Neural stem cell homeostasis is affected in cortical organoids carrying a mutation in Angiogenin

Mutations in Angiogenin (ANG) and TARDBP encoding the 43 kDa transactive response DNA binding protein (TDP‐43) are associated with amyotrophic lateral sclerosis and frontotemporal dementia (ALS‐FTD). ANG is neuroprotective and plays a role in stem cell dynamics in the haematopoietic system. We obtained skin fibroblasts from members of an ALS‐FTD family, one with mutation in ANG, one with mutation in both TARDBP and ANG, and one with neither mutation. We reprogrammed these fibroblasts to induced pluripotent stem cells (iPSCs) and generated cortical organoids as well as induced stage‐wise differentiation of the iPSCs to neurons. Using these two approaches we investigated the effects of FTD‐associated mutations in ANG and TARDBP on neural precursor cells, neural differentiation, and response to stress. We observed striking neurodevelopmental defects such as abnormal and persistent rosettes in the organoids accompanied by increased self‐renewal of neural precursor cells. There was also a propensity for differentiation to later‐born neurons. In addition, cortical neurons showed increased susceptibility to stress, which is exacerbated in neurons carrying mutations in both ANG and TARDBP. The cortical organoids and neurons generated from patient‐derived iPSCs carrying ANG and TARDBP gene variants recapitulate dysfunctions characteristic of frontotemporal lobar degeneration observed in FTD patients. These dysfunctions were ameliorated upon treatment with wild type ANG. In addition to its well‐established role during the stress response of mature neurons, ANG also appears to play a role in neural progenitor dynamics. This has implications for neurogenesis and may indicate that subtle developmental defects play a role in disease susceptibility or onset. © 2024 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

In keeping with its neurogenic and neuroprotective role, mutations in ANG have been identified as a risk factor for amyotrophic lateral sclerosis (ALS) [17][18][19][20][21][22][23][24][25][26] and frontotemporal dementia (FTD) [27].Mutations in TARDBP encoding TDP-43, a 43 kDa RNA processing factor, have also been identified in ALS patients [28].Like ANG, other ALS-FTD-associated proteins with roles in RNA metabolism such as TDP-43, fused in sarcoma (FUS), and heterogenous nuclear ribonucleoprotein A1 (hnRNP A1) are associated with stress granules [29].Since mature neurons are post-mitotic, it seems intuitive to assume the dominant function of these proteins in the adult organism is stress amelioration and that pathologies occur due to impaired stress response.However, ANG also plays a role during development in neurite outgrowth and pathfinding, both in differentiating pluripotent cells and in zebrafish models [1,30].
Silberstein et al [31] identified ANG in a screen for secreted regulators of proliferation and self-renewal in the haematopoietic niche.They showed a dichotomous role for ANG in the regulation of proliferation and selfrenewal in haematopoietic stem/progenitor cells and committed myeloid progenitors [32].ANG is expressed in the developing mouse nervous system and has been shown to play a role in neuronal differentiation of mouse pluripotent cells, as well as in the developing zebrafish nervous system [1,2,30].It is also expressed in the adult mouse subventricular zone and is implicated in neurogenesis during recovery from stroke [33,34].
Here we asked if ANG played a similar a role during neurogenesis using a human induced pluripotent stem cell (iPSC)-derived model.Further, we wondered whether proteins implicated in late-onset neurodegenerative disease also had neurodevelopmental roles and whether defects in these roles contributed to disease susceptibility in later life.
To investigate the mechanisms underlying ALS-FTD due to mutations in the ANG and TARDBP genes, we derived and characterised human iPSCs (hiPSCs) from a family of ALS-FTD cases and an unaffected sibling identified by van Es et al [27].Affected individuals in this kindred have mutations in either ANG or both ANG and TARDBP.Using in vitro directed cortical neuron differentiation and 3D cortical organoid models, we report here the effects of these mutations on both mature and developing neurons.We find that ALS cortical organoids exhibit proliferation defects in addition to the occurrence of abnormally shaped persistent rosettes.We investigated the underlying mechanism affecting proliferation and self-renewal in neural precursor cells (NPCs) in both control and ALS organoids.We also show that these early defects affect the differentiation of mature cortical neuronal subtypes, and furthermore, these mature neurons exhibit a defective stress response.

Ethics approval and consent to participate
Ethical approval was granted by the Medical Ethical Committee of the University Medical Center, Utrecht, Netherlands, and all subjects provided informed consent.All experiments were approved by the local Ethics Approval Committee, University of Bath.

Materials
Sources and details of all materials used are listed in supplementary material, Tables S1-S4.

Isolation of genomic DNA and sequencing of ANG and TARDBP genes
Methods are presented in Supplementary materials and methods.

Reprogramming to iPSCs
Fibroblasts were reprogrammed using the Cytotune2 reprogramming kit following the manufacturer's instructions (Thermo Fisher Scientific, Waltham, MA, USA).Clones were picked between 21 and 23 days after transduction, mechanically disassociated using a pulled glass-pipette, and transferred to individual wells of 24-well plates.Cultures were passaged using collagenase.At least 48 independent clones were picked from the reprogramming of fibroblasts of each genotype.For passaging, cultures were incubated with 1 mg/ml Collagenase IV (Gibco) in DMEM/F12 for 5 min and individual colonies were disassociated to small clumps, dislodged, and transferred to a feeder layer of mitomycin C-treated STO fibroblasts (lab made) on gelatin-coated 10-cm Petri dishes (BD Falcon, Franklin Lakes, NJ, USA).

