The LINC complex is altered in C9-ALS iPSC-derived motor and cortical neurons

To determine whether alterations to the LINC complex play a relevant role in ALS/FTD pathogenesis, we investigated the overall levels and cellular distribution of the main LINC protein components, namely SUN1, SUN2, Nesprin1 (NESP1), and Nesprin2 (NESP2). For this study, we first focused on iPSC-derived motor neurons (iMNs) carrying mutations in the C9ORF72 locus, as this mutation has already been extensively implicated in nuclear envelope disfunctions [20, 21, 22, 23]. iMNs were derived from two pairs of isogenic iPSCs where the patient’s derived C9 HRE had been corrected using CRISPR/Cas9 technology (Supplementary Table 1). Six weeks after differentiation induction, immunostaining with neuronal and MN-specific markers including NeuN, MAP2 and ISL1 confirmed that both the patient mutant lines and their isogenic control lines generated ISL1-positive iMNs with similar efficiency, suggesting the cells had acquired a mature neuronal phenotype (Supplementary Fig. 1). At this time point, we found that the nuclear levels of all SUN and Nesprin proteins tested were severely reduced in C9 MNs compared to control isogenic neurons (Fig. 1 and Supplementary Fig. 2). WB analyses also showed an overall reduction in the whole-cell levels of SUN2 (Fig. 1c) but not SUN1 (Supplementary Fig. 2f). However, we observed a reduction of SUN1 nucleocytoplasm ratio in C9 compared to control isogenic MNs, possibly accounting for this discrepancy (Supplementary Fig. 2c). A similar change was observed for SUN2 but not for either Nesprin 1 or Nesprin 2, suggesting a general reduction in whole cell levels for these proteins (Supplementary Fig. 2).

Fig. 1

LINC complex disruption in C9 iPSCs-derived iMNs. Representative images of SUN2 (a, grays) and Nesprin1 (e, Nesp1, grays) expression in C9 and isogenic control (CTRL) iMNs. Nuclei were identified by DAPI staining (blue), while MAP2 (red) and ISL1 (green) were used as neuronal and motoneuronal markers. The white boxes indicate the neurons enlarged in the panels on the right. Scale bars: 20 μm in main panels, 10 μm in zoomed-in images. The relative quantification of the nuclear mean fluorescence intensity (MFI) for both SUN2 (b) and Nesprin1 (f) shows a significant reduction in the abundance of each protein in C9 iMNs compared to isogenic controls (Mann-Whitney t test, n = 57 and 52 for CTRL and C9-ALS neurons respectively from 4 independent differentiations for SUN2, n = 64 and 40 for CTRL and C9 neurons respectively from 4 independent differentiations for Nesprin1, **** p < 0.0001). c-d. Representative western blot (WB) and quantification of SUN2 levels relative to GAPDH expression shows a significative reduction of total SUN2 levels in both C9 isogenic lines (24a and C52) compared to isogenic counterparts (n = 8 independent experiments for both C9 and controls; Student’s t test, **** p < 0.0001). For all, bars are mean and SEM, while violin plots show the distribution of the data with dashed lines indicating median and quartiles

Since the hexanucleotide expansion in C9ORF72 is also causative for frontotemporal lobar degeneration (FTLD), accounting for about 25% of familial and 5% of sporadic cases [24], we decided to expand our initial observations in C9-ALS iMNs to cortical neurons differentiated from the same iPSC lines as above using the previously published i3 method [25]. Two weeks after the induction of differentiation, all the cells were positive for the neuronal markers NeuN and MAP2, displayed glutamate receptor surface expression, and spontaneous electrical activity (Supplementary Fig. 3), suggesting they had reached a mature state. Under these conditions, we found that the nuclear levels of all the LINC proteins tested were severely reduced in C9 iPSC-derived cortical neurons (i3CNs), similarly to what observed in the iMNs (Fig. 2 and Supplementary Fig. 4). This decrease, quantified by IF assays, was associated to a significant reduction in their whole cell levels, quantified by WB for SUN1 and SUN2 (Fig. 2f and Supplementary Fig. 4f-g). We further confirmed these observations by qualitatively scoring nuclei based on SUN or NESP proteins distribution at the NE. While the linear profile of SUN and NESP proteins fully matched the DAPI or Lamin B (LMNB) profiles in isogenic i3CNs, mutant neurons were characterized by a higher frequency of smaller intensity peaks irregularly distributed across the whole cell (Fig. 2b-d and i-k, and Supplementary Fig. 4b-d, 4i-k). Altogether, these observations suggest that disruption of the LINC complex may be an early phenomenon occurring in C9-ALS.

