Direct Interaction between GGGGCC/AAAAUU Hexanucleotide Repeats and TDP-43/TDP-25

The above analysis was performed in cell lysate. Although this is an advantage of FCCS because the biomolecular interaction can be measured in a solution mixture containing fluorescent molecules, it is not yet clear whether T25 interacts with RNA directly or cellular RNA-binding proteins in complexes of T25 interact with RNA. To confirm the direct interaction with these RNAs, we prepared purified GFP, T43, and T25 through tandem affinity purification using polyhistidine and strep tags. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by silver staining showed that highly purified GFP was obtained; however, several cellular proteins were included in the purified T43 and T25 samples (Figure ​Figure22a). Since T43 is an RNP and binds to RNA with Gq-foldable sequences,^25,33 we identified the copurified proteins with T25 using peptide mass fingerprinting with mass spectroscopy (PMF-MS). Five major proteins copurified with T25 were identified—heat shock protein 70 kDa (HSP70), mitochondrial pyruvate carboxylase (mPC), tubulin, and actin—by PMF-MS, and a GFP-containing fragment was confirmed by Western blotting. Bands corresponding to mPC, tubulin, and actin were also observed with T43 (Figure ​Figure22a). As HSP70 and mPC have been reported to have RNA-binding properties,^34,35 we confirmed that G4/A4 did not bind to these copurified cellular proteins using FCCS. AF-G4/AF-A4 was mixed in the lysate of cells expressing HSP70, β-tubulin, α-actin, or mPC tagged with GFP or yellow fluorescent protein (YFP), and following FCCS analysis under the same conditions as T43/T25 did not show their significant interaction (Figure S3). Since HSP70 is an abundant molecular chaperone and is frequently coprecipitated, the interaction with HSP70 may contribute to maintaining the solubility of TDP-25. In other words, it may be difficult to recover TDP-25 without HSP70 in the soluble fraction. Because the endogenous TDP-43 band was not observed in the purified T25 sample (Figure ​Figure22a), and the interaction between TDP-25 and TDP-43 was not detected using FCCS in cell lysate,¹⁸ it is unlikely that the purified T25 sample contains endogenous TDP-43. Since AF-UG did not bind to T25 in the cell lysate experiments (Figure ​Figure11b), it is unlikely that G4/A4 are recognized by TDP-43 bound to TDP-25. Consequently, even if other RNPs were included in the purified T25 sample, they may not contribute to the interaction analysis with G4/A4. The interaction of purified GFP, T43, and T25 with AF-G4/AF-A4 was then analyzed using FCCS. No positive RCA of the GFP, T43, and T25, when mixed with just AF dye as a negative control, was observed; however, positive RCA of T43 and T25, when mixed with AF-G4 and AF-A4, was observed, and it was higher with AF-G4 than with AF-A4 (Figure ​Figure22b and Figure S4). Dissociation constants (K_d) of these interactions determined from the FCCS analysis were from the submicromolar to micromolar level (Figure ​Figure22c). These K_d ranges are considered reasonable for RNA-protein interactions in solution. Consequently, TDP-43 and TDP-25 interact directly with G4 and A4 and not through other RNPs.

Open in a separate window

Figure 2

Direct interaction of the G4C2/A4U2-repeat RNA with purified TDP-43 and TDP-25. (a) SDS-PAGE of purified GFP, GFP-TDP-43, and GFP-TDP-25, followed by silver staining. Asterisks show purified target proteins. (b) RCA values when AF-labeled RNAs (G4, and A4) folded in 100 mM KCl were mixed with purified GFP monomers (GFP), GFP-TDP-43 (T43), and GFP-TDP-25 (T25) with 100 mM KCl (n = 3; mean ± SE). Significance: **p < 0.01, ***p < 0.001, and ns (p ≥ 0.05). The typical normalized cross-correlation functions are represented in Figure S4. (c) Dissociation constant (K_d) values between T43/T25 (proteins) and AF-labeled G4/A4 (RNA) calculated from Figure ​Figure22b (n = 3; mean ± SE).

Both Truncated RRM2 and IDR Are Responsible for the Interaction between GGGGCC/AAAAUU Hexanucleotide Repeats and C-Terminal Fragments of TDP-43

How does TDP-25 recognize RNA even though it lacks many of the RRMs? To identify the contribution of truncated RRM2 and IDR in TDP-43 CTFs to the interaction with G4/A4, the interactivity of two additional truncated variants of TDP-25 (F220 and C274;³⁶Figure ​Figure33a) with G4/A4 was analyzed using FCCS in Neuro2a cell lysates. Neither F220 nor C274 interacted with G4 in a mixture of Gq and HP, HP-fold G4, and A4 (Figure ​Figure33b), suggesting that both truncated RRM2 and IDR are responsible for the interaction with G4/A4 not through highly conserved nucleic acid binding sequences such as the RRM.

