Distinct cellular toxicity of two mutant huntingtin mRNA variants due to translation regulation

Huntington’s disease (HD) is a neurodegenerative disorder caused by CAG repeat expansion within exon1 of the HTT gene. The gene generates two mRNA variants that carry either a short or long 3′ untranslated region (3′UTR) while encoding the same protein. It remains unknown whether the two mRNA variants play distinct roles in HD pathogenesis. We found that the long HTT 3′UTR was capable of guiding mRNA to neuronal dendrites, suggesting that some long-form HTT mRNA is transported to dendrites for local protein synthesis. To assay roles of two HTT mRNA variants in cell bodies, we expressed mRNA harboring HTT exon1 containing 23x or 145x CAGs with the short or long 3′UTR. We found that mutant mRNA containing the short 3′UTR produced more protein aggregates and caused more apoptosis in both cultured neurons and HEK293 cells, compared with mutant mRNA containing the long 3′UTR. Although the two 3′UTRs did not affect mRNA stability, we detected higher levels of protein synthesis from mRNA containing the short 3′UTR than from mRNA containing the long 3′UTR. These results indicate that the long HTT 3′UTR suppresses translation. Thus, short-form mutant HTT mRNA will be more efficient in producing toxic protein than its long-form counterpart.


Introduction
Huntington's disease (HD), an inherited autosomal dominant neurodegenerative disorder, is characterized by motor disturbance, cognitive loss, and psychiatric manifestations [1]. It is caused by an unstable CAG repeat expansion within exon1 of the HTT gene that encodes huntingtin protein [2]. Corresponding to the CAG genetic expansion, there is an abnormally long polyglutamine (polyQ) tract within the N-terminal region of mutant huntingtin (mHtt). The mHtt exhibits toxic properties that lead to dysfunction and death of neurons, primarily in striatal and cortical areas [3]. The N-terminal fragment of mHtt, which is encoded by HTT exon1, has attracted much attention because it is generated in vivo and can induce HD-like pathology when expressed in mice [4,5].
The expansion of the polyQ tract in mHtt causes misfolding of mHtt and formation of protein aggregates in neuronal nuclei and neuropils in HD patients [5,6]. Once the mHtt concentration a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 reaches a specific level, mHtt monomers initiate formation of dimers, trimers, oligomers, and finally large aggregates [7]. Although it remains unclear if mHtt aggregates are harmful or beneficial, the aggregates reflect accumulation of misfolded mHtt [8]. Misfolded mHtt interferes with a wide range of cellular functions by interacting with both nuclear and cytoplasmic proteins [9][10][11][12].
Many mammalian genes produce multiple transcripts that differ in the length of 3 0 untranslated region (UTR) while encoding the same protein due to alternative polyadenylation [13]. In fact, more than half of all human genes have multiple polyadenylation sites [13,14]. The relative abundance of mRNA isoforms with alternative 3 0 UTRs is highly tissue-specific [15,16]. Long 3 0 UTRs contain additional regulatory elements that can regulate subcellular location of mRNA [16,17], mRNA translation [18][19][20], and subcellular protein localization [21]. Due to the difference in translational location and efficiency, mRNA isoforms with different lengths of 3 0 UTRs have distinct biological functions although they encode the same protein [16,[21][22][23]. The rodent and human HTT genes also produce two populations of mRNA species, one with a short 3 0 UTR (0.64 kb) and the other with a long 3 0 UTR (3.87 kb), and the relative abundance of these two HTT mRNA isoforms varies among tissues [24,25]. It is unknown whether these two HTT mRNAs species have distinct roles in the HD pathogenesis.
In this study we provided evidence that the long HTT 3 0 UTR guides mRNA into neuronal dendrites for local protein synthesis. Although the two forms of HTT mRNA had similar distribution and stability in cell bodies, we found that the short form of HTT mRNA was translated more efficiently than the long form. This led to more protein aggregates and higher apoptosis rate in cells expressing mutant HTT mRNA with the short HTT 3 0 UTR than those with the long HTT 3 0 UTR.

HEK293 cell culture and transfection
HEK293 cells (ATCC) were grown in Dulbecco's modified Eagle's medium (DMEM) (Cellgro) supplemented with 10% fetal bovine serum (Hyclone) and 100 U/ml penicillin/streptomycin (Invitrogen) at 37˚C and 5% CO 2 . For transfection, 2 μl of Lipofectamine 2000 (Invitrogen) and plasmid DNA (0.4 μg/kb) were added to 100 μl of DMEM media, respectively and incubated at room temperature for 5 minutes. The two parts were then mixed and incubated at room temperature for an additional 20 minutes before the mixture was added onto HEK 293 cells.

