP1A (data not shown). We further demonstrated that bacterially expressed Htt-N-15Q and Htt-N-138Q proteins specifically precipitated recombinant full-length InsP3R1 from Sf9 cell lysates (Figure 3A). Since Sf9 lysates do not contain any HAP1 (data not shown), this Htt-N interaction with full-length InsP3R1 appears to be direct. Our GST-IC10 pull-down experiments (Figure 1E) and immunoprecipitation experiments from HAP1−/− cortical lysates (Figure 2C) also support the existence of a direct Htt-InsP3R1 interaction. InsP3R1 associated with Htt-N-138Q more strongly than with Htt-N-15Q (Figure 3A), although this difference was less dramatic than the difference between Htt-23Q and Htt-82Q in GST-IC10 pull-down experiments (Figure 1E).
What are the functional consequences of InsP3R1 association with HAP1 and Htt? To address this, we expressed InsP3R1 in Sf9 cells and reconstituted recombinant InsP3R1 into planar lipid bilayers as previously described (Tang et al., 2003; Tu et al., 2002). Addition of 100 nM InsP3 to the cis (cytosolic) chamber induced low levels of InsP3R1 activity (Figure 3B, second trace, and Figure 3C). Further addition of GST protein or repetitive additions of Htt-N-15Q protein directly to the bilayer had no effect on InsP3R1 activity (Figure 3B, traces 3–5, and Figure 3C). In contrast, addition of Htt-N-138Q protein to the same bilayer resulted in facilitation of InsP3R1 activity (Figure 3B, trace 6, and Figure 3C). On average, InsP3R1 open probability was equal to 0.020 ± 0.008 (n = 16) in the presence of 100 nM InsP3, 0.02 ± 0.01 (n = 6) after addition of Htt-N-15Q, and 0.19 ± 0.05 (n = 4) after addition of Htt-N-138Q. The facilitation of InsP3R1 activity by Htt-N-138Q was most pronounced at low InsP3 concentrations. When InsP3R1 was activated by 2 µM InsP3, additions of GST, Htt-N-15Q, and Htt-N-138Q proteins did not have a significant effect on InsP3R1 open probability (Figure 3D). From these results, we concluded that Htt-N-138Q, but not Htt-N-15Q, sensitizes InsP3R1 to activation by submaximal doses of InsP3.
To evaluate the role of HAP1 in InsP3R1 activation by Htt, we expressed and purified full-length HAP1A as a GST-fusion protein. Addition of HAP1A to the bilayer had no effect on the activity of InsP3R1 in the presence of 100 nM InsP3 (Figure 4A, third trace, and Figure 4B). In contrast to previous experiments (Figures 3B and 3C), addition of Htt-N-15Q to the bilayer pre-exposed to HAP1A facilitated InsP3R1 activity (Figure 4A, fourth trace, and Figure 4B). The following addition of Htt-N-138Q to the same membrane had an additional potentiating effect (Figure 4A, trace 5, and Figure 4B). Similar results were obtained in the experiments when HAP1A was premixed with Htt-N-15Q or Htt-N-138Q prior to addition to the bilayer (data not shown). On average, InsP3R1 open probability was equal to 0.020 ± 0.008 (n = 16) in the presence of 100 nM InsP3, 0.14 ± 0.04 (n = 5) after addition of HAP1A + Htt-N-15Q, and 0.26 ± 0.11 (n = 5) after addition of HAP1A + Htt-N-138Q. From these in vitro functional experiments and our biochemical data, we concluded that HAP1A facilitates activation of InsP3R1 by Htt, most likely by promoting Htt association with InsP3R1 carboxy terminus (Figure 1E).
