Huntington’s Disease Protein Huntingtin Associates with its own mRNA

Htt affinity purifications and RNA isolation

FLAG-Htt affinity purifications were conducted essentially as described [4] with a few modifications. After the 1400× g spin, Triton X-100 was added to S1 lysates to a final concentration of 0.5%. Lysates were incubated at 4°C for 20 minutes, and centrifuged at 14,000× g for 20 minutes to give a total cytoplasmic lysate. EDTA was added to one half of the lysate at a final concentration of 20 mM or an equivalent volume of lysis buffer, and incubated at 4°C for 20 minutes. Protein concentration was measured by Bradford assay and adjusted to 6 mg/mL with lysis buffer prior to binding to 80μL (packed bead volume) of Anti-FLAG M2 agarose beads (Sigma) for 3 hours at 4°C with gentle mixing. Beads were washed and eluted as described [4]. Endogenous Htt immunoprecipitations from Neuro-2a cells was performed also as described [4]. Eluted materials from both types of purifications were extracted with 0.5 mL of TRI-Reagent (Sigma) and processed to isolate RNA and proteins as per manufacturer’s instructions. RNA was further cleaned using a Direct-zol RNA miniprep column (Zymo Research). Eluted RNA was concentrated by ethanol precipitation in the presence of 5μg of linear acrylamide. 20% of eluted proteins were examined for Htt recovery by western blotting and the remainder used to probe for Htt-associated proteins. Mouse brains were obtained in accordance with the guidelines established by the NIH for the care of laboratory animals, and all procedures approved by Institutional Animal Care and Use Committee (Protocol 140308-01).

cDNA synthesis, qPCR oligonucleotides

Total cDNA was generated from 1μg of inputs and the entirety of immunoprecipitations using a cDNA synthesis kit and the 10x primer mix from USB (Affymetrix). Equal volumes of cDNA synthesis reactions were used in RT-qPCR reactions using the 2X SYBR green master mix from USB (Affymetrix). qPCR data was calculated relative to a Gapdh (Htt IP) or Htt (FLAG IP) standard curves from input cDNAs. All qPCR reactions were performed in triplicate and trends were reproduced with experimental replicates. The oligonucleotides used in the PCR reactions were as follows.

Htt 3’ UTR Forward: TAGAGCCATCGAAGGAACTGCTCCC

Htt 3’ UTR Reverse: GCATACAGCAGCAAGGCTTGGACAG

Gapdh Forward: GGTGCTGAGTATGTCGTGGA

Gapdh Reverse: CGGAGATGATGACCCTTTTG

For 3xFLAG-Htt^7Q/7Q mESCs, total RNA and FLAG-IP RNA were extracted with TRI-Reagent. cDNA synthesis and qPCR were carried out as above using the following oligonucleotides.

Htt exon 1-2 Forward: AGGAACCGCTGCACCGA

Htt exon 1-2 Reverse: CTGAGAGACTGTGCCACAATGTT

Htt exon 66-67 Forward: CCAGCCCATGGGTTTCTGCG

Htt exon 66-67 Reverse: AAGCAGCCTGTGGTATGGACTTCC

Oct4 Forward: GAAGCAGAAGAGGATCACCTT

Oct4 Reverse: TTCTTAAGGCTGAGCTGCAAG

MaoA Forward: GCCCAGTATCACAGGCCAC

MaoA Reverse: CGGGCTTCCAGAACCAAGA

Immunofluorescence microscopy

Cells were washed 3x in pre-warmed (37°C) PBS, then fixed in pre-warmed 4% paraformaldehyde for 10 minutes at 37°C. Fixed cells were washed 3x in PBS, then blocked for 1 hour at room temperature in 5% goat serum, 0.25% saponin, in PBS. The visualization of Venus did not require the use of antibodies. All incubations in primary antibodies were carried out overnight at 4°C and secondary incubations done at room temperature for 1 hour. Coverslips were mounted in Dako fluorescent mounting medium. Images were acquired on a Zeiss META510 confocal scanning laser system on an Axiovert 200M microscope using a Plan-Apochromat 100X/1.3 oil-immersion objective lens. Brightness and contrast settings were adjusted image-wide using Adobe Photoshop.

