BIN1 HET and KO cerebral organoids show transcriptional alterations associated with neuronal functional properties

Cerebral organoids (COs) faithfully recapitulate fundamental aspects of the three-dimensional organization of the human brain, including the molecular specification of different neural cell types/subtypes and the generation of complex electrical activity patterns [19, 20]. To investigate the potential role of BIN1 in human neural cells, we generated and characterized COs using isogenic BIN1 wild type (WT), heterozygous (HET) and KO hiPSCs (Fig. 1A, Supplementary Fig. 1, Supplementary Table 1). After 6.5 months of culture, COs were composed of all the major neural cell types identified by the expression of MAP2, GFAP and NESTIN and we did not observe any gross difference in size or morphology of COs among the three genotypes (Supplementary Fig. 2A). Western blot analyses confirmed the reduction and absence of BIN1 protein in BIN1 HET and KO COs, respectively (Supplementary Fig. 2B). We then employed single-nucleus RNA sequencing (snRNA-seq) to further characterize individual cell types/subtypes and investigate possible gene expression alterations associated with reduced BIN1 expression. COs (n = 4 from each genotype) were divided into two halves that were independently processed for western blotting or snRNA-seq. We observed similar expression of general neuronal and glial proteins in these COs (Supplementary Fig. 2C), suggesting a low degree of heterogeneity in these samples. Nevertheless, to further reduce potential batch effects, we pooled COs into a single multiplexed library using Cell Hashing [21]. After sequencing, quality control and demultiplexing, we recovered 4398 singlets that could be grouped into 7 major cell clusters based on the expression of cell type markers SLC1A3 (GLAST), GFAP and TNC (astrocyte); SNAP25, DCX and MAPT (pan-neuronal); SLC17A7 and SLC17A6 (glutamatergic neurons); DLX1, GAD1 and GAD2 (GABAergic neurons); HES6, CCND2 and CDK6 (NPCs); ITGA8 (choroid plexus); and CLIC6 (pigmented epithelium) (Fig. 1B, C). BIN1 expression in COs was mainly detected in glutamatergic neurons and oligodendrocytes (Fig. 1C), similar to the profile described for the human brain [7]—except for brain microglial cells that are not present in COs. We also observed a reduction and enlargement, respectively, in the proportions of glutamatergic neurons and astrocytes in BIN1 KO compared to WT (Fig. 1D; ****p < 0.0001; Chi-square test).

Fig. 1: Similar transcriptional changes in glutamatergic neurons from 6.5-month-old BIN1 HET and KO COs.

A Scheme of the experimental design. B UMAP representation of the different cell subtypes in COs identified using snRNA-seq (n = 4 COs per genotype). C Dot plot representing the expression for BIN1 and key markers used to annotate cell subtypes. D Proportion of cell subpopulations in the three genotypes (****p < 0.0001; Chi-squared test). E Volcano plots representing DEGs comparing HET vs WT or KO vs WT in astrocytes and glutamatergic neurons. DEGs with adjusted p value < 0.05 and |log2FC | >0.25 are shown in red. F Functional enrichment analysis of DEGs commonly identified in HET and KO glutamatergic neurons. GO gene ontology, BP biological processes, CC cellular components.

To identify possible differentially expressed genes (DEGs) in BIN1 KO or HET compared to WT cells, we performed a Wilcoxon test for each major cell type/subtype identified in COs. Consistent with the predominant expression of BIN1 in glutamatergic neurons (Fig. 1D), we identified a high number of DEGs in this cell type both in BIN1 HET (76 genes) and KO (124 genes) compared to WT genotype (Figure1E; Supplementary Fig. 3A). Furthermore, we found that BIN1 HET and KO glutamatergic neurons shared a significant proportion of DEGs and that correlation between gene expression changes in these two independent clones was highly significant (Supplementary Fig. 3B–D). In astrocytes, we also detected 75 DEGs in BIN1 KO, but only 6 DEGs in HET compared to WT (Fig. 1E; Supplementary Fig. 3A). For all other cell types, we observed a maximum of 1–4 DEGs in the comparison between BIN1 KO vs WT or HET vs WT (Supplementary Table 2). These observations suggest that both BIN1 null (KO) and partial deletion (HET) affect similar biological processes in glutamatergic neurons in a dose-dependent manner. Accordingly, similar GO terms were enriched for DEGs identified in BIN1 KO or HET glutamatergic neurons, including several terms associated with synaptic transmission (Fig. 1F). For BIN1 KO glutamatergic neurons, we also identified GO terms associated with ion channel complex and calcium ion binding (Supplementary Fig. 3E; Supplementary Table 3), further suggesting that reduced BIN1 expression leads to specific transcriptional changes associated with functional properties of this neuronal subtype. DEGs identified in astrocytes of BIN1 KO compared to WT COs significantly enriched for terms associated with nervous system development and cell migration (Supplementary Table 3).