Since amino-terminal Htt fragments were used in bilayer experiments shown above, we determined the effect of full-length Htt on InsP3R1 activity. We generated baculoviruses expressing full-length Htt-23Q and Htt-82Q and used them in coinfection experiments with RT1 baculovirus encoding InsP3R1. In immunoprecipitation experiments we found that InsP3R1 and Htt-23Q/82Q formed a complex when coexpressed in Sf9 cells (Figure 5A), which do not contain any detectable HAP1. As such, interaction of the InsP3R1 with Htt23Q/82Q in Sf9 cells appears direct, consistent with our earlier findings. Similar to pull-down data gathered with the GST-Htt-N protein (Figure 3A), binding of Htt to full-length InsP3R1 was only modestly modulated by polyQ expansion.
Microsomes prepared from Sf9 cells coinfected with RT1 and Htt-23Q or Htt-82Q baculoviruses were fused to planar lipid bilayers. No channel activity was observed in control experimental conditions (Figures 5B and 5D, first trace). Addition of 100 nM InsP3 induced only low levels of channel activity in InsP3R1 coexpressed with Htt-23Q (Figure 5B, second trace, and Figure 5C), but resulted in dramatic activation of InsP3R1 coexpressed with Htt-82Q (Figure 5D, second trace, and Figure 5E). Elevation of InsP3 concentration to 2 µM increased activity of InsP3R1 coexpressed with Htt-23Q (Figure 5B, third trace, and Figure 5C) and had no additional effect on activity of InsP3R1 coexpressed with Htt-82Q (Figure 5D, third trace, and Figure 5E). From these experiments we concluded that sensitivity of InsP3R1 to activation by InsP3 is increased by formation of a complex with Htt-82Q full-length protein, but not with Htt-23Q full-length protein.
Striatal MSN neurons are the most selectively and severely affected in HD (Vonsattel et al., 1985). To test the in vivo functional effects of Httexp on InsP3R1, we established primary MSN cultures from E18 embryonic rats (Mao and Wang, 2001). Over 90% of striatal neurons are projection GABAergic MSNs (Gerfen, 1992). A large fraction of cells in our cultures (>90%) were strongly positive for GAD65 marker in immunostaining experiments (data not shown), confirming their identity as MSNs (Chesselet et al., 1993; Mao and Wang, 2001). We transfected MSNs at 20 days in vitro (DIV) with full-length Htt-23Q, Htt-82Q, or Htt-138Q expression plasmids. To identify transfected cells, the Htt plasmids were cotransfected with enhanced green fluorescent protein (EGFP)-expressing plasmid. During transfections, the Htt:EGFP plasmid ratio was kept at 3:1 to ensure that every GFP-positive cell was transfected with Htt-expressing plasmid. In control experiments, MSNs were transfected with the EGFP plasmid alone. Only GFP-positive cells were compared in our analysis of different Htt constructs.
In contrast to striatal interneurons, MSNs abundantly express phospholipase C (PLC)-linked mGluR1/5 receptors (Mao and Wang, 2001, 2002; Tallaksen-Greene et al., 1998). To stimulate InsP3R1-mediated Ca2+ release, we challenged Fura-2-loaded MSN neurons with 10 µM 3,5-dihydroxyphenylglycine (DHPG), a specific mGluR1/5 receptor agonist (Mao and Wang, 2002; Schoepp et al., 1999). To exclude the contribution of N-methyl-D-aspartate receptors (NMDAR) and L-type Ca2+ channels to the observed Ca2+ signals and to simplify the analysis, the imaging experiments were performed in Ca2+-free media containing 100 µM EGTA (see Experimental Procedures for details). The local Ca2+ concentration in these experiments is estimated from the ratio of Fura-2 signals at 340 nm and 380 nm excitation wavelengths as shown by pseudocolor images. Representative data with EGFP, EGFP + Htt-23Q, EGFP + Htt-82Q, and EGFP + Htt-138Q transfected MSN neurons are shown on Figure 6. The transfected cells were identified by GFP imaging (Figure 6, first column, arrows) prior to collecting quantitative Fura-2 340/380 ratio data. We noticed that prior to application of DHPG, the basal Ca2+ levels were slightly elevated in Htt-transfected cells when compared to control cells (Figure 6, second column). On average, the basal 340/380 ratio was equal to 0.43 ± 0.02 (n = 14) for EGFP-transfected cells (Figures 7A and 7I), 0.51 ± 0.02 (n = 18) for EGFP + Htt-23Q, 0.55 ± 0.02 (n = 29) for EGFP + Htt-82Q, and 0.54 ± 0.015 (n = 21) for EGFP + Htt-138Q (Figures 7C, 7E, 7G, and 7I). The basal Ca2+ levels were significantly (p < 0.01, unpaired t test) higher in Htt-transfected MSNs as compared to control MSNs. These results are in agreement with elevated basal Ca2+ levels observed in hippocampal neurons from YAC46 transgenic HD mice (Hodgson et al., 1999).