Open in a separate window

Figure 3

Truncated RRM2 is required for the interaction with G4C2/A4U2-repeat RNA. (a) Primary structures of human TDP-43 (1–414 amino acids), TDP-25 (220–414 amino acids), F220 (220–262 amino acids), and C274 (274–262 amino acids). Numbers show the amino acid position. (b) RCA values when AF-labeled RNAs (G4, and A4) that were folded in 100 mM KCl (Gq + HP) or 100 mM LiCl (HP) were mixed with GFP monomers (GFP), GFP-TDP-43 (T43), GFP-TDP-25 (T25), GFP-F220 (220), and GFP-C274 (274) with 100 mM KCl (n = 3; mean ± SE). Significance: *p < 0.05, **p < 0.01, ***p < 0.001, and ns (p ≥ 0.05).

GGGGCC/AAAAUU Hexanucleotide Repeats Prevent Aggregation of TDP-25 and Ameliorate Its Cytotoxicity

Next, to demonstrate the activity of G4 and A4 that prevent the aggregation of TDP-25, G4/A4 tagged with bacteriophage-derived M13 sequences (M13-G4 and M13-A4, respectively) were expressed in Neuro2a cells by transfection of plasmid DNA that transcribes such RNAs under the human H1 promoter. The expression of M13-G4, M13-A4, and M13-tagged 24-mer UG repeat RNA (M13-UG), and only the M13 tag as a control (M13N) in Neuro2a cells was confirmed by reverse transcription PCR (RT-PCR) (Figure S5a). The population of cells expressing M13-G4 and M13-A4 that harbor T25 aggregates in the cytoplasm was decreased, while that of cells expressing M13N and M13-UG was not (Figure ​Figure44a). Although the expression levels of T25 in the soluble fraction of the cell lysates were not changed in cells expressing M13-G4, M13-A4, M13-UG, or M13N compared to those in empty vector transfection, the levels of T25 in the insoluble fraction significantly decreased in cells expressing M13-G4 and M13-A4 (Figure ​Figure44b and Figure S6a). The total abundance of T25 in whole cell lysates was not changed in the expression of M13-G4 and M13-A4 (Figure S6b). These results suggest that G4/A4 can be an aggregation suppressor for TDP-25, similar to a cytoplasmic molecular chaperone HSP70 for TDP-25 aggregation.³⁷ Furthermore, RNA that binds to TDP-43, as in the case of UG repeat RNA, is not a sufficient condition for the prevention of T25 aggregation. Although these G4/A4 are supposed to, the next question is whether the G4/A4 introduced into cells disturbs the intracellular transcriptome, resulting in the indirect suppression of T25 aggregation through the upregulation of stress-induced proteins, such as heat shock proteins. To this end, transcriptomic analysis was performed using RNA-seq when M13-G4, M13-A4, and M13-UG were expressed with T25 in Neuro2a cells. No transcripts with a >2- and <0.5-fold change in the significant expression ratio (p < 0.05) were observed in both mRNA and noncoding RNA (ncRNA) in the comparison of M13-G4/A4 with M13-UG (Figure ​Figure44c). Therefore, it is unlikely that up-regulated chaperone-like genes by expression of M13-G4/A4 inhibit T25 aggregation indirectly; in other words, we suggest that these RNAs directly prevent T25 aggregation.

Open in a separate window

Figure 4

Reduced aggregates of GFP-TDP-25 in Neuro2a cells expressing G4C2/A4U2-repeat RNA. (a) Confocal fluorescence images of GFP-TDP-25 (T25) in Neuro2a cells (left). The percentage of cells containing cytoplasmic aggregates of T25 when an empty vector (empty) or M13-tagged RNA was expressed (M13N, M13-UG, M13-G4, and M13-A4) (right) (the number of trials = 3; mean ± SE). (b) Western blotting of Neuro2a cells expressing T25 when M13-tagged RNAs were expressed. Typical blots of T25 in soluble and insoluble fraction, and α-tubulin in the soluble fraction (left). The quantified intensity ratio of T25 in the insoluble fraction compared with that in the empty vector transfection (the number of trials = 3; mean ± SE) (right). The gel loading the insoluble fraction of the cell lysate stained with Coomassie brilliant blue as an internal control was represented in Figure S6a. The blots of total T25 abundance in whole cell lysate were represented in Figure S6b. (c) Volcano plot of mRNA (left) and ncRNA (right) expression change when M13-G4 or -A4 was expressed in Neuro2a cells compared with M13-UG (the number of trials = 4). Areas of nonsignificant change are shaded in gray (0.5–2-fold change and p ≥ 0.05). (d) Normalized relative fluorescence intensity (RFI) indicating mobile fraction of T25 in the cytoplasmic aggregates determined using FRAP (measured cell number = 9; mean ± SE). (e) Population of T25-expressing cells stained with propidium iodide (PI) as a dead cell marker (the number of trials = 3; mean ± SE). (f) Population of T25-expressing cells stained with DRAQ7 as a dead cell marker using FACS (the number of trials = 3; mean ± SE). (a, b, e, f) Significance: *p < 0.05, **p < 0.01, and ***p < 0.001.