RNA isolation and RT-PCR
Total RNA was prepared from cells lysed in Trizol (Gibco) following the manufacturer's instructions.RNA was treated with DNase (Ambion DNAfree) and reverse transcribed using Revert-Aid H minus firststrand cDNA synthesis kit (Fermentas, Waltham, MA, USA) using oligodT primers.
RT-PCR was performed in 30-μl reaction volumes on 10 ng cDNA template using GoTaq G2 system (Promega, Madison, WI, USA) with variable MgCl 2 and 0.5 mM of each primer.Cycling parameters were as follows: 95 C for 2 min followed by 35 cycles of 95 C for 30 s, 60 C for 30 s and 72 C for 30 s, extension at 72 C for 1 min before holding at 10 C. Primer sequences are listed in supplementary material, Table S2.

Immunofluorescence of cells adhered to cover slips
Cells grown on coverslips were fixed for 20 min in either ice-cold 4% PFA or methanol:acetic acid (3:1).Fixed cells on coverslips were dehydrated through an ethanol series and stored in 70% ethanol for long-term storage.Fixed cells on coverslips were rehydrated to PBS prior to staining and blocked for 1 h in blocking solution [0.1% gelatin (Sigma-Aldrich), 1% FBS (Labtech) and 0.5% Triton X-100 (BDH, VWR, Radnor, PA, USA) in PBS], and samples were incubated overnight at 4 C with the appropriate primary antibodies diluted in blocking solution.The unbound primary was washed off by 4Â 10-min washes in PBST (0.1% Triton X-100 in PBS).Cells were incubated for 2 h with secondary antibodies diluted in blocking buffer.Further to incubation with the secondary antibody, the coverslips were washed (4Â 10 min) with PBST and two further washes with just PBS, with the last wash containing 0.1 μg/ml DAPI.Coverslips were mounted with Mowiol (Sigma-Aldrich) and cells imaged using a Leica DM5500B microscope with a DFC 360FX camera.Details of the antibodies used are presented in supplementary material, Table S3.

Histology, immunohistochemistry, and immunofluorescence
Embryoid bodies (EBs) to be sectioned were transferred to 30% sucrose in PBS and kept at 4 C overnight.EBs were then embedded in OCT (Raymond A. Lamb, London, UK), snap frozen on dry ice, and stored at À80 C. Sections were cut using a Leica cryostat, transferred to chrome alum (Sigma-Aldrich) and gelatin (Sigma-Aldrich) subbed slides (SuperFrost Menzel Glaser, Braunschweig, Germany), and stored at À20 C.
Sections were allowed to warm to room temperature (RT) prior to immunostaining and washed twice in PBS to remove OCT.Antigen retrieval was performed in boiling sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) for 15 min followed by cooling and two washes with PBS.After blocking for 1 h in blocking buffer, samples were incubated overnight at 4 C with primary antibodies diluted in blocking buffer.This was followed by 4Â 10 min washes in PBST (0.1% Triton X-100 in PBS) followed by incubation for 2 h at RT with appropriate secondary antibodies diluted in blocking buffer.Unbound secondary antibody was removed by washing with PBST (4Â 10 min) followed by two further washes with PBS with the last wash containing 0.1 μg/ml DAPI.Stained sections were mounted in Mowiol (Sigma-Aldrich) and imaged using a Leica DM5500B microscope with a DFC 360FX camera.Antibodies and dilutions are listed in supplementary material, Table S3.

Haematoxylin and eosin (H&E) staining
Sections of cortical organoids were washed twice in water before incubation in Mayers haematoxylin (Sigma-Aldrich) for 5 min.After two washes, sections were blued in dilute ammonia water followed by a further two washes in water.Haematoxylin-stained sections were dehydrated through 30, 50, and 70% ethanol, for 2 min each, and stained in 1% Eosin Y (Sigma-Aldrich, in 70% ethanol) for 5 min, followed by rapid washes in 70 and 95% ethanol, then two 100% ethanol washes for 5 min each.Sections were cleared by two washes of 2 min each in Histoclear (National Diagnostics, Atlanta, GA, USA) prior to mounting in dibutylpthalate polystyrene xylene.

O-propargyl-puromycin (OPP) incorporation and detection
For cells grown on coverslips, medium was changed 2 h prior to OPP incorporation, or upon the addition of ANG/ANG mutant or ANG inhibitor at the indicated times.Cultures were incubated for 30 min with 10 μM OPP (Jena Bioscience, Jena, Germany).After OPP incorporation, cells were fixed in 4% PFA in PBS for 15 min at RT followed by two washes in PBS.Cells on coverslips were stained for additional markers using the protocol described above.
For the detection of incorporated OPP, the Click-iT reaction was set up in 1Â reaction buffer, additive and copper protectant (Thermo Fisher Scientific) together with 1:400 Alexa Fluor 488 picolyl azide following the manufacturer's instructions.After 30 min at RT, coverslips were washed twice with PBS and mounted in Mowiol.Images were acquired as 3D stacks from four random fields for each treatment from three independent experiments.Z-Stacks were deconvoluted in Leica LAS AF software (Leica Microsystems, Milton Keynes, UK).Deconvoluted stacks were subjected to flat-field correction using BaSiC in Fiji [42,43] (https://github.com/marrlab/BaSiC, accessed 10 July 2018; https://imagej.net/software/fiji/, accessed 10 July 2018), followed by maximum intensity projection.Nuclear masks were created using the DAPI channel.Mean grey values for each nucleus was measured in the OPP and proliferating cell nuclear antigen (PCNA) channels.Nuclear PCNA texture analysis was performed using the Fiji plugin GLCM Texture Too (T.Cornish, https://github.com/cornish/GLCM-TextureToo/, accessed 12 October 2019).

