Polyglutamine Expansion Reduces the Association of TATA-binding Protein with DNA and Induces DNA Binding-independent Neurotoxicity^,

Abstract

TATA-binding protein (TBP) is essential for eukaryotic gene transcription. Human TBP contains a polymorphic polyglutamine (polyQ) domain in its N terminus and a DNA-binding domain in its highly conserved C terminus. Expansion of the polyQ domain to >42 glutamines typically results in spinocerebellar ataxia type 17 (SCA17), a neurodegenerative disorder that resembles Huntington disease. Our recent studies have demonstrated that polyQ expansion causes abnormal interaction of TBP with the general transcription factor TFIIB and induces neurodegeneration in transgenic SCA17 mice (Friedman, M. J., Shah, A. G., Fang, Z. H., Ward, E. G., Warren, S. T., Li, S., and Li, X. J. (2007) Nat. Neurosci. 10, 1519–1528). However, it remains unknown how polyQ expansion influences DNA binding by TBP. Here we report that polyQ expansion reduces in vitro binding of TBP to DNA and that mutant TBP fragments lacking an intact C-terminal DNA-binding domain are present in transgenic SCA17 mouse brains. polyQ-expanded TBP with a deletion spanning part of the DNA-binding domain does not bind DNA in vitro but forms nuclear aggregates and inhibits TATA-dependent transcription activity in cultured cells. When this TBP double mutant is expressed in transgenic mice, it forms nuclear inclusions in neurons and causes early death. These findings suggest that the polyQ tract affects the binding of TBP to promoter DNA and that polyQ-expanded TBP can induce neuronal toxicity independent of its interaction with DNA.

TATA-binding protein (TBP)3 is required for transcriptional initiation by the three major RNA polymerases (RNAP I, II, and III) in eukaryotic nuclei. As a component of distinct multi-subunit transcriptional complexes, TBP is involved in the expression of most eukaryotic genes (1, 2). Recruitment of the TBP-containing TFIID complex to the TATA box, the AT-rich cognate DNA-binding sequence for TBP that is generally positioned 28–34 nucleotides upstream of the transcription start site (3), is an initial and crucial step in the formation of the preinitiation complex at the promoters of some protein-encoding genes transcribed by RNAP II (3, 4). Further assembly of the preinitiation complex requires TFIIB, a linchpin that directly contacts TBP and the RNAP II-interacting protein TFIIF (4). The presence of the large Mediator complex, which can contribute to preinitiation complex assembly (5) and also connects upstream activator proteins with RNAP II, may be necessary for both basal and activated transcription by RNAP II (6). However, multiple components of the general transcription machinery, including TBP and TFIIB, are also direct targets of transcriptional activators (4).

Unlike the TBP C terminus, which mediates virtually all of the transcriptionally relevant interactions involving TBP hitherto characterized (7) and is highly conserved among eukaryotes (8), the N terminus is evolutionarily divergent and shows sequence conservation only in vertebrates (9). The primary structure of the C terminus consists of two imperfect direct repeats (amino acids 143–202 and 233–293 in mouse TBP) separated by a basic linker region. Each repeat constitutes half of the quasi-symmetrical molecular saddle structure assumed by the TBP C terminus. The concave surface of the saddle interacts with DNA, whereas the convex surface is involved in numerous protein interactions (7). A role for the TBP N terminus, and the polyQ domain in particular, in transcriptional activation at TATA-containing promoters has been suggested previously (10, 11). In humans, the polyQ tract normally contains 25–42 glutamine residues (12). Expanded repeats of >42 glutamines generally result in SCA17 (13–16), although reduced penetrance has been observed in the range of 43–49 glutamines (17).

The clinical phenotype of SCA17 is heterogeneous but often includes ataxia, dementia, and psychiatric symptoms. Epilepsy is also seen in some cases. SCA17 can phenocopy Huntington disease and is alternatively named Huntington disease-like 4, or HDL4 (17–19). Neurodegeneration in SCA17 is frequently widespread but most prominent in the cerebellum. Atrophy of the striatum, thalamus, cerebral cortex, inferior olive, and nucleus accumbens also have been reported in SCA17 (14, 18, 20, 21). Immunohistochemical examination of postmortem brain tissue from SCA17 patients has demonstrated the presence of ubiquinated intranuclear inclusions (14, 20), a neuropathological hallmark of polyglutamine expansion disorders (22).