Antibodies

For western blotting mouse monoclonal anti-Htt mAB2166 (Millipore) was used at 1:1000, anti-RACK1 (Gnb2l1) (BD Transduction Laboratories) at 1:2500, anti-Rps6 2217 (Cell Signaling Technology) at 1:500, anti-Myo5a (Cell Signaling Technology) 1:1000, anti-β-tubulin (Covance) 1:1000, anti-β-actin ac15 (Santa Cruz Biotechnology) 1:400, and anti-GFP (Santa Cruz Biotechnology) at 1:500 in TBST. For immunofluorescence, mouse monoclonal anti-FLAG M2 (Sigma) was used at 1:1000, rabbit polyclonal anti-Htt (Enzo Life Sciences) at 1:1000, and anti-Fmr1 (Millipore) mAB2160 at 1:500. For immunoprecipitations 2μg of anti-Htt (Enzo Life Sciences) was used to immunoprecipitate Htt from Neuro-2a cells or 2μg of rabbit IgG (Santa Cruz Biotechnology) as the negative control.

Plasmids

The MS2 binding sites of the bacteriophage MS2 protein were cloned downstream of the Renilla luciferase gene in pRL-TK vector (Promega). PCR-cloned Htt 3’ UTR sequences from a mouse genomic clone containing the entire Htt 3’ UTR was inserted downstream of the MS2 binding sites to generate the MS2bs reporter constructs. pNLS-MS2-Venus was obtained from [10] and the FLAG-Ago2 plasmid was obtained from Addgene. Htt 3’ UTR sequences were also subcloned downstream of pEGFP-C1 (Clontech) to generate the GFP reporter constructs.

Cell culture and transfections

Mouse Neuro-2a cells were cultured in 10% FBS in DMEM (Mediatech) and transiently transfected using Lipofectamine 2000 (Invitrogen) under recommended conditions. Mouse Htt knock-in striatal precursor lines [11] were cultured in 10% FBS in DMEM and grown at 33°C. 3μg of total plasmid DNA/well of a 6-well plate was used in MS2 transfections: 3μg pNLS-Venus-MS2 in control transfections, 800 ng pNLS-MS2-Venus and 2.2μg pRL-MS2bs-Htt-short 3’ UTR, and 600 ng pNLS-MS2-Venus and 2.4μg pRL-MS2bs-Htt-long 3’ UTR. Transfections were performed for 24 hours prior to fixation. For GFP reporter assays, Neuro-2a cells were split into 6 cm plates. At 24 hours, Htt-17Q or CMV-empty control plasmid was transfected at a concentration of 3μg/plate. At 30 hours, cells were split split equally into 4–25 mm plates. After 48 hours, equimolar amounts of GFP reporters were transfected: 1.18μg GFP-SV40-polyA, 2.1μg GFP-Htt-long 3’ UTR, 1.33μg GFP-Htt-short 3’UTR. Cells were harvested after 72 hours and 20μg of soluble lysates were analyzed by SDS-PAGE followed by immunoblotting with anti-GFP and anti-Htt antibodies and visualized and quantified using Odyssey imager (LI-COR). siRNA transfections were performed as described. Transfection of rat cortical neurons and co-localization of endogenous Htt with Htt 3’ UTR were performed as previously described [6, 12].

Experiments with mouse embryonic stem cell (mESC) lines

Three lines were used: Wild type, 3xFLAG-Htt^7Q/7Q and 3xFLAG-Htt^140Q/7Q. mESCs were cultured on gelatin-coated plates in standard ES medium (DMEM supplemented with 15% ES-Cult FBS, 1% sodium pyruvate, 1% NEAA, 1% penicillin-streptomycin, 0.1 mM 2-mercaptoethanol, and 1000 U/ml LIF). Cells were lysed in total lysis buffer20 mM Tris-HCl (pH 8.0), 137 mM NaCl, 1 mM EDTA, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl₂, 1 mM DTT, 1 mM PMSF, and 1X protease inhibitor cocktail]. Sixty milligrams of total protein was immunoprecipitated with anti-FLAG M2 affinity beads (Sigma A2220) for 4 hours on a nutator at 4°C. The beads were first washed with total lysis buffer in a column, and then second wash with total lysis buffer w/o Triton X-100. FLAG-Htt complexes were eluted with 3X FLAG peptide (200 ng/μl, Sigma F4799), and concentrated by TCA (trichloroacetic acid) overnight. SDS-PAGE was performed by loading 40μg of total protein for each input sample and 5% of total eluate for each FLAG-IP sample. Membranes were probed using the following primary antibodies: Mouse anti-FLAG-M2 (Sigma F-1804), mouse anti-Htt2166 (Millipore MAB2166) and mouse anti-polyQ (clone 3B5H10, Sigma P1874). IRDye 800CW goat anti-mouse secondary antibody was used (LI-COR Biosciences). Fluorescent signal was scanned using the LI-COR Odyssey machine.