Ten micromolar DHPG corresponds to a threshold concentration for mGluR1/5 receptor activation in MSN neurons, and only a small response to DHPG application at this concentration was observed in control MSNs transfected with EGFP plasmid alone (Figure 6, first row, and Figures 7A and 7B) and in MSN neurons transfected with Htt-23Q plasmid (Figure 6, second row, and Figures 7C and 7D). In contrast, significant response to 10 µM DHPG was observed in MSNs transfected with Htt-82Q plasmid (Figure 6, third row, and Figures 7E and 7F). Even stronger response to this agonist was observed in MSNs transfected with Htt-138Q plasmid (Figure 6, fourth row, and Figures 7G and 7H). Paired t test analysis revealed that peak 340/380 ratios were significantly (p < 0.001) higher than basal 340/380 ratios in Htt-82Q and Htt-138Q transfected neurons, but not in Htt-23Q and EGFP-only transfected neurons (Figure 7I). Thus, we concluded that overexpression of full-length Httexp sensitizes InsP3R1 to InsP3 in MSN neurons. The effects of Httexp on InsP3R1 function were most pronounced at threshold levels of stimulation, as control, Htt-, and Httexp-transfected MSN neurons responded in a similar manner to 500 µM DHPG (data not shown).
Htt is a substrate for cleavage by caspases, and Htt proteolysis may be an early step in the pathogenesis of HD (Wellington et al., 2000, 2002). Removal of caspase 3 and 6 cleavage sites in the Htt quintuple caspase-resistant mutants [Htt(R)] reduces Httexp proteolysis and toxicity in apoptotically stressed neurons (Wellington et al., 2000). Does sensitization of InsP3R1 by Httexp depend on prior cleavage by caspases? To answer this question, we cotransfected MSN neurons with EGFP + Htt(R)-15Q or EGFP + Htt(R)-138Q and analyzed responses of transfected cells to 10 µM DHPG. We discovered that Htt(R)-138Q, but not Htt(R)-15Q, significantly potentiated DHPG-induced Ca2+ release in MSNs (Figure 7I). Importantly, the potentiating effects of caspase-resistant Htt(R)-138Q and the cleavable form of Htt-138Q were similar (Figure 7I). From these results, we concluded that Httexp does not require proteolytic cleavage by caspases 3 and 6 to have potentiating effects on InsP3R1-mediated Ca2+ release in MSN neurons.
Programmed neuronal death (apoptosis) underlies the symptoms of many neurodegenerative disorders, including Alzheimer’s, Parkinson’s, and Huntington’s disease (Mattson, 2000). Ca2+ plays an important role in neuronal signaling (Berridge, 1998), and perturbed Ca2+ homeostasis is one of the key steps during initiation of the apoptotic program in affected neurons (Mattson and Chan, 2001). Abnormalities in ER-mediated Ca2+ signaling due to a mutation of presenilin 1 have been linked to the development of Alzheimer’s disease (Mattson and Chan, 2001). Several lines of experimental evidence point to a connection between HD and aberrant neuronal Ca2+ signaling. Abnormally high cytosolic Ca2+ levels were detected in CA1 pyramidal neurons from the YAC46 HD mouse model (Hodgson et al., 1999). The lymphoblast mitochondria of HD human patients and brain mitochondria from the YAC72 HD mouse model display abnormal Ca2+ homeostasis (Panov et al., 2002). The derangement in Ca2+ signaling occurs early (Panov et al., 2002) and can therefore be invoked as an initiating event in the pathogenesis of HD, followed by activation of caspases in response to high Ca2+ levels (Juin et al., 1998; Zeron et al., 2002). In addition to promoting apoptosis, increased caspase activity would be expected to catalyze Httexp proteolysis (Wellington et al., 2002), resulting in the generation of toxic amino-terminal Httexp fragments (Hackam et al., 1998). Abnormally high Ca2+ levels would also be expected to promote Httexp proteolysis via calpain-dependent mechanisms (Goffredo et al., 2002; Kim et al., 2001).
