Altered Distributions of Gemini of Coiled Bodies and Mitochondria in Motor Neurons of TDP-43 Transgenic Mice

Altered distributions of Gemini of coiled bodies and mitochondria in motor neurons of TDP-43 transgenic mice

Xiu Shana,b,1, Po-Min Chianga,b,1, Donald L. Pricea,b,c,d, and Philip C. Wonga,b,c,2

aDepartment of Pathology, bDivision of Neuropathology, cDepartment of Neuroscience, and dDepartment of Neurology, The Johns Hopkins University School of Medicine, Baltimore, MD 21205-2196

Edited* by Richard L. Huganir, The Johns Hopkins University School of Medicine, Baltimore, MD, and approved August 4, 2010 (received for review March 16, 2010) TAR DNA-binding protein-43 (TDP-43), a DNA/RNA-binding protein possibility that altered RNA metabolism or RNA processing may involved in RNA transcription and splicing, has been associated with underlie and contribute to motor neuron degeneration (2). the pathophysiology of neurodegenerative diseases, including ALS. Despite recent advances in development of TDP-43 transgenic However, the function of TDP-43 in motor neurons remains unde- models in mice (11, 12) and flies (13), no experimental evidence fined. Here we use both gain- and loss-of-function approaches to is currently available to support the view that TDP-43 partic- determine roles of TDP-43 in motor neurons. Mice expressing hu- ipates in pathways that regulate RNA processing in motor neu- man TDP-43 in neurons exhibited growth retardation and prema- rons. To begin to address this issue, we generated mice either ture death that are characterized by abnormal intranuclear lacking endogenous TDP-43 or expressing human TDP-43 in inclusions composed of TDP-43 and fused in sarcoma/translocated neurons, including motor neurons. Here we provide evidence to in liposarcoma (FUS/TLS), and massive accumulation of mitochon- support TDP-43 in regulating the physiology of motor neurons, dria in TDP-43-negative cytoplasmic inclusions in motor neurons, including those that impact on the proper distributions of mi- lack of mitochondria in motor axon terminals, and immature neu- tochondria in the cytoplasm and of fused in sarcoma/trans- romuscular junctions. Whereas an elevated level of TDP-43 disrupts located in liposarcoma (FUS/TLS) and SMN-associated Gemini the normal nuclear distribution of survival motor neuron (SMN)- of coiled bodies (GEMs) in the nucleus. Together with results NEUROSCIENCE associated Gemini of coiled bodies (GEMs) in motor neurons, its from our TDP-43 conditional knockout mouse model, our find- absence prevents the formation of GEMs in the nuclei of these cells. ings implicate a critical role of TDP-43 in controlling the for- Moreover, transcriptome-wide deep sequencing analysis revealed mation of SMN-associated GEMs that may impact on RNA fi that a decrease in abundance of neuro lament transcripts contrib- metabolism in motor neurons. uted to the reduction of caliber of motor axons in TDP-43 mice. In concert, our findings indicate that TDP-43 participates in pathways Results critical for motor neuron physiology, including those that regulate Growth Retardation, Muscle Weakness, and Death in TDP-43 Transgenic the normal distributions of SMN-associated GEMs in the nucleus and Mice. Several lines (W1, W2, and W3) of mice expressing wild-type mitochondria in the cytoplasm. human TDP-43 (hTDP-43) were generated using the Thy1.2 promoter (Fig. S1), which is capable of driving expression post- ALS | RNA metabolism | frontotemporal lobar degeneration with natally in neurons, including motor neurons. TDP-43 mice ubiquitinated inclusions | fused in sarcoma/translocated in liposarcoma | exhibited retardation of development when compared with non- survival motor neuron transgenic littermates (Fig. 1A). The severities of this phenotype correlated with the copy number of the transgene: Mice from line dentified first as a regulator of HIV gene expression, TAR W1, with the highest transgene copy number (Fig. S1), were IDNA-binding protein (TDP-43) is a DNA/RNA-binding protein markedly smaller than their nontransgenic littermates (Fig. 1A) that contains two RNA-recognition motifs and a glycine-rich C- and die within 3 wk of age, whereas mice derived from W2 and – terminal domain thought to be important for mediating protein W3, two lines harboring lower numbers of transgenes (Fig. S1), protein interactions (1, 2). Although TDP-43 has been implicated exhibited growth retardation to a lesser extent (Fig. 1A), and most as a key factor regulating RNA splicing of human cystic fibrosis of those mice grew to adulthood. Interestingly, male TDP-43 mice transmembrane conductance regulator (CFTR) (3), Apolipopro- exhibited ≈20% reduction in body weight at 4 wk of age when tein A-II (4), and Survival Motor Neuron (SMN) (5), the im- compared with nontransgenic male littermates (Fig. 1B). We ob- portance of TDP-43 in the central nervous system had not been served that transgenic males derived from both W2 and W3 lines fi demonstrated until it was identi ed as a component of ubiquiti- exhibited a more severe phenotype when compared with trans- nated protein aggregates in cases of amyotrophic lateral sclerosis genic female littermates, an outcome that appears to be related to (ALS) and frontotemporal lobar degeneration with ubiquitinated inclusions (FTLD-U) (6). In both of these diseases, TDP-43 is