Given the well characterized transcriptional involvement of TBP, investigation of the effect of polyQ expansion on TBP function could provide important insight into the molecular pathogenesis of SCA17 as well as transcription dysregulation that is known to occur in other polyglutamine diseases (23). To this end, we recently established transgenic mouse models of SCA17 that express polyQ-expanded TBP under the control of the mouse prion promoter (24). These mice show striking neurological symptoms and die at 2–9 months of age. We also found that polyQ expansion enhances the interaction of TBP with TFIIB and reduces the expression of HSPB1, which is important for neuritic integrity and neuronal survival (24). However, the influence of polyQ expansion on the association of TBP with DNA remains to be investigated.

In the current study, we demonstrate that polyQ expansion reduces in vitro binding of TBP to TATA box DNA. We also show that N-terminal TBP fragments, which harbor the expanded polyQ tract but lack an intact C-terminal DNA-binding domain, are present in transgenic SCA17 mouse brains. polyQ-expanded TBP that is incapable of binding DNA formed nuclear inclusions and caused a severe neurological phenotype in transgenic mice. Together, our results indicate that polyQ-expanded TBP is inhibitory to TATA-dependent transcription when it is unable to bind DNA productively and can induce neurotoxicity independent of DNA binding.

EXPERIMENTAL PROCEDURES

Antibodies and Reagents—Polyclonal antibodies against TBP included N-12, which has an N-terminal epitope (Santa Cruz), and EM192. The antigen used to generate EM192 was a glutathione S-transferase fusion protein that contained a C-terminal fragment of TBP (amino acids 204–259 in mouse TBP). Two mouse monoclonal antibodies against TBP, the N-terminal (amino acids 1–20 in mouse and human TBP) antibody 1TBP18 (QED, San Diego, CA) and the polyglutamine-reactive antibody 1C2 (Chemicon), were also utilized in these studies. Other reagents included polyclonal antibodies against TFIIB (C-18, Santa Cruz) and His tag (H15, Santa Cruz) and Hoechst 33258 (Molecular Probes) for nuclear labeling. Full-length and truncated TBP cDNA constructs encoding polyQ tracts of different length (13–105 glutamines) were described previously (24).

Recombinant Protein Purification—His₆-tagged TBPs were generated by subcloning cDNAs from pRK5 vector into pET28a vector. Recombinant proteins were expressed in Escherichia coli BL21 (DE3) by induction with 1 mm isopropyl d-thiogalactopyranoside for 1.5 h at room temperature. Hexahistidine TBPs were purified by nickel-nitrilotriacetic acid-agarose chromatography as described previously (25).

Electromobility Shift Assays (EMSAs)—Double-stranded, 45-bp oligonucleotides containing the adenovirus early 1B TATA box were labeled with [γ-³²P]ATP and T4 polynucleotide kinase. His₆-tagged TBPs were mixed with radiolabeled oligonucleotides (<1 × 10⁵ cpm) in 20 μl of binding buffer (20 mm HEPES, pH 7.9, 80 mm KCl, 5 mm MgCl₂, 0.2 mm dithiothreitol, 0.1 mm EDTA, 5% glycerol, 0.1% Nonidet P-40, 0.5 μg poly(dI-dC), and 0.05 μg of bovine serum albumin) and incubated at room temperature for 30 min. For competition reactions, 100-fold molar excess of unlabeled probe was included. DNA-protein complexes were resolved on a 4% polyacrylamide gel (60:1) in 0.5× Tris borate-EDTA buffer. The gel was dried and subjected to autoradiography.

Cell Culture—Human embryonic kidney (HEK) 293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin.

Cerebellar granule neurons were cultured from postnatal (P7-P9) transgenic SCA17 mice as described previously (24). After culturing for 15 days in vitro, cerebellar neurons were fixed with 4% paraformaldehyde and stained with TBP antibodies.