Are the biophysical molecular dynamics of TDP-25 within the aggregates then changed in the expression of G4/A4? To investigate this issue, we performed fluorescence recovery after photobleaching (FRAP) analysis in live Neuro2a cells was performed. TDP-25 was quite immobile and formed amyloids in the aggregates;^18,36 however, the expression of M13-G4/A4 did not alter the immobility of T25 (Figure ​Figure44d), suggesting that G4/A4 may act on TDP-25 in the diffusive states in the cytoplasm to prevent aggregation before the formation of deposited aggregates. As a possible reason why G4/A4 cannot act on the deposited aggregates, HSP70 overexpression does not dramatically increase the mobile population of TDP-25 in the aggregates.^18,37 HSP70 is colocalized and dynamically interacts with TDP-25 aggregates;³⁷ however, RNA is not stained inside the cytoplasmic TDP-25 aggregates.¹⁸ This unremarkable difference in TDP-25 mobility in the aggregates, even when overexpressed HSP70 interacts with the aggregates, is consistent with no significant changes in TDP-25 mobility in the aggregates when M13-G4 and M13-A4 were expressed, as shown here.

Next, to demonstrate that G4/A4 without any tag directly prevents TDP-25 aggregation, the aggregate formation of T25 in Neuro2a cells introduced with RNA was analyzed. Transfection of synthetic 24-mer UG repeat RNA (UG) and siRNA against the Renilla luciferase mRNA (siRL)³⁸ as controls preintroduction of T25-expression vector did not change the population of cells containing T25 aggregates in the cytoplasm compared to mock transfection; however, the direct transfection of synthetic G4/A4 preintroduction of T25-expression vector significantly decreased the proportion of cells containing T25 aggregates (Figure S7a,b). The levels of T25 in the insoluble fraction decreased in G4- and A4-transfected cells (Figure S7b,c). These results confirm the hypothesis that G4/A4 can be used as an aggregation suppressor for T25 by introducing it into cells. Furthermore, the transfection of synthetic G4/A4 postintroduction of T25-expression vector significantly decreased the proportion of cells containing T25 aggregates and the amount of T25 in the insoluble fraction (Figures S7f–k); however, the efficiency of reduction of T25 aggregates during G4/A4 to UG transfection was higher with pre-T25 transfection than with post-T25 transfection (Figures S7). Therefore, G4/A4 may inhibit the aggregation and assembly process before TDP-25 forms aggregates in the cytoplasm that can be viewed by microscopes. These RNAs would maintain the oligomeric states of TDP-25 and/or prevent its intermolecular assembly.

To demonstrate whether M13-G4 and M13-A4 expression ameliorates TDP-25 cytotoxicity, we quantified the population of propidium iodide (PI)-stained cells using fluorescence microscopy and DRAQ7-stained cells using fluorescence-activated cell sorting (FACS), indicating dead cells. The population of PI- and DRAQ7-positive and T25-expressing cells was dramatically reduced in M13-G4/A4-expressing cells compared to M13N or M13-UG expression and empty vector transfection as controls (Figure ​Figure44e,f), suggesting that G4/A4 plays a role in the reduction of cytotoxicity in TDP-25-expressing cells.

Because G2 but not A2 interacted with TDP-25 (Figure ​Figure11d), we next investigated whether G2/A2 can reduce the TDP-25 aggregate. M13-tagged G2/A2 (M13-G2/A2) was expressed as much as M13-G4/A4 in Neuro2a cells (Figure S8a). M13-G2/A2 expression in Neuro2a cells did not change the population of cells that harbor T25 aggregates and the levels of T25 in the insoluble fraction (Figure S8b). G2 is believed to be too short to form the Gq structure efficiently, while it is not entirely incapable of forming Gq as intermolecular assemblies. The interaction strength of AF-G2 with T25 was reduced by approximately 60% compared to AF-G4 (Figure ​Figure11e). Therefore, for effective inhibition of the TDP-25 aggregation by r(G4C2) in the cell, a certain degree of interaction strength with TDP-25, which corresponds to that of four repeats of r(G4C2), may be required. Furthermore, r(A4U2) may also require four repeats to efficiently suppress the TDP-25 aggregation.