Fig. 2

LINC complex disruption in C9-ALS i3CNs. a. Representative images of SUN1 (grays) in C9-ALS and control i3CNs. DAPI (blue) identified the nuclei, while Phalloidin (green) labeled the actin cytoskeleton. The white boxes indicate the neurons enlarged in the panels on the right. b-d. Plots of line profiles of SUN1 and DAPI intensities normalized to the max intensity in control cells. Mutant cells display frequent mislocalization of SUN1 to either the nucleoplasm or cytoplasm (arrows), quantified in d. The yellow dashed lines in a indicate the lines used for the profile plots (Mann-Whitney t test, n = 4, *p < 0.05). e. The quantification of the nuclear mean fluorescence intensity (MFI) of SUN1 in C9 i3CNs shows a significant reduction in its abundance relative to isogenic controls (Mann-Whitney t test, n = 46 CTRL and C9 neurons respectively from 4 independent differentiations, ****p < 0.0001). f-g. Representative western blot (WB) and quantification of SUN1 levels relative to Histone 3 (H3) expression shows a significative reduction of total SUN1 levels in C9 lines compared to isogenic counterparts (n = 8 independent experiments for both 24a and C52 iPSC isogenic pairs; Student’s t test, **** p < 0.0001). h. Representative images of Nesprin2 (Nesp2, grays) staining pattern in i3CNs from C9 and Ctrl iPSC lines. DAPI (blue) identified the nuclei, while LaminB (LMNB, green) labeled the nuclear lamina. The white boxes indicate the neurons enlarged in the panels on the right. i-k. Plots of line profiles of Nesprin2 and DAPI intensities normalized to the max intensity in control cells. Mutant cells display frequent mislocalization of Nesprin2 to either the nucleoplasm or cytoplasm (arrows), quantified in k. The yellow dashed lines in h indicate the lines used for the profile plots (Student’s t test, n = 4, *p < 0.05). l. The quantification of the Nesprin2 relative nuclear MFI shows a significant reduction in its abundance in C9 i3CNs compared to isogenic controls (Mann-Whitney t test, n = 31 and 38 for CTRL and C9 neurons from 4 independent differentiations, * p < 0.05). Scale bars: 20 μm in the main panels, 10 μm in zoomed-in images. For all, bars are mean and SEM, while violin plots show the distribution of the data with dashed lines indicating median and quartiles

SUN1 and SUN2 levels are reduced in C9-ALS iPSC-derived spinal cord organoids

Given the importance of the LINC complex in sensing mechanical properties of the cell’s intracellular and extracellular environment, which are not fully reproduced in a 2D cell culture system, we decided to assess LINC complex alterations in spinal cord organoids, which were generated using established protocols [26]. As expected, after 25 days of differentiation we could clearly identify NeuN-positive neurons and ISL1-positive MNs within each organoid, while GFAP-positive cells appeared around day 50 (Supplementary Fig. 5), at which point the organoids were considered to have reached a mature state. When we quantified SUN1 and SUN2 protein distribution and NE abundance by IF, we found a significant reduction of their nuclear levels in ISL1-positive C9 iMNs at day 50 (Fig. 3g and n), which was also coupled by the alteration of their distribution, as shown by the analysis of their line profiles. In fact, we found that mutant neurons were characterized by a loss of high-intensity peaks at the periphery of the DAPI profile and an increase in intranuclear peaks (Fig. 3b-e, and 3i-l). This was confirmed by the blind assessment of SUN1 and SUN2 frequency of mislocalization from the NE (Fig. 3f, m). An overall reduction of both SUN1 and SUN2 protein levels was also observed by WB using whole protein extracts of 120-day-old organoids (Figure o-r). Altogether these results confirm the impairment of the LINC complex in a 3D environment such as C9 iPSCs-derived spinal cord organoids.