Altered expression of activity-related genes in BIN1 KO and HET COs

Neuronal firing patterns (such as tonic and burst firing) play a key role in the transcriptional regulation of a particular set of genes designated activity-related genes (ARGs) [22]. While neurons stimulated with brief patterns of electrical activity transcribe rapid primary response genes (rPRGs) or early response genes (ERGs), those stimulated with sustained patterns of electrical activity express delayed primary response genes (dPRGs), secondary response genes (SRGs) or late response genes (LRGs) (Fig. 2A) [23, 24]. Using Cell-ID [25], we quantified the enrichment for ARGs signatures (Supplementary Table 4) in COs at single-cell resolution as an indirect readout of neuronal electrical activity patterns in this model. We first confirmed that ARG signatures were predominantly enriched in neurons (Fig. 2B). Then, we quantified the proportion of glutamatergic or GABAergic neurons significantly enriched for such specific responses in gene signatures (padj < 0.05; hypergeometric test). We observed a significantly higher proportion of glutamatergic neurons enriched for dPRGs and LRGs both in BIN1 HET and KO, as well as SRGs in BIN1 KO compared to WT glutamatergic neurons, whereas the proportion of glutamatergic neurons enriched for rPRGs and ERGs was reduced in BIN1 HET and KO (Fig. 2C). In sharp contrast, the only difference observed in GABAergic neurons was a reduction in the proportion of cells enriched for SRGs (Supplementary Fig. 4). These results suggest that reduced BIN1 expression in glutamatergic neurons triggers neuronal firing patterns towards sustained activity leading to a higher expression of late-response ARGs.

Fig. 2: Altered expression of activity-related genes and reduced synaptic numbers in BIN1 HET and KO glutamatergic neurons.

A Scheme indicating the different sets of ARGs regulated by brief and sustained patterns of electrical activity [23, 24]. rPRGs: rapid primary response genes; dPRGs: delayed primary response genes; SRGs: secondary response genes; ERGs early response genes, LRGs late response genes, Exc glutamatergic neurons, Inh GABAergic neurons. B Feature plots showing the enrichment score of single cells for ARG signatures. Enrichment scores correspond to the –log10(padj) of the Cell-ID-based enrichment test. C Proportions of glutamatergic neurons enriched for the different ARG signatures according to genotype (*p < 0.05; **p < 0.01; ***p < 0.001; Chi-squared test). D Immunohistochemistry for HOMER1 (red), SYP (green) in 6.5-month-old BIN1 WT, HET and KO COs. E–G Quantifications of the SYP and HOMER1 spot density, and the percentage of SYP assigned by HOMER1 spots in BIN1 WT, HET and KO COs (#p < 0.1; *p < 0.05; **p < 0.01; Dunn’s multiple comparison test; n = 3 COs per genotype).

Lower numbers of synaptic puncta in BIN1 HET and KO COs compared to WT

Transcriptional alterations in BIN1 HET and KO glutamatergic neurons are also suggestive of synaptic dysfunction, which is an early hallmark of AD pathology [26]. We thus sought to determine whether reduced BIN1 expression could affect synaptic connectivity in COs. Using immunohistochemistry to detect the expression of the pre-synaptic protein Synaptohysin-1 (SYP) and post-synaptic protein HOMER1, we were able to quantify the frequency of putative synaptic contacts (% SYP assigned) in COs (see methods). We observed a significant reduction in the percentage of SYP assigned both in BIN1 HET and KO compared to WT, mainly due to a reduction in the number of post-synaptic spots expressing HOMER1 (Fig. 2D–G).