DNA constructs
To detect the subcellular distribution of endogenous HTT mRNAs in rat neurons, we cloned a segment of the HTT coding sequence for production of riboprobes through PCR amplification of rat cDNAs. The sequences of PCR primers are as follows: 5 0 -GATT GAATTC cactgctggacagattccga-3 0 (forward) and 5 0 -GATT CTCGAGactggtatgatgtggtatcacc-3 0 (reverse). The italic part was a random sequence, GAATTC was Eco RI restriction site, and CTCGAG was Xho I restriction site.
To generate constructs pExon1Q23-Myc-A, pExon1Q145-Myc-A, pExon1Q23-Myc-A Ã B, and pExon1Q145-Myc-A Ã B for expression of Myc-tagged exon1-encoded N-terminal fragment of huntingtin, we digested plasmids pcDNA3.1-Exon1Q23-Myc and pcDNA3.1-Ex-on1Q145-Myc (Cat #: CH00036 and CH00046, Coriell Institute for Medical Research) with Pme I and then inserted either the short or long human HTT 3 0 UTR. In pExon1Q23-Myc-A Ã B and pExon1Q145-Myc-A Ã B, the first polyadenylation site was mutated from the sequence TTTAAA to GAATTC, such that the two constructs produce only long 3 0 UTR HTT mRNA. For live cell imaging, the Kaede-containing constructs were created by digesting these four Exon1 constructs with Xho I and Age I to replace the Myc-encoding sequence with the Kaedeencoding sequence, which was PCR amplified with pKaede-N1 (A gift from Michael Davidson, Addgene plasmid #54726) [26] as the template using XhoI-and AgeI-flanked primers: 5 0ctcgagatatgagtctgattaaaccagaaatgaag-3 0 (forward) and 5 0 -accggtttacttgacgttgtccggc-3 0 (reverse).
To generate lentiviral constructs for expression of Myc-tagged and exon1-encoded huntingtin fragment, we digested the pUltra vector (A gift from Malcolm Moore, Addgene plasmid #24129) [27] with AgeI and XmaI to remove the 0.8-kb GFP sequence, made the linearized vector have blunt ends, and then inserted the blunt-ended fragment of Exon1Q23 (or Q145)-Myc released from pcDNA3.1-Exon1Q23 (or Q145)-Myc with BamHI and PmeI, generating pUltra-Exon1Q23 (or Q145)-Myc. To generate pUltra-Exon1Q23 (or Q145)-Myc-A construct, we digested pUltra-Exon1Q23 (or Q145)-Myc with BamHI and SalI, and then inserted a sequence that encodes the short HTT 3 0 UTR with compatible restriction sites. To add the long HTT 3 0 UTR to pUltra-Exon1Q23 (or Q145)-Myc, we digested the plasmid with EcoRI and SalI and then inserted a sequence that encodes the long HTT 3 0 UTR with compatible restriction sites. Because the first polyadenylation site was mutated from the sequence TTTAAA to GAATTC, which accidently created an EcoRI restriction site, the final constructs lacked the sequence for the short 3 0 UTR and were termed pUltra-Exon1Q23 (or Q145)-Myc-B.

Production of lentivirus
Lentivirus was produced using a standard transient calcium phosphate transfection protocol [28]. In brief, 5 x 10 6 HEK293 cells were plated in each 10 cm dish coated with 100 μg/ml of PDL. When the culture reached 80-90% confluence, cells were transfected with 1.41 μg/kb/ dish of a lentiviral construct, 7.34 μg/dish of pMDL (packaging plasmid expressing gag and pol), 3.96 μg/dish of pVSVG (packaging plasmid expressing envelop protein VSVG) and 2.84 μg/dish of pREV (packaging plasmid expressing Rev). Sixteen hours post-transfection, medium was replaced with fresh DMEM supplemented with 2% FBS and 1% glutamate. Then every 24 hours, media containing lentivirus were collected, and viral particles were concentrated by precipitation with 50% polyethylene glycol 6000 following centrifugation. The virus pellet was resuspended in Neural Basal medium, aliquoted, and stored at −80˚C. The titer of a viral preparation was determined on HEK293 cells after a serial dilution.