What are the molecular mechanisms linking the Httexp mutation with increased Ca2+ levels in affected neurons? Httexp alters mitochondrial Ca2+ homeostasis by directly interacting with the mitochondrial membrane (Panov et al., 2002). While this interaction likely contributes to mitochondrial dysfunction, it remains unclear how intracellular Ca2+ is affected by Httexp at the mitochondria. An alternative molecular mechanism has been suggested by the recent description of a Htt-PSD95-NMDAR complex (Sun et al., 2001) and the ability of Httexp to potentiate NMDAR function (Chen et al., 1999; Sun et al., 2001; Zeron et al., 2002). In addition to influx via NMDAR and voltage-gated Ca2+ channels, neuronal Ca2+ is also largely influenced by release from intracellular stores in the endoplasmic reticulum (ER) upon activation of class 1 metabotropic glutamate receptors (mGluR1/5) (Pin and Duvoisin, 1995). The alterations in ER enzymes that have been observed in HD brains (Cross et al., 1985) might be caused by abnormal Ca2+ release from the ER (Korkotian et al., 1999). Here, we describe a novel mechanism linking Httexp to InsP3R1-mediated Ca2+ release from the ER in striatal MSNs.
We have identified a protein complex that contains InsP3R1, Htt, and HAP1. This complex was discovered through the identification of HAP1A as a binding partner of the InsP3R1 cytosolic carboxy-terminal tail in the Y2H screen (Figures 1A and 1B). In biochemical experiments, we found that Htt directly interacts with the InsP3R1 cytosolic carboxy-terminal tail and that binding to this limited region of InsP3R1 was highly dependent on both the presence of HAP1A and polyQ expansion within Htt (Figure 1E). Htt also binds to full-length InsP3R1, but this interaction is not strongly modulated by polyQ expansion (Figures 3A and and5A)5A) or the presence of HAP1A (Figures 2C, ,3A,3A, and and5A5A).
The effects of Htt and HAP1A on InsP3R1 function are largely consistent with our biochemical analysis. In planar lipid bilayer experiments, the Httexp amino terminus or full-length Httexp, but not wild-type Htt, greatly facilitated InsP3R1 activity at 100 nM InsP3 (Figures 3B, 3C, 5D, and 5E). However, the amino terminus of wild-type Htt facilitated InsP3R1 activity in the presence of HAP1A (Figures 4A and 4B). Thus, InsP3R1 is sensitized to InsP3 in conditions when Htt binds to the InsP3R1 carboxy terminus (Httexp or Htt with HAP1A), but not in conditions when Htt does not bind to InsP3R1 carboxy terminus (Htt alone). From these results we concluded that association of Httexp or Htt-HAP1A with InsP3R1 carboxy terminus sensitizes InsP3R1 to activation by InsP3.
The molecular and cellular basis for selective vulnerability of MSNs in HD remains elusive. One proposed mechanism is centered on the finding that MSNs primarily express the NR1A/NR2B subtype of NMDAR (Kuppenbender et al., 1999; Landwehrmeyer et al., 1995; Zeron et al., 2002). Importantly, Httexp appears to selectively enhance the activity of NR1A/NR2B channels, relative to other NMDAR subtypes, suggesting that increased NMDAR function might be specifically occurring in MSNs of HD patients (Chen et al., 1999; Sun et al., 2001; Zeron et al., 2001). Our results suggest another possible explanation for selectivity in HD involving the enrichment of mGluR5, a member of the group I mGluRs, in MSNs (Kerner et al., 1997; Mao and Wang, 2001, 2002; Tallaksen-Greene et al., 1998). Stimulation of group I mGluR in MSNs leads to the generation of InsP3 and release of Ca2+ via InsP3R1 (Mao and Wang, 2002). In addition, stimulation of group I mGluR is known to potentiate NMDAR activity in neurons, most likely via a PKC-dependent pathway (Calabresi et al., 1999; Pisani et al., 2001; Skeberdis et al., 2001).