Next, we investigated whether APP processing and Tau phosphorylation, which have been previously associated with BIN1 and are known to modulate neuronal electrical activity [10, 15, 27], could also be altered in BIN1 HET and KO COs. To this end, we measured the intracellular levels of full-length APP and APP β-CTF (as a readout of amyloidogenic APP processing), total and phosphorylated TAU proteins by western blotting. Besides a trend for reduced TAU expression, likely explained by the reduced proportion of neurons (Fig. 1), we did not detect any significant differences in the intracellular levels of APP, APP β-CTF, TAU or phospho-TAU (Ser202, Thr205) in BIN1 HET and KO compared to WT COs (Supplementary Fig. 5). Altogether, these results suggest that reduced BIN1 expression could alter neuronal functional properties without significantly affecting APP or Tau metabolism in COs.

Cell-autonomous role of BIN1 in the regulation of neuronal gene expression

Our results in COs suggest that reduced BIN1 expression affects mainly glutamatergic neurons. However, at least in BIN1 KO COs, we cannot completely rule out an effect of BIN1 deletion in astrocytes that could indirectly impact glutamatergic neurons. Therefore, to unambiguously probe the cell-autonomous effect of BIN1 deletion on the electrical activity and gene expression of human glutamatergic neurons, we generated BIN1 WT or KO pure neuronal cultures by direct lineage-reprogramming of hiPSC-derived neural progenitor cells NPCs (hiNPCs) using doxycycline-inducible expression of ASCL1 (Fig. 3A; see Supplementary Methods). After validation of the efficient lineage-reprogramming of hiNPCs into highly pure neurons (hereafter ASCL1-hiNs) (Fig. 3B), we added exogenous human cerebral cortex astrocytes to trophically support functional neuronal maturation and synaptic connectivity [28]. Using snRNA-seq after 4 weeks of differentiation we identified 5583 cells (n = 3 independent culture batches) clustered into two main glutamatergic neuron (GluNeu-I and II), one GABAergic neuron (GABANeu), one immature/unspecified neuron (UnspNeu), two astrocytes (Astro-I and II) and one proliferative NPC groups (Fig. 3C, D). Sample-level differential gene expression analysis using DESeq2 [29], revealed 99 DEGs ( | log2FC | >0.25 and FDR < 0.05) in BIN1 KO GluNeu-II compared to WT, but only two in GluNeu-I and one in immature neurons (Fig. 3E; Supplementary Table 5, Supplementary Fig. 6). As observed in COs (Fig. 1H–I), GO term enrichment analysis revealed a significant enrichment for terms associated with synaptic transmission, ion channel activity and calcium signaling pathways (Fig. 3F; Supplementary Fig. 6; Supplementary Table 6). The percentage of GluNeu-II enriched for late-response ARGs was slightly greater in BIN1 KO compared to WT, but without statistical differences (Fig. 3G, H). Exogenously added human astrocytes co-cultured with BIN1 WT and KO hiNs also showed a low number of DEGs (11 in Astro-I; Supplementary Table 5), likely reflecting an astrocyte reaction to primary changes in hiNs in response to BIN1 deletion.

Fig. 3: Neuronal-specific BIN1 KO cell-autonomously elicit transcriptional changes.

A Scheme of the Ascl1-induced hiNs experiments. B Images showing BIN1 WT and KO hiNs 7 days after the beginning of doxycycline treatment immunolabeled for neuronal markers MAP2 and TUBB3 and astrocyte marker GFAP and stained with DAPI. Scale bar: 50 μm. C UMAP representation of the different cell subtypes identified in ASCL1-hiNs cultures using snRNA-seq (n = 3 independent culture batches). D Dot plot representing expression of key markers used to annotate cell subtypes. E Volcano plot representing DEGs comparing BIN1 KO vs WT glutamatergic neurons (GluNeu-II). Differential expression analysis was performed using DEseq2 on the sample level gene expression matrix, including the experiment batch as covariates. DEGs with adjusted p value < 0.05 and |log2FC | >0.25 are shown in red. Gene labels are shown for calcium- and synapse-related genes. F Functional enrichment analysis of DEGs identified in glutamatergic neurons. Bar plots representing the top 15 the enriched GO terms in each category at adjusted p value < 0.01. G UMAP representing GluNeu-II enrichment for dPRGs and SRGs (Tyssowski et al., 2018). Red points indicate enrichment with adjusted p value < 0.05, hypergeometric test using CellID. H Dot plots showing percentage of GluNeu-II cells enriched for dPRGs and SRGsby genotype in 3 independent cultures.