Fibroblast stress assays
Patient-derived fibroblasts were seeded at a density of 10 4 cells/cm 2 in triplicate in 96-well plates (BD Falcon) in fibroblast growth medium.Medium was changed after 24 h to include either tert-butylhydroperoxide (tBH), paraquat, or sodium arsenate (all Sigma-Aldrich) at indicated concentrations.Where indicated, purified Angiogenin (200 ng/ml) was added 30 min prior to the addition of the stress agent.Viability after treatment was determined by Promega 96-well nonradioactive cell titre kit following the manufacturer's instructions.

Western blotting
Fibroblasts were grown on tissue culture dishes (BD Falcon) in complete fibroblast medium.The cultures were treated at 90% confluency with either 1 mM hydrogen peroxide for 1 h for induction of stress or medium alone as a control.Cells were lysed in lysis buffer [2% SDS (Sigma-Aldrich), 10% glycerol (BDH), 60 mM Tris (Sigma-Aldrich) and 100 mM dithiothreitol (Sigma-Aldrich), pH 6.8].Protein concentrations were determined using a Biorad Protein Assay (Bio-Rad Laboratories, Watford, UK), and 40 μg of each lysate was loaded onto a 10% Tris-Tricine gel with unstained low molecular weight markers (Fermentas).Resolved proteins were electro-blotted onto a PVDF membrane at 30 V for 1.5 h.Membranes were incubated in blocking solution [5% Marvel milk powder in PBS (Thermo Fisher Scientific) with 0.1% Tween 20 (Sigma-Aldrich)] for 1 h at RT. Blocked membranes were incubated with primary antibody in blocking solution overnight at 4 C. Membranes were washed in PBST (4Â 10 min) followed by incubation with an appropriate peroxidase-conjugated secondary antibody for 1 h at RT.The unbound secondary antibody was washed off in PBST (4Â 10 min) prior to chemiluminescent detection using the ECL Kit (Sigma-Aldrich).See supplementary material, Table S3, for details of antibodies.

Analysis of single-cell mRNA sequencing data
The UCSC Cell Browser interactive viewer (https:// cells.ucsc.edu/,accessed 12 October 2019) was used to examine ANG expression in the 'cortex-dev' dataset (https://cells.ucsc.edu/?ds=cortex-dev, accessed 12 October 2019) generated by Nowakowski et al (2017) [42].This data set spans human corticogenesis and neuronal differentiation from progenitor cells to post-mitotic neurons.Cells expressing ANG were binned by cell-type cluster and the median expression as transcripts per million (TPM) calculated.The distribution of ANG-expressing cells was binned by gestational time (age in weeks).

Statistical information
Each experiment was repeated at least three times using two independent cell lines of the same genotype.In organoid experiments, at least four different fields from four individual organoids were analysed in each experiment.In experiments using monolayer cell culture, images were acquired from at least four random fields in each experiment.Images for quantification were acquired at equal gain and exposure, flat field corrected using the BaSiC ImageJ plugin, and further analysed using Fiji as described under each method [43,44].Where possible, data points representing the quantification of results from each independent experiment or organoid are presented as dot plots with an overall mean ± SEM.Where fluorescent intensity analyses were performed, plots based on single-cell fluorescence intensity are presented pooled with the median and interquartile ranges.

Fibroblasts from ALS-affected kindred carry ANG and TDP-43 protein mutants
Skin fibroblasts were isolated from three members of an ALS-FTD affected kindred [27] -ALS04624 (ALS24) fibroblasts carry the ANG K17I variant, ALS01364 (ALS64) carry both ANG K17I and TDP-43 N352S , and finally the unaffected kin fibroblasts CON0327 (CON) carry mutations in neither gene (supplementary material, Figure S1A,B).We showed previously that the K17I variant exhibited greatly reduced RNase activity and lacked neuroprotective activity [45,46].Both the ALS fibroblast lines were more vulnerable to stress and showed altered TDP-43 levels and processing (supplementary material, Figure S1C-G).The addition of ANG WT protein protected these fibroblasts from stress to a limited extent (supplementary material, Figure S1C-E, dashed).

Reprogramming and characterisation of patientderived iPSCs
Five individual iPSC clones derived from each fibroblast line that showed good ES-like morphology were further characterised.Each showed robust expression of pluripotency markers, comparable to the human ESC line Shef1 (Figure 1A and supplementary material, Figure S2).Robust expression of a panel of 12 pluripotency-associated transcripts was seen by RT-PCR at levels comparable to Shef1 (Figure 1B).Clones negative for the reprogramming cassette were used for further experiments (Figure 1C).Two iPSC clones from each donor were tested for their ability to form derivatives of the three germ layers.Representative markers and morphologies were successfully identified in embryoid bodies formed from these clones through H&E staining and immunohistochemistry (supplementary material, Figure S3).These clones, Control (CON-2 and CON-5), ANG mutant (ALS24-1 and ALS24-5), and ANG and TDP-43 mutant (ALS64-2 and ALS64-4), were taken forward to investigate the effects of the mutations in iPSC-derived cortical organoids.

