Miniature brain models with mutated actin genes show how cells cause small brain size

To broaden our view of the cortical malformations found in BWCFF-S patients, we reviewed the literature on the neurological manifestations in patients carrying pathogenic variants (or likely pathogenic variants) of ACTB and ACTG1 genes (Riviere et al, 2012; Verloes et al, 2015; Di Donato et al, 2016; Di Donato et al, 2014; Accogli et al, 2020; Eker et al, 2014; Rossi et al, 2003). We combined this literature review with an analysis of brain imaging data of 29 not previously reported patients carrying pathogenic variants of ACTB (20 patients) and ACTG1 (nine patients). We confirmed the previous observation (Verloes et al, 2015) that malformations of cortical development were present in almost 70% of BWCFF-S patients. Figure 1 shows representative brain MRI images illustrating malformations typically associated with the BWCFF-S in comparison to the images of a healthy individual (Fig. 1A–C). Two of the patients (Fig. 1D–F,J–L) each carry a point mutation in ACTB which results in a single amino acid substitution, of which the BWCFF-S patient with the actin variant NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile (Fig. 1D–F) will serve as source for one set of the cerebral organoids studied here (see below). The other two patients (Fig. 1G–I,M–O) each carry a point mutation in ACTG1 which also results in a single amino acid substitution, of which the BWCFF-S patient with the actin variant NM_001614.5↗ (ACTG1):c.608 C > T p.Thr203Met (Fig. 1G–I) will serve as source for another set of the cerebral organoids studied here (see below).

The patient with the ACTB variant p.Thr120Ille was reported previously (Di Donato et al, 2014) and had microcephaly with a head circumference of –2.5 SD and severe intellectual disability with an IQ < 40. Briefly, the brain malformations in this patient included frontal predominant pachygyria (Fig. 1E) and periventricular nodular heterotopia located bilaterally in the occipital horns of the lateral ventricles (Fig. 1F). The patient carrying the variant ACTG1:p.Thr203Met was non-ambulant, showed severe microcephaly of –5.2 SD, and had no speech development with an IQ < 40. Brain MRI demonstrated frontotemporal pachygyria (Fig. 1H,I) accompanied by agenesis of the corpus callosum and cerebellar hypoplasia (Fig. 1G). Figure 1J–L shows a brain MRI from a second patient carrying a pathogenic ACTB variant, NM_001101.5↗ (ACTB):c.571 A > G, p.Lys191Glu. This patient had microcephaly of –3.5 SD, and the brain MRI showed bilateral anterior-predominant pachygyria, with very prominent perivascular spaces (Fig. 1K). A second patient carrying a pathogenic ACTG1 variant is presented in Fig. 1M–O. This patient was previously reported (Accogli et al, 2020) and carries the mutation NM_001614.5↗ (ACTG1):c.767 G > A, p.Arg256Gln, located at a mutational hotspot. Briefly, the brain MRI images of this patient demonstrated partial agenesis of the corpus callosum (Fig. 1M) and a very specific pattern of cortical malformations consistent with frontal predominant pachygyria associated with a thin posterior band heterotopia (Fig. 1N,O).

Prior to generating cerebral organoids from BWCFF-S patient-derived iPSCs, we first determined whether the expression patterns of βCYA and γCYA in human cerebral organoids generated from control iPSCs would be similar, in principle, to those in fetal human neocortex, which we considered to be a prerequisite for using cerebral organoids as a model for BWCFF-S. As shown by immunohistochemistry in Appendix Figs. S2 and S3 for 12 and 16 post conception weeks (pcw) fetal human neocortex tissue, respectively, both βCYA and γCYA showed a widespread distribution across the developing cortical wall, from the VZ all the way to the marginal zone, with the peak of immunoreactivity in the apical-most region of the VZ, which is known to harbor the apical adherens junction (AJ) belt.

To investigate whether cerebral organoids show similar widespread βCYA and γCYA expression patterns as observed in fetal human neocortex tissue, we generated cerebral organoids from two human control iPSC lines, SC102A and CRTDi011-A. Immunohistochemistry of these organoids at day 30 of culture revealed an, in principle, similar widespread distribution of both βCYA and γCYA across the wall of the ventricle-like structures, from the ventricular surface to the outer surface (Appendix Fig. S4, c1 and c2). These ventricle-like structures are fluid-filled cavities surrounded by layers of cells within a given brain organoid, corresponding to the in vivo brain ventricles. The cellular layers surrounding these structures in cerebral organoids include a VZ and a subventricular zone (SVZ)/neuronal layer (NL), roughly corresponding to the different zones and layers of the developing cortical wall. As in fetal tissue, within the VZ of the organoids, βCYA and γCYA immunoreactivity were concentrated towards the apical surface.