BIN1 KO leads to alteration in the electrical activity pattern of ASCL1-hiNs

The transcriptional changes observed in our 2D and 3D models could suggest a role of BIN1 in regulating functional properties of glutamatergic neurons. To directly address this possibility, we used multi-electrode arrays (MEA) to record and quantify multi-unit activity (MUA) in ASCL1-hiNs. As previously described in spontaneously differentiated hiPSC-derived neuronal cultures [30], ASCL1-hiNs cells exhibited a diverse range of spontaneous activity patterns, including regular discharges, population bursts and period activity (Supplementary Fig. 7A). In this respect, we found a conspicuous change in the temporal organization of MUA after BIN1 deletion (Supplementary Fig. 7A), mainly characterized by a greater number of spike bursts at 4 weeks (Supplementary Fig. 7D). These alterations may result from changes at the single cell or the population level (different number of neurons contributing to each electrode, for example). To disentangle these possibilities, we used waveform-based spike sorting to examine the functional consequences of BIN1 deletion at the single neuronal level (Fig. 4A). We identified a similar number of single units per recording electrode between genotypes (WT: 4.92 ± 2.34; KO: 5.27 ± 2.45), indicating that BIN1 deletion does not affect the density of active neurons within culture. However, we observed reduced single-unit activity (SUA) frequency (Fig. 4B) and higher SUA amplitude (Fig. 4C) in BIN1 KO compared to WT ASCL1-hiNs. Interestingly, we could not detect significant changes in the number of bursts per neuron (WT: 11.01 ± 6.71; KO: 10.36 ± 8.59), although both the burst duration and the number of spikes within a burst were significantly lower in BIN1 KO compared to WT ASCL1-hiNs (Fig. 4D, E). We also observed a prominent temporal disorganization of BIN1 KO hiNs activity by computing the array-wide spike detection rate (ASDR, Fig. 4G), which reveals the strength of the synchronized population activity, and the autocorrelograms of SUAs (Fig. 4H, I), which allows the apprehension of synchronized periodicity. These analyses revealed that most spikes of BIN1 WT neurons were organized in bursts occurring at periodic intervals of about 8-10 s, whereas the spikes of BIN1 KO neurons were randomly distributed, leading to a higher percentage of spikes occurring outside of bursts compared to WT neurons (Fig. 4J).

Fig. 4: Disorganization of neuronal activity in BIN1 KO ASCL1-hiNs.

A Raster plots showing the decomposition of multi-unity activity (MUA, black lines) into single-unit activity (SUA, colored lines) using spike waveform clustering. Quantification of single-neuron firing rate (B; **p = 0.0034), spike amplitude (C; *p = 0.0106), burst duration (D; ****p < 0.0001) and number of spikes per burst (E; ****p < 0.0001) at 4 weeks (Mann–Whitney test; n = 5 independent experiments; WT: 376 neurons; KO: 416 neurons). F Raster plots showing SUA recorded from 5 different electrodes of BIN1 WT (left) or KO (right) ASCL1-hiNs cultures after 4 weeks of differentiation. G Array-wide spike detection rate (ASDR) plots based on SUA recorded in BIN1 WT and KO ASCL1-hiNs cultures. Each line represents one independent culture batch. Normalized autocorrelogram heatmap (H, each line refers to one SUA) and averaged correlation (I) for all SUAs recorded in 5 independent BIN1 WT and KO ASCL1-hiNs cultures. J Percentage of spikes outside of bursts (*p = 0.0417, Mann–Whitney test).

Altered electrical activity of BIN1 KO ASCL1-hiNs is associated with normal synapse numbers and altered TAU phosphorylation

The changes in neural network activity observed in BIN1 KO hiNs could be explained by, among other things, a reduced synaptic connectivity, as observed in our long-term COs cultures (Fig. 2). To test this possibility, we first quantified the number of synaptic contacts in BIN1 WT and KO ASCL1-hiNs cultures. In contrast with COs, we did not detect any significant differences in the number of putative synaptic contacts (% SYP assigned) in BIN1 KO compared to WT ASCL1-hiNs, neither after 4 nor 6 weeks of differentiation (Fig. 5A–D). Next, we quantified the number and activity of glutamatergic synapses by using real-time imaging of ASCL1-hiNs expressing the glutamate sensor iGLUSnFr [31]. In accordance with our observations based on immunocytochemistry, we did not detect differences neither in the number of glutamatergic synapses (active spots) nor in the frequency of events (change in fluorescence levels in active spots) in BIN1 KO compared to WT ASCL1-hiNs (Supplementary Fig. 8; Supplementary Movies 1 and 2), indicating that changes in neuronal activity observed in our cultures are not related to changes in synaptic transmission.