ALS cortical organoids show altered progenitor frequency and organisation
The neuroepithelial marker nestin (NES) is expressed throughout early-stage organoids, with numerous rosettes of radially arranged neural progenitors present in both control and ALS organoids (Figure 2A).Between days 20 and 30, cells positive for the cortical SVZ marker TBR2 appear at the superficial zone of the rosettes (Figure 2A), followed by expression of doublecortin (DCX), a marker of immature migratory neurons [47] (Figure 2B).Cortical differentiation progresses across the time course with increased Neurofilament-m positive (NF-M) neurites and expression of the cortical subtype markers TBR1 (early-born layer VI), CTIP2 (layer V), and CUX1 (late-born layer II/III) (Figure 2C-E) [48].
Organoids of each genotype were competent to form the expected cortical cell identities; however, we noted substantial differences in the organisation and frequency of the progenitors.Rosettes remained more frequent at later stages in ALS organoids and appeared disorganised in size and shape compared to controls.DCX + and TBR2 + intermediate progenitors were not restricted to the superficial regions of ALS rosettes.In all cases, the ALS64 double mutant appeared more severely affected than the ALS24 single-mutant organoids.We therefore sought to better characterise the effect on the progenitor population and then subsequently in mature neurons.
Rosettes were easily identifiable by H&E staining (Figure 2F).The initial frequency and mean areas of ALS24 rosettes were comparable to that in CON organoids, while ALS64 rosettes were more frequent but smaller (Figure 2G,H, p < 0.05).The frequency of rosettes in ALS organoids remained higher than that in the CON organoids to the later stages of differentiation.Throughout the time course, ALS24 rosettes were significantly larger ($10-20%) than CON rosettes, and their overall frequency remained higher in comparison.We further characterised the organoids by comparing the frequency of cells expressing proliferative markers in conjunction with the neuroectodermal markers SOX2 and PAX6 and the neuronal marker NEUN.

Increased frequency of SOX2 progenitors in ALS cortical organoids
By day 20, most cells in organoids expressed SOX2 (Figure 3A,B), which reduced in frequency from day 30 onwards.However, ALS organoids consistently maintained a higher frequency of SOX2 + cells in comparison to CON organoids.By day 50 only 30.1% SOX2 + cells persisted in CON organoids, whereas 72.1% and 64.1% of SOX2 + cells were observed in ALS24 and ALS64 organoids ( p < 0.005).As SOX2 + cell frequency decreased, the frequency of PAX6 + cells increased to a 62.5% peak by day 40 in CON organoids but remained around 10% lower from day 40 onwards in ALS organoids (p < 0.05).However, the frequency of NEUN + cells did not differ significantly between organoids by day 50, rising to 53.2% in CON, 45.1% in ALS24, and 42.1% in ALS64 organoids.

Persistence of proliferative progenitors in ALS cortical organoids
The increased number of SOX2 + progenitors seen in the rosettes of later-stage ALS organoids was reflected in an increased frequency of mitotic cells (Figure 3A).The frequency of PCNA + and phosphohistone 3 (PH3 + ) in organoids of all genotypes was comparable at day 20 but differed from day 30 onwards.Consistently, fewer PH3 + cells (11.6%) were present in CON rosettes in comparison to both ALS24 and ALS64 rosettes (23.1 and 20.1%) (Figure 3C, p < 0.05).This was also observed in day-50 ALS organoids in which rosettes remained more frequent and contained a higher proportion of PCNA + and PH3 + cells.
The positioning of cells undergoing mitosis in rosettes was also disrupted in ALS organoids.While PCNA + cells were distributed throughout the rosette, PH3 + cells in CON rosettes were typically positioned at the apical domain of the rosette (Figure 3A,D, within the dashed lines).In contrast, significantly more PH3 + cells were distributed throughout ALS24 and ALS64 organoid rosettes away from the apical domain (Figure 3A, arrows), with means of 6.8 and 5.1% within ALS24 and ALS64 rosettes versus 1.2% in CON rosettes (Figure 3E, p < 0.05).This proportion increased up to a maximum of around 26% on day 50 for both ALS24 and ALS64.Non-apical dividing cells in a rosette may be indicative of a differentiating progenitor leaving the rosette; however, this did not correspond with the increased frequencies of SOX2/PAX6 + in ALS rosettes.Therefore, the ectopic location of these cells in rosettes in the ALS organoids is likely to be due to defects in the cellular organisation of the rosette.