Reporter Assays—pEGFP-C3 containing the CMV promoter was used to assess activated transcription. For experiments with full-length TBP, HEK293 cells were co-transfected with the reporter and TBP-31Q, TBP-71Q, or empty pRK5. For experiments with truncated TBP, HEK293 cells were co-transfected with the reporter and TBP-13Q-T, TBP-71Q-T, or empty pRK5. Co-transfections were done in triplicate in a 12-well plate using Lipofectamine. The cells were harvested in 500 μl of cold 1× PBS 48 h after transfection, and 50 μl of resuspended cells was combined with 50 μl of 1× PBS in a 96-well black bottom plate. Samples from each co-transfection were plated in duplicate, and green fluorescence was detected with a FLOUstar Galaxy microplate reader (BMG LABTECH, Offenburg, Germany).

Immunostaining—Immunofluorescence and immunohistochemistry were performed as described previously (24). Fluorescent images were acquired on a Zeiss microscope (Axiovert 200 MOT; Carl Zeiss Imaging) equipped with a digital camera (Hamamatsu Orca-100) and Openlab software (Improvision Inc). Both 40× and 63× objectives were used for image acquisition. Immunostaining of mouse brain sections with 1C2 (1:20,000) was performed after treatment with 88% formic acid for 10 min at room temperature.

Western Blotting—Brain tissue samples were homogenized in radioimmune precipitation assay buffer (50 mm Tris-HCl, pH 8.0, 150 mm NaCl, 1 mm EDTA, pH 8.0, 1 mm EGTA, pH 8.0, 0.1% SDS, 0.5% sodium deoxycholate, and 1% Triton X-100), sonicated for 10 s before and after 30 min incubation on a rotating apparatus at 4 °C, and ultimately clarified at 3000 rpm for 5 min. Two hundred micrograms of the supernatant was used for Western blotting. For recombinant proteins and brain lysates, Western blotting was performed using 4–20% Tris-glycine polyacrylamide gels (Invitrogen).

Transgenic TBP-105Q-T Mice—Truncated TBP cDNA containing a 105-CAG/CAA repeat (TBP-105Q-T) was inserted into the XhoI cloning site of an expression vector that utilizes the murine prion promoter (24). The TBP-105Q-T construct was microinjected into the male pronucleus of fertilized oocytes from FVB mice at the Emory Transgenic Facility. Founder mice were identified by PCR analysis of tail DNA using primers that flanked the TBP CAG/CAA repeat: forward, 5′-cca cag cct att cag aac acc-3′; and reverse, 5′-aga agc tgg tgt ggc agg agt gat-3′. Positive founders were backcrossed onto the FVB strain background and also mated with B6CBAF1/J mice. All of the mice were bred and maintained in an animal facility at Emory University in accordance with institutional guidelines.

Statistics—Student's t test and one-way analysis of variance with Tukey's post hoc test were used to determine statistical significance between groups. p < 0.05 was considered significant in all cases.

RESULTS

polyQ Expansion Inhibits the Binding of TBP to DNA—The TBP N terminus is normally antagonistic to the formation of a stable TBP-TATA box complex (26). Because altered TBP conformation induced by expansion of the N-terminal polyQ domain could affect DNA binding, we compared the in vitro interaction of normal and polyQ-expanded TBP with TATA box DNA. Recombinant, His₆-tagged TBPs that contained either 31 or 71 glutamines were expressed in bacteria and purified by nickel chromatography (Fig. 1A). Soluble proteins were incubated with a radiolabeled oligonucleotide that included the adenovirus E1b TATA box (27). Mutant TBP (71Q) shifted less probe than normal TBP (31Q) (Fig. 1B). Western blotting confirmed that the observed difference in DNA binding was not due to aggregation of TBP-71Q under gel shift conditions (Fig. 1C). These results indicate that expansion of the polyQ domain can reduce the intrinsic binding of TBP to a cognate promoter sequence. polyQ-expanded TBP That Is Unable to Bind DNA Inhibits Gene Transcription—Because some TBP mutants that cannot bind DNA efficiently in vitro are still able to facilitate transcription initiation (28), we evaluated the effect of polyQ tract length on TBP-mediated transcription. We previously generated truncated TBP (TBP-T) mutants with an internal deletion (amino acids 204–259 in mouse TBP) in the DNA-binding domain (Fig. 2A; Ref. 24). Specifically, the deletion included the basic linker region and part of the second direct repeat in the C terminus. Because even small deletions in the basic segment are known to compromise the interaction of TBP with TATA box DNA (29), we first tested the DNA binding capability of truncated TBP (71Q-T) with an expanded polyQ (71Q) domain. As expected and unlike full-length TBP-71Q (71Q-F), 71Q-T was unable to bind TATA box DNA in vitro (Fig. 2B).