In striatal MSNs, the presence of full-length Httexp sensitizes Ca2+ release in response to subthreshold concentrations of the mGluR1/5 agonist DHPG (Figures 6 and and7).7). We propose that sensitizing influences of Httexp on InsP3R1 and the NR1A/NR2B subtype of NMDAR (Chen et al., 1999; Sun et al., 2001; Zeron et al., 2001) have a synergistic effect on glutamate-induced Ca2+ signals in MSNs of HD individuals (Figure 8). In MSNs of HD patients, the activation of both mGluR5 and NMDAR by low concentrations of glutamate released by corticostriatal projection neurons leads to supranormal Ca2+ signals (Figure 8B) when compared to normal MSNs (Figure 8A), resulting in neuronal dysfunction and apoptosis. Overall, our findings provide support for the hypothesis that perturbation of neuronal Ca2+ signaling mediated by NMDAR and InsP3R1 may be a primary cause of MSN selectivity in HD. In our model, the toxic “gain of Httexp function” (Tobin and Signer, 2000) corresponds to the polyQ-dependent ability of Httexp to sensitize NMDAR and InsP3R1-mediated Ca2+ signals in MSN neurons (Figure 8B). In addition, our model points to mGluR5 as potential target for pharmacological treatment of HD.
The carboxy-terminal region of rat InsP3R1 (Mignery et al., 1990) (amino acids D2590–A2749) was amplified by PCR and cloned into pLexN vector (IC bait). The yeast two-hybrid screen of a rat brain (P8-P9) cDNA library (kind gift of Dr. T. Südhof) with the IC bait and liquid yeast two-hybrid assays were performed as previously described (Maximov et al., 2003).
The following rat InsP3R1 (Mignery et al., 1990) baits in pLexN vectotor were generated by PCR (listed by encoded amino acid and residue numbers): IC = D2590–A2749, IC_G2736X = D2590–G2736, IC3 = R2676–A2749, IC4 = Q2714–A2749, IC5 = D2590–Q2714, IC6 = D2590–R2676, IC7 = D2590–L2646, IC8 = D2590–F2627, IC9 = L2646–A2749, IC10 = F2627–A2749. The InsP3R1 expression constructs in pGEX-KG are GST-IC8 = D2590–F2627 and GST-IC10 = F2627–A2749. The full-length clones of rat HAP1A and HAP1B (Li et al., 1995) were amplified by RT-PCR from rat brain mRNA and cloned into the pVp16-3 yeast two-hybrid prey vector, pCMV-HA mammalian expression vector, and pGEX-KG bacterial expression vector. Full-length Htt plasmids Htt-23Q (HD-FL-23Q) and Htt-82Q (HD-FL-82Q) in pRc/CMV expression vector were kindly provided by Dr. Christopher A Ross (Cooper et al., 1998). The full-length Htt-15Q and Htt-138Q plasmids (Wellington et al., 2000) were cloned into the pCI expression vector (Promega). The caspase 3 and 6 resistant Htt quintuple mutant plasmids Htt(R)-15Q and Htt(R)-138Q in pRc/CMV vector have been previously described (Wellington et al., 2000). The Htt-N expression constructs in pGEX-KG are Htt-N-15Q/138Q = M1–K158 of human Htt.