Fig. 3

SUN1 and SUN2 alterations in spinal organoid iMNs. Representative images of SUN1 (a, grays) and SUN2 (h, grays) staining in spinal organoids. Islet1 (ISL1, green) expression was used to identify iMNs, while DAPI (blue) labeled the nuclei. b-f. Representative images and line profiles of SUN1 distribution in iMNs from control (b-c) and C9 mutant (d-e) organoids show a disruption in its localization at the NE, which was more frequently observed in iMNs from mutant organoids (quantified in f; Student’s t test, n = 4, **p < 0.01). The yellow dashed lines in a indicate the lines used for the profile plots. g. Quantitative analysis of SUN1 nuclear levels shows a significant reduction in C9 organoids compared to isogenic control (Student’s t test, n = 71 and 31 from 4 independent experiments, *p < 0.05). i-m. Representative images and line profiles of SUN2 distribution in iMNs from control (i-j) and C9 mutant (k-l) organoids show a disruption in its localization at the NE, which was more frequently observed in iMNs from mutant organoids (quantified in m; Student’s t test, n = 4, **p < 0.01). n. Quantitative analysis of SUN2 nuclear levels shows a significant reduction in C9 organoids compared to isogenic control (Student’s t test, n = 71 and 47 from 4 independent experiments, **p < 0.01). o-r. Representative blots and quantification of SUN1 (o-p) and SUN2 (q-r) levels from whole lysates of spinal organoids from C9 and control iPSCs shows a significant reduction in their overall levels (Student’s t test, n = 4 in p, n = 7 and 8 in r, *p < 0.05, ****p < 0.0001). GAPDH was used as loading control. For all, bars are mean and SEM, while violin plots show the distribution of the data with dashed lines indicating median and quartiles. Scale bars: 20 μm in a and h, 10 μm in b, d, i, and k

LINC complex is disrupted in motor neurons of sporadic and C9-ALS spinal cord

Since our data in 2D and 3D culture systems of both cortical and motor neurons strongly suggested that disruption to the LINC complex may be relevant to ALS/FTD disease pathogenesis, we decided to confirm such findings using patient-derived postmortem tissues. We first investigated the distribution of SUN and NESP proteins in the spinal cord of 5 sALS and 3 familial C9-ALS samples compared to 5 healthy controls (Supplementary Tables 2 and 3). These patients were all diagnosed with definite ALS and presented with a combination of upper and lower motor neuron signs and concomitant pathology. Only one C9-ALS patient also presented signs and symptoms related to FTD pathology. Given that SUN and NESP immunoreactivity creates a distinct ring around the nuclei of neurons, we scored morphologically defined motor neurons in the ventral horn blindly, based on the presence of a complete perinuclear ring at their equatorial plan (Fig. 4 and Supplementary Fig. 6). We found a significant higher percentage of MNs presenting a weaker or undetectable nuclear localization (i.e. “disrupted” staining) of both SUN and NESP proteins in MNs of sALS and C9-ALS compared with healthy controls, while no significant difference was noted between sALS and C9-ALS (Fig. 4g-h, Supplementary Fig. 6g-h).

Fig. 4

LINC complex disruption in sALS and C9-ALS spinal cordpostmortem specimens. a-f. Spinal cord sections from control (a, d), sALS (b, e) and C9-ALS (c, f) patients were stained with antibodies specific for SUN1 (a-c) and Nesprin1 (d-f). Hematoxylin and eosin counterstains were used to identify the nucleus and cytoplasm, respectively. The black boxes identify the motor neurons enlarged in the insets. Scale bars: 100 μm. g-h. The frequency of disrupted NE staining for both SUN1 (g) and Nesprin1 (h) was quantified blindly in at least two sections from each patient’s tissue. A significant increase in the percentage of cells with disrupted staining was observed in both sALS and C9-ALS spinal cords compared to controls. Each dot represents the mean of at least two sections for each case, horizontal lines show mean and standard deviation (one-way ANOVA with Tukey post hoc test, n = 5, 5, and 3, *p < 0.05, **p < 0.01, ns = not significant)