depleted from the nuclei but accumulates in the ubiquitinated Author contributions: X.S., P.-M.C., and P.C.W. designed research; X.S. and P.-M.C. inclusions of affected neurons, suggesting that loss of normal performed research; X.S., P.-M.C., D.L.P., and P.C.W. analyzed data; and X.S., P.-M.C., function of TDP-43 as a nuclear protein or, alternatively, gain of D.L.P., and P.C.W. wrote the paper. a toxic function by TDP-43 aggregates play significant roles in the The authors declare no conflict of interest. pathogenesis of ALS and FTLD-U (1). Moreover, the identifi- *This Direct Submission article had a prearranged editor. cation of mutations in TDP-43 that are linked to both sporadic Freely available online through the PNAS open access option. and familial ALS (2, 7, 8) provides evidence that TDP-43 directly Data deposition: The data reported in this paper have been deposited in the Gene Ex- contributes to the pathogenesis of these neurodegenerative dis- pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE22351). orders. However, the exact mechanisms by which mutant TDP-43 1X.S. and P.-M.C. contributed equally to this work. contributes to ALS remain elusive. Interestingly, recent discov- 2To whom correspondence should be addressed. E-mail: [email protected]. FUS/TLS eries of mutations in , a gene that encodes another This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. RNA-binding protein, linked to ALS (9, 10) offer the intriguing 1073/pnas.1003459107/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1003459107 PNAS Early Edition | 1of6 Downloaded by guest on September 27, 2021 Fig. 1. Early postnatal growth retardation in mice expressing wild-type TDP-43. (A) Decrease in size of wild-type TDP-43 transgenic (tg) male mice derived from independent founders W1 and W3; asterisks indicate, respectively, transgenic mice at 14 and 21 d of age. (B) Body weights of 4-wk- Fig. 2. Pathological abnormalities in spinal cords of 3-wk-old W3 TDP-43 old male TDP-43 transgenic mice from line W3 were compared with non- mice. (A and B) Hematoxylin and eosin staining reveals eosinophilic aggre- transgenic (ntg) littermates. Note significant reduction in body weights of gates in cell bodies of motor neurons in spinal cords of transgenic (tg) mice TDP-43 transgenic mice (ntg, n = 13; W3 tg, n = 15; P < 0.0001). Error bars (B, arrows); such structures were not identified in nontransgenic (ntg) lit- indicate SEM. (C) Accumulation of TDP-43 in spinal cords of W3 mice at 4 wk termates (A). Boxes at bottom right are enlarged micrographs showing of age. A mouse monoclonal antibody recognizing human-specific TDP-43 a healthy motor neuron (A) and a neuron bearing a large cytoplasmic ag- was used to determine the level of transgene expression, whereas a rabbit gregate marked by an arrowhead (B). (C–E) The antibody recognizing human antibody against both mouse and human TDP-43 was used to compare the TDP-43 specifically reveals that the transgene is extensively expressed in level of transgene expression with that of endogenous TDP-43. f, female; m, spinal cord neurons of transgenic mice (D and E); no immunoreactivity is male. (D) Densitometric analysis of TDP-43 protein levels in 4-wk-old W3 detected in nontransgenic mice (C). hTDP-43 is localized in the nucleus and mice using the rabbit antibody against both mouse and human TDP-43 forms intranuclear granular structures (E, arrowheads) in some neurons (female tg, n = 3; male tg, n = 6; ntg littermates, n = 4). Compared with that bearing cytoplasmic aggregates (E, arrow), which are identified by eosin of mouse endogenous TDP-43, the levels of human TDP-43 are, respectively, counterstain. (F and G) Immunohistochemical analysis with ubiquitin anti- 1.3- and 3.6-fold in 4-wk-old W3 transgenic females (P = 0.0011) and males body shows that the level of ubiquitination is elevated in the spinal cords of (P < 0.0001). Error bars indicate SEM. transgenic mice (G) when compared with that in nontransgenic mice (F). Arrows point to neurons with eccentric nuclei, indicating the presence of cytoplasmic aggregates, and those neurons are heavily stained with ubiq- a higher (2- to 3-fold) accumulation of TDP-43 in transgenic males C D TDP-43 uitin, particularly within the nuclear compartments. (H and I) Double im- (Fig. 1 and ). Male mice abruptly developed severe munofluorescence analyses using antisera against HSP60, a mitochondrial tremor, abnormal reflex of hindlimbs (Fig. S1), and gait abnor- marker, and hTDP-43 suggest the abnormal accumulation of mitochondria malities within a short time window, ranging from postnatal day 14 (I, arrowheads) in transgene-expressing neurons delineated by dashed lines. to day 18. Female TDP-43 mice did not show such significant Arrows in I point to staining of hTDP-43 in the nucleus. Note the normal reduction in body weight and developed fine tremor only after 3 distinct distribution of HSP60 in the cytoplasm of motor neurons of non- mo of age. Because transgenic mice from lines W2 and W3 transgenic mice (H). (J–L) Electron micrographs of spinal motor neurons show exhibited similar behavioral phenotypes and pathology in addition massive accumulation of mitochondria (L; arrows denote mitochondria) to comparable levels of transgene expression, we focused our within large, cytoplasmic aggregates in transgenic mice (K; arrows point to a cytoplasmic aggregate). [Scale bars, 20 μm(A–B and E–I), 1 μm(J–L).] subsequent analyses using line W3.