Cell Lines

Murine neuroblastoma Neuro-2a cells (#CCL-131; ATCC, Manassas, VA, USA) and Flp-In T-REx 293 cells (Thermo Fisher, Waltham, MA, USA) were maintained in DMEM (D5796; Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% FBS (12676029, Thermo Fisher), 100 U/mL penicillin G (Fujifilm Wako Chemicals), and 0.1 μg/mL streptomycin (Fujifilm Wako Chemicals) as described previously.¹⁸

Construction of Plasmids

The plasmids for the expression of monomeric enhanced GFP (meGFP)-tagged TDP-25 (pGFP-TDP-25), 220–262 amino acids CTF (F220; truncated RRM2), and 274–414 amino acids CTF (C274) were used as per the protocol established in previous reports.^18,36 The synthetic DNA fragment for a variant of TDP-43 that lacks nuclear localization and exporting signal sequences and carries amino acid substitutions of C173S/C175S (TDP-43CS-mNLS/NES),⁴² was ordered via DNA fragment synthesis services (Eurofins Genomics, Tokyo, Japan). The fragment was inserted into the HindIII and BamHI sites of pmeGFP-N1 (pTDP43CS-GFP). For purification of meGFP-tagged wild type TDP-43, synthetic cDNA fragment for polyhistidine tag and Strep-tag (hereafter, underlined primer sequences denote restriction enzyme sites for cloning) (5′-TGTACAGTCACCATCATCACCATCACTGGAGCCACCCGCAGTTCGAAAAATAAGCGGCCGC-3′) was inserted into the BsrGI and NotI sites of wild type TDP-43 fused with meGFP at the C-terminal, as used in our previous study (pTDP43-GFP-HS).¹⁸ For purification of GFP-TDP25 and GFP monomers, synthetic cDNA fragment for polyhistidine tag and Strep-tag (5′-GCTAGCCACCATGCACCATCATCACCATCACTGGAGCCACCCGCAGTTCGAAAAACTACCGGT-3′) was inserted into the NheI and AgeI sites of pGFP-TDP-25 and pmeGFP-N1 (pHS-GFP-TDP-25 and pHS-GFP, respectively). The synthetic DNA fragment for M13-tagged RNA (M13N, M13-UG, M13-G4, M13-A4, M13-G2, and M13-A2) (Table S1) was ordered using DNA oligonucleotide synthesis services (Thermo Fisher, Waltham, MA, USA). The fragments were inserted into the BglII and HindIII sites of pSUPERIOR.neo (pSUPn-M13N, M13-UG, pSUPn-M13-G4, pSUPn-M13-A4, pSUPn-M13-G2, and pSUPn-M13-A2). An expression plasmid for GFP-tagged HSP70 (pHSP70-GFP) was modified from that for red fluorescent protein-tagged HSP70 in a previous report.³⁷ Expression plasmids for yellow fluorescent protein (YFP)-tagged β-actin (pEYFP-Actin) and GFP-tagged α-tubulin (pEGFP-Tub) were purchased from Takara Bio (Shiga, Japan). An expression plasmid for GFP-tagged mitochondrial pyruvate carboxylase (pCMV3-PC-GFPSpark; HG15615-ACG) was purchased from Sino Biological Japan (Kanagawa, Japan).

RNA Synthesis

All synthetic RNAs were ordered using RNA oligonucleotide synthesis services (AJINOMOTO BioPharma, Osaka, Japan). All sequences used in this study are presented in Table S2. For fluorescence assay, the 5′ end of the RNAs were labeled with Alexa Fluor 647, a fluorescent dye, followed by purified by high performance liquid chromatography using oligonucleotide synthesis and purification services (AJINOMOTO BioPharma). For introduction into the cells, synthetic RNAs with no modification (AJINOMOTO BioPharma) were used. The sequence of siRNA against renilla luciferase mRNA (siRL) was obtained from a reference.³⁸

Circular Dichroism (CD)

CD spectra of synthetic RNAs in an HD-containing buffer were recorded in a 1 mm path length cuvette at 20 °C using a spectropolarimeter (J-820, Jasco, Tokyo, Japan) equipped with a Peltier temperature controller in a 220–340 nm measurement range, 0.5 nm pitch wavelength, 1 nm bandwidth, 100 nm/min scanning speed. Spectral contribution from the buffer was subtracted from all CD spectra.