Cortical neuron culture
Primary cortical neurons, isolated from E18.5 Sprague Dawley rat embryos, were cultured at a density of 2.0×10 5 cells per well in 12-well plates, according to procedures described previously [16]. Briefly, all embryos were removed after timed pregnant rats were euthanized using CO 2 . Isolated cortices were digested with 20 U/ml of papain in Hank's Balanced Salt Solution (HBSS) at 37˚C for 20 min. Dissociated cells were plated onto 15-mm-diameter coverslips coated with poly-D-lysine (100 μg/ml) in DMEM supplemented with 10% FBS and 100 U/ml penicillin/streptomycin at 37˚C and 5% CO 2 . Three to four hours after plating, media was replaced with Neurobasal media (Invitrogen) supplemented with 2% B27, 0.5mM L-glutamine, 0.125 mM glutamate, and 1% penicillin-streptomycin. At 5 days in vitro (DIV5), lentivirus was added to culture, and neurons were cultured for another 20 days before being fixed for subsequent immunocytochemistry. All animal procedures were approved by the Scripps Florida Institutional Animal Care and Use Committee (protocol # 13-003).

Fluorescence in situ hybridization
Fluorescence in situ hybridization (FISH) of primary rat cortical neurons and HEK293 cells on coverslips was performed using DIG-labeled riboprobes and the TSA Plus Fluorescein System (PerkinElmer) according to previously described procedures [16]. After hybridization, cells were treated with RNase A at 37˚C to remove unannealed probes. Probe concentrations were 500 ng/ml for HTT mRNA and 100 ng/ml for GFP mRNA.

RNA isolation and RT-qPCR
HEK293 cells were transfected with pExon1Q145-Myc-A and pExon1Q145-Myc-A Ã B. At 24h post transfection, total RNA was isolated using TRIzol Reagent (Invitrogen) according to manufacturer's instruction. To remove any contaminating genomic and plasmid DNA, total RNA was treated with RNase-free DNase for 30 minutes at 37˚C, followed by phenol/chloroform extraction. cDNA was synthesized in the presence with oligo(dT) and a reverse primer complementary to short HTT 3 0 UTR using M-MuLV Reverse Transcriptase (NEB, Cat # M0253). Real-time PCR was performed using the Power SYBR Green PCR master mix (Roche, Cat # 04913914001) with specific gene primers, and gene expression was presented relative to 18S rRNA using 2 −ΔΔCt method. The sequences of real-time PCR primers are as follows: 5 0 -GAGG ATCTGAATATGCATACCGG-3 0 (forward) and 5 0 -CATAGGGACCAAGCTGGC-3 0 (reverse) for the target gene; 5 0 -ACAGGATTGACAGATTGA-3 0 (forward) and 5 0 -TATCGGAATTAACCAG ACA3 0 (reverse) for human 18S rRNA.

Subcellular fractionation and Western blotting
HEK293 cells were lysed on ice for 60 min in lysis buffer containing 10 mM Tris (pH 7.4), 1% Triton X-100, 150 mM NaCl, 10% glycerol, and freshly added protease inhibitors (Roche Complete Protease Mini, Cat # 4693159001) and phosphatase inhibitors (PhosStop pellets, Sigma Aldrich, Cat # 4906845001). Cell lysates were centrifuged at 15,000 × g for 20 min at 4˚C. Supernatant was collected as the soluble fraction. The pellet was washed 3 times with lysis buffer and centrifuged at 15,000 × g for 5 min each at 4˚C. The pellet was resuspended in lysis buffer supplemented with 4% SDS, sonicated 3 times, boiled for 30 min, and collected as the insoluble fraction. Protein concentration was determined using Lowry Protein Assay (Bio-Rad). For Western blot analysis, proteins were subjected to 4%-12% bis-tris midi gels (Bio-Rad) and transferred to nitrocellulose membrane. Membrane was blocked with Odyssey Blocking Buffer (Fisher Scientific). The following primary antibodies were used: mouse anti-Myc (Cell Signaling, Cat # 2276, 1:1000 dilution) and mouse anti-beta tubulin (Sigma-Aldrich, Cat # T8328, 1:8000 dilution). Appropriate IRDye infrared secondary antibodies were purchased from LI-COR Biosciences and used at a dilution of 1:10000. Odyssey Infrared Imaging System (LI-COR Biosciences) was used to detect the signal of target proteins.