FIGURE 1.

Polyglutamine expansion inhibits the intrinsic binding of TBP to DNA. A, Coomassie staining of purified His₆-tagged TBPs (31Q and 71Q, arrows) used in EMSAs. B, representative EMSA showing that polyQ-expanded TBP binds less TATA box DNA than normal TBP. Recombinant TBPs (130–220 ng) were incubated with a radiolabeled 45-mer probe containing the adenoviral E1b TATA box, and protein-DNA complexes were resolved on a 4% polyacrylamide gel. C, a Western blot probed with an N-terminal TBP antibody (N-12) demonstrating that equivalent amounts of soluble TBP were used in EMSAs. Note the absence of aggregated TBP-71Q protein in the stacking gel (bracket) after incubation under gel shift conditions.

FIGURE 2.

DNA binding capability of polyQ-expanded TBP determines its effect on TATA-dependent transcription activity. A, schematic structure of full-length (71Q-F) and truncated (71Q-T) His₆-tagged TBPs with an expanded polyQ domain (71Q). hTBP, human TBP; mTBP, mouse TBP. B, representative EMSA showing that TBP-71Q-T, unlike TBP-71Q-F, was unable to interact with TATA box DNA. Binding reactions contained an increasing amount of purified recombinant proteins (150, 225, and 340 ng), and Western blotting with anti-His confirmed that a similar amount of TBP-F and TBP-T was used (lower panel). Control, no TBP added. C, schematic layout of the EGFP reporter used to assess activated, TATA-dependent transcription (top panel). Reporter assays showing the effect of full-length TBP with 31Q or 71Q (middle panel) and truncated TBP with 13Q or 71Q (bottom panel) on activated transcription are presented below. The data are expressed as the means + S.E. (n = 3). *, p < 0.05; **, p < 0.01; ***, p < 0.001.

To assess activated transcription in the presence of mutant TBPs, we used an EGFP reporter gene driven by the CMV promoter with an intact enhancer region. The reporter construct was co-transfected in HEK293 cells with TBP-F or TBP-T containing a normal (13Q or 31Q) or expanded (71Q) polyQ domain (Fig. 2C, top panel). Although TBP-31Q-F mildly stimulated expression from the strong CMV promoter, activation of the promoter by TBP-71Q-F was significantly higher (Fig. 2C, middle panel). These results are consistent with the previous finding that polyQ expansion enhanced TBP-mediated activation of a CREB-dependent reporter gene (30). Importantly, we found that truncated TBPs inhibited the activity of the CMV-EGFP reporter (Fig. 2C, bottom panel). Moreover, polyQ expansion (71Q) significantly enhanced the inhibitory effect of TBP-T on activated transcription. These results suggest that when TBP is unable to bind DNA productively, it may cause toxicity by inhibiting TATA-dependent transcriptional activity.