Cell Culture and Transfection

Neuro2a (2.0 × 10⁵) or Flp-In T-REx 293 cells (4.0 × 10⁵) were spread in a 35 mm glass-bottom dish for microscopic experiments (3910–035; IWAKI-AGC Technoglass, Shizuoka, Japan) or a 35 mm plastic dish for cell lysis (150318; Thermo Fisher) 1 day before transfection. Before spreading Flp-In T-REx 293 cells, the dishes were coated with 0.01 mg/mL poly-l-lysine. The mixture of pGFP-TDP-25 (1 μg) or pTDP43CS-GFP (1 μg) and pSUPERIOPR.neo (1 μg), pSUPn-M13N (0.25 μg mixed with 0.75 μg of the empty vector to normalized the RNA expression level), pSUPn-M13-G4 (1 μg), pSUPn-M13-A4 (1 μg), pSUPn-M13-G2 (1 μg), or pSUPn-M13-A2 (1 μg) were transfected using 5.0 μL of Lipofectamine 2000 (Thermo Fisher). After overnight incubation for transfection, the medium was replaced. Neuro2a and Flp-In T-REx 293 cells were cultured for 24 and 48 h after transfection, respectively, before microscopic analysis or cell lysis. For protein purification, Neuro2a cells (3.0 × 10⁶) were spread in a 150 mm plastic dish (TR4003; NIPPON Genetics, Tokyo, Japan) 1 day before transfection. A mixture of 14 μg of salmon sperm DNA (BioDynamics Laboratory Inc., Tokyo, Japan) and 16 μg of expression plasmid DNA (pTDP-43-GFP-HS, pHS-GFP-TDP-25, or pHS-GFP) was diluted in a 150 mM NaCl solution, followed by the addition of 77.5 μg polyethylenimine (PEI) (Fujifilm Wako Chemicals) in a 10 mM Tris-HCl buffer (pH 8.0). The mixture was added to the dishes after 15 min of incubation. After 24 h of incubation, the cells were lysed for protein purification. For purification, 30, 30, and 2 dishes were prepared for TDP-43, TDP-25, and GFP monomers, respectively. For sequential transfection of RNA and plasmids, Neuro2a cells (2.0 × 10⁵) were spread in a 35 mm glass-bottom dish (IWAKI-AGC Technoglass) 1 day before the first transfection. For RNA transfection pre-T25 vector introduction, after the cells were transfected with RNA (25 pmol) using 5.0 μL of Lipofectamine RNAi MAX (Thermo Fisher), they were incubated for 7 h. After the medium exchange, the cells were transfected with pGFP-TDP-25 (1 μg) using 2.5 μL of Lipofectamine 2000 (Thermo Fisher). The cells were cultured for 24 h before microscopic analysis. For RNA transfection post-T25 vector introduction, after the cells were transfected with pGFP-TDP-25 (1 μg) using 2.5 μl of Lipofectamine 2000 (Thermo Fisher), the cells were incubated for 25 h. After the medium exchange, the cells were transfected with RNA (25 pmol) using 5.0 μL of Lipofectamine RNAi MAX (Thermo Fisher). The cells were cultured for 24 h before the microscopic analysis.

Protein Purification

GFP-tagged protein-expressing Neuro2a cells were collected in tubes after trypsinization. Lysis buffer containing 50 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 1% Noidet P-40 (Fujifilm Wako Chemicals), and 1× protease inhibitor cocktail (P8340, Sigma-Aldrich) was added to a tube placed on ice, and the supernatant was recovered after centrifugation at 20,400 × g for 10 min at 4 °C. Ni-NTA agarose beads (25 μL slurry for GFP monomers and 75 μL slurry for TDP-43 and TDP25; Fujifilm Wako Chemicals) were added to the supernatant, followed by rotation and incubation for 60 min. The beads were washed three times in a buffer containing 50 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 20 mM imidazole, and 0.2% Noidet P-40. After the proteins were eluted in a buffer containing 50 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 250 mM imidazole, and 0.2% Noidet P-40, the protein solutions were applied to a spin column containing a Strep-Tactin XT Superflow suspension (Zymo Research, Irvine, CA, USA). After the column was washed in a buffer containing 50 mM HEPES-KOH (pH 7.5), 150 mM NaCl, and 0.2% Noidet P-40, the proteins were eluted in a buffer containing 50 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 0.2% Noidet P-40, and 50 mM biotin (Fujifilm Wako Chemicals). The proteins were concentrated in a buffer containing 20 mM HEPES-KOH (pH 7.5) and 0.2% Noidet P-40 using a dialysis spin column (Amicon Ultra-10; Merck, Darmstadt, Germany). The concentrations of purified fluorescent proteins were measured by using fluorescence correlation spectroscopy (FCS). The purified proteins were immediately used for the FCCS measurements. For mass spectroscopic analysis, the purified proteins were separated on a 5–20% gradient gel (ATTO, Tokyo, Japan) using SDS-PAGE, followed by silver staining using silver stain reagents (423413; Cosmo Bio, Tokyo, Japan). The bands of interest were cut and analyzed using MALDI-TOF-MS, followed by peptide mass fingerprinting, which was performed using an outsourced service (Genomine, Pohang, South Korea).