Live cell imaging
HEK293 cells were grown on glass-bottomed 35x35mm dishes and transfected with 0.6 μg/kb of pcDNA3.1-Exon1Q23 (or Q145)-Kaede-A or A Ã B. Twenty-four hours after transfection, time-lapse imaging was performed using a 40X oil immersion objective under a Nikon C2 + confocal microscope (Nikon Instruments Inc.) equipped with a stage-top chamber (INUG2A-TIZ, Tokai Hit Co.). The chamber was humidified and maintained at 37˚C with 5% CO 2 . To monitor Kaede synthesis, we first photo-converted the existing Kaede by exposing cells to 405 nm laser with 120 gain and 20% laser power (Nikon Intensilight C-HGFI) for 2 minutes. Newly emerging green-form Kaede was then imaged under 488 nm of excitation laser and a 525/50 nm emission filter at 10-min intervals for 8 hours. The intensity of Kaede fluorescence at the plasma was determined using Nikon software analyzer.

Statistical analysis
Data are reported as means ± SEM. The significance of differences was analyzed using Student's t test or two-way ANOVA with post-hoc Bonferroni correction ( Ã p < 0.05, ÃÃ p < 0.01, and ÃÃÃ p < 0.001). A p value of less than 0.05 was considered statistically significant.

Differential subcellular localization of two HTT mRNA variants in neurons
Alternative polyadenylation produces two HTT mRNA species either with a short or long 3 0 UTR [24,25]. Using RNA probes derived from the rat HTT coding region, we performed fluorescence in situ hybridization (FISH) on cultured rat cortical neurons. We could detect FISH signals in both cell bodies and distal dendrites (Fig 1A), indicating that some HTT mRNA molecules are transported to dendrites. To investigate whether one of the two HTT mRNA variants is preferentially transported to dendrites, we cloned the sequences encoding the two HTT 3 0 UTRs from human genomic DNA. We termed the sequence between the stop codon and the first polyadenylation site "A" and the sequence between the first polyadenylation site and the second polyadenylation site "B". Three constructs were generated to link the GFP-coding sequence with a sequence encoding the 3 0 UTR for bovine growth hormone (BGH), sequence A (encoding the short HTT 3 0 UTR) or sequence AB (encoding the long HTT 3 0 UTR) (Fig 1B). We transfected the three constructs into cultured rat cortical neurons and detected expressed GFP transcripts using FISH with RNAs derived from the GFP coding sequence as probes. As shown in Fig 1C, GFP transcripts were only detectable in cell bodies and proximal dendrites of neurons transfected with either the GFP-BGH or GFP-A construct. Conversely, GFP transcripts were present in cell bodies and distal dendrites of neurons transfected with the GFP-AB construct. We measured GFP mRNA levels along one main dendrite of each transfected neuron. GFP mRNA levels in dendrites were significantly higher in neurons expressing GFP-AB than in neurons expressing either GFP-BGH or GFP-A (Fig 1D). There was no significant difference between neurons expressing GFP-BGH and GFP-A in GFP mRNA levels along the main dendrite. These results suggest that the vast majority of HTT mRNA localized in dendrites should be the long isoform.
Short 3 0 UTR HTT mRNA produces more mutant huntingtin aggregates than the long 3 0 UTR isoform in neurons The two HTT 3 0 UTRs could have impact on mutant huntingtin aggregation by regulating stability, translation, and/or subcellular localization of HTT mRNA. To test this hypothesis, we created four lentiviral constructs to express the huntingtin N-terminal fragment in neurons (Fig 2A). The constructs harbor the normal HTT Exon1 sequence linked to either the short HTT 3 0 UTR (LV-Exon1Q23-Myc-A) or the sequence unique to the long HTT 3 0 UTR (LV-Ex-on1Q23-Myc-B), or a mutated HTT Exon1 linked to either the short HTT 3 0 UTR (LV-Ex-on1Q145-Myc-A) or the sequence unique to the long HTT 3 0 UTR (LV-Exon1Q145-Myc-B). The normal Exon1 has a sequence encoding a 23-glutamine tract (Q23), whereas the mutated Exon1 contains a sequence encoding a 145-glutamine tract (Q145). In addition, there is a Myc tag linked to Exon1-encoded huntingtin fragment (Fig 2A).
