Importance of the Q/N-rich Segment for Protein Stability and Activity of Endogenous Mouse TDP-43

TAR DNA-binding protein 43 kDa (TDP-43), a nuclear protein, plays an important role in the molecular pathogenesis of amyotrophic lateral sclerosis (ALS). TDP-43 aggregation and translocation out of the nucleus are crucial factors in ALS. TDP-43 aggregation results from its resistance to degradation, to which the long-disordered C-terminal region (CTR) is thought to contribute. The CTR has two Gly, aromatic, and Ser-rich (GaroS) segments and an amyloidogenic core divided into a hydrophobic patch and a Gln/Asn (Q/N)-rich segment. Although TDP-43 lacking the CTR is known to be unstable, as observed in knock-in mice, it is unclear which of these segments contributes to the stability of TDP-43. Here, we generated 12 mouse lines lacking the various sub-regions of CTR by genome editing and compared the protein stability, activity, and subcellular localization of TDP-43. We demonstrated the functional diversity of the four segments of CTR, �nding that the presence of Q/N-rich segment greatly restored the protein stability and activity of TDP-43. In addition, we found that the second GaroS deletion did not affect protein stability and mouse development.


Introduction
TAR DNA-binding protein 43 kDa (TDP-43), a member of the heterogeneous nuclear ribonucleoprotein family, has emerged as a key player in the molecular pathogenesis of amyotrophic lateral sclerosis (ALS) 1,2 . TDP-43 mainly localizes in the nucleus and is involved in RNA metabolism, including splicing, that promotes cystic brosis transmembrane conductance regulator (CFTR) exon 9 skipping 3 . Moreover, TDP-43 is essential for mouse development [4][5][6] . In ALS, cytoplasmic aggregation of TDP-43 is a pathological feature that is often accompanied by nuclear loss of TDP-43 7,8 . Because the amount of nuclear TDP-43 is strictly regulated by binding of TDP-43 to the 3′-untranslated region of its own mRNA, thereby forming an autoregulatory negative feedback loop [9][10][11][12][13] , nuclear loss of TDP-43 can induce the upregulation of TDP-43 mRNA and produce a vicious cycle that leads to the perturbation of TDP-43 homeostasis 14,15 . This imbalance of aggregation-prone TDP-43 plays a role in the pathological mechanism of ALS.
A long-disordered C-terminal region (CTR), known as a prion-like domain 2 , is also thought to be important for the stability of TDP-43 and its resistance to degradation. The CTR can be divided into four subregions, including a rst Gly, aromatic, and Ser-rich (GaroS1) segment; a hydrophobic patch (HP); a Gln/Asn (Q/N)-rich segment; and a second GaroS2 segment ( Fig. 1) 16,17 . In the CTR, an amyloidogenic core of amino acid (aa) residues 311-360 has been identi ed to be critical for aggregation 16 . In addition, the CTR is a hotspot for ALS-causing TARDBP mutations. These mutations accelerate aggregate formation, with the result that fragmented and phosphorylated C-termini of TDP-43 accumulate in the affected tissues of ALS patients [18][19][20] . Recently, TDP-ΔC (aa 274-414) knock-in mice were generated, and they showed embryonic lethality of homozygous mice, and that TDP-43 lacking CTR is unstable 21 .
However, it is unclear which of the segments of the CTR contribute to the stability of TDP-43, indicating the importance of analyzing mice with deletions of more localized regions within the CTR. In this study, we generated 12 mouse lines lacking the various sub-regions of the CTR by genome editing and examined their effects on protein stability, activity, and subcellular localization of TDP-43 to clarify the functional diversity of the C-terminal sub-regions.

Results
Protein stability and activity restoration are dependent on the length of the TDP-43 CTR.
