Introduction
Amyotrophic lateral sclerosis (ALS) is the most common adult-onset motor neuron disease, characterized by progressive muscle atrophy and weakness that eventually results in respiratory failure and death. Over 90% of ALS cases are not linked to clear familial inheritance and have no defined cause (i.e., sporadic ALS or sALS). The remaining 10% of ALS cases are associated with pathogenic mutations in one of more than 30 different genes [
1]. One of the most common genetic causes for both ALS and frontotemporal dementia (FTD), a closely related neurodegenerative condition that shares both genetic and pathological signs with ALS, is a
GGGGCC hexanucleotide repeat expansion (HRE) in the
C9ORF72 gene (C9-ALS) [
2]. This mutation accounts for about 40% of familial and 7% of sporadic ALS cases [
3]. Although the molecular determinants of neuronal degeneration in ALS/FTD remain generally unclear, protein aggregation, most prominently of the RNA-binding protein TDP-43 [
4], and alterations to nuclear homeostasis [
5] have been accepted as common pathomechanisms underlying most familial and sporadic forms of the disease. In fact, alterations to the nuclear lamina and to the nuclear pore complex (NPC), the largest multiprotein assembly encompassing the inner and outer nuclear membranes, have been widely reported in several in vivo and in vitro models of ALS/FTD, including postmortem patients’ tissue. Of note, we have recently shown that mechanical strain on the nucleus, exerted by the cytoskeleton via the linker of nucleoskeleton and cytoskeleton (LINC) complex, can lead to NPC injury, ruptures of the nuclear envelope (NE), and accumulation of DNA damage [
6], significantly contributing to disease pathology. One of the cell’s main mechanosensors, the LINC complex is the second largest transmembrane complex present in the NE. While its modular protein composition is variable in different cell types and conditions, the LINC complex generally consists of a trimer of SUN proteins (SUN1 and SUN2) spanning the inner nuclear membrane, connected to a trimer of Nesprin proteins (Nesprin 1–4) embedded in the outer nuclear membrane [
7]. The cytoplasmic domains of Nesprins bind to different cytoskeletal components, while their intermembrane Klarsicht, Anc-1, Syne homology (KASH) domains bind to SUN proteins, which are in turn bound to the nuclear lamina and NPCs [
8]. Because of these multiple interactions, the LINC complex effectively couples the cytoskeleton to NPCs and the nuclear lamina, resulting in nuclear exposure to cytoskeletal forces [
9]. These connections act as a fundamental determinant of nuclear morphology and size, chromatin organization, and nucleolar arrangement [
10,
11].
Mutations in components of the LINC complex are associated with several human diseases, including laminopathies [
12], skeletal muscle diseases [
13], neurodevelopmental diseases [
14,
15], neurodegenerative conditions such as cerebellar ataxia with motoneuronal involvement [
16] and juvenile ALS [
17,
18]. In addition, reduced nuclear localization of SUN1 protein was recently described in sALS iPSCs-derived neurons in vitro [
19], suggesting that this complex may play a relevant role in ALS/FTD pathogenesis. In this study, we show that multiple protein components of the LINC complex are misregulated in different in vitro ALS models, as well as in patient’s postmortem tissue. Furthermore, we found that relative nuclear and nucleolar sizes are significantly reduced in sALS and C9-ALS neurons. Interestingly, nuclear and nucleolar shrinkage occurs even in the absence of common ALS pathological markers but correlates strongly with the degree of disruption of the LINC complex, suggesting that this pathological phenotype depends on the LINC complex and may precede TDP-43 mislocalization as an early sign of cell dysfunction.
Discussion
Recent studies in several models of sporadic and familial ALS and FTD, including mutant
C9ORF72, TDP-43, FUS and
PFN1, have pointed to the loss of nuclear envelope (NE) and NPC integrity, as well as disruption of nucleocytoplasmic transport, as key drivers of disease initiation and progression [
23,
30,
31,
32]. While many of the molecular players and pathways that lead to such dysfunctions are currently unknown, our previous work has pointed to the nucleus-cytoskeleton connection as a possible mediator of this process. In fact, we have previously shown that positive or negative modulation of actin polymerization can lead to NPC injury and nuclear import defects [
32]. Furthermore, we recently showed that increased mechanical tension on the cell’s nucleus can lead to breaks in the NE, NPC injury, and DNA damage [
6]. The LINC complex is a structural component of the nuclear envelope and a fundamental element in the mechanical transmission of physical stimuli to the nucleus. Through its interactions with the cellular cytoskeleton, nuclear lamina, and NPC, the LINC complex plays a pivotal role in the maintenance of nuclear homeostasis in response to mechanical strain [
33]. Together, these observations prompted us to directly investigate the role of the LINC complex in ALS/FTD disease.