Abnormal Distribution of Mitochondria in Motor Neurons of TDP-43 fi arrows). However, ubiquitin immunoreactivity was more intensive Transgenic Mice. Consistent with our observation of motor de cits in the nuclear as compared with the cytoplasmic compartment, in these lines of mice, histological analyses revealed eccentric suggesting that the cytoplasmic inclusions may not be composed of nuclei with abnormal eosinophilic aggregates in cell bodies of highly ubiquitinated proteins. To ascertain the composition of motor neurons in spinal cord (Fig. 2 A and B and Fig. S2A) and these cytoplasmic aggregates, we performed immunocytochemical brainstem. To investigate whether the transgenic product was analyses using a series of markers for various organelles. Markers a component of the eosinophilic aggregates, sections of spinal of endoplasmic reticulum and Golgi were not associated with cord and brainstem were stained with antibodies recognizing B specifically human TDP-43. Although human TDP-43 can be these cytoplasmic aggregates (Fig. S2 ), whereas several mito- localized to nuclei of motor neurons, we failed to observe TDP- chondrial markers including mitochondrial chaperonin HSP60 I 43 immunoreactivity associated with these cytoplasmic inclusions (Fig. 2 ) and mitochondrial voltage-dependent anion channel (Fig. in mice ranging from 3 wk (Fig. 2 C–E) to 3 mo of age. In- S3) localized to these cytoplasmic inclusions in many motor neu- terestingly, some of the human TDP-43-immunoreactive nuclei rons expressing human TDP-43, strongly indicating that these were eccentric, suggesting the presence of abnormal cytoplasmic eosinophilic aggregates are composed, in part, of accumulations of aggregates in these motor neurons (Fig. 2E, arrow). Moreover, mitochondria. Ultrastructural analysis of motor neurons from 3- human TDP-43 often was localized to two prominent intra- wk-old TDP-43 mice confirmed these observations. Mitochondria nuclear structures associated with eccentric nuclei in motor are normally evenly distributed within the cell bodies of neurons neurons (Fig. 2E, arrowheads). Increased ubiquitin immunore- (Fig. 2J), but motor neurons from TDP-43 mice displayed cyto- activity was present in motor neurons of spinal cord (Fig. 2 F and plasmic inclusions composed of massive accumulation of mito- G) and brainstem, particularly in those neurons with eccentric chondria (Fig. 2 K and L). These observations suggest that nuclei, indicating the presence of cytoplasmic inclusions (Fig. 2G, elevated levels of TDP-43 impact on the intracellular trafficking of

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1003459107 Shan et al. Downloaded by guest on September 27, 2021 organelles and consequently lead to abnormal distributions of acetylcholine receptors (AChR) in a multiperforating pattern mitochondria in motor neurons. characteristic of normal NMJ (Fig. 3DTopand Fig. S4). In con- To observe directly the distributions of mitochondria in dif- trast, in the NMJ of mtCFP;TDP-43 mice, AChRs form a plaque- ferent compartments of motor neurons, TDP-43 mice were cross- like pattern (Fig. 3D Bottom and Fig. S4) similar to that described bred to Thy1-mitoCFP mice, in which a subpopulation of mito- in a mouse model of spinal muscular atrophy (SMA), a motor chondria are fluorescently labeled with CFP in neurons (14). As neuron disease of infancy and childhood (15). Such abnormalities expected, CFP-labeled mitochondria marked motor neurons and are associated with abnormal synaptic transmission in these SMA were visualized in processes in the ventral horns of nontransgenic mice (15). These observations suggest the possibility that synaptic mice (Fig. 3A). In contrast, mtCFP;TDP-43 compound mice transmission is altered at the NMJ of TDP-43 mice. This in- showed mitochondria clustered within inclusions of motor neu- terpretation is consistent with the weakness and reduction in size rons (Fig. 3B, arrows). The observations that mitochondria of muscle fibers observed in the TDP-43 mice (Fig. 3E). Quanti- mainly accumulated within inclusions of cell bodies and were tative analysis revealed ≈20% reduction in cross-sectional area of sparsely distributed in neuronal processes in the mtCFP;hTDP- muscle fibers in TDP-43 mice as compared with that of non- 43 compound mice suggest the possibility that trafficking of or- transgenic littermate controls (Fig. 3E), indicating that impaired ganelle, particularly mitochondria, is impaired in these nerve transmission at the NMJ may underlie weakness observed in these cells. If this is the case, nerve terminals of these mice may be TDP-43 mice. deficient in mitochondria. Consistent with this speculation, we observed a marked reduction of mitochondria (as indicated by the TDP-43 Regulates SMN-Associated GEMs in Motor Neurons. To de- CFP signal intensity) at nerve terminals of neuromuscular junc- termine how elevated levels of TDP-43 in the nucleus lead to tions (NMJ) in mtCFP;TDP-43 compound mice (Fig. 3C). In motor neuron dysfunction, we asked whether aspects of TDP-43- TDP- muscle sections of control mice, the normal juxtaposition of the related nuclear functions are altered in motor neurons of 43 pre- and postsynaptic terminals is reflected in the clustering of mice. Immunocytochemical analysis of TDP-43 in motor neurons with cytoplasmic inclusions revealed a striking abnormal localization of TDP-43 in the nuclear compartment that is usually associated with two conspicuous intranuclear aggregates (Fig. 2E and Fig. S5). To begin to identify the protein components of these TDP-43-containing nuclear inclusions, we cos-