GST-IC8 and GST-IC10 proteins were expressed in the BL21 E. coli strain and purified on glutathione-agarose beads. HA-HAP1A and HA-HAP1B were expressed in COS7 cells by DEAE-dextran transfection (Sambrook et al., 1989). Htt-23/82Q proteins were expressed in HEK293 cells by calcium-phosphate transfection (Sambrook et al., 1989). 48 hr after transfection, COS7 or HEK293 cells were collected with ice-cold PBS and solubilized for 30 min at 4°C in extraction buffer A (1% CHAPS, 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4 [pH 7.2], 5 mM EDTA, 5 mM EGTA, and protease inhibitors). Extracts were clarified by centrifugation for 20 min at 100,000 × g and incubated for 1 hr at 4°C with GST-IC8 or GST-IC10 proteins. Beads were washed three times with the extraction buffer A. Attached proteins were analyzed by Western blotting with anti-Htt or anti-HA monoclonal antibodies.
Full-length rat InsP3R1 (RT1)-encoding baculoviruses were previously described (Tu et al., 2002). Full-length Htt plasmids Htt-23Q and Htt-82Q (Cooper et al., 1998) were subcloned into pFastBac1 vector (Invitrogen), and Htt-23Q/82Q baculoviruses were generated using Bac-to-Bac system (Invitrogen). The Spodoptera frugiperda (Sf9) cells (100 ml) were infected with RT1 viruses or coinfected with RT1:Htt-23Q/82Q viral mixtures (3:1) and collected by centrifugation 72 hr postinfection. The RT1-infected cells were solubilized in extraction buffer A, cleared by centrifugation (100,000 × g), and mixed with an equal volume of HA-HAP1A or HA-HAP1B-expressing COS7 lystates prepared as described above for 2 hr at 4°C. The mixture was precipitated with anti-InsP3R1 polyclonal antibody (T443) attached to protein A-Sepharose beads and analyzed by Western blotting with anti-HA monoclonal antibodies. The cells coinfected with RT1 and Htt-23Q/82Q baculoviruses were used to prepare microsomes for bilayer experiments as previously described (Tu et al., 2002). Obtained microsomes were solubilized in extraction buffer A, cleared by centrifugation (100,000 × g in TL-100), used in immunoprecipitation experiments with anti-InsP3R1 polyclonal antibodies, and analyzed by Western blotting with anti-Htt monoclonal antibodies.
Rat and mouse brain tissues were isolated, homogenized, and solubilized for 1.5 hr at 4°C in extraction buffer A. The lysate was clarified by 20 min centrifugation at 100,000 × g and utilized in pull-down experiments with GST-IC8 and GST-IC10 proteins or in the immunoprecipitation experiments with anti-InsP3R1 polyclonal antibodies (T443) performed as described above. GST-IC8 and GST-IC10 proteins (200 µg/ml final concentration) were included in the immunoprecipitation reactions as indicated. An additional 0.5 M KCl wash step was included in the immunoprecipitation experiments with mouse samples as indicated in the text. The precipitated fractions were analyzed by Western blotting with anti-Htt and anti-HAP1 monoclonal antibodies.
Single-channel recordings of recombinant InsP3R1 (RT1) expressed in isolation or coexpressed with Htt-23Q/82Q proteins were performed as previously described (Tang et al., 2003; Tu et al., 2002) at 0 mV transmembrane potential using 50 mM Ba2+ (trans) as a charge carrier. The cis (cytosolic) chamber contained 110 mM Tris dissolved in HEPES (pH 7.35), 0.5 mM Na2ATP, pCa 6.7 (0.2 mM EGTA + 0.14 mM CaCl2) (Bezprozvanny et al., 1991). InsP3R1 were activated by addition of 100 nM InsP3 or 2 µM InsP3 (Alexis) to the cis chamber as indicated in the text. GST, Htt-N-15Q, Htt-N-138Q, and HAP1A proteins were expressed in BL21 E. coli, purified on gluthathione beads, eluted with reduced glutathione, dialyzed overnight against cis recording buffer (110 mM Tris/HEPES [pH 7.35]), and added in 1 µl volume (0.3 mg/ml protein with addition of 0.02 mM ruthenium red) directly to the cis side of the bilayer containing InsP3R1 without stirring. Exposure of InsP3R1 to the test proteins was terminated 2–3 min after addition by stirring the cis chamber for 30 s (1:3000 dilution of test protein stocks). The InsP3R1 single-channel currents were amplified (Warner OC-725), filtered at 1 kHz by a low-pass eight-pole Bessel filter, digitized at 5 kHz (Digidata 1200, Axon Instruments), and stored on computer hard drive and recordable optical discs. For off-line computer analysis (pClamp 6, Axon Instruments), currents were filtered digitally at 500 Hz. For presentation of the current traces, data were filtered at 200 Hz.