Fig. 5: Similar synaptic density and higher TAU phosphorylation in BIN1 KO compared to WT ASCL1-hiNs.

A, B Immunocytochemistry using the astrocyte marker GFAP, neuronal marker MAP2, pre-synaptic marker SYP and post-synaptic marker HOMER1 in BIN1 WT ASCL1-hiNs after 4 weeks of differentiation in a three-chamber microfluidic device. Scale bar = 200 μm. Rectangular box in (A) is magnified in (B), allowing the identification of putative synaptic contacts. Fraction of SYP spots assigned by HOMER1 spots in MAP2 processes at 4 (C) and 6 weeks (D) ASCL1-hiNs cultures (n = 8 microfluidic devices per genotype from 2 independent differentiations). E Western blot for total TAU protein C-terminal (TAU-C), phosphorylated (p)-TAU at Ser202, Thr205 (AT8) and β-ACTIN in 4-week-old ASCL1-hiNs cultures. F Quantification of TAU-C/β-ACTIN, p-TAU/ β-ACTIN and p-TAU/TAU-C levels in BIN1 KO ASCL1-hiNs normalized to WT (*p = 0.0379 for p-TAU/ β-ACTIN and p = 0.0262 for p-TAU/TAU-C; Mann–Whitney test; n = 7 independent culture batches). Uncropped plots are shown in Supplementary Fig. 12.

Taking advantage of this culture system comprising enriched neuronal populations, we also sought to confirm whether BIN1 deletion could be associated with changes in APP processing or Tau phosphorylation. To this end, we measured the extracellular levels of amyloid-beta (Aβ) peptides, as well as the intracellular levels of full-length APP and APP β-CTF, total and phosphorylated TAU proteins in ASCL1-hiNs cultures. Like COs, we did not detect any significant differences neither in the extracellular levels of Aβ1-x or Aβ1-42, nor in the intracellular levels of APP or APP β-CTF in BIN1 KO compared to WT ASCL1-hiNs (Supplementary Fig. 9). However, in contrast with our observations in COs, we observed significantly higher levels of phospho-TAU (Ser202, Thr205) relative to β-ACTIN and total TAU in in BIN1 KO compared to WT ASCL1-hiNs (Fig. 5E, F). Together, these observations may suggest that BIN1 deletion primarily impairs neuronal intrinsic properties regulating electrical activity and Tau phosphorylation prior to detectable changes in synaptic numbers (observed only in long-term cultures – Fig. 2) and independently of alterations in APP processing.

BIN1 regulates neuronal Ca2+ dynamics through Cav1.2

In neurons, electrical activity is always accompanied by an influx of Ca2+ ions, which play a fundamental role in the regulation of neuronal firing and activity-dependent gene transcription [32]. We therefore postulated that reduced BIN1 expression in human glutamatergic neurons could primarily affect Ca2+ dynamics, as previously suggested for cardiomyocytes [33]. To directly test this possibility, we first studied Ca2+ dynamics in BIN1 WT and KO ASCL1-hiNs using real-time calcium imaging experiments. We observed spontaneous, synchronous calcium transients among adjacent cells both in BIN1 WT and KO ASCL1-hiNs cultures (Supplementary Movies 3 and 4). By quantifying calcium spike transients (>2 standard deviations above the noise level), we showed a significantly higher frequency of Ca2+ transients in BIN1 KO compared to WT ASCL1-hiNs (Fig. 6A, B, D). Moreover, the dynamics of individual Ca2+ transients in BIN1 KO were qualitatively different from WT ASCL1-hiNs (Fig. 6C). These differences could be quantitatively measured by a longer time to reach the maximum intracellular Ca2+ levels and to recover baseline levels (Fig. 6E, F).

Fig. 6: Altered frequency of calcium transients in BIN1 KO ASCL1-hiNs.