Rosettes in ALS organoids are misshapen due to abnormal orientation of division
Rosettes in ALS mutant organoids appeared malformed and misshapen in comparison to CON organoids, often lacking the dense radial array of nuclei and well-defined lumen.The centrosomal protein pericentrin (PCNT) was ANG mutations and stem cell homeostasis 415 used to identify the bases of primary cilia projecting from the apical domain of cells bordering the lumen in rosettes [49,50].Many small clusters of PCNT + cells begin to organise in day-20 CON organoids (supplementary material, Figure S4A), while larger, less organised patches of PCNT + cells were observed in ALS24 and ALS64 organoids.Rosettes form by day 30 in organoids of all three genotypes.CON rosettes contained a distinct lumen with a continuous ring of PCNT + centrosomes at the apical surface and well-defined apical boundaries  until day 50, at which point they dissipated.In contrast, at day 30, the apical domains of ALS24 and ALS64 rosettes appeared disorganised with a much thicker band of apical cells and increasingly disorganised shape (supplementary material, Figure S4A,D).The mean circularity of rosettes in CON organoids remained consistently high (above 0.85, where 1.00 is a circle), while ALS rosette circularity was reduced to around 0.80 for ALS24 or 0.67 for ALS64 at day 50 (supplementary material, Figure S4B, p < 0.05).Division of neural progenitors in the rosette occurs at the apical domain, with the plane of division perpendicular to the lumen.This results in the characteristic densely packed radially arranged nuclei in the rosette (supplementary material, Figure S4D).Nuclei in CON organoid rosettes at day 30 orientate 70-90 from the apical domain (supplementary material, Figure S4E).In contrast, the apical nuclei of ALS rosettes showed no robust orientation in this manner.Whether these defects are causative or arise as a result of the altered progenitor dynamics will require further understanding of the underlying mechanism.

Altered protein synthesis in ANG mutant ALS hiPSC neural progenitors
ANG plays a role regulating global protein synthesis and has been shown to be involved in regulating this to maintain haematopoietic stem cell quiescence [32].In light of this, we investigated whether proteostasis was altered during neural differentiation of iPSCs carrying ANG mutations.We characterised protein synthesis at different stages of differentiation using an O-propargylpuromycin (OP-Puro) incorporation assay [51,52].In this assay, pulsed OP-Puro is incorporated into nascent polypeptides and can be visualised through a click reaction with fluorescent substrates.
Shef1 hESC, CON-2 hiPSC, and two ALS-ANG hiPSC lines (ALS24-1 and ALS24-5) by day 6 of differentiation gave rise to neuroectodermal cells with similar protein synthesis levels and PCNA distribution.On day 12 of differentiation, raised honeycombed rosettelike structures were present which were positive for SOX2 and NES and showed high levels of PCNA.Cells in the rosettes showed the highest levels of ANG expression during differentiation (supplementary material, Figure S5A), consistent with published expression data [42] (supplementary material, Figure S5B-D).They also showed high levels of protein synthesis, though this was significantly lower in both ALS cultures compared to CON cultures (Figure 4AB, p < 0.05).Replating in media lacking the neural precursor cell (NPC) growth factors EGF, FGF2, or LIF resulted in the differentiation of CON NPCs to DCX-positive intermediate progenitor cells (IPCs) by day 20 (Figure 4A).This was accompanied by a substantial reduction in both protein synthesis and PCNA expression.In contrast, multiple small rosettes could still be observed in ALS cultures on day 14 after replating.PCNA and protein synthesis levels remained comparable to day 12 NPCs (Figure 4B,C), suggesting this identity persisted for longer in both ALS24 clones despite the absence of NPC growth factors.

Protein synthesis can be rescued by treatment with ANG
ANG is readily taken up by cells from media, and we exploited this property to assess whether ANG or its variants had an effect on protein synthesis.Undifferentiated pluripotent stem cells (PSCs), neural progenitors (NPCs), and intermediate progenitors (IPCs) derived from CON-2 and ALS24-1 were treated with ANG WT , catalytically inactive mutant ANG K40I , or ANG inhibitor NCI-65828 [53] (Figure 4D).Protein synthesis and PCNA levels were quantified at each stage (Figure 4E and supplementary material, Figure S6D,C).ANG WT significantly increased protein synthesis in undifferentiated CON-2 and ALS24-1 and rescued levels in ALS24 NPCs closer to those of the control NPCs (p < 0.05, Figure 4E).No change in PCNA levels in PSCs was observed; however, ANG WT resulted in significantly increased PCNA in both CON and ALS NPCs (supplementary material, Figure S6C).In contrast, treatment with both ANG K40I and NCI-65828 consistently led to a significant reduction in the level of protein synthesis in PSCs and NPCs in each genotype.PCNA levels in CON NPCs were reduced to levels comparable to untreated ALS NPCs.No robust effects were seen in IPCs, though ANG WT resulted in increased protein synthesis and PCNA levels in CON-2 cells.
The proportion of S-phase cells in a population can be determined by cell cycle-dependent changes in the distribution of PCNA [54].A significantly larger number of ALS24 NPCs appear to be in S-phase (supplementary material, Figure S7A, p < 0.05).Treatment with ANG K40I or inhibitor resulted in significantly higher frequencies of S-phase cells in both control and ALS NPCs (supplementary material, Figure S7B, p < 0.05).This, together with the lack of difference in total PCNA + cells, indicates a prolonged S-phase in ALS24 NPCs rather than rapid proliferation since it was not accompanied by an increase in the frequency of PCNA + or PH3 + cells.This suggests that the increase in ALS NPC frequency may be due to increased selfrenewal, rather than increased proliferation.