We also found a strong correlation between the frequency of SUN1 and SUN2 staining alteration in the spinal cords of our cohorts (r = 0.88, p < 0.0001), supporting the power of our analysis (Supplementary Fig. 7a). Furthermore, we noted a strong correlation between Nesprin and SUN proteins disruption in each patient (Supplementary Fig. 7b-e), which hinted at a possible causal link between the two phenotypes. In fact, and in accordance with previous literature [27], we found that SUN1 localization at the NE is necessary for Nesprin2 protein nuclear localization. Knocking down SUN1 in HEK293 cells led to a severe depletion of Nesprin2 from the nucleus and alterations to nuclear shape (Supplementary Fig. 8), which suggests that loss of SUN proteins from the NE may be the first pathologic change that leads to the loss of Nesprins and the disruption of LINC function.

SUN1 and SUN2 distribution is disrupted in the motor cortex of ALS postmortem tissues

As ALS patients commonly present clinically with a combination of upper and lower motor neuron signs and concomitant pathology [28], we next performed additional immunohistochemical staining of SUN1 and SUN2 on sections from the brain motor cortex region of each case. As described above, we scored cortical neurons in the motor cortex based on the presence of a complete perinuclear ring at their equatorial plan. For both SUN1 and SUN2, we observed a significant increase in the frequency of nuclei lacking a complete perinuclear ring in sALS or C9-ALS compared with healthy controls (Fig. 5). Furthermore, we noticed a higher degree of SUN2 staining disruption in C9-ALS compared to sALS tissues (Fig. 5h). Immunofluorescence analysis of brain tissue also allowed us to determine that SUN1 protein loss from the nuclear envelope is mostly restricted to MAP2-positive neurons, with other MAP2-negative cells mostly spared (Supplementary Fig. 9). To further prove the strength of our analyses, we looked at the relationship between the disruption of SUN1 and SUN2, finding a strong degree of correlation between these two variables (r = 0.75, p = 0.0030, Supplementary Fig. 10a). We also evaluated the relationship between SUN proteins pathology across the corticospinal neural axis. For both proteins, we found a strong correlation in the degree of disruption between spinal cord (SC) and brain for each patient (SUN1: r = 0.72, SUN2: r = 0.75), with the control group clearly clustering separately from both ALS cohorts (Supplementary Fig. 10b, c).

Fig. 5

Disruption of SUN proteins in sALS and C9-ALS brain postmortem specimens. a-f. Sections of the brain motor cortex from control (a, d), sALS (b, e) and C9-ALS (c, f) patients were stained with antibodies specific for SUN1 (a-c) and SUN2 (d-f). Hematoxylin and eosin counterstains were used to identify the nucleus and cytoplasm, respectively. The black boxes identify the neurons enlarged in the insets. Scale bars: 100 μm. g-h. The frequency of disrupted NE staining for both SUN1 (g) and SUN2 (h) was quantified blindly in at least two sections from each patient’s tissue. A significant increase in the percentage of cells with disrupted staining was observed in both sALS and C9-ALS spinal cords compared to controls. A significant higher frequency of disruption in C9-ALS neurons was detected for SUN2. A similar trend was observed for SUN1, but it did not reach statistical significance. Each dot represents the mean of at least two sections for each case, horizontal lines show mean and standard deviation (one-way ANOVA with Tukey post hoc test, n = 5, 5, and 3, *p < 0.05, **p < 0.01, ***p < 0.001, ns = not significant)

Since we found that both SUN1 and SUN2 proteins were mislocalized in cortical neurons and motor neurons in vitro, we wondered whether similar alterations could be found in the patients’ postmortem tissues. To that end, we assessed the intracellular distribution of SUN proteins by IF, using MAP2 as a cytoplasmic marker and DAPI as nuclear reference, and comparing their linear profiles between control and ALS samples (Fig. 6 and Supplementary Fig. 11). As expected, the linear profile of both SUN1 and SUN2 in control neurons showed two high intensity peaks at the periphery of the nucleus (Fig. 6a-c and Supplementary Fig. 11a-c), in accordance with their localization at the NE. In sALS and C9-ALS neurons, SUN1 and SUN2 distribution showed irregular profiles lacking high intensity perinuclear peaks, confirming our blind analysis of the immunohistochemistry data (Fig. 6d-i and Supplementary Fig. 11d-i). Furthermore, a greater frequency of high-intensity peaks was noted outside of the nucleus. Together, these data demonstrate a severe impairment in the localization of SUN proteins to the NE in the motor cortex of ALS patients, leading to mislocalization and accumulation of both proteins in the cytoplasm of these cortical neurons.