Given this widespread distribution of both βCYA and γCYA in the control cerebral organoids, we checked the expression of canonical markers in these organoids under the present conditions, again using immunohistochemistry. Consistent with previous reports (Kadoshima et al, 2013; Lancaster et al, 2013; Lancaster and Knoblich, 2014), the control cerebral organoids after 30 days in culture were found to express markers of proliferating progenitors (SOX2), neurogenic basal progenitors (TBR2), and newborn neurons (Tuj1) including deep-layer neurons (Ctip2) (Appendix Figs. S5–S7, c1 and c2). Immunostaining for nestin revealed the fibers of radially organized progenitors in the VZ, with densely packed nuclei as seen by DAPI staining or SOX2 immunostaining (Appendix Fig. S5, c1 and c2).

Next, we sought to model BWCFF-S in cerebral organoids. To this end, we collected fibroblasts from two BWCFF-S patients, carrying either the actin variant NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile (Fig. 1D–F) or the actin variant NM_001614.5↗ (ACTG1):c.608 C > T p.Thr203Met (Fig. 1G–I), and reprogrammed them into iPSCs. The two sets of iPSCs exhibited normal karyotype and expressed typical pluripotency markers (Appendix Figs. S8 and S9). We generated cerebral organoids from two clones of each patient-derived iPSC line.

Next, we explored whether cerebral organoids generated from the BWCFF-S patient-derived iPSCs showed a similar widespread βCYA and γCYA distribution and canonical marker expression as the control organoids and hence could be suitable models to study the disrupted early cortical development characteristic of such patients. Indeed, immunohistochemistry at day 30 of culture revealed that both types of organoids, expressing either the ACTB variant or the ACTG1 variant, showed a similar widespread distribution of both βCYA and γCYA across the various zones of the wall of the ventricle-like structures (Appendix Fig. S4). For the ACTG1 variant-expressing organoids, a concentration of βCYA and γCYA immunoreactivity at the apical surface of the VZ could be observed. Moreover, using western blot analysis, we found that both βCYA and γCYA have a similar relative abundance in BWCFF-S compared to control cerebral organoids (Appendix Fig. S10A,B). Also, phalloidin staining showed no apparent effects on F-actin in the BCWFF-S condition compared to control (Appendix Fig. S10C). Like control organoids, both the ACTB or ACTG1 variant-expressing organoids showed SOX2, TBR2, Tuj1, NeuN, Ctip2, and nestin (Appendix Figs. S5–S7) immunoreactivity in the typical locations. We conclude that the cerebral organoids generated from the BWCFF-S patient-derived iPSCs can be suitable models to study the microcephaly observed in these patients.

In the above-described immunostainings, we had noticed that the ventricle-like structures in the BWCFF-S organoids appeared to be smaller than those in the control organoids. We therefore decided to quantitatively compare various size parameters of the BWCFF-S organoids vs. control organoids. As the first step in the analyses, we quantified the entire size of the cerebral organoids. At 30 days of cerebral organoid culture, the two types of BWCFF-S organoids were significantly smaller compared to control organoids (Fig. 2A,C). Moreover, at 50 days of organoid culture, a similar size difference was observed (Fig. 2B,D), suggesting that both the control and the two types of BWCFF-S cerebral organoids were growing at a similar speed between day 30 and day 50 of culture. These data are in line with the microcephaly observed in the two patients and in at least half of the whole BWCFF-S cohort (Verloes et al, 2015), and indicate that BWCFF-S cerebral organoids can recapitulate features of the patients’ phenotype.

To better understand the size reduction observed in BWCFF-S cerebral organoids, we performed single-cell RNA sequencing (scRNA-seq) on patient-derived organoids and control organoids at 30 and 50 days of culture. In addition, we analyzed cerebral organoids derived from control CRTDi011-A iPSCs, in which we introduced the ACTB Thr120Ile mutation using CRISPR/Cas9. In the following, we first focus on the patient-derived BWCFF-S cerebral organoids before turning to the CRISPR/Cas9-edited organoids later in the results.