Increased frequency of late-born cortical neurons in ALS organoids
Although ALS iPSC-derived NPCs seemed to favour self-renewal under conditions that promoted differentiation in CON NPCs (Figure 3B), we did not observe a significant reduction in the numbers of mature neurons in cortical organoids.However, within this mature population we found a bias towards later-born cortical neurons in ALS organoids with increased occurrence of CTIP2 + and CUX1 + neurons (Figure 5A,B, p < 0.05).This bias was small but accompanied by a significant reduction in early-born TBR1 + neurons (p < 0.05).No effect was seen on SATB2 + neuron frequency, though  this could have been due to the overall low frequency of these late-born neurons.

Increased apoptosis during stress in layer V neurons
Extended treatment (16 h) of cortical organoids with sodium arsenite (SA) led to the loss of cortical neurons (Figure 5B).In particular, CTIP2 + layer V neurons were more vulnerable in comparison to TBR1 + , CUX1 + , or SATB2 + neurons in ALS24 and ALS64 organoids (Figure 5B, p < 0.05).Significantly more cells were positive for the apoptotic marker cleaved caspase 3 (CC3) in ALS organoids after SA treatment, as well as significantly more CC3 + /CTIP2 + Increased neuronal cell death and impaired stress granule formation can be rescued by ANG The neuroprotective effects of ANG are thought to be due, at least in part, to its ability to induce stress granule (SG) formation in response to stress [8,46].Under normal culture conditions, SGs are very rarely observed in neurons (Figure 6A,B), and the addition of ANG alone does not induce their formation (not shown).
Treatment with SA for 16 h resulted in neuronal death with the viability of ALS24-derived neurons reduced to a third below that of the control neurons (Figure 6C, p < 0.05).The viability of both control and ALS cortical neurons was rescued by the addition of ANG WT 30 min prior to SA treatment.ANG WT rescued ALS-mutant neuron survival to levels similar to controls without ANG WT treatment but below that of controls treated with ANG WT (Figure 6C, p < 0.05).SG induction was observed after 1 h exposure to SA (Figure 6B).Significantly fewer SGs were present in SA-treated ALS-mutant neurons in comparison to control neurons (p < 0.05, supplementary material, Figure S6D).The addition of ANG WT prior to SA treatment significantly rescued SG frequency in both ALS and control neurons (p < 0.05, Figure 6D).Corresponding with the survival data, SG frequency in ALS24 neurons treated with ANG WT was rescued to levels similar to CON and Shef1 neurons without ANG WT treatment (p < 0.05, Figure 6D).