The rat medium spiny neuronal (MSN) cultures on poly-D-lysine (Sigma) coated 12 mm round glass coverslips were established by following published procedures (Mao and Wang, 2001). The 5 µM of cytosine arabinoside (AraC, Sigma) was added at 2–4 DIV to inhibit glial cell growth. At 20 DIV the MSN cultures were transfected by the calcium-phosphate method (Maximov and Bezprozvanny, 2002) with EGFP-C3 plasmid (Clontech) or a 1:3 mixture of EGFP:Htt plasmids as indicated in the text. 48 hr after transfection, the MSN neurons were loaded with 5 µM Fura2-AM (Molecular Probes) in artificial cerebrospinal fluid (ACSF) (140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 2 mM CaCl2, 10 mM HEPES [pH 7.3]) for 45 min at 37°C. For imaging experiments the coverslips were mounted onto a recording/perfusion chamber (RC-26G, Warner Instrument) maintained at 37°C (PH1, Warner Instrument), positioned on the movable stage of an Olympus IX-70 inverted microscope, and perfused with ACSF media by gravity flow. Following GFP imaging, the coverslip was washed extensively with Ca2+-free ACSF (omitted CaCl2 from ACSF and supplemented with 100 µM EGTA). In Ca2+ imaging experiments the MSN cells were intermittently excited by 340 nm and 380 nm UV light (DeltaRAM illuminator, PTI) using a Fura-2 dichroic filter cube (Chroma Technologies) and 60× UV-grade oil-immersed objective (Olympus). The emitted light was collected by an IC-300 camera (PTI), and the images were digitized by ImageMaster Pro software (PTI). Baseline (6 min) measurements were obtained prior to bath application of 10 µM or 500 µM 3,5-DHPG (Tocris) dissolved in Ca2+-free ACSF. The DHPG solutions were prewarmed to 37°C before application to MSNs. Images at 340 and 380 nm excitation wavelengths were captured every 5 s and shown as 340/380 image ratios at time points as indicated. Background fluorescence was determined according to manufacturer’s (PTI) recommendations and subtracted.
The following monoclonal antibodies were used: anti-HA (HA.11) from Covance, anti-Htt (mAB2166) from Chemicon International, Intramonoclonal anti-HAP1 1B6 is a kind gift of Dr. Claire-Anne Gutekunst (Chan et al., 2002), GAD65 antibodies from BD Pharmingen. Polyclonal anti-InsP3R1 T443 antibody was previously described (Kaznacheyeva et al., 1998). Secondary HRP-conjugated anti-rabbit and anti-mouse antibodies were from Jackson Imunoresearch.
We thank Thomas C. Südhof for advice with yeast two-hybrid screen and the gift of a rat brain cDNA library and Phyllis Foley for administrative assistance. We thank Thomas C. Südhof for rat InsP3R1 cDNA, Claire-Anne Gutekunst for HAP1 monoclonal antibodies, and Christopher A Ross for HD-FL-23Q and HD-FL-82Q plasmids. I.B. is supported by the Robert A. Welch Foundation, the Huntington’s Disease Society of America, the Hereditary Disease Foundation, and NIH R01 NS38082. M.R.H. is supported by the Canadian Institutes of Health Research, the Hereditary Disease Foundation, and the Huntington’s Disease Society of America and holds a Canada Research Chair in Human Genetics.