In this study, we interrogated three different ALS/FTD models, including iPSC-derived cortical and motor neurons, spinal cord organoids, and
postmortem sALS and C9-ALS patient-derived brain and spinal cord samples, and identified severe alterations in the nuclear levels and cellular distribution of the four most abundant components of the LINC complex: SUN1, SUN2, Nesprin1, and Nesprin2. Albeit limited in the number of samples, our analyses of
postmortem tissues provided important insights into the relevance of LINC complex disruption to human disease. However, patient-derived samples often represent a late stage of the disease with extensive degeneration, which may not exhibit molecular or cellular signatures directly associated with the initiating events that caused the disease. By contrast, neurons derived from iPSCs, while more embryonic-like and lacking normal tissue connections, can provide insights into the earliest stages of neurodegeneration, shedding light on the pathogenic events that drive disease initiation. By combining both approaches, our data rigorously show that cellular and/or NE levels for all four LINC proteins tested were severely reduced both in cortical and motor neurons. A similar reduction in SUN1 levels have been recently reported in iPSC-derived spinal neurons [
19], which confirms and strengthens our observations. While these observations were limited to one time point, future studies will be required to investigate the timing of appearance and progression of the observed phenotypes in in vitro as well as in vivo models, and their relevance to disease pathogenesis as well as to the normal cellular aging process. We also found that spinal motor neurons in the
postmortem tissues displayed a higher degree of disruption compared to cortical neurons, but the levels of LINC complex pathology between spinal cord and brain from each individual patient correlated strongly, indicating a parallel evolution, albeit of different magnitude, of this phenotype in different compartments of the CNS affected by ALS/FTD. Our comprehensive approach also allowed us to investigate the relationship between the inner nuclear membrane SUN proteins and the outer nuclear membrane Nesprin proteins. Interestingly, we found a strong correlation between SUN and Nesprin proteins’ alteration across all models, suggesting that SUN protein localization at the NE is necessary for Nesprins assembly in the LINC complex [
27], and that loss of SUN proteins may trigger Nesprin mislocalization and loss of LINC complex activity.
Through its interaction with the cytoskeleton and nuclear lamina, the LINC complex plays a fundamental role not only in the modulation of nuclear positioning, but also in regulating nuclear morphology and in promoting changes and adaptations to chromatin organization [
29]. For instance, acute depletion of SUN1 was shown to alter nuclear morphology, chromatin organization, and nucleolar distribution [
11], while deletion or overexpression of different LINC proteins was proven to alter the ratio between nuclear and cytoplasmic volumes [
34]. It has been recently demonstrated that nuclear and nucleolar shrinkage occurs in both C9-ALS and sALS spinal MNs independently of many other common pathological changes, including TDP-43 mislocalization or HRE RNA foci [
35]. Nucleolar size alterations have also been observed in C9-ALS patient lymphocytes, fibroblasts and patient iPSC-derived neurons [
36]. Surprisingly, in these cellular models the nucleoli were found to be larger when compared to controls, suggesting that either aging or the non-physiological conditions of 2D cell culture may affect the phenotype differently. However, the cause of this phenomenon is not well understood. Our analyses of the impact of LINC complex alterations on nuclear morphology showed a strong correlation between SUN1/SUN2 disruption and the reduction of nucleolar and nuclear size in ALS MNs. While MNs with SUN1 and SUN2 normally localized at the NE displayed nuclear and nucleolar sizes comparable to the MNs of non-neurological controls, altered SUN1/SUN2 staining strongly correlated with smaller nucleoli and an overall reduced nuclear volume. While a strong causal link cannot be established at this point, our data suggested that alterations to the LINC complex may be a key element driving this phenotype. Interestingly, we were able to observe nuclear and nucleolar alterations both in patient-derived tissues and 3D spinal organoids, while 2D-cultured motor or cortical neurons lacked a well-defined and mature nucleolus and did not display obvious alterations to nuclear morphology under our experimental conditions. While 2D cultures are a valuable tool that allow for the isolation of cell-autonomous events by limiting confounding variables, they nonetheless represent a reductionist model that lacks the complex interactions of cells with their environment, including cell-to-cell and cell-extracellular matrix (ECM) mechanical interaction. For our study in particular, such interactions were particularly relevant, given the well-established role of the LINC complex in transmitting forces from the ECM to the nucleus to modulate cell mechanical properties and function. Spinal cord organoids thus offered a more complex model that could better recapitulate both tissue cell diversity and complex mechanical interactions between cells and their surroundings [
37].