Our scRNA-seq analysis identified 17 distinct clusters, representing different progenitor and neuronal populations but also mesenchymal-like, neural crest, and neuroectodermal-like cells expressing characteristic markers (Figs. 3A and EV1A,B; Appendix Fig. 11). Moreover, progenitor and neuronal populations could be distinguished based on their cell cycle phase, with neuronal clusters enriched in G1/G0 phase cells, while progenitor clusters were enriched in S/G2/M phase cells (Fig. 3B). To better understand the cell type composition, we combined clusters of the same cell types leading into six categories: radial glia (RG), basal progenitors (BPs), neurons, neuroectodermal-like cells, neural crest and mesenchymal-like cells (Figs. 3C and EV1C). Using this approach, we found that the percentage of RG relative to total cells was reduced in BWCFF-S cerebral organoids at 30 days of culture compared to controls (Fig. 3C). Moreover, we found an increased percentage of BPs in BWCFF-S ACTB Thr120Ile patient-derived cerebral organoids compared to controls. In BWCFF-S ACTG1 Thr203Met organoids, the BP percentage was comparable to controls. However, it is important to note that neural crest and neuroectodermal-like cells are overrepresented in BWCFF-S ACTG1 Thr203Met organoids (Fig. 3C). To better assess BP output, we calculated the BP/RG ratio and found a strong increase in this parameter in day 30 BWCFF-S cerebral organoids, indicating an elevated BP output (Fig. 3D).

On day 50 of cerebral organoid culture, the relative contribution of RG in both control and BWCFF-S organoids was reduced, while the percentage of neurons increased (Figs. 3C and EV1C), in line with the neuron production between day 30 and 50. Interestingly, while there was a clear reduction in the proportion of neurons in BWCFF-S ACTG1 Thr203Met organoids, BWCFF-S ACTB Thr120Ile patient-derived cerebral organoids showed a comparable proportion of neurons to the control (Fig. EV1C). However, two points need to be considered in this context. First, when examining individual neuronal populations, the relative contribution of dorsal neurons was reduced in both BWCFF-S ACTG1 Thr203Met as well as ACTB Thr120Ile cerebral organoids (Fig. EV1B). Second, due to the strong size reduction of BWCFF-S cerebral organoids (Fig. 2B,D), even if the relative contribution of neurons was similar in control and BWCFF-S organoids (Fig. EV1C), total neuron numbers would still be reduced in BWCFF-S cerebral organoids. In summary, the scRNA-seq data indicate a strong effect of the BWCFF-S mutations on RG, affecting BPs and subsequently neurons.

Due to the strong reduction in the RG proportion in BWCFF-S organoids compared to control in our scRNA-seq data, we next analyzed the ventricle-like structures of these organoids, as they represent the compartment of apical radial glia as the primary RG and VZ progenitor type. These analyses revealed that the size of the ventricle-like structures and the morphology of the VZ in the two types of BWCFF-S organoids differed significantly from the controls (Fig. 4). To better characterize this abnormal morphology, we quantified different parameters of the VZ in the control and the BWCFF-S cerebral organoids at 30 days of organoid culture, as already at this time point, the size difference was obvious. For this purpose, we used DAPI-stained control and BWCFF-S cerebral organoids and first determined the radial thickness of the VZ (Fig. 4A). We found a significantly reduced VZ radial thickness for the two types of BWCFF-S organoids compared to control organoids (Fig. 4B), indicating a thinner VZ likely caused by a reduced cell number. We next quantified the perimeter of the VZ, which also was smaller in the BWCFF-S VZ compared to the VZ of control organoids (Fig. 4C). Moreover, to corroborate the reduction in VZ size in the two types of BWCFF-S organoids compared to control, we measured the total area occupied by the VZ and found a significant reduction of VZ area in the two types of BWCFF-S cerebral organoids compared to control (Fig. 4D). To have an additional measurement which is independent of the placement of the basal boundary of the VZ, we quantified the length of the apical surface and found a significant reduction in the two types of BWCFF-S cerebral organoids compared to control (Fig. 4E). Interestingly, while these general VZ parameters were changed in both BWCFF-S conditions (ACTB and ACTG1), the number of ventricle-like structures was only reduced in the actin variant NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile (Fig. 4F).

Despite the reduction in the above-mentioned VZ parameters, the density of cell nuclei in the VZ was similar for the two types of BWCFF-S organoids as compared to control organoids (Fig. 4G,H). Considering these VZ parameters together, these data strongly suggested that the reduction in the former VZ parameters (Fig. 4A–E) of the BCWFF-S organoids was likely caused by a decrease in VZ progenitors. Indeed, quantification of the DAPI-stained nuclei revealed a substantial decrease of the number of VZ progenitors per ventricle-like structure in the BWCFF-S organoids as compared to control organoids (Fig. 4I). Such a decrease can have various causes, including diminished proliferation of VZ progenitors or their increased apoptotic cell death. However, caspase 3 staining showed that apoptotic cells were rare in both BWCFF-S and control organoids (Fig. EV2).