polyQ-expanded TBP Fragments Lacking an Intact C-terminal DNA-binding Domain Form Nuclear Inclusions in SCA17 Mouse Brains—Proteolysis of several polyQ disease proteins has been shown to generate polyQ-containing protein fragments (31). Accordingly, proteolytic cleavage of polyQ-expanded TBP might generate truncated TBP fragments that lack at least part of the C-terminal DNA-binding domain. To provide in vivo evidence for this possibility, we performed Western blot analysis of brain lysates from SCA17 mice expressing TBP-71Q-F (24) with antibodies against different regions of TBP (Fig. 3A). Western blotting of cerebral cortical tissues with 1C2, which reacts with the N-terminal polyQ domain and has a preference for expanded polyQ, revealed smaller TBP fragments (indicated by asterisk in left panel of Fig. 3B) in SCA17 but not wild-type samples. The small fragments as well as oligomeric TBP-71Q could be detected with a different N-terminal TBP antibody (1TBP18) that reacts with the first 20 amino acids of TBP (Fig. 3B, middle panel). However, an antibody to the TBP C terminus (EM192) did not detect the small fragments on the same blots (Fig. 3B, right panel). Analysis of cerebellar extracts from wild-type and transgenic mice also revealed the specific presence of mutant TBP and its degraded products in TBP-71Q-F samples (Fig. 3C). The specificity of the EM192 antibody was verified by antigen preabsorption (supplemental Fig. S1) and its selective immunoreactivity with transfected TBPs (supplemental Fig. S2). These results support the idea that TBP fragments containing the expanded polyQ domain but lacking an intact C-terminal DNA-binding domain are present in the SCA17 transgenic mouse brain.

FIGURE 3.

polyQ-expanded TBP fragments that lack an intact C terminus are present in SCA17 mouse brains. A, schematic representation of mouse TBP showing the epitopes of different antibodies used to evaluate expression. B, Western blot analysis of lysates from cerebral cortex of a symptomatic TBP-71Q-F mouse (line 16, 7 months of age; Ref. 24) and a wild-type (WT) littermate. Note that the polyglutamine (1C2, left panel) and N-terminal TBP (1TBP18, middle panel) antibodies recognized transgenic TBP (arrowheads) and lower molecular weight TBP bands (asterisks) that were specific to TBP-71Q-F mouse brain samples. The lower bands were not detected by an antibody against C-terminal TBP (EM192, right panel). C, Western blot analysis of lysates from cerebella of TBP-71Q-F mice and wild-type littermates (line 16, 7 months of age; Ref. 24). Arrowhead, full-length transgenic TBP; asterisk, transgenic TBP fragments. Note that the expanded polyQ tract (71Q) retarded the migration of transgenic proteins in the gel and that high molecular weight TBP-71Q-F oligomers could be detected with an N-terminal antibody (1TBP18, bracket) in A and C. The arrows indicate endogenous TBP in all panels.

Next, we examined whether mutant TBP without an intact C-terminal DNA-binding domain could form nuclear aggregates. We transiently transfected TBP-105Q-F in HEK293 cells and performed immunofluorescent labeling with antibodies to N-terminal and C-terminal TBP. Most aggregates that were detected by the antibody against the TBP N terminus were not labeled by the C-terminal TBP antibody (Fig. 4A). However, some transfected cells contained aggregates that could be labeled by both TBP antibodies (Fig. 4B), indicating that the immunoreactivity of the C-terminal TBP epitope is not compromised by aggregate conformation. Rather, these data suggest that the C-terminal epitope generally is not present in TBP-105Q aggregates. To substantiate these findings in vivo, we performed the same immunostaining experiments in cultured cerebellar granule cells from SCA17 mice expressing TBP-105Q-F (24). Whereas the antibody to N-terminal TBP reacted with nuclear TBP inclusions, the antibody to C-terminal TBP only labeled diffuse TBP in granule cell nuclei (Fig. 4C). Together, these results indicate that mutant TBP fragments lacking an intact C-terminal DNA-binding domain can form nuclear inclusions in the SCA17 brain.

FIGURE 4.

Nuclear inclusions in SCA17 cell models contain N-terminal TBP. A, double immunofluorescent staining of HEK293 cells transfected with TBP-105Q-F. Most TBP aggregates (arrows) could be labeled with an N-terminal (1TBP18) but not a C-terminal (EM192) TBP antibody. B, some TBP-105Q-F-transfected cells contained aggregates (arrows) that could be labeled by both N-terminal and C-terminal antibodies. C, representative double immunofluorescent staining of cultured cerebellar granule neurons (15 days in vitro) from TBP-105Q-F mice with antibodies against N-terminal (1TBP18) or C-terminal TBP (EM192). Note that only the antibody against N-terminal TBP labeled nuclear TBP inclusions (arrows). Scale bars, 10 μm.