To determine the segment necessary to avoid embryonic lethality of TDP-ΔC knock-in mice 21 , we established 12 mouse lines using the zinc nger nuclease (ZFN) and CRISPR/Cas9 systems, which were used to target the Q/N-rich (N345-Q346) and GaroS2 (Y374) segments, respectively (Fig. 1, Supplementary Figs. S1, S2). The 12 mouse lines were categorized into ve groups based on the putative protein structures in the CTR (Fig. 1): (1) deletion within a 6-aa region of the Q/N-rich segment (ZE13A/Δ9 and ZE13B/Δ18), (2) deletion within a 6-aa region of the GaroS2 segment (CRY9A/Δ3 and CRY8A/Δ18), (3) frame-shift deletions of the GaroS2 segment (CRY8B/Δ8, CRY9B/Δ29), (4) deletion of most of Q/N-GaroS2 (ZE10/Δ10, ZM39P/Δ295, and ZM87/Δ2), and (5) a large deletion including the GaroS1-HP segment (ZM39A/Δ261). All heterozygous mice were fertile and showed no obvious motor phenotypes during daily handling. Heterozygous intercrosses of each mouse line showed that homozygous mice were viable in Groups 1, 2, and 3, whereas they were embryonic lethal in Groups 4 and 5 (Table 1). Since embryonic lethality was recovered in Group 3 compared to Groups 4 and 5, the GaroS1-HP-Q/N-rich segment is thought to be necessary and su cient for mouse development. To estimate the stability of mutant proteins, we performed subcellular fractionation of eight-week-old mouse cerebrum and western blot analysis using the N-260 anti-TDP-43 antibody. In Group 1 homozygous mice, mutant TDP-43 was observed mainly in the nucleus with a mobility shift consistent with a 3 or 6 aa deletion (Fig. 2a). In Group 3 heterozygous and homozygous mice, the intensity of the band corresponding to the GaroS2 deletion was comparable to that of wild-type (WT) TDP-43 depending on the number of alleles (Fig. 2b). However, mutant bands corresponding to most of the Q/N-GaroS2 deletion were not observable in the nuclear or cytoplasmic fractions of Group 4 heterozygous mice ( Fig. 2c). Because protein stability was restored in Group 3 compared to Group 4, similar to embryonic lethality, the Q/N-rich segment is required for TDP-43 stability and activity.
These results indicate that TDP-43 lacking most of Q/N-GaroS2 has residual activity only in ZM87/Δ2 mice. Thus, we evaluated cerebellar tissue from ZM87/Δ2 mice at E19.5 and detected weaker bands both in heterozygotes and homozygotes compared to those of WT ( Supplementary Fig. S3b). Furthermore, the ZM39P/Δ295 band was detected in unfertilized oocytes at an intensity that was comparable to those of WT and the ZM39A/Δ261 bands ( Supplementary Fig. S3c), suggesting that TDP-43 lacking most of Q/N-GaroS2 is rapidly degraded in the cerebrum at eight weeks of age.
The GaroS2 segment negatively regulates its own activity.
Since TDP-43 strictly regulates its own mRNA levels through autoregulation 9-13 , its residual activity can be estimated by evaluating its own mRNA level. Thus, we performed qRT-PCR analysis using the following ve primer sets: Hprt1 (internal reference), λpolyA (external control), Tardbp exon 3 to 4 (Exon 3-4), Tardbp exon 5 to 6 (Exon 5-6), and Tardbp exon 6 (WT Exon 6). Exon 3-4 and Exon 5-6 recognize both WT and mutant alleles, whereas one of the WT Exon 6 primers targets deleted sequences induced by genome editing, thus WT Exon 6 recognizes the WT allele alone. In Group 1 homozygous mice, the total expression levels of Tardbp mRNA did not change in comparison with those of WT mice (Fig. 2d). However, the total expression levels were signi cantly increased in Groups 3 (1.247 to 1.448) and 5 (1.853 to 1.974), indicating compensation for the decrease in TDP-43 autoregulatory activity. In sharp contrast, the total expression levels in both Group 4 heterozygotes and homozygotes were signi cantly decreased (0.850 to 0.887) compared to those of WT mice (Fig. 2e). These results suggest that the residual activity of mutant TDP-43 is increased due to the deletion of the GaroS2 segment, which presumably acts as a negative regulator.
CTR contributes to nuclear localization.
As shown in Fig. 2c, Group 5 heterozygous mice (ZM39A/Δ261) exhibited a strong signal for nonfunctional TDP-43 lacking the GaroS1-HP segment, which was detected in the cytoplasmic fraction.