RNA-Seq analysis

RNA-Seq libraries were made using the SMARTer Ultra Low RNA kit for Illumina (Clontech, Cat # 634935 and 634936), using 1 to 2 ng of RNA. Libraries were run on an Illumina 2500, using a paired end 50 protocol. Sequencing reads were demultiplexed and converted to FASTQ format using Illumina Bcl2FastQ software. Reads were aligned to the mouse genome (build mm10/GRCm38) using the splice-aware STAR aligner [13]. PCR duplicates were removed using the Picard toolkit (http://broadinstitute.github.io/picard). HTSeq package [14] was utilized to generate counts for each gene and exon based on number of aligned reads overlapping them. These counts were used to test for differential exon usage using negative binomial generalized linear models and for visualization as implemented by the DEXSeq R package [15].

Purified endogenous Htt complex contains its own mRNA

In previous studies we utilized FLAG-tagged wild type (F7Q/F7Q) and mutant (F140Q/+) Htt knock-in mice [16] to isolate Htt-containing protein complexes from juvenile mouse brains. Co-purifying proteins identified by mass spectrometry revealed many RNA binding proteins and translation factors, suggesting a new role for Htt in post-transcriptional gene regulatory pathways [4]. The finding prompted us to analyze RNA that might be present in the Htt protein complexes. Surprisingly, microarray and RNA-Seq analyses of co-purifying mRNAs consistently identified the Htt mRNA as the most enriched RNA in our FLAG-Htt purifications from mouse brains compared with the negative control (FLAG-IP eluates of normal mouse brains expressing no FLAG-tagged proteins). Multiple RT-qPCR experiments on co-eluted RNA from FLAG affinity purifications confirmed the drastic enrichment of Htt mRNA in F7Q and F140Q/+ purifications compared with the control. Comparison of the recovery of Htt mRNA to other mRNAs (Oct 4, MaoA, as examples) in the FLAG-IP eluates clearly demonstrated selective retention of the Htt mRNA in the FLAG immunoprecipitates (Supplementary Figure 1). No other mRNAs showed enrichment above the levels found in the control immunoprecipitate, which represented non-specific binding. We wondered whether this interaction might be an artifact of the purification scheme used to purify Htt as the FLAG tag is positioned at the N-terminus of wild type and mutant alleles. It is possible that our Htt purifications recovered Htt mRNA co-translationally; i.e., incompletely translated FLAG-Htt was immunoprecipitated and as such was associated with the Htt mRNA solely through the whole translational apparatus. To rule out this possibility, we dissociated ribosomes prior to FLAG affinity purification by the addition of EDTA to total cytoplasmic lysates [17]. Addition of EDTA did not significantly affect the amount of FLAG-Htt recovered from F7Q affinity purifications, but reduced F140Q/+ recovery by approximately 50% (Fig. 1a, lanes 13–16). However, EDTA pre-treatment nearly completely abolished the wild type and mutant Htt interactions with ribosome-associated proteins Rps6 and Gnb2l1 (Rack1) (Fig. 1a, lanes 13–16). In contrast, the Htt-Myo5a interaction was not completely abolished by this treatment. These data show that EDTA addition does not prevent FLAG-mediated purification of FLAG-Htt and that interactions of Htt with ribosomal proteins is largely dependent upon the integrity of the whole ribosome.