We infected cultured rat cortical neurons at DIV5 with each of the four lentiviral vectors and then examined Myc immunoreactivity in MAP2-marked neurons at DIV15, DIV20, and DIV25. As revealed by Myc immunoreactivity, no protein aggregates were detected in neurons infected with two lentiviral vectors expressing the normal huntingtin N-terminal fragment (LV-Exon1Q23-Myc-A and LV-Exon1Q23-Myc-B) at any time point (Fig 2B, S1 Fig and S2  Fig). In these neurons the expressed protein was evenly distributed in the soma and dendrite. However, the mutant huntingtin fragment appeared in forms from diffused molecules to aggregates in transduced neurons (Fig 2C, S1 Fig and S2 Fig). At DIV25, many more neurons transduced by LV-Exon1Q145-Myc-A displayed mutant huntingtin aggregates than neurons transduced by LV-Exon1Q145-Myc-B (Fig 2C and 2D). Furthermore, aggregates were significantly larger in neurons transduced by LV-Exon1Q145-Myc-A than in neurons transduced by LV-Exon1Q145-Myc-B (Fig 2C and 2E). These results indicate that the short form of mutant HTT mRNA is more efficient in producing mutant huntingtin aggregation in neurons. Short 3 0 UTR HTT mRNA also produces more mutant huntingtin aggregates than the long 3 0 UTR isoform in HEK293 cells One of the reasons why the long form of mutant HTT mRNA produces fewer protein aggregates than the short form in neurons could be due to dendritic localization of the long-form HTT mRNA and thus its reduced levels in cell bodies. If this is true, the two forms of mutant HTT mRNA should produce comparable amounts of protein aggregates in non-neuronal HEK293 cells. To test this hypothesis, we created four mammalian expression constructs to express the N-terminal fragment of huntingtin in HEK293 cells (Figs 3A and 4A). The expression cassettes in these constructs are identical to those in the lentivral constructs except for the sequence encoding the long 3 0 UTR. The sequence (termed A Ã B) encodes a full-length long HTT 3 0 UTR and contains mutations at the first polyadenylation site, such that it produces only the long 3 0 UTR.
We transfected the four expression constructs into HEK293 cells and examined the distribution of the expressed Htt fragment at 48h and 72h after transfection using fluorescent immunocytochemistry with antibodies against Myc and Htt. Although the Htt antibody also  ( Fig 3B and 3C). In contrast, the mHtt fragment with a 145Q tract translated from mRNA with either the short or long HTT 3 0 UTR aggregated to form cytoplasmic and nuclear inclusions in some transfected cells 48 h and 72 h after transfection (Fig 4B). Inclusions could be Huntingtin mRNA variants found in cells expressing the mHtt fragment even at 24 h after transfection (S3 Fig). As observed in neurons (Fig 2), more cells transfected with pExon1Q145-Myc-A (expressing mRNA with the short HTT 3 0 UTR) had aggregates than those transfected with pExon1Q145-Myc-A Ã B (expressing mRNA with the long HTT 3 0 UTR) (Fig 4C). Furthermore, aggregates were significantly larger in cells expressing mutant HTT mRNA with the short 3 0 UTR than in cells expressing mutant HTT mRNA with the long 3 0 UTR (Fig 4D). These results show that the short form of mutant HTT mRNA is more efficient in producing mHtt aggregation than its long-form counterpart in HEK293 cells. They also indicate that dendritic localization of HTT mRNA is not the reason why the long HTT 3 0 UTR reduces mHtt aggregation in neurons.