One important and surprising observation was that most of the LINC-dependent phenotypes we observed, including protein mislocalization and nuclear morphology alterations, were independent from any TDP-43 related pathology. While rare co-aggregation of SUN1 or SUN2 with TDP-43 was observed, the majority of mislocalized cytoplasmic SUN proteins formed discrete clumps that did not contain TDP-43, and most TDP-43 skein-like aggregates were negative for SUN proteins. Similarly, nuclear and nucleolar shrinkage did not correlate with TDP-43 nuclear depletion or aggregation, which instead correlated with an overall reduction in cellular volume. Shrinkage of nuclear and nucleolar size has been previously described in ALS
postmortem tissue [
35,
38] and prp-TDP-43
A315T mouse model, seemingly preceding neuronal loss and motor defects [
38]. Our novel data propose a mechanistic explanation for this early sign of disease and suggest that LINC complex defects may be an early event in the pathogenic cascade in ALS/FTD, representing a parallel pathological event that adds to the burden of disease caused by TDP-43-related dysfunction.
In conclusion, our results, obtained in complementary in vivo and in vitro models, provide a robust body of evidence pointing at the alteration of the LINC complex as an early ALS phenotype conserved in the late stages of the disease, making the LINC complex a possible target for both biomarker or therapy development.
Methods
iPSCs culture
The protocols met all ethical and safety standards for work on human cells. The human iPSC lines used in this study are identified in Supplementary Tables
1 and were fully characterized in our lab. They were maintained on Matrigel-coated plates in StemFlex Medium (Gibco), and passaged every 4–6 days using 0.5mM EDTA in Ca
2+ and Mg
2+-free 1x PBS (Thermo Fisher Scientific).
Cortical neuron (i3CN) differentiation
To differentiate iPSCs to i
3CNs, we used the method described by Dr. Michael E. Ward [
25]. This method relies on the integration of a gene expression cassette in the safe-harbor CLYBL1 locus and produces > 90% pure neuronal populations. To integrate the cassette, two independent pairs of isogenic
C9ORF72 iPSCs were transfected with a plasmid containing the coding sequence for the NGN2 transcription factor, NLS-mApple fluorescent protein, and blasticidin resistance gene, flanked by homology arms to the CLYB1 genomic locus (Addgene plasmid # 124,229), and ribonucleoprotein particles containing the Cas9 nuclease and a site-specific gRNA (Synthego). Colonies were positively selected based on blasticidin (10 µg/ml) resistance and the expression of NLS-mApple. To minimize the effects of inter-clonal variation, all cells positive for the cassette integration were pooled together and expanded as neuroprogenitors. To induce differentiation, 2 µg/ml doxycycline was added to the medium for 2 days, while proliferating cells were killed off with BrdU treatment. Cells were than plated on poly-ornithine and poly-lysine coated coverslips and switched to neuronal medium (Neurobasal, 2% B27, 1% N2, 1% NEAA, 1 µg/ml laminin). To verify full differentiation, positivity for NeuN and Tuj1 and negativity for Oct4 and Sox2 was assessed.