We next measured the levels of Htt mRNA present in FLAG-Htt purifications by quantifying it relative to the amount of Htt mRNA present in 1μg of total input RNA, which is expressed as a numerical ratio (Fig. 1b). FLAG IP of both wild type (F7Q/F7Q) and mutant (F140Q/+) Htt protein recovered significant amounts of Htt mRNA compared with negative control. Although addition of EDTA drastically reduced the enrichment ratio of Htt mRNA for both wild type and mutant, these ratios were still 12 and 24 times greater than that of the negative control, for wild type and mutant,respectively (Fig. 1b, EDTA-treated). These data show that a pool of the Htt protein-Htt mRNA association occurs independently of ribosomal integrity. Furthermore, the enrichment of Htt mRNA in EDTA-treated samples compared with the negative control argues against the possibility that the Htt protein-Htt mRNA association is an experimental artifact. To further substantiate the Htt protein-Htt mRNA interaction, we immunoprecipitated endogenous Htt from Neuro-2a cells and measured the amount of Htt 3’ UTR present within control (IgG) and Htt immunoprecipitates (Htt) by RT-qPCR. Htt immunoprecipitates contained 3 times as much Htt 3’ UTR as the negative control, indicating that Htt protein-Htt mRNA association does not occur through the FLAG tag.

Htt protein co-localizes with Htt mRNA in neuronal cells

We have previously shown that Htt co-localizes with the 3’ UTR of β-actin and Bdnf mRNA in cultured rat neurons [8]. We hypothesized that the 3’ UTR of Htt is likely to contain element(s) responsible for associating with Htt protein. To confirm this possibility we examined the location of Htt with respect to a reporter construct harboring the short and long versions of the Htt 3’ UTR in a mouse immortalized striatal precursor cell lines expressing wild-type Htt (STHdh^7Q/7Q) or mutant Htt (STHdh^111Q/111Q) [11]. Our assays employed the bipartite MS2 reporter system [10, 18] to visualize Htt protein with respect to a transfected mRNA containing the short and long versions of the Htt 3’ UTR (Fig. 2a), preceded by the Renilla luciferase coding sequence and 24 tandem copies of the bacteriophage MS2 binding sequence. Control experiments indicated that Venus signal was primarily confined to the nucleus in wild-type and mutant STHdh cells transfected with the NLS-MS2-Venus reporter alone (Fig. 2b, Venus panels), indicating that the presence of the NLS on the Venus reporter confined Venus to the nucleus. In contrast, wild-type cells (Fig. 2b, STHdh^7Q/7Q) expressing both the NLS-MS2-Venus and an MS2 binding site-tagged Htt short 3’ UTR displayed numerous large and small cytoplasmic puncta. Many of these cytoplasmic puncta co-localized with wild-type Htt (Fig. 2b, short UTR, arrowheads), evidenced by the appearance of yellow spots in the merged image. A recent cross-linking immunoprecipitation study of fragile-X mental retardation protein (Fmrp) identified Htt RNA as a target of Fmrp binding in mouse brains [17]. Consistent with this finding, we found co-localization of Fmrp with the Htt 3’ UTR in this assay (Fig. 2b, short UTR, blue-green spots). Furthermore, we observed several Htt short 3’ UTR puncta, which were labeled for both Htt and Fmrp, suggesting that both proteins are capable of associating with the Htt short 3’ UTR (Fig. 2b, upper panel, short UTR, arrow). Not surprisingly, wild-type cells expressing NLS-MS2-Venus and MS2 binding site-tagged Htt long 3’ UTR constructs also displayed numerous cytoplasmic puncta, which also stained positively for Htt and Fmrp (Fig. 2b, long UTR, arrow and arrowhead). These data demonstrate that the 3’ UTR of Htt is sufficient to attract Htt protein to an exogenous mRNA in wild-type cells.

Mutant cells (Fig. 2b, STHdh^111Q/111Q) expressing NLS-MS2-Venus and MS2 binding site-tagged Htt short 3’ UTR also displayed cytoplasmic distribution of Venus signal, although the puncta were often larger and more concentrated than those observed in wild-type cells. Like wild-type cells, Htt short 3’ UTR as reported by Venus signal was frequently found to co-localize with both Fmrp and Htt, particularly at the periphery of the cell. This peripheral localization of all three elements was also observed in mutant cells in which the localization of the long version of the Htt 3’ UTR was monitored (Fig. 2b, long UTR). Finally, we transfected MS2 binding site-tagged Htt long 3’ UTR into rat cortical neurons and demonstrated co-localization with endogenous rat Htt by immunofluorescence(Fig. 2c).