Soluble and insoluble mHtt in cells expressing short or long 3 0 UTR HTT mRNA
We reasoned that more protein may be produced from short 3 0 UTR HTT mRNA than long 3 0 UTR HTT mRNA, leading to more and larger protein aggregates in cells or neurons expressing mutant HTT mRNA with the short 3 0 UTR than those with the long 3 0 UTR. Mutant huntingtin is present as both soluble and insoluble protein in cells [29]. We prepared lysates from HEK293 cells transfected with either pExon1Q145-Myc-A or pExon1Q145-Myc-A Ã B, separated lysates into soluble and insoluble fractions, and analyzed proteins in the two fractions. Immunoblotting analysis with Myc antibodies revealed that the mHtt fragment with a calculated molecular weight of 30 kDa was mainly present as a 60-kDa entity in the soluble fraction (Fig 5A), likely due to formation of SDS-resistant dimers. The levels of soluble mHtt were comparable in the two groups of samples ( Fig 5B). Conversely, the insoluble fraction contained high molecular weight of protein species, which are likely oligomers of the mHtt N-terminal fragment (Fig 5C). The level of large mHtt oligomers was 4.7-fold higher in cells transfected with pExon1Q145-Myc-A than those transfected with pExon1Q145-Myc-A Ã B (Fig 5D). This result indicates that cells expressing short 3 0 UTR HTT mRNA produce much more mHtt protein, the majority of which is present as insoluble oligomers, compared with cells expressing long 3 0 UTR HTT mRNA.
No effect of the two HTT 3 0 UTRs on levels of HTT mRNA in cells Cells transfected with pExon1Q145-Myc-A could have higher levels of Exon1 HTT mRNA, which leads to elevated production of mHtt, compared with cells transfected with pEx-on1Q145-Myc-A Ã B. We performed FISH to determine levels of GFP mRNA in HEK293 cells transfected with either pGFP-A or pGFP-AB (Fig 1B). After FISH, we stained cells with phalloidin to display the cellular cytoskeleton structure. Construct pGFP-AB should produce two GFP mRNAs with either the short or long HTT 3 0 UTR. Because long 3 0 UTR HTT mRNA is more abundant than the short variant in vivo [24,25], GFP mRNA with the long 3 0 UTR should be the major variant in cells transfected with pGFP-AB. As shown in Fig 6A, GFP mRNA showed no obvious difference in subcellular distribution in cells transfected with the two GFP constructs. Furthermore, GFP mRNA levels were comparable in cells transfected with the two GFP constructs (Fig 6B). To further examine the effect of the HTT 3 0 UTRs on mRNA levels, we performed real-time RT-PCR to measure HTT exon1 mRNA in HEK293 cells transfected with either pExon1Q145-Myc-A or pExon1Q145-Myc-A Ã B. Again, we found no significant impact of the HTT 3 0 UTRs on mRNA levels (Fig 6C). These results indicate that mRNA abundance is not the reason why cells transfected with pExon1Q145-Myc-A have an elevated level of mHtt oligomers. Because the two GFP constructs have the same promoter and thus transcription should be the same, these results also suggest that the two HTT 3 0 UTRs do not affect mRNA stability.

Regulation of translation by HTT 3 0 UTRs
We then investigated whether mRNA with the short HTT 3 0 UTR is translated at a higher rate than mRNA with the long HTT 3 0 UTR, which leads to faster accumulation of mHtt. To this end, we employed photoconvertible Kaede protein to examine the effect of two HTT 3 0 UTRs on protein synthesis rate in HEK293 cells. Upon UV light irradiation, Kaede protein is converted from its native green-fluorescence form (Kaede-green) to an irreversible red-fluorescence form (Kaede-red) [30]. We generated four constructs with the short or long HTT 3 0 UTR to express Kaede-tagged normal or mutant Htt N-terminal fragment (Fig 7A). We transfected HEK293 cells with the four Kaede constructs, irradiated transfected cells with UV light for 2 min, and then monitored green fluorescence from newly synthesized Kaede using time-lapse imaging (Fig 7B). Analysis of fluorescence intensity revealed that Kaede was synthesized at higher rates from the constructs with the short HTT 3 0 UTR than those with the long HTT 3 0 UTR regardless of polyQ tract length in the tagged Htt N-terminal fragment ( Fig  7C and 7D and S1-S4 Videos). These results suggest that the additional sequence in the long HTT 3 0 UTR over the short HTT 3 0 UTR (i.e. sequence B) suppresses translation, leading to reduced production of mHtt oligomers in cells transfected with pExon1Q145-Myc-A Ã B in comparison with those with pExon1Q145-Myc-A. Mutant HTT mRNA with the short 3 0 UTR is more toxic than the one with the long 3 0 UTR Mutant HTT mRNA with the short 3 0 UTR is translated at a higher rate and produces more misfolded mHtt, potentially causing more cell death, compared with its counterpart with the long 3 0 UTR. To test this hypothesis, we transfected HEK293 cells with pExon1Q145-Myc-A or pExon1Q145-Myc-A Ã B and then detected apoptosis in transfected cells by fluorescence immunocytochemistry using antibodies against Myc and activated caspase 3 ( Fig 8A and S4  Fig). Apoptosis rate in cells expressing mHtt from short 3 0 UTR mRNA was drastically higher than that from long 3 0 UTR mRNA (Fig 8B). This result indicates that the short variant among the two HTT mRNA species present in cells is more toxic when the CAG repeat in the HTT gene is expanded.