Truncated, polyQ-expanded TBP Forms Nuclear Aggregates That Sequester TFIIB—We have shown previously that TBP aggregates in cell and mouse models of SCA17 sequester the general transcription factor TFIIB (24). Because truncated TBP fragments may form inclusions in SCA17 brains, we asked whether a TBP double mutant that is unable to bind DNA can still sequester TFIIB in aggregates. Truncated TBP containing a nonexpanded polyQ tract (13Q) did not form obvious aggregates upon overexpression in HEK293 cells (Fig. 5A, top panel). In contrast, transfection of truncated, polyQ-expanded TBPs (71Q-T and 105Q-T) resulted in the formation of nuclear aggregates that could be detected with an antibody to N-terminal TBP (N-12) (Fig. 5A, middle and bottom panels). 1C2 staining did not label truncated TBP aggregates (Fig. 5A, left panels), consistent with the preferential reactivity of this antibody for soluble polyQ (32). Also, unlike the predominantly nuclear localization of full-length mutant TBPs (Fig. 5B; Ref. 24), soluble truncated TBPs with expanded polyQ tracts showed substantial cytoplasmic distribution in transfected cells (Fig. 5, A and B). This mislocalization might be partially attributable to the presence of a predicted nuclear localization signal (Predict-NLS program) (33) in the deleted C-terminal region of truncated TBPs. Importantly, truncated TBPs harboring a polyQ expansion aggregated more extensively than full-length TBPs with expanded polyQ tracts (Fig. 5B).

FIGURE 5.

Truncated, polyQ-expanded TBP that is unable to bind DNA aggregates extensively in transfected cells. A, double immunofluorescent staining of HEK293 cells transfected with truncated TBPs containing polyQ tracts of different length using antibodies to TBP (N-12) and polyglutamine (1C2). Note that anti-TBP detected nuclear TBP aggregates, whereas 1C2 showed only diffuse staining. B, immunofluorescent expression analysis of full-length (TBP-F, upper panel) and truncated (TBP-T, lower panel) TBPs in HEK293 cells. Transfected cells were labeled with an N-terminal TBP antibody (N-12). Note that TBP-T forms more aggregates than TBP-F. Scale bars, 10 μm.

To investigate whether nuclear aggregates formed by truncated TBPs could recruit TFIIB, we performed double immunolabeling of transfected TBP-71Q-T or TBP-105Q-T and endogenous TFIIB in HEK293 cells. The staining showed that truncated TBPs with expanded polyQ tracts sequestered endogenous TFIIB in nuclear aggregates (Fig. 6). Thus, despite its inability to bind DNA in vitro, truncated, polyQ-expanded TBP could still interact aberrantly with TFIIB.

FIGURE 6.

Nuclear aggregates formed by truncated, polyQ-expanded TBP sequester TFIIB. Double immunofluorescent staining of HEK293 cells expressing TBP-71Q-T (upper panel) and TBP-105Q-T (lower panel) with anti-TBP (1TBP18) and anti-TFIIB. Note that endogenous TFIIB localized to nuclear TBP aggregates (arrows). Scale bar, 10 μm.

Transgenic Mice Expressing Truncated, polyQ-expanded TBP Show Nuclear Inclusions and Severe Neurological Symptoms—To examine rigorously whether mutant TBP can induce neurotoxicity when it lacks the ability to bind DNA, we generated transgenic mice expressing TBP-105Q-T under the control of the mouse prion promoter. Similar to our previous efforts to generate SCA17 transgenic mice (24), few positive founders (5 of 50 pups, or 10%) expressing TBP-105Q-T were obtained. Transgenic progeny from three breeding founders showed severe neurological symptoms, including failure to thrive, body weight loss, hyperactivity, clasping, seizure, and tremor, before early death at the age of 3–8 weeks (data not shown). We were unable to maintain the TBP-105Q-T mouse lines because of the severe neurological phenotype and short lifespan.