Although the non-functional TDP-43 had a bipartite nuclear localization signal (NLS, aa 82-98, Fig. 1) 22,23 , it was not observed in the nuclear fraction. We performed immunohistochemical analysis of the spinal anterior horn region stained with N-260 anti-TDP-43 antibody, but the mis-localization of mutant TDP-43 was not detected (Fig. 3a). To con rm whether a similar result occurs in human TDP-43, we performed transient expression analysis using HeLa cells with enhanced green uorescent protein (EGFP) tagged to the N-terminus of human TDP-43 constructs lacking the GaroS1 segment (dGaroS1), a region equivalent to ZM39A/Δ261 (d261CTR) and the entire CTR (dCTR) (Fig. 3b). Similar to the mouse immunohistochemical analysis, three mutant and WT TDP-43s were localized in the nuclei of living cells, whereas EGFP alone was diffusely localized in cell bodies (Fig. 3c). In contrast to the uorescence imaging, subcellular fractionation followed by western blotting showed that the three mutant TDP-43s were strongly detected in the cytoplasmic fraction (Fig. 3d). Although the reason is unclear, a similar discrepancy has been reported using constructs with a near total deletion of CTR (aa 314-414) 23,24 .
Taking these reports into account, large deletions of CTR appear to disturb the nuclear localization only when assessed by subcellular fractionation and western blotting.

Discussion
We Our results strongly suggest that mTDP-Ss are rapidly degraded in the cerebrum. In addition to protein stability, we found that the GaroS2 segment is not essential, but potentially acts as a negative regulator of its own autoregulatory activity.
The protein structure in solution exhibits two canonical α-helices (321-330 and 335-343) in the HP segment 16 , which overlaps a highly conserved region (CR, 320-340) anked by two intrinsically disordered regions (IDR1 and IDR2) 26  It has been shown that removing the GaroS2 segment has no apparent effect on either the TDP-43 activity of CFTR exon 9 skipping or intra-cellular localization in cultured cells 23,33 . These reports support our results showing no apparent phenotype in CRY9B/Δ29 mice lacking the GaroS2 segment. However, there are several mutations in the GaroS2 segment that cause familial ALS, and the Y374X mutation, similar to the CRY9B/Δ29 mutation, has been found in sporadic ALS 34 . Furthermore, two steric zippers (aa 370-375 and 396-402) are present in the GaroS2 segment 30 . It is noteworthy that the GaroS2 segment is subjected to phosphorylation at a minimum of seven serine residues (aa 389, 393, 395, 403, 404, 409, and 410) [18][19][20] , which may serve to prevent excessive aggregation 31 . Considering that disordered proteins are generally regulated by post-translational modi cations (PTMs) 35 , it may be that the GaroS2 segment is not essential for basic TDP-43 activity, but rather regulates its own activity, presumably due to PTMs including phosphorylation.
We used western blotting to show that human TDP-43 with larger CTR deletions localized mainly in the cytoplasmic fraction, whereas the protein constructs were present in the nuclei of living cells when imaged using uorescence microscopy. In a pioneering study using U2OS cells and CTR deletion constructs, the observed immuno uorescence patterns of subcellular localizations were variable, whereas biochemical fractionation showed that cytoplasmic localization was dependent on the length of the CTR deletion 23 . Recently, TDP-43 was reported to demix in the cytoplasm as well as the nucleus using CTR 36 .
The demixing of RNA-free TDP-43 in the nucleus induces anisotropic intranuclear liquid spherical shells (anisosomes), and the role of the demixing is speculated to tether RNA binding proteins, including TDP-43, near the site of transcription 37 . Furthermore, a nuclear export signal (aa 239-250) in TDP-43 appears to be non-functional, and a signi cant amount of TDP-43 is exported to cytoplasm through passive diffusion 38,39 . These observations raise the possibility that TDP-43 with large CTR deletions can be imported to the nucleus, but leak out during biochemical fractionation due to insu cient nuclear retention. Although further analysis and mouse models are required to understand the full functional diversity of the TDP-43 C-terminal sub-regions, we expect that the results presented here will help elucidate the pathogenesis of ALS and perhaps lead to the development of new ALS treatments.

Methods
Generation of ZE, ZM, and CRY mouse lines.
CompoZr Custom ZFNs (Sigma-Aldrich, St. Louis, MO) were designed to target Tardbp exon 6 and manufactured as ZFN binding (uppercase) and cutting (lowercase) sites (GTTAGCCAGCCAGCAGaaccagTCGGGCCCATCTGGGA). The ZFN mRNAs were microinjected into the pronuclei of B6C3F1/Jcl fertilized oocytes to generate ZE lines according to the manufacturer's instructions. Next, a ssDNA oligonucleotide (Supplementary Fig. S1a) was co-injected with the ZFN mRNAs as a donor template to generate the ZM lines carrying large deletions.