Fig. 6

SUN1 is mislocalized in the cytoplasm in cortical neurons of sporadic and C9-ALS patients. Representative images of motor cortex sections from control (a, b), sALS (d, e) and C9-ALS (g, h) patients stained with antibodies specific for SUN1 (red) and MAP2 (grays). DAPI (blue) was used to label nuclei. White boxes in a, d, and g identify neurons enlarged in b, e, and h. Line profile plots show the marked difference of SUN1 distribution in sALS (f) and C9-ALS (i) cortical neurons compared to controls (c). The yellow dashed lines in a indicate the lines used for the profile plots. Scale bars: 20 μm in main panels, 10 μm in zoomed-in images

LINC complex disruption contributes to altered nuclear morphology in ALS neurons

Since the LINC complex is a known regulator of nuclear morphology and size through its interaction with the nuclear lamina and cytoskeleton [29], we wondered whether the observed alterations in the levels and localization of its main components could impact the control of nuclear size. To test that possibility, we first quantified the area of the nucleolus at its equatorial plane in our 2D and 3D cultures and normalized it to the area of the whole nucleus (relative nucleolar area). While we did not observe any major alteration in iPSC-derived MNs grown on a solid 2D substrate (not shown), we found that ISL1-positive C9 MNs grown in the organoids had significantly smaller nucleoli compared to their isogenic controls (Supplementary Fig. 12a-c), in agreement with previously reported in vivo observations. When we performed a similar analysis in the spinal cord samples, we again observed in both sALS and C9-ALS motor neurons a significant reduction in the absolute nucleolar area (13.99 ± 4.75 μm for sALS, 9.73 ± 3.93 μm for C9-ALS and 22.71 ± 4.77 μm for CTRL) and a trend toward smaller nuclei (166.8 ± 40.80 μm for sALS, 156 ± 51.36 μm for C9-ALS and 204.5 ± 40.61 μm for CTRL) compared to non-neurological controls (Supplementary Fig. 12d-f). No difference was observed in the average cytoplasmic area (879.7 ± 183.2 μm for sALS, 826.6 ± 373.8 μm for C9-ALS and 881.4 ± 167.3 μm for CTRL).

To find out a possible causal relationship between LINC complex disruption and nuclear morphological features, we compared nuclear and nucleolar areas between ALS neurons with normal and disrupted SUN staining. For this analysis, we normalized the nuclear area to the area of the cytoplasm, and the nucleolar area to the area of the nucleus, to avoid confoundings due to differences in cell body or nuclear size (Fig. 7). Interestingly, we found that both sALS and C9-ALS neurons with disrupted SUN1 or SUN2 staining had significantly smaller nucleus to soma ratios compared to control neurons (Fig. 7b, d). An even bigger difference was found when the normalized nucleolar area was compared between control and SUN1 or SUN2-disrupted ALS MNs (Fig. 7d, e). Importantly, we found that sALS or C9-ALS MNs with intact SUN1 or SUN2 staining had instead comparable nuclear and nucleolar areas to the healthy controls (Fig. 7b-e). While it is not possible to fully draw strong causal vs. correlative conclusions from postmortem tissue, these data suggested that LINC complex disruption may be driving severe alterations to nuclear and nucleolar homeostasis.