Fluorescence Cross-Correlation Spectroscopy

Before FCCS measurement, 10 μM Alexa Fluor 647-labeled RNAs were heat-denatured and cooled (7 °C/min) in a buffer containing 20 mM Tris-HCl (pH 8.0) and 100 mM KCl for the mixture of Gq and HP or LiCl for the only HP as described previously.³¹ The tertiary structure of these RNA was confirmed using circular dichroism and nondenatured electrophoresis.³¹ Neuro2a cells were lysed in a buffer containing 20 mM HEPES-KOH (pH 7.5), 1% Noidet P-40, and a 1 x protease inhibitor cocktail (Sigma-Aldrich). The supernatants of the lysates were recovered after centrifugation at 20,400 × g for 10 min at 4 °C. The supernatant was mixed with 100 nM RNA in KCl, MgCl₂, heparin, or HD-containing buffer and incubated for 5 min before the measurement. Double-strand 500 bp DNA as a positive control (PC-DNA) for FCCS was amplified by PCR with ATTO488- and ATTO647N-labeled primers and lambda DNA as a PCR template and was purified using MicroSpin S-400 HR columns (GE Healthcare, Chicago, IL, USA) as previously reported.³² The mixture of ATTO488- and ATTO647N-labeled primers (ATTO488-λDNA-R and ATTO647N-λDNA-F, respectively) was used as a negative control (NC-DNA) for FCCS. The DNA sequences as PC-DNA, ATTO488-λDNA-R, and ATTO647N-λDNA-F are represented in Table S3. FCCS was performed using LSM 510 META + ConfoCor3 (Carl Zeiss) with a C-Apochromat 40 × /1.2NA W. UV–VIS-IR water immersion objective lens (Carl Zeiss) as previously reported.¹⁸ The photons were recorded over 100 s. Curve fitting analysis for the acquired auto- and cross-correlation function was performed using ZEN software (Carl Zeiss) with a model for two-component 3D diffusion involving a one-component exponential blinking fraction in the fitting for autocorrelation function, as shown in the following eq 1, and with a model for two-component 3D diffusion without a blinking fraction (T = 0 in eq 1) in the fitting for cross-correlation function.

1

where G(τ) represents an auto- or cross-correlation function of the time interval τ; τ_D, i represents the diffusion time (i = 1 or 2); F_i represents the fraction of the component (i = 1 or 2); N represents the average number of particles in the confocal detection volume; T and τ_T represent the exponential blinking fraction and its meantime, respectively (these are zero for cross-correlation function); s represent the structural parameter determined by analyzing the standard fluorescent dyes (ATTO 488 and Cy5) on the same day. The relative cross-correlation amplitude (RCA) was calculated using eq 2.

2

where N_C and N_G represent the number of interacting and total green molecules, respectively. Dissociation constants (K_d) were calculated as reported previously.⁴⁸

RT-PCR

Total RNA of cells expressing M13-tagged RNA was isolated using an RNeasy Mini Kit (QIAGEN, Venlo, Netherlands) according to the manufacturer’s instructions. The RNA was eluted in diethyl pyrocarbonate-treated ultrapure water. To detect M13-tagged RNAs, total RNA (1.9 μg) was used to synthesize first-strand cDNA using the Mir-X miRNA First-Strand Synthesis Kit (Takara Bio Inc.) according to the manufacturer’s instructions. PCR was performed using a 0.25 U PrimeSTAR HS DNA polymerase (Takara Bio Inc.) with 0.2 mM dNTP mixture, 2 pmol of M13-forward primers (5′-gtaaaacgacggccagt-3′), 2 pmol of M13-reverse primers (5′-gagcggataacaatttcacacagg-3′), and template cDNA (corresponding to 15 ng of total RNA). Amplification was performed by using a PC320 thermal cycler (Astec-Bio, Fukuoka, Japan). The products were separated on a 5–20% polyacrylamide gel (ATTO) in Tris-glycine buffer. The gel was stained with SYBR Gold (Thermo Fisher), and gel images were acquired by using a Limited-STAGE II gel imager with blue LED light (AMZ System Science, Osaka, Japan).