tained spinal cord sections of TDP-43 mice using antisera NEUROSCIENCE directed against a variety of nuclear markers along with a human TDP-43-specific antibody. Although we did not observe the colocalization of TDP-43 with ubiquitin in these nuclear inclusions (Fig. 4A, arrowheads), we discovered that TDP-43-immunoreactive nuclear aggregates contained FUS/TLS, an RNA- binding protein (Fig. 4B, arrowheads, and Fig. S6) recently linked to cases of ALS (9, 10). In addition, the central cores of these TDP-43 nuclear inclusions associate with SC35 (Fig. 4C, arrowheads), a marker of non-snRNP (small nuclear ribonu- cleoprotein) splicing speckles (16). These results suggest that increased levels of TDP-43 induce its association with FUS/TLS and SC35, proteins involved in RNA metabolism. To examine whether elevated levels of TDP-43 impact on pathways that are involved in RNA splicing, we assessed the distributions of the SMN complex in relation to GEMs and to Cajal bodies, two nuclear structures containing high concentrations of the SMN protein (17, 18), which is linked to SMA (19–21). The SMN complex is part of a large multimeric protein assembly essential for biogenesis of snRNPs required for pre-messenger RNA splicing (22). Although previous studies of TDP-43 and SMN showed colocalization of these two proteins in the nucleus of transient transfected cells (23), we failed to detect colocalization of TDP-43 with SMN-associated GEMs in motor neurons of TDP- 43 mice. Rather, our immunocytochemical analysis using SMN antibody showed a significant increase in the number of GEM Fig. 3. Altered distribution of mitochondria in motor neurons and abnor- bodies in motor neurons of TDP-43 mice (Fig. 4D,arrows).The mal neuromuscular junctions in 3-wk-old W3 TDP-43 mice. (A and B) The distribution of CFP-labeled mitochondria normally observed in motor neu- nontransgenic motor neurons usually harbor two GEMs. In the rons of mtCFP tg mice (A; two motor neurons are delineated by dashed lines) transgene-expressing motor neurons, the number of GEMs varies but reorganized and confined within large cytoplasmic inclusions (B, arrows; widely from one to eight. The number is significantly high in asterisks denote nuclei of motor neurons) in compound mtCFP;TDP-43 tg neurons with eccentric nuclei and cytoplasmic inclusions (Fig. 4D). mice (B; three such motor neurons are delineated by dashed lines). (C)A However, such neurons account for 5–30% of total neurons in decreased level of mitochondria is observed at nerve terminals of neuro- cervical or lumbar spinal cord. We observe on average three or muscular junctions of double transgenic mice (Lower) compared with mtCFP four GEMs in TDP-43-expressing motor neurons (Fig. 4F). Be- mice (Upper). Note the lack of mitochondria invested into the presynaptic cause GEM bodies dynamically shuttle between the nucleolus and α terminals; postsynaptic terminals are marked by staining with -bungrotoxin the nucleoplasm (24, 25), we examined the distributions of GEMs (α-BTX). Three mtCFP and four mtCFP;TDP-43 double transgenic mice were TDP-43 fl examined. (D and E) Alteration of postsynaptic distribution of AChR on muscle in motor neurons of mice. Double immuno uorescence fibers (D) and a decrease of muscle size (E) are observed in TDP-43 transgenic staining of GEMs (using SMN antisera) and the nucleoli (using mice. Quantitative analysis (E)shows≈20% reduction of cross-sectional area of antisera specificforfibrillarin) disclosed that whereas GEMs are muscle fibers in transgenic mice compared with nontransgenic littermate normally distributed with one or two discretely associated with the controls (P < 0.001, n = 4). Error bars indicate SEM. (Scale bars, 20 μm.) nucleolus in neurons of nontransgenic mice, SMN is present dif-

Shan et al. PNAS Early Edition | 3of6 Downloaded by guest on September 27, 2021 To further examine the role of TDP-43 in motor neurons, we used a complementary loss-of-function approach to delete the gene encoding TDP-43. Because TDP-43 is required for embryogenesis (26, 27), we generated and characterized conditional knockout mice with floxed Tardbp alleles (28). Immunofluores- cence analysis revealed, in addition to localization of SMN in the cytoplasm, two prominent SMN-containing GEMs as expected in each spinal motor neuron of control mice (Fig. 4G Upper). However, we failed to find such SMN-containing GEMs in motor neurons of conditional Tardbp knockout mice, although SMN can be localized in the cytoplasm (Fig. 4G Lower). To confirm that GEMs failed to form in the nucleus of neurons lacking TDP- 43, we immunostained spinal cord sections using antisera against other components of GEMs. Whereas both Gemin 2 and Gemin 8 can be localized to GEMs in motor neurons of wild-type mice (Fig. 4 H and I Upper), no GEMs were identified using these markers in conditional Tardbp knockout mice (Fig. 4 H and I Lower), indicating that TDP-43 is required for the formation of GEMs in the nucleus of motor neurons. Taken together, results from both loss- and gain-of-function studies of TDP-43 converge to support the idea that TDP-43 is critical for the generation of SMN-containing GEMs, and that alteration of TDP-43 could impact on pathways that control RNA splicing.