Discussion
Similar to many human genes, the HTT gene produces two mRNA variants that differ in the length of 3 0 UTR while producing the same protein due to alternative polyadenylation. Our results indicate that long 3 0 UTR HTT mRNA can be transported to neuronal dendrites for local translation. Furthermore, we found that short 3 0 UTR HTT mRNA is translated at a higher rate than the long isoform in cell bodies, such that cells expressing short 3 0 UTR mutant HTT mRNA display more mHtt aggregates and a higher rate of cell death compared to those expressing the long isoform. These findings suggest that the two isoforms of HTT mRNA play distinct roles in the HD pathogenesis. No studies have been reported to examine dendritic localization of HTT mRNA. We were able to detect HTT mRNA in dendrites of cultured rat cortical neurons, suggesting that Htt protein can be locally synthesized in dendrites. The observation that the long, but not short, HTT 3 0 UTR is capable of guiding artificial GFP mRNA to dendrites of transfected neurons indicates that dendritic HTT mRNA should be the long isoform. Among its many cellular roles, Htt is involved in cargo trafficking in neuronal processes [31]. It would be interesting to determine whether normal HTT mRNA in dendrites regulates dendritic cargo trafficking and whether mutant HTT mRNA in dendrites plays any special role in the HD pathogenesis.
The current study was focused on distinct roles of mutant HTT mRNA in cell bodies. We found that mutant HTT mRNA with the short 3 0 UTR produced much more mHtt protein in HEK293 cells, which led to more mHtt aggregates and cell death, than its counterpart with the long 3 0 UTR, although the two mRNA species had comparable stability. Immunocytochemistry against activated caspase 3 only detected a small percentage of apoptotic cells in the cultures expressing mHtt; however, an apoptotic cell contains activated caspase 3 only for a short period of time. If we assume that caspase 3 is active over a 2-hour window during apoptosis, 0.15% cells positive to activated caspase 3 in the cultures expressing mutant HTT mRNA with the short 3 0 UTR would mean that 1.8% cells die over a day. The number is significant, given that HD takes many years to develop. Because the long HTT 3 0 UTR contains all sequence of the short HTT 3 0 UTR, low translation of long 3 0 UTR HTT mRNA should be due to translation suppression activities of the additional 3 0 UTR sequence. This explanation is consistent with previous reports that long 3 0 UTRs suppress translation of Bdnf mRNA [20,32]. Long 3 0 UTR Bdnf mRNA is largely sequestered into translationally dormant ribonucleoprotein particles at rest, whereas short 3 0 UTR Bdnf mRNA is actively translated to maintain basal production of brain-derived neurotrophic factor (BDNF). Upon neuronal activation, the long Bdnf 3 0 UTR, but not the short Bdnf 3 0 UTR, imparts rapid and robust activation of translation [20]. Similarly, long 3 0 UTRs have been reported to negatively affect translation of mRNAs for GluR2, IMP-1, Cyclin D2, and DICER1 [19,32].
3 0 UTRs often harbor several binding sites for diverse trans-acting factors, including RNAbinding proteins (RBPs) and microRNAs (miRNAs) [33,34]. Variation in 3 0 UTR length can alter translation regulation by inclusion or exclusion of RBP-or miRNA-binding sites. Longer 3 0 UTR more likely possess such sites, or more of them, and the mRNA will therefore be more likely subjected to negative regulation. Indeed, a systematic examination of 3 0 UTRs found that 52% of miRNA target sites are located downstream of the first poly A site [35]. Furthermore, it was reported that mRNAs with longer 3 0 UTRs contained a 2.1-fold higher number of miRNA target sites than those with shorter 3 0 UTRs in T cells [18]. Consequently, the amount of protein generated by an mRNA is highly related with its 3 0 UTR length, usually transcripts with shorter 3 0 UTRs producing higher levels of protein [18]. We speculate that the additional 3.2-kb sequence in the long HTT 3 0 UTRs contains binding sites for miRNAs and/or RBPs that suppress translation. Further investigations are needed to decode the "language" of the HTT 3 0 UTR, such as identification of cis-regulatory elements in the long HTT 3 0 UTR and trans-acting factors. Our findings suggest that the short form of mutant HTT mRNA is more toxic than the long form. Reducing production of the short form of mutant HTT mRNA by manipulating the polyadenylation process could likely delay the onset of HD.