Confocal Microscopy and Cell Counting

GFP-expressing cells were observed using an LSM 510 META microscope (Carl Zeiss, Jena, Germany) with a Plan-Apochromat 10×/0.45NA objective (Carl Zeiss). GFP was excited at a wavelength of 488 nm. The excitation beam was divided using an HFT405/488. Fluorescence was corrected via a 505–570 nm band-pass filter (BP505–570) and then acquired using a photomultiplier tube. The pinhole diameter was set to 72 mm. Transmission bright-field images were acquired simultaneously with the GFP fluorescence images. Aggregate-positive cells were counted by using the Fiji-ImageJ software. The population of cells that harbored cytoplasmic aggregates was normalized to the number of GFP-positive cells. Multiple fluorescent images were acquired at randomly selected locations, and then GFP-positive and aggregate-positive cells were counted using the all acquired images. Aggregate-positive cells were counted when the brightness of the structures was higher than that in the cytoplasm. The total number of GFP-positive cells was counted, ranging from 200 to 400 per trial. Statistical analysis of three independent transfections was performed (Student’s t test), and the results are plotted in the graph (trial number = 3; mean ± SE). Propidium iodide (PI) staining for determining dead cells was performed as previously reported.¹⁸ The population of PI-positive cells was normalized to the total number of GFP-positive cells. The total number of cells that were counted ranged from 392 to 735 for TDP-25 and from 761 to 1327 for TDP-43 per trial from the randomly acquired multiple fields of fluorescent images. Statistical analysis of three independent trials was performed (Student’s t test), and the results are plotted in the graph (trial number = 3; mean ± SE).

Fluorescence Recovery after Photobleaching

Photobleaching experiments were performed with an LSM 510 META system using a C-Apochromat 40×/1.2NA W Korr. UV–VIS-IR M27 objective lens (Carl Zeiss), as previously reported.^18,36,37 GFP was excited and photobleached at 488 nm. The GFP fluorescence was collected through a band-pass filter (BP505–570). The pinhole size was set to 200 μm, the zoom factor was set to 5×, and the interval time for image acquisition was set to 10 s. The X- and Y-scanning sizes were 512 pixels, and GFP was photobleached at 488 nm with 100% transmission for less than 1.92 s. Relative fluorescence intensities (RFIs) were measured using the Fiji-ImageJ software and calculated as reported previously.^18,36,37 Normalized RFIs were calculated using the mean RFI of T25 and T43CS in the aggregates with an empty vector transfection just after photobleaching.

Cell Lysis, Solubility Assay, and Filter Retardation Assay

Cells expressing T25 and T43CS were lysed and fractionated as reported previously.¹⁸ Briefly, cells were lysed in a lysis buffer containing 25 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 1% Noidet P-40, 0.1% SDS, 0.25 U/mL benzonase (Sigma-Aldrich), and 1× protease inhibitor cocktail (Sigma-Aldrich). After centrifugation (20,400 × g, 10 min, 4 °C), the fractionated supernatants were recovered, and the pellets were solubilized in PBS containing 1 M urea. To recover the whole cell lysates without the fractionation, cells were lysed in PBS containing 1 M urea, 1% SDS, 0.25 U/mL of benzonase, and 1× protease inhibitor cocktail (Sigma-Aldrich). SDS-PAGE sample buffer-mixed lysates were denatured at 98 °C for 5 min. The samples were applied to a 12.5% polyacrylamide gel (ATTO) and subjected to electrophoresis in an SDS-containing buffer. Proteins were transferred onto polyvinylidene difluoride membranes (Cytiva, Marlborough, MA, USA). The membrane was blocked with a PBS-T buffer containing 5% skim milk for 1 h. Horseradish peroxidase-conjugated anti-GFP (#598–7, MBL, Tokyo, Japan) or anti-α-tubulin antibodies (#HRP-66031, Proteintech, Rosemont, IL, USA) were reacted in CanGet Signal Immunoreaction Enhancer Solution 1 (TOYOBO, Osaka, Japan). To detect endogenous HSP70, an anti-HSP70 antibody (#10995–1-AP, Proteintech) was reacted in the blocking buffer and then reacted with a horseradish peroxidase-conjugated antirabbit IgG antibody (#111–035–144, Jackson ImmunoResearch, West Grove, PA, USA). Chemiluminescent signals were acquired using a ChemiDoc MP imager (Bio-Rad, Hercules, CA, USA). As an internal control for the insoluble fraction, the insoluble fractions were applied to a 12.5% polyacrylamide gel (ATTO) and subjected to electrophoresis in an SDS-containing buffer, followed by staining with a Coomassie brilliant blue R-250 (Fujifilm Wako Chemicals). Images of the stained gels were obtained by using a Limited-STAGE II gel imager with transparent white light (AMZ System Science). For the filter retardation assay, the cells were lysed in a PBS buffer containing 1% SDS and 0.25 U/mL of benzonase (Sigma-Aldrich). The lysates were applied to a cellulose acetate membrane with a pore size of 0.2 μm (Advantec Toyo, Tokyo, Japan) by using a Bio-Dot SF blotter (Bio-Rad). The membrane was blocked in PBS-T buffer containing 5% skim milk for 1 h. The GFP on the membrane was detected by Western blotting, as described above. All of the intensities were measured using Fiji-ImageJ. After subtracting the values in the background region where there are no bands, the ratio value of the intensity of T43CS/T25 in the insoluble fraction or trapped on the filter to the control was calculated.