Immunohistochemical examination of transgenic mouse brains revealed widespread expression of the double mutant TBP transgene (Fig. 7), comparable with that of full-length mutant TBP (TBP-105Q) in SCA17 transgenic mouse brains (24). TBP-105Q-T was intensely labeled by 1C2 after treatment of brain sections with formic acid (Fig. 7), a solvent that can expose the inaccessible 1C2 epitope in polyQ inclusions (34). Notably, nuclear inclusions were detected in various brain regions in TBP-105Q-T transgenic mice (arrows in Fig. 7). The neuronal intranuclear inclusions formed by TBP-105Q-T and the severe neurological symptoms of these transgenic mice support the idea that mutant TBP can induce neurotoxicity without binding to DNA.

FIGURE 7.

Truncated, polyQ-expanded TBP accumulates in nuclei and forms nuclear inclusions in the brains of transgenic mice. Striatal sections (top panel) from wild-type, transgenic SCA17 (TBP-105Q-F; Ref. 24), and truncated TBP (TBP-105Q-T) transgenic mice were labeled with 1C2. Representative 1C2 staining of cerebral cortex, hippocampus, and cerebellum from a TBP-105Q-T transgenic mouse is included (middle panel). The arrows indicate nuclear inclusions, which are conspicuous in the high magnification micrographs (bottom panel). Immunohistochemical analysis was performed on mice at 7 weeks of age. Scale bars, 10 μm.

DISCUSSION

Previous in vivo studies have provided strong evidence that the toxicity of mutant polyQ proteins is a gain-of-function phenomenon (22). Moreover, it was recently shown that polyQ expansion promotes neuropathology in SCA1 by altering native interactions, as opposed to mediating novel interactions, involving the disease protein ataxin-1 (35). Because the function of TBP has been well characterized, SCA17 offers an ideal model for investigation of how polyQ expansion influences the normal function of disease proteins and the contribution of these effects to the molecular pathogenesis of polyQ diseases.

Among the various transcriptionally relevant interactions involving TBP, the binding of TBP to the TATA box is an initial and rate-limiting step for the assembly of the preinitiation complex at the promoters of some RNAP II-transcribed genes (4). Although our recent studies have shown that polyQ expansion enhances the interaction of TBP with its normal interacting partner TFIIB and affects the expression of HSPB1, a small heat shock protein that confers neuronal protection (24), the influence of polyQ expansion on the interaction of TBP with promoter DNA has not been addressed.

In the present study, we show that an expanded polyQ domain can inhibit the intrinsic association of TBP with TATA box DNA in vitro. This finding suggests that the interaction of the TBP C terminus with DNA might be regulated by the length of the N-terminal polyQ domain. Given the variable length of the polyQ tract across the vertebrate phylum and among humans, this regulatory feature could have important implications for normal transcriptional diversity. Alternatively, the decreased interaction of TBP-71Q-F with promoter DNA in vitro may reflect the strong influence of polyQ expansion on protein conformation.

The inhibitory effect of an expanded polyQ domain on the TBP-DNA interaction might be expected to suppress gene expression. However, consistent with an earlier report (30), we found that full-length mutant TBP stimulates a TATA-dependent transcription reporter in cultured cells. Because transcription activity involving TBP crucially depends on its association with other transcription factors (7), our results suggest that simply reducing the intrinsic DNA binding capability of TBP may not be sufficient to inhibit TBP-mediated transcription. Consistent with this idea, microarray experiments using SCA17 mouse brains did not reveal global down-regulation of gene expression (24). Moreover, previous findings indicated that cooperative interaction of TBP with other transcription factors can override the normally antagonistic effect of the N terminus on DNA binding by TBP (28).

Another interesting finding in our study is that truncated, polyQ-expanded TBP, which is unable to bind DNA, appears to be more toxic than full-length TBP with a polyQ expansion. Truncated TBP inhibited TATA-dependent transcriptional reporter activity, and the level of inhibition increased with polyQ tract length. Although overexpressed truncated TBP with a normal repeat could also inhibit transcription reporter activity in vitro, products of normal TBP degradation that are generated at the endogenous level apparently do not form nuclear inclusions and probably are unstable in vivo. Consequently, TBP fragments harboring nonexpanded polyQ tracts would not be pathogenic.