Fig. 7

LINC complex disruption correlates with nuclear morphological alterations in ALS MNs. a. Representative images of spinal motor neurons from control (top), sALS (middle), and C9-ALS (bottom) with normal or abnormal SUN1 or SUN2 staining. The dashed yellow lines indicate the cells contour, white dots identify the nucleus, and yellow dots highlight the nucleolus. b-e. The quantification of the relative nuclear (b, d) and nucleolar (c, e) area in cells categorized based on the presence of a normal or abnormal SUN1 (b, c) or SUN2 (c, e) nuclear staining shows a significant correlation between LINC complex disruption and smaller nuclear and sub-nuclear structures (one-way ANOVA with Tukey post hoc test, n = 22, 47, and 15 in b and c, n = 15, 26, and 10 in d and e, *p < 0.05, **p < 0.01, ****p < 0.0001, ns = not significant)

We next wondered what role, if any, could TDP-43 nuclear depletion and aggregation, the main pathological hallmark of ALS/FTD, have in driving LINC complex disruption and the LINC-associated decrease in nuclear and nucleolar size. To address this important question, we focused our analysis on the motor cortex, since we observed a broader range of LINC complex pathology in this tissue compared to the spinal cord, where most MNs presented with severe alterations of all LINC proteins. When we quantified the co-occurrence of TDP-43 nuclear depletion and SUN1 disruption in MAP2-positive cortical neurons, we found that almost the totality of the ALS neurons with mislocalized TDP-43 also presented with SUN1 perinuclear disruption (Supplementary Fig. 13), supporting a strong correlation between the two events. However, we also found that about 20% of neurons with normally localized TDP-43 presented alterations in SUN1 distribution (Supplementary Fig. 13), suggesting that either LINC complex alterations precede TDP-43 mislocalization, or that the two phenomena are independent from each other.

Based on these results, we decided to evaluate changes to the nuclear morphology (i.e., area and circularity) and soma size in cortical neuron divided into four sub-groups: (1) Control neurons, used as a reference (2), ALS neurons with normal SUN1 and TDP-43 staining (3), ALS neurons with disrupted SUN1 staining but normal TDP-43 distribution, and (4) ALS neurons with disrupted SUN1 staining and abnormal cytoplasmic TDP43 mislocalization. Similar to what we observed in spinal MNs, we found that ALS neurons with normal SUN1 distribution displayed nuclear parameters indistinguishable from non-neurological controls (Fig. 8). Disruption of SUN1 staining instead correlated strongly with a significant reduction in nuclear circularity and area, which was irrespective of changes in TDP-43 localization (Fig. 8c and Supplementary Fig. 14b). However, our analyses also revealed that TDP-43 cytoplasmic mislocalization not only further reduced absolute nuclear size (Fig. 8c), but also led to a significant reduction in the overall size of the cell soma (Supplementary Fig. 14a), suggesting that TDP-43 nuclear depletion and cytoplasmic aggregation may lead to general cell toxicity which impacts many different aspects of cellular homeostasis. This observation also explains the paradoxical trend of the relative nuclear size in ALS neurons, which is heavily reduced in neurons with disrupted SUN1 but normal TDP-43 staining but seems to return to control levels in neurons with cytoplasmic mislocalization of TDP-43 (Fig. 8b). Overall, our results suggest that LINC complex disruption may be an early event that leads to the pathological alteration of nuclear homeostasis, contributing to ALS/FTD pathogenesis.

Fig. 8

SUN1 nuclear loss is a main contributor to cellular morphological alterations. (a) Representative images of cortical neurons categorized based on the presence of nuclear (Nuc.) or cytoplasmic (cyto.) TDP-43 (green) and normal (Norm.) or abnormal (Abn.) SUN1 staining (red). MAP2 (grays) was used as a neuronal marker, while DAPI (blue) labeled the cell nucleus. Scale bar: 10 μm. (b) Quantification of the relative nuclear to cytoplasmic area shows that neurons with abnormal SUN1 distribution have significantly smaller nuclei compared to all other groups (one-way ANOVA with Tuckey post hoc test, n = 22, 21,29, 20, ****p < 0.0001). (c) Quantification of the absolute nuclear area of cortical neurons shows that all ALS neurons have smaller nuclei compared to healthy controls. However, loss of SUN1 from the NE (categorized as abnormal) further impacts nuclear size, regardless of TDP-43 localization (one-way ANOVA with Tuckey post hoc test, n = 22, 21,31, 22, *p < 0.05, **p < 0.01, ****p < 0.0001)