Discussion
Onset of neurodegenerative disease is most commonly associated with ageing.While defects in the cellular processes essential for adult neuron survival are critical at later stages, we show here that degeneration may be primed by defects occurring during development.We found that ALS-FTD-associated mutations in ANG led to changes in the timing and frequency of cortical neuron subtype specification due to alterations in the dynamics of neural stem cell self-renewal.Although the source of iPSC lines characterised here are of a single family, we were able to both rescue and phenocopy key aspects with ANG protein variants and inhibitors in control iPSC and ESCs.
We observed increased frequency of NPCs and rosettes in ALS cortical organoids.In addition, using a stepwise monolayer neuronal differentiation model, we observed a reduction in protein synthesis in ALS NPCs with a larger proportion of these NPCs in S-phase.A longer S-phase is associated with self-renewal of NPCs rather than differentiation in the developing murine nervous system [55,56].It is likely that the ANG K17I variant leads to a lengthened S-phase and increased selfrenewal of NPCs.The ANG K17I variant results in both a significant loss of RNase activity and impaired nuclear translocation [46], both of which are essential for its function [57].
Since the RNase activity of ANG is essential for its role in transcriptional remodelling through stress granule induction, it is possible a similar mechanism is able to change the expression profile of a cell during differentiation.We found that ANG protein levels during differentiation are at their highest at the rosette stage and do not vary in level between NPCs in rosettes, suggesting that it may be required to prime NPCs to respond to differentiation cues rather than directly promoting differentiation.Post-transcriptional control of SOX2 and PAX6 expression through translational repression is essential for adult neural stem cell differentiation [58] and translational remodelling occurs on a large scale as differentiation progresses [59].
ANG is involved in adult stem cell dynamics in the haematopoietic system [31,32], where it regulates haematopoietic stem cell (HSC) quiescence by restricting proliferation.ANG also subsequently promotes proliferation in myeloid (but not lymphoid) progenitors [32].It is difficult to directly compare our results with these findings in HSCs as there is no real quiescent population present in our developmental neurogenesis models.Perhaps the closest analogue to NPCs are the myeloid progenitor cells, where ANG promotes protein synthesis and proliferation.It has been suggested that adult neurogenesis is altered in ALS-FTD [60].ANG is also upregulated during adult neurogenesis in mice [61] and is expressed in the mouse subventricular zone after ischaemic stroke in the neuroblasts [34].
Defective NPC dynamics in ANG mutants resulted in altered frequencies of layer-specific neurons in organoids, an effect somewhat similar to neurogenesis dysfunction shown to affect layer specification in mice.Sustained self-renewal of cortical NPCs through over-expression of PAX6 results in an increased production of upper layer neurons [62].In contrast, the loss of SuFu (Suppressor of Fused) led to reduced NPC self-renewal and increased lower layer neuronal identities [63].Mutations in several genes associated with FTLD also resulted in neurogenesis defects in vitro.Diseaseassociated mutations in the MAPT gene led to reduced NPC proliferation and a reduction in glutamatergic neuron frequency [64], while patient-derived hiPSCs carrying progranulin (GRN) mutations were reportedly unable to differentiate efficiently to cortical neurons [65].
In addition to defects in neurogenesis, we also observed an increased sensitivity to stress leading to the death of cortical neurons.Layer V CTIP2 + neurons and, to a lesser extent, CUX1 + layer II/III neurons showed increased apoptosis in response to stress.Increased susceptibility in these cell types was demonstrated previously during disease.In FTLD, a subset of CTIP2 + layer V neurons termed Spindle or von Economo neurons (VENs) are reportedly the earliest neuronal subtype to undergo degeneration [66,67].Upper layer neurons expressing CUX1 have also been shown to be more susceptible to degeneration in comparison to deeper layer neurons [68].The loss of corticospinal motor neurons (CSMNs) from layer V of the motor cortex ANG mutations and stem cell homeostasis  is a characteristic ALS pathology.As with VENs, CSMNs are also part of the CTIP2 + pool [48].Loss of neurons from layers II/III and V of the motor cortex has been reported in C9orf72 and TDP-43 mouse models of ALS-FTD [69,70].Layer V and CSMNs receive afferent connections from layer II/III [71], and mis-specification of earlier stages may also result in circuits more vulnerable to degeneration.Together, these reports suggest the layer-specific sensitivity observed in our model fits well with the expected ALS-FTD pathology, where ALS is the primary diagnosis and agrees with the clinical presentation of the patient kindred from which these hiPSC lines were derived [27].
The impaired induction of stress granules increases the sensitivity of ANG K17I neurons to stress.This effect is seen across all neuronal subtypes and the patient fibroblasts from which the iPSC lines were derived.This suggests additional factors underlie the vulnerability of particular types of neurons in ALS-FTD.Stress granule induction could be both rescued and elevated by the addition of WT ANG to both ALS ANG K17I and control neurons, with a concomitant increase in survival during stress.ANG-mediated neuroprotection of motor neurons can occur in a paracrine fashion through astrocytes [72]; however, here astrocytes are mostly absent, indicating a neuronal intrinsic effect.Both autocrine and paracrine neuroprotective mechanisms likely co-exist; ANG may act immediately in neurons under stress and then further recruit astrocytes in a relationship similar to that seen between hypoxic tissue and endothelial cells during angiogenesis.
The increasing identification of genetic and environmental factors linked to disease risk highlights the multifactorial and multistep nature of ALS-FTD [73,74].In addition to its neuroprotective role, ANG appears to function at multiple points in the developing nervous system besides progenitor cell proliferation [1,30].Knock-out or mutation of other genes implicated in neurodegeneration in mouse models also causes adult and neurodevelopmental defects [75][76][77][78].The degree to which neurodevelopmental defects such as those reported here manifest in vivo and how they may contribute to the multistep process required for disease onset requires further work.If so, early intervention targeted at factors like ANG could potentially reduce risk for affected families or those identified through genetic screening.However, the mechanisms by which ANG interacts with other ALS-FTD-associated proteins and pathways in the mature and developing nervous system remain to be investigated.

Figure 1 .
Figure 1.ALS kindred-derived iPSC lines express markers associated with pluripotency.Representative iPSC lines derived from fibroblasts obtained from an unaffected family member (CON0327), an ANG K17I mutation-carrying patient (ALS04624), and a patient with both ANG K17I and TDP-43 N352S mutations (ALS01364).(A) Phase contrast (Phase Con.), immunostaining for pluripotency-associated factors, and alkaline phosphatase (Alk.Phos.) expression of colonies at passage 7. Scale bars, 100 μm.(B) RT-PCR for pluripotency-associated transcripts for five representative clones of each genotype.(C) RT-PCR for episomal reprogramming transcripts to identify transgene-free cell lines.The hES Shef1 cell line was used as positive control; (À) negative control.

Figure 2 .
Figure 2. Control and ALS hiPSCs are competent to form cortical neurons during organoid differentiation.Cortical organoids formed from control iPSCs (CON), ALS ANG K17I iPSC (ALS24), or ALS ANG K17I /TDP-43 N352S iPSC (ALS64) were generated over 50 days, fixed at the indicated time points, sectioned, and immunostained for pan-neuronal markers and layer-specific cortical markers.(A) Nestin (NES) for neuro-ectoderm and immature neurons and TBR2 for cortical SVZ progenitors.(B) Doublecortin (DCX) for immature neurons.(C) Neurofilament-m (NF) for mature neurons and TBR1 for early-born layer VI cortical neurons.(D) CTIP2 for layer V cortical neurons.(E) CUX1 for layer II/III cortical neurons.(F) H&E staining of cryosections showing organisation of rosettes at days 30 and 50.Arrows indicate rosettes.Scale bars, 100 μm.(G and H) Mean rosette frequency and area in organoids over the 50-day time course.*p < 0.05, **p < 0.005, asterisk colour indicates cell line series.Data compared by ANOVA with Bonferroni's post hoc test.Error bars show SEM.