Identification of Potential Downstream Targets of TDP-43 in Spinal Cord. Our findings implicating a role of TDP-43 in the regulation of RNA metabolism led us to hypothesize that alteration in RNA transcription and splicing may occur in our TDP-43 mice. We therefore asked whether perturbations in RNA metabolism are observed in spinal cords of TDP-43 mice using a transcriptome- wide differential RNA expression (RNA-Seq) approach (29, 30). cDNA libraries generated from mRNA extracted from spinal Fig. 4. TDP-43 regulates SMN-associated GEMs in motor neurons. (A–C) cords of three 3-wk-old TDP-43 mice (line W3) and three non- fl Double immuno uorescence analyses of components of intranuclear TDP- transgenic littermates were sequenced. We identified 313 genes 43-immunoreactive granules in spinal motor neurons of 3-wk-old W3 P < transgenic mice (n = 3) and control littermates (n = 3) using antisera against with changes in splicing pattern [ value E-10; Gene Expres- TDP-43 (green channel) and a series of nuclear markers (red channel). Rep- sion Omnibus (GEO) database, accession no. GSE22351]; 2,017 resentative double immunofluorescence staining showed a TDP-43-positive genes were differentially expressed (P value < E-10; GEO ac- neuron (A) lacked ubiquitin (Ub) immunoreactivity in intranuclear granules cession no. GSE22351). Some of these genes, such as glial fibrillar (A, arrowheads), but was immunoreactive with antisera against either Fused acidic protein (GFAP), are expressed exclusively in nonneuronal in Sarcoma (FUS; B, arrowheads) or a marker of non-snRNP splicing speckles cells. This is not a surprise, because increased astrogliosis was (SC35; C, arrowheads). (D and E) Analysis of the number and distribution of observed in 3-wk-old TDP-43 mice (Fig. S7). Because the trans- the Gemini of coiled bodies (GEMs) using SMN antibody. Whereas motor gene is driven by the neuron-specific Thy1.2 promoter, we believe neurons in nontransgenic mice showed two SMN-containing GEMs (arrows; that the primary pathogenic events occur within neurons of TDP- D, ntg), the number of GEMs increases significantly in TDP-43 transgenic 43 mice (arrow; D, W3 tg). Error bars indicate SEM. In contrast to motor neurons mice. Interestingly, the top 30 affected genes (expressed in from nontransgenic littermates in which one GEM is associated with the neurons) showed perturbations in both differential expression nucleolus [arrows, E (ntg, SMN/Fl)], motor neurons that harbor cytoplasmic and alternative splicing (see list in Tables S1 and S2, re- aggregates from transgenic mice showed SMN is present diffusely within the spectively), and many of these genes are involved in the regula- nucleolus, identified by fibrillarin antibody (E), and SMN-containing GEMs are tion of cellular architecture. For example, neurofilament mRNAs confined to the perinucleolar region (E, arrow). All sections are stained with are decreased significantly in TDP-43 mice. Given the importance DAPI to mark nuclei. [Scale bars, 10 μm(A–E).] (F) Quantitative measurement of neurofilament proteins in determining axonal caliber and its on the number of GEMs in neurons. Counts were based on 20 serial sections roles in ALS (31), we determined whether down-regulation of < – for each spinal cord. P 0.0001, n = 3. Error bars indicate SEM. (G I) Absence members of this gene family had functional consequences in of GEMs in spinal motor neurons lacking TDP-43. Immunofluorescence TDP-43 mice. Consistent with the reduction of levels of these analysis using antisera against SMN (G), Gemin 2 (H), and Gemin 8 (I) local- fi ized two GEMs (indicated by arrows) in the nucleus of motor neurons of mRNAs, the protein levels of NF-M and NF-L were signi cantly control mice (+/+, upper). No GEM can be identified in the nucleus of motor decreased in TDP-43 mice (Fig. 5 A and B). Importantly, analysis neurons derived from conditional Tardbp knockout mice (−/−, lower) using of the ventral roots from the lumbar region of spinal cords antisera against SMN (G), Gemin 2 (H), or Gemin 8 (I); 100 motor neurons revealed a decrease in the number of large-caliber axons in TDP- from three control or three conditional Tardbp knockout mice were exam- 43 transgenic mice, an observation consistent with the reduced ined. [Scale bars, 20 μm(G–I).] G2, Gemin 2; G8, Gemin 8. levels of NF proteins (Fig. 5 D and E). Taken together, these results provide additional support for the roles of TDP-43 in regulation of RNA splicing and RNA transcription of a subset of fusely within the entire nucleolus and SMN-containing GEMs are genes, including those members of the neurofilament family. confined to the perinucleolar region of motor neurons in TDP-43 mice (Fig. 4E). Similar results were observed using sections de- Discussion rived from spinal cords of three transgenic mice and three litter- Findings from our gain- and loss-of-function studies converge mate controls. The integrity of GEMs was confirmed by the to establish that TDP-43 plays a critical role in the regulation of identification of these same nuclear structures with other essential SMN-containing GEMs that impact on RNA metabolism in motor components of GEMs, including Gemin 2 and Gemin 8. neurons. Moreover, our observations of aberrant TDP-43- and FUS/