Truncated TBP with expanded polyQ tracts formed more nuclear aggregates than full-length mutant TBP in transfected cells. Because aggregates formed by truncated, polyQ-expanded TBP can also sequester TFIIB, it is likely that mutant TBP becomes more toxic when it is unable to bind DNA productively but retains the ability to associate with other transcription factors. The continued interaction with TFIIB, which can be strengthened by poly-expansion (24), might undermine the assembly of the preinitiation complex and thereby inhibit transcription activity in a dominant negative fashion. The in vivo toxicity of truncated, polyQ-expanded TBP was indicated by the severe neurological phenotype that it caused in transgenic mice. Whereas we previously established SCA17 mice that express full-length mutant TBP (24), we were unable to maintain TBP-105Q-T transgenic mice.

Based on these and previous findings (24), we propose a model for the molecular pathogenesis of SCA17 (Fig. 8). When TBP is soluble, a long polyQ tract may antagonize dimerization, a mechanism of negative autoregulation (36), but enhance interactions with other transcription factors, such as TFIIB, to facilitate TBP recruitment to promoter DNA. The enhanced protein-protein interactions would compensate for the reduced intrinsic DNA binding capability of TBP. However, as is universally observed in the polyQ diseases (22), expanded polyQ causes TBP to misfold and aggregate over time. TBP oligomers or aggregates, which may consist of proteolytic TBP fragments, presumably are unable to bind promoter DNA but can still interact with TFIIB and/or other transcription factors, a situation that could negatively affect gene transcription by reducing the promoter occupancy of the affected proteins.

FIGURE 8.

Model for mutant TBP-mediated transcriptional dysregulation in SCA17. Increased interaction of soluble polyQ-expanded TBP with TFIIB and/or other transcription factors may allow for recruitment of the former to certain TATA-containing promoters. Because this soluble form of TBP is not inherently defective, its recruitment can stimulate TATA-dependent transcriptional activity (upper panel). However, mutant TBP can no longer productively interact with promoter DNA after proteolytic processing and/or misfolding. TBP aggregates can sequester particular transcription factors, such as TFIIB, and thereby reduce their availability at certain promoters. Decreased transcriptional activity is the likely consequence of aberrant interaction of misfolded TBP with basal transcription factors or activator proteins (lower panel).

The finding that mutant TBP can induce neurotoxicity independent of its association with DNA has important implications for the pathogenesis of SCA17. Proteolytic degradation of mutant TBP probably generates small, polyQ-containing fragments that lack a functional C-terminal DNA-binding domain. Accumulation of these mutant TBP fragments in neuronal nuclei could lead to neuropathology. Indeed, we observed small TBP fragments, which were not labeled by an antibody to C-terminal TBP, in SCA17 mouse brains. In support of this finding, most nuclear aggregates formed after transfection of full-length mutant TBP could only be labeled by an antibody to N-terminal TBP. Moreover, N-terminal but not C-terminal TBP was detected in TBP-105Q-F neuronal intranuclear inclusions. Despite lacking any intrinsic DNA binding capability, these polyQ-containing TBP fragments may continue to interact with other transcription factors to cause toxicity.

This toxic consequence could be relevant to the selective neurodegeneration in SCA17. TBP is ubiquitously expressed like other polyQ disease proteins but is probably more critical for cellular function and survival. Consistently, SCA17 is a rare disease (17, 37) and is caused by a relatively small pathogenic range of CAG/glutamine repeats (43–66 units) (38). These facts suggest that limited expansion of the polyQ domain can alter the critical function of TBP to induce severe toxicity. The difficulty of establishing SCA17 transgenic mouse lines is consistent with this idea (24).

It remains largely unclear why polyQ expansion in ubiquitously expressed proteins only affects neuronal viability and causes selective neurodegeneration in each polyQ disease. Because the function of TBP has been well characterized, investigation of SCA17 mouse models should provide insight into this important issue. Because mutant TBP can cause neurotoxicity independent of its essential DNA binding function, its selective toxicity might be due to cell type-specific processing and/or abnormal interactions that involve or affect proteins with restricted expression patterns. Further examination of these possibilities would help elucidate the mechanism(s) underlying selective neurodegeneration in SCA17 and other polyQ diseases.