Figure 3 .
Figure 3. ALS mutant organoids show persistence of early neural progenitor cells and neural rosettes.(A) Cortical organoids formed from control iPSCs (CON), ALS ANG K17I iPSCs (ALS24), or ALS ANG K17I /TDP-43 N352S iPSCs (ALS64) were analysed at the indicated time points by immunostaining of cryosections.Dashed lines indicate apical domain in rosettes and white arrows indicate ectopic PH3 + cells.Scale bars, 100 μm.(B) Quantification of the frequency of markers SOX2, PAX6, and NEUN.(C) Quantification of the frequency of the proliferative markers PCNA and the proportion of PCNA + cells also PH3 + .(D) Schematic showing a generalised structure of a neural rosette with a 30-μm 'apical' ventricular zone-like domain adjacent to the lumen.(E) Quantification of proportion of PH3-positive rosette cells found within apical zone, elsewhere in rosette (non-apical), or outside rosette entirely (non-rosette).*p < 0.05, **p < 0.005.Where used, asterisk colour indicates marker series.Data compared by ANOVA with Bonferroni's post hoc test.Data acquired from a minimum of nine organoids from three independent experiments and presented as percentage of total DAPI-positive nuclei present.Error bars show SEM.

Figure 4 .
Figure 4. Protein synthesis and PCNA levels during differentiation.O-propargyl-puromycin (OPP) incorporation as a proxy for protein synthesis in differentiating Shef1 hESCs and control iPSCs (CON) or ALS ANG K17I iPSCs (ALS24).(A) OPP incorporation was performed for 30 min prior to fixation followed by immunostaining for PCNA expression and click chemistry detection of OPP at indicated time points.Scale bars, 100 μm.(B) The mean level of OPP incorporation and (C) PCNA expression level were determined at the indicated time points from the mean nuclear grey values of individual cells in four random fields from three independent experiments.*p < 0.05, log e -transformed data compared by ANOVA with Bonferroni's post hoc test.Error bars show SEM.(D) Control hiPSCs (CON-2) and ALS ANG K17I iPSCs (ALS24-1) grown on Matrigel in E8 medium (PSC) and differentiated to rosette-stage neural progenitor cells (NPC, day 12).Cells were treated with 200 ng/ml WT ANG, 200 ng/ml ANG K40I , or 80 μM NCI-65828 for 6 h prior to fixation where indicated.Immature progenitor cells (IPC, day 20) OPP and PCNA data in supplementary material, Figure S6.(E) Violin plots of OPP incorporation as a proxy for protein synthesis in individual cells.Mean grey values measured from individual nuclei.* increased and † decreased p < 0.05.PSC and IPC data transformed (Log e x) and compared by ANOVA with Bonferroni's post hoc test.NPC data transformed [Log e (255 À x)] and compared by ANOVA with Bonferroni's post hoc test.Red circle shows median value of population.3,000-4,000 cells per plot from four random positions in three independent experiments.

Figure 5 .
Figure 5. Bias towards late-born cortical neuron subtypes in ALS cortical organoids and sensitivity to stressors.(A) Immunofluorescence for layer-specific cortical neuron markers TBR1 (layer VI), CTIP2 (layer V), CUX1, and SATB2 (layer II/III) in cortical organoids formed from CON-2, ALS24-1, and ALS64-2 iPSCs.Scale bars, 100 μm.(B) Quantification of cortical neuron subtypes with or without exposure to 0.5 μM sodium arsenite (SA) for 16 h prior to fixation and staining for subtype markers.Subtype markers quantified as a proportion of DAPI-positive nuclei in three organoids from three independent experiments.Each circular point represents an organoid; diamonds show means with SEM error bars.Data compared by ANOVA with Bonferroni's post hoc test.*p < 0.05.

Figure 6 .
Figure 6.Impaired stress granule assembly in ALS-24 cortical neurons can be rescued by WT ANG.(A) Immunofluorescence for cortical neurons differentiated from Shef1 hESCs, CON-2, ALS ALS24-1, and ALS24-5 hiPSCs stained for Tuj1 and NF-M and with DAPI.(B) Immunofluorescence for eIF3b and TIAR in cortical neurons differentiated from Shef1 hESCs, CON, ALS24-1, and ALS24-5 hiPSCs shows characteristic redistribution into stress granules in the presence of 0.5 μM sodium arsenite (SA) added to induce stress for 1 h prior to fixation.(C) Quantification of viability by MTT assay after prolonged SA treatment, with and without WT ANG.WT ANG was added 30 min prior to and during a 16-h SA treatment where indicated.p < 0.05.(D) Quantification of frequency of stress granules present in cells after brief SA treatment, with and without WT ANG.WT ANG was added 30 min prior to and during a 1-h SA treatment where indicated.Dashed lines indicate mean 'WT + SA' values of viability and SG frequency for reference.Data compared by ANOVA with Bonferroni's post hoc test.p < 0.05.
Authors.The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.www.pathsoc.orgJ Pathol 2024; 262: 410-426 www.thejournalofpathology.com See figure legends for more specific details.Where data were skewed, transformation was performed either by Log e x or [Log e (255 À x)], as indicated in the appropriate figure legends.Normality was determined using the D'Agostino-Pearson test.Data were tested directly using ANOVA and Bonferroni's or Dunnett's post hoc test.Statistical analysis was performed using either SPSS 23 or RStudio (https://posit.co/products/open-source/rstudio/,accessed 12 October 2019).