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1003459107 Shan et al. Downloaded by guest on September 27, 2021 pression vector which would impact on the spatial pattern and level of expression of the transgene could account for the dis- parity. Although a mild increase in the level of cleaved caspase-3 was observed in a subset of transgene-expressing motor neurons, these cells did not exhibit evidence of apoptotic nuclei (Fig. S8), indicating that apoptosis is not a primary factor leading to motor neuron abnormalities in our TDP-43 mice. Interestingly, TDP-43- positive cytoplasmic inclusions were also not observed in mice expressing an ALS-linked TDP-43 mutant (12), an observation that was interpreted to suggest the possibility that mutant TDP-43 may play a pathogenic role in the nuclear compartment. However, it was shown recently that a rat transgenic model expressing human mutant TDP-43 exhibits cytoplasmic TDP-43 inclusions in motor neurons (34). Although mitochondrial dysfunction has been proposed to be involved in the pathogenesis of ALS (35), it is not clear as to how an increase in TDP-43 leads to the massive accumulation of mitochondria in cell bodies of motor neurons observed in TDP- 43 mice. We speculate that components of the axonal transport system, including molecular motors, may be altered in motor neurons of the TDP-43 mice. Interestingly, kinesin-associated proteins Kif3a and KAP3, thought to be a determinant of rate of disease progression in sporadic ALS (36), are localized within Fig. 5. Decreased levels of neurofilament proteins and reduction in caliber cytoplasmic inclusions of motor neurons in TDP-43 mice (Fig. of motor axons in TDP-43 mice. (A and B) Protein blot analysis of neuro- S2C). Although we have not excluded the possibility of non- filament from spinal cords of nontransgenic (ntg) and W3 transgenic (tg) physiological effects due to increased expression of human TDP- mice (A) showed, respectively, a 23% and a 25% reduction in the protein 43 in mice, future efforts are necessary to clarify these issues and < level of NF-M and NF-L in 3-wk-old W3 transgenic mice (B; n =5;*P 0.0164 NEUROSCIENCE < – it will be of interest to explore the ways in which aberrant traf- and **P 0.0001). Error bars indicate SEM. (C E) Analysis of caliber of axons ficking, altered axonal transport, and abnormalities of mito- in ventral roots. Note the reduction in number of large-caliber axons and TDP43 a concomitant increase in the number of small-caliber axons in 3-wk-old W3 chondria are associated with -linked ALS. transgenic mice (D and E; n = 3) when compared with nontransgenic litter- Considering the importance of GEMs/SMN in the regulation of mates (C and E; n = 3). Error bars indicate SEM. Scale bars, 20 μm(C and D). RNA metabolism, the impact of TDP-43 on this nuclear structure implies that TDP-43 could mediate RNA metabolism indirectly through the GEMs/SMN pathway. However, our recent studies TLS-positive nuclear inclusions, abnormal accumulation of mito- using conditional Tardbp knockout ES cells suggested that TDP- chondria in motor neurons, immature neuromuscular junctions, 43 also regulates a large set of RNAs (28). Supporting this notion and atrophy of skeletal muscle in TDP-43 mice strongly support is a recent report documenting that HDAC6 is a target of TDP-43 the view that normal levels of TDP-43 are crucial for the main- in nonneuronal cells (37). Taken together, these findings suggest tenance of neuronal physiology, including that of motor neurons. that TDP-43 are critical for RNA processing, either directly At present, it is not clear how ALS-associated TDP-43 mutants through association with target RNAs or indirectly through the lead to motor neuron degeneration. Although it is plausible that regulation of mediators essential for RNA metabolism, such as loss of nuclear TDP-43 function is due to sequestration of ALS- SMN-associated GEMs, FUS/TLS, or SC35. Indeed, even with the associated mutant TDP-43 in the cytoplasm of motor neurons (1), background mRNA signals of nonneuronal cells, our deep se- our findings also raise the possibility that aspects of RNA me- quencing analysis of spinal cords from TDP-43 mice has identified tabolism are perturbed or compromised by ALS-linked TDP-43 a set of potential downstream targets of TDP-43, including neu- mutants in motor neurons. It is interesting that TDP-43 forms rofilaments. Interestingly, NF-L mRNA has been shown to be intranuclear granules containing splicing factors and FUS/TLS, associated with TDP-43 (38). Future deep sequencing studies a predominantly nuclear protein with structural homology to using motor neurons derived from TDP-43 transgenic and condi- TDP-43 (2), and is linked to ALS (9, 15). Recruitment of FUS/ tional knockout mouse models are predicted to reveal a more TLS into TDP-43-immunoreactive nuclear structures in motor comprehensive and interesting set of motor neuron-specific, neurons indicates that TDP-43 and FUS/TLS may be functionally downstream genes impacted by TDP-43. related and, when mutated, are involved in the same pathogenic Materials and Methods pathways in ALS. Although mice expressing an ALS-linked mu- TDP-43 tant TDP-43 exhibit evidence of motor neuron disease (12), it is Generation of Human Transgenic Mice. The complete human TDP-43 not clear whether this is simply due to increased expression of this cDNA was subcloned into a mouse Thy1.2 expression cassette and sub- sequently injected into C57BL/6;SJL hybrid mouse embryos by the Transgenic protein in mice. As increased levels of wild-type TDP-43 are as- Facility at The Johns Hopkins University School of Medicine. Integration of the sociated with cellular toxicity in yeast (32) and mammalian cells transgene into the mouse genome was determined by Southern blot analysis. (33) and motor neuron disease phenotype in mice (11), future A 200-bp human TARDBP DNA probe, which shares 92% homology to mouse comparative studies of mice expressing wild-type TDP-43 com- Tardbp gene, was used in the Southern blot to identify both the endoge- parable to that of mutant TDP-43 are required to clarify this issue. nous gene and transgene. The signal intensities were measured with