FACS Analysis

RNA-transfected Neuro2a and 293 cells were prepared the same as for cell lysis and confocal microscopy. Cells were trypsinized and suspended in a medium containing FBS. After centrifugation (1,000 × g, 3 min, 25 °C), the cells were suspended in Opti-MEM I (Thermo Fisher). The cell concentration was adjusted to 6.0 × 10⁵ cells/ml after cell counting. Subsequently, 3 μM DRAQ7 (BioStatus, Leicestershire, UK) was added for dead cell staining. Immediately before FACS analysis, the cells were passed through a cell strainer (no. 352235, Corning Incorporated, Corning, NY, USA), and fluorescence of DRAQ7-positive cells was detected using a 668–722 nm band-pass filter with 638 nm excitation on a JSAN desktop cell sorter (Bay Bioscience, Brookline, MA, USA). The percentage of dead cells was determined based on higher fluorescence intensity.

RNA-seq and Data Processing

Total RNA of the cells expressing M13-tagged RNA was recovered in the same manner as that of RT-PCR. The total RNA concentration was measured using a Quantus Fluorometer with a QuantiFluor RNA system (Promega). The quality of total RNA was examined using a fragment analyzer system with an Agilent HS RNA kit (Agilent Technologies). Libraries were prepared using an MGIEasy RNA Directional Library Prep Set (MGI Tech Co., Ltd.) according to the manufacturer’s instructions. The concentration of the libraries was measured using a Qubit 3.0 Fluorometer with a dsDNA HS assay kit (Thermo Fisher). The quality of the libraries was analyzed using a fragment analyzer with a dsDNA 915 reagent kit (Agilent Technologies). Circular DNA was created using the libraries and an MGIEasy Circularization kit (MGI Tech Co., Ltd.), according to the manufacturer’s protocol. DNA Nano Ball (DNB) was prepared using a DNBSEQ-G400RS high-throughput sequencing kit (MGI Tech Co., Ltd.) according to the manufacturer’s protocol. DNBs were sequenced using a DEBSEQ-G400 sequencer (MGI Tech Co., Ltd.). After removing the adaptor sequence using Cutdapt ver. 1.9.1 software, Pair reads that were less than 40 bases and a quality score of less than 20 were removed using a sickle ver. 1.33 software. Mapping was performed by referring to a murine genome reference (GRCm39) using the hisat2 ver. 2.2.1 software. The number of reads was counted using the featureCounts ver. 2.0.0 software. Reads per kilobase per million (RPKM) and transcripts per million (TPM) values were calculated. After removing transcripts with low read counts (<19), TPM values derived from total RNA obtained four times, independently, were averaged. The mean fold change and p-values were calculated for cells expressing M13-G4/A4 in comparison with cells expressing M13-UG. Volcano plots were prepared by using Microsoft Excel.

A.F. was supported by the Support for Pioneering Research Initiated by the Next Generation (SPRING) program by the Japan Science and Technology Agency (JST) at Hokkaido University (JPMJSP2119). M.K. was supported by a Japan Society for Promotion of Science (JSPS) Grant-in-Aid for Scientific Research (B) (22H02578) and Grant-in-Aid for Challenging Research (Exploratory) (22K19886). A.K. was supported by grants from Japan Agency for Medical Research and Development (AMED) (JP23gm6410028 and JP22ym0126814); by grants from JSPS Grant-in-Aid for Scientific Research (24H02286; 22H04826; 18K06201; and 16KK0156); by a grant from Hokkaido University Office for Developing Future Research Leaders (L-Station); by a grant from Hoansha Foundation; by a grant from Hagiwara Foundation of Japan; and by a grant from Nakatani Foundation. We would like to thank Ms. Asuka Murata and Mr. Koya Yoshizawa for technical support, Editage (www.editage.com) for English language editing, and the Open Facility, Global Facility Center, Creative Research Institution, Hokkaido University for using a ChemiDoc MP imager and a JSAN desktop cell sorter.