Htt represses gene expression through its 3’ UTR sequence

Next we sought to determine the effect that the 3’ UTR of Htt might have on the expression of a linked GFP reporter. To do this, we generated GFP expression constructs containing either the SV40 polyadenylation signal alone, the short Htt 3’ UTR, or the long Htt 3’ UTR (Fig. 3a). These constructs were transfected into Neuro-2a cells at equal molar ratios and GFP expression levels monitored by western blotting. Addition of the Htt short 3’ UTR to the GFP expression construct led to a 57% reduction in GFP protein (Fig. 3b), while the long 3’ UTR had an even larger inhibitory effect, reducing GFP expression by 95% compared with the empty control. These data agree with the results from [19] and support an inhibitory role for the 3’ UTR of Htt on protein expression.

Given that Htt likely associates with its own mRNA and that much of this association is dependent upon the presence of intact ribosomes (Fig. 1a, b), we wanted to determine the effect that overexpression of Htt had on the levels of GFP expression in the context of the different Htt 3’ UTRs. Overexpression of full-length wild-type Htt (Htt17Q) led to a nearly 50% decrease in the levels of GFP-Htt short when compared with the control without exogenous Htt expression (Fig. 3c). The already low levels of GFP-Htt long did not change with Htt overexpression.

To further test the possibility that Htt acts to negatively regulate the levels of protein produced from mRNA containing the Htt 3’ UTR, we examined the effects of Htt knockdown on the expression of the GFP reporter constructs. siRNA-mediated knockdown of Htt led to a nearly 10-fold decrease in the levels of Htt compared with the control knockdown (Fig. 3d, Htt level), and a 50% increase in GFP-Htt short, and a 40% increase in GFP-Htt long expression levels compared with their levels in the control knockdown (Fig. 3d, GFP level). These results suggest that Htt has a role in the maintenance of its protein levels through the 3’ UTR.

RNA-Seq analysis of Htt co-purifying RNA identifies mis-spliced Htt exon 1 mRNA

We performed RNA-Seq analysis of the same set of FLAG-immunoprecipitated samples from mouse brains previously subjected to microarray analysis. We indeed detected Htt mRNA sequences across the entire length of the gene at levels orders of magnitude greater than in the control immunoprecipitate (from normal mouse brains expressing no FLAG-tagged proteins to control for non-specific binding to the FLAG antibody) in both wild type and mutant Htt immunoprecipitates (Fig. 4a). Furthermore, analysis of sequence reads in FLAG immunoprecipitates from F140Q/+ brains showed presence of nearly equal number of reads corresponding to endogenous wild-type mouse allele and FLAG-tagged knock-in allele, which can be distinguished by a short human sequence present in the knock-in allele (data not shown). This provides further compelling evidence that Htt protein-Htt mRNA association does not occur through the FLAG tag or through pull down of nascent FLAG-Htt polypeptide. A recent report identified the presence of mis-spliced Htt mRNA containing the sequence of Htt exon 1 and beginning of intron 1 specifically in cells and tissues expressing mutant Htt protein [20]. This prompted us to look for mRNA corresponding to Htt intron 1. We indeed detected sequence reads from intron 1 only in Htt immunoprecipitates obtained from 140Q-expressing mouse brains (Fig. 4b, c). These results indicate that Htt protein complex contains both full-length and mis-spliced Htt mRNA.

Consistent with the above results, RNA-Seq analysis of FLAG-Htt140Q immunoprecipitates from mouse embryonic stem cells expressing the FLAG-tagged protein were also found to contain full-length and exon1 Htt mRNA (data not shown). Furthermore, immunoblotting of the FLAG-Htt immunoprecipitates demonstrated the presence of a polypeptide migrating at ∼80 kDa (aberrant migration due to expanded polyQ sequence) detected by α-FLAG and α-polyQ antibodies (Fig. 5a, c), but not α-Htt (2166) monoclonal antibody whose epitope is located C-terminal to the region encoded by exon 1 (Fig. 5b). This polypeptide was found only in cells expressing mutant Htt, confirming the existence of the N-terminal mutant Htt fragment encoded by the mis-spliced Htt mRNA.