To generate GaroS2-de cient mice, the guide RNA (gRNA) target sequence (TCTGGAAATAATTCCTACAGtgg, lowercase indicating protospacer adaptor motif, PAM) was designed to target Tardbp exon 6 encoding the N-terminal region of the GaroS2 segment. The gRNA was synthesized commercially (Thermo Fisher Scienti c, Waltham, MA). The gRNA and CAS9 protein (Integrated DNA Technologies, Coralville, IA) were electroporated in fertilized zygotes of C57BL/6JJcl mice using previously described methods with minor modi cations 40 . Electroporation was performed in HEPESbuffered Whitten's medium using platinum plate electrodes on tempered glass (LF501PT-1; BEX, Tokyo, Japan). The pulse conditions of CUY21EDIT II (BEX) were 30 V (3 ms pulse duration, 97 ms interval) × 7 times. After electroporation, zygotes were incubated in Whitten's medium for 24 h at 37°C and 5% CO 2 , and the surviving two-cell-stage embryos were transferred into the oviducts of pseudo-pregnant female mice.
The genome-edited founder mice were crossed with C57BL/6JJcl mice to obtain heterozygous mouse lines. The genotyping conditions are described in Supplementary Table S1. All animal experiments were carried out in accordance with the guidelines of the National Institutes of Health, and the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, and were approved by the Dean of Kitasato University School of Medicine based on judgment by the Institutional Animal Care and Use Committee (Approval no. 2020-061) in compliance with ARRIVE guidelines (https://arriveguidelines.org).
All mice were maintained under speci c pathogen-free conditions with free access to CE-2 standard food (CLEA, Tokyo, Japan) and ultra ltered water. For the mouse lines where embryonic lethality was expected, including ZM39P/Δ295, the analysis at E12.5 was prioritized to reduce the number of mice.
Subcellular fractionation and western blotting. Total RNA preparation and qRT-PCR analysis.
To monitor the measurement process, we spiked external RNA (λpolyA + RNA-A) from an External Standard Kit for qPCR (Takara, Shiga, Japan) into the cerebral homogenate at a concentration of 1.8 × 10 6 copies/mg brain tissue, and simultaneously extracted total RNA from the cerebrum using RNAiso Plus (Takara). First-strand cDNA was synthesized using a SuperScript VILO cDNA synthesis Kit (Invitrogen, Waltham, MA). Real-time qRT-PCR was performed using SYBR Premix Ex Taq II and a Thermal Cycler Dice Real-time System (Takara). Each cDNA sample was measured in duplicate applying the ΔΔCT method. Statistical analyses were performed using the REST2009 program (http://www.gene-quanti ca tion.de/rest-2009.html). The primer sets are listed in Supplementary Table S1. Immunohistochemistry.
Cell culture and uorescence imaging.
Declarations Figure 1 Schematic diagram of mouse TDP-43 and putative protein structures with various CTR deletions. TDP-43 contains 414 aa residues and consists of a bipartite nuclear localization signal (NLS), two RNA recognition motifs (RRM), and a long-disordered CTR. CTR is divided into four sub-regions, the GaroS1 (yellow), HP (green), Q/N-rich (blue), and GaroS2 segments (purple); the aa sequences are depicted by one-letter codes according to the TDP-43 CTR. The genome-edited mice were categorized into ve groups based on the putative CTR structures. Each deleted region is drawn as a red dotted line. The putative aa sequence in each mouse line is shown in black and red letters indicating WT and substituted aa residues, respectively. The gaps in the aa sequences are represented as red dashes. The characteristics of each group determined in this study are shown on the right. Mouse, embryonic lethality of homozygous mice.
Protein, protein stability in the eight-week-old mouse cerebrum. mRNA, total expression level of the Tardbp mRNA. NE, not examined. Note that the putative protein structures of ZM93/Δ3 and ZM91/Δ1 mice (gray) are similar to those of ZM13A/Δ9 and ZE10/Δ10, respectively, and were not used in this analysis.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. 20211125SupplToshiyaSatoR.docx