Bio-Rad Although phenotypes reminiscent of cases of ALS including TDP- Quantity One and compared between transgene and endogenous Tardbp to fi 43-positive cytoplasmic inclusions, abnormal accumulations of 25- assess the copy number of the transgene. Three founders were identi ed – and bred with C57BL/6 mice. The copy number was assessed on F1 mice of and 35-kDa TDP-43 fragments, and 25 30% loss of neurons were each founder. The animal use protocol was approved by the Animal Care reported in mice expressing wild-type TDP-43 (11), we failed to and Use Committee of the Johns Hopkins Medical Institutions. observe any of these findings in our TDP-43 mice. Although the reasons for such striking differences between these two sets of SDS/PAGE and Immunoblotting. Mouse tissues were homogenized in RIPA mice is unclear at present, factors including the design of the ex- buffer with 1% SDS, resolved by 4–12% bis-Tris gel (Invitrogen), and trans-

Shan et al. PNAS Early Edition | 5of6 Downloaded by guest on September 27, 2021 ferred onto PVDF membrane (Millipore). Immunoblotting was performed Visualization of Mitochondria in Spinal Motor Neurons and Nerve Terminals. using the following antibodies: mouse-anti-TDP-43 mAb (clone 2E2-D3; Thy1-hTDP-43 transgenic mice were mated with Thy1-mtCFP transgenic mice Novus Biologicals), rabbit anti-TDP-43 pAb (Proteintech Group), mouse-anti- (The Jackson Laboratory, no. 006617). Double transgenic mice were identi- actin (Sigma), mouse-anti-neurofilament light chain mAb (clone NR4; fied by tail biopsy genotyping of hTDP-43 and CFP. mtCFP mice were used as Sigma), and mouse-anti-neurofilament 160 kDa (clone NN18; Millipore). the control for double transgenic mice. Anesthetized mice were transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer Immunohistochemistry. Mice under anesthetization by an i.p. injection of 15% (pH 7.4). Spinal cord and tibialis anterior muscle were isolated, postfixed in chloral hydrate were perfused transcardially with PBS (pH 7.4) and sub- the same fixative for 4 h, and cryoprotected in 30% sucrose in PBS (pH 7.4) sequently by 4% paraformaldehyde in phosphate buffer (pH 7.4). Organs overnight. The spinal cord tissues were cut into cross-sections at 10-μm were removed, postfixed in the same fixative overnight, and embedded in thickness and muscle tissues were cut longitudinally at 50 μm. Mouse-anti- paraffin blocks. Sagittal sections of brains and cross-sections of spinal cords TDP-43 mAb was applied to identify the transgene-expressing neurons in (10 μm) were used for immunohistochemical analysis. The following anti- spinal cord of hTDP-43 transgenic mice. Alexa Fluor 594 was used as the bodies were used: mouse-anti-TDP-43 mAb (clone 2E2-D3) and rabbit-anti- secondary antibody. The muscle sections were incubated with Alexa Fluor α ubiquitin (Dako). The immunoreactivity was visualized by a Vectastain Elite 594-conjugated -bungrotoxin (Invitrogen) at 1:1,000 dilution and sub- ABC Kit (Vector Laboratories) and diaminobenzidine. sequently washed with PBS three times in 1 h. Sections were mounted and visualized by Zeiss 510 Meta confocal microscope. Immunofluorescence. Tissues fixed by 4% paraformaldehyde in phosphate fl buffer (pH 7.4) were cryoprotected by 30% sucrose in PBS (pH 7.4). Frozen Histology. Muscle tissues were dissected and ash-frozen in freezing iso- μ sections (10 μm) of spinal cord tissues from three or more nontransgenic and pentane. Cryosections were cut at 10- m thickness and stained with hema- fi transgenic mice, respectively, were processed for double immunofluores- toxylin and eosin. Myo ber area was measured using ImageJ software (W.S. cence using the following antibodies: mouse-anti-TDP-43 mAb (clone 2E2- Rasband, National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij). fi D3), rabbit-anti-TDP-43 pAbs (Proteintech Group; Abcam), rabbit-anti-FUS Quanti cation of axonal caliber in the ventral roots of lumbar spinal cord pAb (Proteintech Group), rabbit-anti-fibrillarin pAb, mouse-anti-germin2, sections, which were stained with toluidine blue, was performed using mouse-anti-SC35 mAbs (Abcam), rabbit-anti-HSP60 pAb (Cell Signaling), ImageJ software. The frequency distribution of axonal calibers was plotted mouse-anti-SMN, germin8 mAbs (Sigma). Alexa Fluor 488 and 594 (Invi- (nontransgenic, n = 3 mice, 3,981 axons; transgenic, n = 3 mice, 1,988 axons). trogen) were used as secondary antibodies to visualize proteins. Sections were mounted and visualized by an Olympus IX71 fluorescence microscope Statistical Analysis. Biochemical and morphological data were analyzed by ’ or Zeiss 510 Meta confocal microscope. Excel and GraphPad Prism using unpaired Student s t tests. The quantitative data in this study were expressed as the mean ± SEM. Electron Microscopy. Anesthetized mice were perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) and postfixed with 4% ACKNOWLEDGMENTS. We thank V. Nehus, J. Ling, and F. Davenport for technical support and Y. H. Jeong, S. Sisodia, J. Rothstein, and J. Nathans for paraformaldehyde with 2% glutaraldehyde in 0.1 M phosphate buffer (pH helpful discussions and critical reading of the manuscript. This study was 7.4) overnight. Tissues were then washed in PBS, dehydrated, and embedded supported in part by the Muscular Dystrophy Association (P.C.W.), The μ in Epon. Thick (1 m) and thin (100 nm) sections were stained, respectively, Robert Packard Center for ALS Research (P.C.W.), The Johns Hopkins with toluidine blue and lead citrate/uranyl acetate. The images were Neuropathology Consolidated Gift Fund, and National Institute of Neuro- obtained using a Hitachi 7600 transmission electron microscope. logical Disorders and Stroke Grant R01 NS41438 (to P.C.W.).