Motor neuron (iMN) differentiation
iMNs were generated following the protocol published by Du et al. [
39] with minimal modification. Briefly, confluent iPSC cultures were detached using 0.5mM EDTA and plated 1:4 on Matrigel ES-coated plates in StemFlex Medium (Gibco) supplemented with 10µM ROCK inhibitor (SellChem). The following day, cells were switched to neural induction media (DMEM/F12 and Neurobasal medium at 1:1 ratio, 1% B27, 0.5% N2, 1% PenStrep, 1% Glutamax; ThermoFisher Scientific) supplemented with ascorbic acid (0.1 mM, Sigma-Aldrich), CHIR99021 (3 µM, Tocris), DMH1 (2 µM, Tocris) and SB431542 (2 µM, Tocris). On day 7 and 13, cells were dissociated with Accutase (Millipore) and plated on Matrigel growth factor reduced (MaGR, Corining) in neural induction media further supplemented with 0.1 µM retinoic acid (Sigma-Aldrich) and 1 µM purmorphamine (Calbiochem). Valproic acid (0.5mM, Tocris) was added on day 13. To induce MN differentiation, cells were dissociated with Accutase (Millipore) and cultured in suspension for 6 days on an horizontal shaker in neural medium supplemented with 0.5 µM retinoic acid and 0.5 µM purmorphamine. On day 25, cells were dissociated into single cell suspension with Accutase (Millipore) and plated on MaGR-coated plates in neural medium with 0.5 µM retinoic acid, 0.5 µM purmorphamine and 0.1 µM Compound E (Calbiochem) until day 35.
Spinal cord organoids
Spinal cord organoids were generated following published protocols with minor modifications [
26]. We first dissociated iPSCs into single cells, seeded 30,000 cells per well in Nunclon Sphera ultralow attachment 96-well plates (ThermoFisher Scientific) in neural induction medium supplemented with 10 ng/mL FGF (Corning), 20 ng/mL EGF (Corning) and 10 µM ROCK inhibitor. Neuralization was induced via treatment with 3 µM CHIR99021, 2 µM DMH1 and 2 µM SB431542. Media was changed every other day and supplementaed with 0.1 µM retinoic acid from day 3 to induce caudalization. At day 10, organoids were embedded in 1% Matrigel Matrix for Organoid Culture (Corning) and kept in slow (70 rpm) horizontal agitation. From day 10, neural media was further supplemented 0.5 µM retinoic acid and 1 µM purmorphamine, while 0.1 mM ascorbic acid, 20 ng/mL BDNF, 10 ng/mL GDNF were added after day 20 and till day 120.
HEK293 cell culture and transfection
HEK293 cells were grown in DMEM media supplemented with 10% FBS. For immunofluorescence experiments, cells were plated on poly-lysine coated coverslips (1 mg/ml; Millipore Sigma) and allowed to recover for 24 h. Cell transfection was carried out using TurboFect reagent (ThermoFisher Scientific) according to manufacturer’s instructions using following plasmids: pLV[shRNA]-EGFP-U6 > hSUN1 (target sequence: TTCATGGACGAGGGCATATAC) or pLV[shRNA]-EGFP/Puro-U6 > Scramble_shRNA (VectorBuilder). Cells were processed for immunofluorescence 48 h after transfection.
Immunofluorescence and image acquisition of in vitro models
Motor or cortical iPSC-derived neurons were seeded on Matrigel (Corning) or poly-D-Lysine (Millipore Sigma) coated glass coverslips at the density of 300 cell/mm2. Cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.2% Triton-X 100 for 15 min. Cells were blocked with 5% bovine serum albumin before hybridization with the appropriate antibodies overnight at 4 °C. Anti-mouse and anti-rabbit donkey secondary antibodies conjugated with either Alexa Fluor 647, Rhodamine-Red X, or Alexa Fluor 488 (Jackson Immunoresearch and ThermoFisher Scientific) were hybridized for 1 h at room temperature.
Organoids were washed in PBS, fixed in 4% PFA at 4 °C overnight, and incubated in 30% sucrose for cryopreservation for at least 16 h. Organoids were embedded into OCT and preserved at -80 °C. Sections were sliced at 10 μm thickness using a cryostat (Leica), permeabilized with 0.1% Triton-X 100 for 10 min and incubated in blocking buffer (5% FBS, 1% BSA, 0.01% Triton-X 100) for 2 h. Slides were incubated with primary antibodies (see Supplementary Table
4) overnight at 4 °C, followed by 2 h hybridization with appropriate secondary antibodies at room temperature.
Organoids sections and cells coverslips were mounted onto a glass slide using Prolong Gold mounting medium (ThermoFisher Scientific) and imaged using a widefield microscope (Leica DMi8 Thunder) equipped with a cooled CMOS camera (DFC9000 GTC). Images were acquired as Z-stacks (0.21 μm step size) using a x63 lens unless otherwise specified.