A Snapshot of a 4-week-old ASCL1-hiNs culture labeled with Oregon green BAPTA. B Representative plot of fluorescence changes over time in 1000 frames. C Representative traces showing the fluorescence changes in BIN1 WT and KO ASCL1-hiNs. Red dashed lines indicate the time to reach the fluorescence maximal intensity (raising time - t1) and to return to baseline (recovery time - t2). D Quantification of calcium transients in BIN1 WT and KO ASCL1-hiNs (****p < 0.0001; Mann–Whitney test; n = 3 independent cultures for each genotype; number of active cells per condition: 754 (WT), 1006 (KO)). E, F Quantification of rising time (t1) and recovery time (t2) for calcium transients (**p = 0.0022; ****p < 0.0001; Mann–Whitney test). G Images showing PLA spots using anti-BIN1 and anti-Cav1.2 antibodies in 4-week-old BIN1 WT and KO hiNs. Cells were also immunolabeled for the neuronal marker MAP2 (green), the astrocyte marker GFAP (white), and stained with DAPI (blue). H Western blot for Cav1.2 (without and with blocking peptide) and β-ACTIN in 4-week-old ASCL1-hiNs cultures. I Quantification of Cav1.2/β-ACTIN levels in BIN1 WT and KO ASCL1-hiNs cultures (*p = 0.0286; Mann–Whitney test; n = 4 independent culture batches). J Western blot for Cav1.2 and β-ACTIN in the total protein extracts from 4-week-old ASCL1-hiNs treated with KCl (+) or vehicle (–). Plot shows the quantification of Cav1.2 normalized by β-ACTIN. K Western blot for Cav1.2, Cav1.3 and EEA1 in the endosomal protein extracts from 4-week-old ASCL1-hiNs treated with KCl (+) or vehicle (–). L Plot shows the optical density of these proteins (****p < 0.0001; Chi-square test; n = 6 independent cultures for each genotype/treatment pooled). Uncropped plots are shown in Supplementary Figs. 13, 14. M Auto-correlograms of 4-week-old BIN1 WT and KO hiNs treated or not with 50 nM Nifedipine (NIF) for 2 weeks. N Percentage of spikes outside of bursts (WT or WT + NIF vs KO or KO + NIF: ****padj < 0.0001; KO vs KO + NIF: *padj = 0.0124; Dunn’s multiple comparison test; n = 3 independent culture batches). O Average firing rates. P Comparison of the enrichment for GO terms in genes upregulated in KO vs WT and KO + nifedipine (NIF) vs WT GluNeu-II using fast gene set enrichment analysis (FGSEA). The FGSEA results are shown for the top15 GO terms with reduced enrichment in BIN1 KO ASCL1-hiNs treated with nifedipine (NIF) or vehicle (Control) compared to WT cells.

In human heart failure, BIN1 expression is reduced, leading to an impairment in Cav1.2 trafficking, calcium transients, and contractility [33]. Thus, we sought to determine if BIN1 could also interact and regulate LVGCC expression in human neurons. To this end, we performed proximity ligation assay (PLA) to probe a possible interaction between BIN1 and Cav1.2 or Cav1.3, the two LVGCCs expressed in ASCL1-hiNs (Supplementary Fig. 10). We observed a widespread BIN1-Cav1.2 PLA signal (Fig. 6G) and, to a lesser extent, a BIN1-Cav1.3 PLA signal in neurons (Supplementary Fig. 10). Next, we quantified neuronal LVGCC protein by western blotting and observed higher total Cav1.2 expression in BIN1 KO compared to WT ASCL1-hiNs (Fig. 6H–I). Protein expression of neither Cav1.3, nor the members of the Cav2 family (Cav2.1, Cav2.2 and Cav2.3) were altered in the same cultures (Supplementary Fig. 10), suggesting a specific regulation of Cav1.2 expression by BIN1.