1. Forman MS, Trojanowski JQ, Lee VM (2007) TDP-43: A novel neurodegenerative 20. Zhang Z, et al. (2008) SMN deficiency causes tissue-specific perturbations in the proteinopathy. Curr Opin Neurobiol 17:548–555. repertoire of snRNAs and widespread defects in splicing. Cell 133:585–600. 2. Lagier-Tourenne C, Cleveland DW (2009) Rethinking ALS: The FUS about TDP-43. Cell 21. Burghes AH, Beattie CE (2009) Spinal muscular atrophy: Why do low levels of survival 136:1001–1004. motor neuron protein make motor neurons sick? Nat Rev Neurosci 10:597–609. 3. Buratti E, et al. (2001) Nuclear factor TDP-43 and SR proteins promote in vitro and in 22. Maniatis T, Tasic B (2002) Alternative pre-mRNA splicing and proteome expansion in vivo CFTR exon 9 skipping. EMBO J 20:1774–1784. metazoans. Nature 418:236–243. 4. Mercado PA, Ayala YM, Romano M, Buratti E, Baralle FE (2005) Depletion of TDP 43 23. Wang IF, Reddy NM, Shen CK (2002) Higher order arrangement of the eukaryotic overrides the need for exonic and intronic splicing enhancers in the human apoA-II nuclear bodies. Proc Natl Acad Sci USA 99:13583–13588. gene. Nucleic Acids Res 33:6000–6010. 24. Dundr M, et al. (2004) In vivo kinetics of Cajal body components. J Cell Biol 164: – 5. Bose JK, Wang IF, Hung L, Tarn WY, Shen CK (2008) TDP-43 overexpression enhances 831 842. exon 7 inclusion during the survival of motor neuron pre-mRNA splicing. J Biol Chem 25. Sleeman JE, Trinkle-Mulcahy L, Prescott AR, Ogg SC, Lamond AI (2003) Cajal body 283:28852–28859. proteins SMN and Coilin show differential dynamic behaviour in vivo. J Cell Sci 116: – 6. Neumann M, et al. (2006) Ubiquitinated TDP-43 in frontotemporal lobar degen- 2039 2050. 26. Sephton CF, et al. (2010) TDP-43 is a developmentally regulated protein essential for eration and amyotrophic lateral sclerosis. Science 314:130–133. early embryonic development. J Biol Chem 285:6826–6834. 7. Sreedharan J, et al. (2008) TDP-43 mutations in familial and sporadic amyotrophic 27. Wu LS, et al. (2010) TDP-43, a neuro-pathosignature factor, is essential for early lateral sclerosis. Science 319:1668–1672. mouse embryogenesis. Genesis 48:56–62. 8. Kabashi E, et al. (2008) TARDBP mutations in individuals with sporadic and familial 28. Chiang PM, et al. (2010) Deletion of TDP-43 down-regulates Tbc1d1, a gene linked to amyotrophic lateral sclerosis. Nat Genet 40:572–574. obesity, and alters body fat metabolism. Proc Natl Acad Sci USA, 10.1073/pnas.1002176107. 9. Vance C, et al. (2009) Mutations in FUS, an RNA processing protein, cause familial 29. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: A revolutionary tool for amyotrophic lateral sclerosis type 6. Science 323:1208–1211. transcriptomics. Nat Rev Genet 10:57–63. 10. Kwiatkowski TJ, Jr, et al. (2009) Mutations in the FUS/TLS gene on chromosome 16 30. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B (2008) Mapping and cause familial amyotrophic lateral sclerosis. Science 323:1205–1208. quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 5:621–628. 11. Wils H, et al. (2010) TDP-43 transgenic mice develop spastic paralysis and neuronal 31. Bruijn LI, Miller TM, Cleveland DW (2004) Unraveling the mechanisms involved in inclusions characteristic of ALS and frontotemporal lobar degeneration. Proc Natl motor neuron degeneration in ALS. Annu Rev Neurosci 27:723–749. – Acad Sci USA 107:3858 3863. 32. Johnson BS, McCaffery JM, Lindquist S, Gitler AD (2008) A yeast TDP-43 proteinopathy 12. Wegorzewska I, Bell S, Cairns NJ, Miller TM, Baloh RH (2009) TDP-43 mutant model: Exploring the molecular determinants of TDP-43 aggregation and cellular transgenic mice develop features of ALS and frontotemporal lobar degeneration. toxicity. Proc Natl Acad Sci USA 105:6439–6444. – Proc Natl Acad Sci USA 106:18809 18814. 33. Zhang YJ, et al. (2009) Aberrant cleavage of TDP-43 enhances aggregation and 13. Li Y, et al. (2010) A Drosophila model for TDP-43 proteinopathy. Proc Natl Acad Sci cellular toxicity. Proc Natl Acad Sci USA 106:7607–7612. USA 107:3169–3174. 34. Zhou H, et al. (2010) Transgenic rat model of neurodegeneration caused by mutation 14. Misgeld T, Kerschensteiner M, Bareyre FM, Burgess RW, Lichtman JW (2007) Imaging in the TDP gene. PLoS Genet 6:e1000887. axonal transport of mitochondria in vivo. Nat Methods 4:559–561. 35. Liu J, et al. (2004) Toxicity of familial ALS-linked SOD1 mutants from selective 15. Kong L, et al. (2009) Impaired synaptic vesicle release and immaturity of recruitment to spinal mitochondria. Neuron 43:5–17. neuromuscular junctions in spinal muscular atrophy mice. J Neurosci 29:842–851. 36. Landers JE, et al. (2009) Reduced expression of the Kinesin-Associated Protein 3 16. Spector DL (2001) Nuclear domains. J Cell Sci 114:2891–2893. (KIFAP3) gene increases survival in sporadic amyotrophic lateral sclerosis. Proc Natl 17. Battle DJ, et al. (2006) The SMN complex: An assembly machine for RNPs. Cold Spring Acad Sci USA 106:9004–9009. Harb Symp Quant Biol 71:313–320. 37. Fiesel FC, et al. (2010) Knockdown of transactive response DNA-binding protein (TDP- 18. Gall JG (2000) Cajal bodies: The first 100 years. Annu Rev Cell Dev Biol 16:273–300. 43) downregulates histone deacetylase 6. EMBO J 29:209–221. 19. Lefebvre S, et al. (1995) Identification and characterization of a spinal muscular 38. Strong MJ, et al. (2007) TDP43 is a human low molecular weight neurofilament (hNFL) atrophy-determining gene. Cell 80:155–165. mRNA-binding protein. Mol Cell Neurosci 35:320–327.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1003459107 Shan et al. Downloaded by guest on September 27, 2021