Image analysis
Immunofluorescence images were deconvolved using an adaptative blind deconvolution algorithm (Autoquant X3, Media Cybernetics) before analysis. Fluorescence intensity levels were quantified using Fiji [
40] (US National Institutes of Health). First, 3D stacks were compressed to 2D images using the maximal intensity projection algorithm. Second, a mean filter (with radius 2.0 pixels) was applied to the DAPI nuclear staining channel to discretely separate single nuclei. Finally, mean fluoresce levels (MFI) of the target proteins were measured within discrete region of interests (ROI) established by gating single nuclei.
For the analysis of linear profiles of SUN and Nesprin proteins, the equatorial plane of the cells was manually identified, and a horizontal line was drawn across the center of the DAPI nuclear staining and through the entire cell. Linear profiles of DAPI and LINC complex proteins staining were generated using Fiji Plot Profile function, normalized to controls, and then merged on the same graph to be compared.
To evaluate qualitative alterations in the nuclear distribution of LINC proteins, a blinded analysis was performed on 3D stacks of individual optical slices. To avoid misinterpretation of staining profiles, no 3D to 2D compression of the images was performed as suggested [
41,
42]. For LINC complex analysis, abnormal staining was considered if the signal was not uniformly distributed around the nucleus with the presence of gaps. For all experiments, raw values were normalized to the mean of the control condition. For all experiments, data acquired from both iPSC lines were analyzed and graphed aggregated. Graphs and analysis of data separated by line is shown in Supplementary Fig.
15.
Western blots
Cells were centrifuged at 250 x g for 10 min at 4 °C and processed in lysis buffer (10 mM Tris-HCL pH 8, 100 mM NaCl, 1 mM EDTA pH 8, 1% NP40) supplemented with protease inhibitor (Complete EDTA-free, Roche) and phosphatase inhibitor (phosphoSTOP, Roche) and sonicated (40 kHz for 10s; Sonifier® SFX150, Emerson) to ensure uniform and complete lysis. Protein extracts (30 µg) were resolved by SDS-PAGE on Mini 4 − 20% Novex Tris-Glycine Gels (Thermo Fisher). Primary antibodies were incubated overnight at 4 °C (Supplementary Table
4). Secondary antibodies conjugated with IRDye® infrared fluorophores (LI-COR) were incubated for 1 h at room temperature. Blots were visualized using the Odyssey Infrared Imaging System (LI-COR). For all experiments, data acquired from both iPSC lines were analyzed and graphed aggregated. Graphs and analysis of data deparated by line is shown in Supplementary Fig.
15.
Immunohistochemistry and immunofluorescence of human postmortem samples
Human postmortem motor cortex and cervical spinal cord paraffin sections were obtained from TargetALS biorepository. Clinical data from the cases are summarized in Supplementary Tables
2 and Supplementary Table
3. Sections were de-paraffinized in xylene (Sigma-Aldrich) and re-hydrated with scaling dilutions of ethanol (ThermoFisher Scientific) to MilliQ water. Endogenous peroxidase blocking was conducted with 0.3% H
2O
2 in Methanol solution for 40 min (ThermoFisher Scientific) followed by re-hydration with scaling concentrations of methanol. Antigen retrieval was performed with 10 mM citric acid pH 6 at 100 °C for 15 min. After an hour of cool-down slides were washed with PBS and incubated with blocking buffer (1% BSA, 5% FBS, 0.01% Triton-X 100 in 1x PBS) followed by Avidin/Biotin blocking (Vector laboratories). Slides were incubated overnight at 4 °C with primary antibodies (see Supplementary Table
4) and at room temperature for 1 h with secondary biotinylated antibody (anti-rabbit IgG (H + L), Jackson Laboratories). For immunoreactivity detection, slides were incubated with Vectastain Ekite ABC reagent (Vector laboratories) for 30 min and then with ImmPACT DAB substrate kit (Vector laboratories) from 2 to 5 min. After counterstain with hematoxylin and/or eosin, slides were de-hydrated and mounted on microscope cover glass (Globe Scientific) with mounting medium (Epredia). Sections were imaged using a widefield microscope (Leica DMi8 Thunder) equipped with a color digital camera (DMC5400) and a 20x objective. For quantification of nuclear, nucleolar, and somatic areas, hematoxylin was used as nuclear marker while eosin was used to identify the whole cytoplasm. For immunofluorescence assays of
postmortem tissue sections, we followed the same protocol already described for organoids sections after the initial steps of de-paraffinization and de-hydration.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.