Notably, LVGCCs are key regulators of neuronal firing [30] and activity-dependent internalization of these channels is a key mechanism in firing homeostasis [34]. We thus set out to investigate whether BIN1 deletion could impair this mechanism in human neurons. We stimulated ASCL1-hiNs with KCl 65 nM for 30 min to induce neuronal depolarization and collected total and endosomal proteins for analysis. We confirmed a higher global level of Cav1.2 in BIN1 KO compared to WT ASCL1-hiNs that was independent of KCl treatment (Fig. 6J). We also confirmed that the treatment with KCl 65 nM for 30 min was sufficient to increase the endosomal fraction expressing the early endosome antigen 1 (EEA1) and the internalization of Cav1.3 in both BIN1 WT and KO ASCL1-hiNs (Fig. 6K–L). However, while Cav1.2 expression in the endosomal fraction was 50% higher after KCl treatment in BIN1 WT, this rise was only 10% in BIN1 KO ASCL1-hiNs (Fig. 6K–L). These results indicate that BIN1 interacts and regulates the activity-dependent internalization of Cav1.2 in human neurons.

Treatment with the calcium channel blocker nifedipine partly rescues electrical and gene expression alterations in BIN1 KO ASCL1-hiNs

To investigate whether the network dysfunctions observed in BIN1 KO ASCL1-hiNs could be related to the altered expression of Cav1.2 in these cells, we treated cultures with a physiologically relevant concentration (50 nM) of the Cav1.2 blocker nifedipine [35] for 2 weeks and recorded neuronal activity using MEA electrophysiology. We observed a partial recovery of the oscillatory pattern of neuronal electrical activity observed in WT cells (Fig. 6M). Interestingly, the percentage of spikes outside bursts was not affected by nifedipine treatment in BIN1 WT but was significantly lower in BIN1 KO ASCL1-hiNs (Fig. 6N), indicating a partial recovery of burst organization. To note, no difference in firing rates was observed whatever the models and conditions (Fig. 6O). After 2 weeks of nifedipine treatment (4 weeks of differentiation), we also performed snRNA-seq experiments and recovered a total of 1537 cells (n = 2 independent culture batches), which were mapped into the 7 clusters described earlier (Fig. 3; Supplementary Fig. 11). Using the Wilcoxon test, we found that nifedipine treatment down-regulated several genes in BIN1 KO ASCL1-hiNs, especially in the GluNeu-II population (Supplementary Table 7). The downregulated genes after nifedipine treatment were enriched for LRGs of excitatory neurons, suggesting rescue of the altered activity-related gene expression (Supplementary Fig. 11D). Furthermore, several GO terms associated with ion channel activity and synapse transmission, which are found enriched in BIN1 KO vs WT GluNeu-II population, show consistently reduced enrichment in nifedipine treated KO cells (Fig. 6P; Supplementary Table 8). Altogether, these data support the view that BIN1 contributes to the regulation of electrical activity and gene expression through the regulation of Cav1.2 expression/localization in human neurons.

Molecular alterations in BIN1 KO organoids and ASCL1-hiNs are also present in glutamatergic neurons of AD patients

We finally sought to evaluate whether molecular alterations in our neural models may recapitulate some of those observed in the brains of AD cases. For this purpose, we used a publicly available snRNA-seq dataset generated from the entorhinal cortex (EC) and superior frontal gyrus (SFG) of AD patients at different Braak stages [36]. We first observed a progressive and significant decrease in BIN1 mRNA levels in glutamatergic neurons of both brain regions (Fig. 7A), suggesting that reduced BIN1 expression in this cell type may be a common feature occurring in the AD pathology progression. We then compared DEGs identified in BIN1 KO glutamatergic neurons (either from COs or ASCL1-hiNs) with those identified in the same cell subtype of AD brains (Supplementary Table 9). Remarkably, DEGs identified in BIN1 KO glutamatergic neurons (either from COs or ASCL1-hiNs) showed a statistically significant overlap with DEGs detected in this cell population of AD brains at different Braak stages (Fig. 7B). In astrocytes, however, a similar significant overlap could only be observed between COs and AD brains. GO analysis based on DEG overlap between BIN1 KO ASCL1-hiNs and AD brain glutamatergic neurons indicated significant enrichment for pathways associated with glutamate receptor activity and gated channel activity (Fig. 7C; Supplementary Table 8). Similarly, DEG overlap between BIN1 KO COs and AD brain glutamatergic neurons was significantly enriched for genes associated with glutamate receptor activity, gated channel activity and calcium ion binding (Fig. 7D; Supplementary Table 8). No significant enrichment was observed for DEG overlap between BIN1 KO COs and AD brain astrocytes (data not shown). Therefore, gene expression alterations in BIN1 KO hiNs and glutamatergic neurons of AD patients overlap and converge to biological processes associated with calcium homeostasis.