Quantification of Ki-67– and PH3–positive cells in the VZ revealed no significant differences in the proportion of these cells over total VZ cells between the BWCFF-S and control organoids (Figs. 4J,K and EV3). However, these cycling cell markers do not distinguish between the various modes of cell division in the VZ, that with regard to the two classes of progenitors found in the VZ (see terminology in “Methods”) apply only to APs; these modes of AP division may be either (i) symmetric–proliferative (1 AP – > 2 APs), resulting in an increase in AP and hence VZ progenitor number; (ii) asymmetric–self-renewing (1 AP – > 1 AP + 1 cell that leaves the VZ, typically a newborn BP), resulting in the maintenance of AP number but a transient increase in VZ progenitor number; or (iii) self-consuming (1 AP – > 2 cells that leave the VZ), resulting in a decrease of AP number but a transient increase in VZ progenitor number (Florio and Huttner, 2014; Heide and Huttner, 2023).

We therefore explored whether the mode of AP division was altered in the BWCFF-S organoids as compared to control organoids. Clues in this regard can be obtained from an analysis of the orientation of the metaphase plate and of the cleavage plane of mitotic APs in anaphase. The cleavage plane orientation of mitotic APs normally is predominantly vertical (i.e., parallel to the apical-basal axis of the cell); such a cleavage plane orientation is required for symmetric–proliferative divisions of APs but is also compatible with asymmetric–self-renewing divisions of these progenitors. In contrast, oblique cleavage plane orientation is typically associated with asymmetric–self-renewing divisions, and a horizontal cleavage plane orientation of mitotic APs (i.e., perpendicular to their apical-basal axis) is associated with either asymmetric–self-renewing divisions, or consumptive divisions yielding two non-AP daughters (Huttner and Kosodo, 2005; Mora-Bermudez and Huttner, 2015; Xing et al, 2021). Thus, while cleavage plane orientation does not strictly dictate daughter cell fate in the mammalian cerebral cortex, symmetric proliferative AP divisions require a vertical cleavage plane, as this is the only orientation that equally partitions the apicobasal cytoarchitecture, from the basal lamina contact to the apical endfoot, a hallmark of APs.

We first used high-resolution images of DAPI- and α-tubulin-stained AP metaphase nuclei to determine the mitotic AP metaphase plate orientation (Fig. 5A,B). We categorized AP metaphase plate orientation into three groups: (1) 90–61° as vertical orientation, (2) 60–31° as oblique orientation, and (3) 30–0° as horizontal orientation. We found a switch from predominantly vertical metaphase plate orientation to oblique and horizontal metaphase plate orientation in the mitotic APs of BWCFF-S cerebral organoids in comparison to those of control organoids (Fig. 5A,B). Orientation of the metaphase plate and later of the cleavage plane depends, among other things, on the abundance of astral microtubules (Mora-Bermudez and Huttner, 2015; di Pietro et al, 2016). Astral microtubules reaching the apical and basal cell cortex are more abundant in cells dividing in a vertical cleavage plane orientation than in cells dividing in an oblique or horizontal orientation. A reduced number of these microtubules would be indicative of increased oblique and horizontal cleavage plane orientation (Mora-Bermudez et al, 2014; Vargas-Hurtado et al, 2019; Da Silva et al, 2021). Accordingly, we next tested if the increase in oblique and horizontal metaphase plate orientation in mitotic APs of BWCFF-S cerebral organoids is caused by a reduced number of astral microtubules reaching the apical and basal cell cortex. For this purpose, we quantified the number of apical and basal (Fig. 5C, left) as well as central (Fig. 5C, right) astral microtubules in mitotic APs of control and BWCFF-S cerebral organoids. However, we did not observe any differences in the number of astral microtubules between the two conditions (Fig. 5C).

In light of this result and of the fact that mitotic spindles and metaphase plate can alter their orientation during metaphase (Haydar et al, 2003), we asked if the increase in oblique and horizontal metaphase plate orientation would culminate in a corresponding switch in cleavage plane orientation in anaphase. To analyze cleavage plane orientation, we used high-resolution images of the DAPI- and SOX2-stained AP anaphase nuclei at the ventricular surface (stained for ZO-1), and determined mitotic AP cleavage plane angles relative to the apical surface (Fig. 5D,E). Similar to the analysis of metaphase plate orientation, we categorized AP anaphase cleavage plane angles into three groups: (1) 90–61° as vertical cleavage, (2) 60–31° as oblique cleavage, and (3) 30–0° as horizontal cleavage. We found that in contrast to the predominantly vertical cleavage plane orientation of mitotic APs in control cerebral organoids, mitotic APs of BWCFF-S cerebral organoids exhibited a nearly randomized or majoritarian horizontal cleavage plane orientation (Fig. 5F,G), in line with the results of the analysis of the metaphase plate orientation. Symmetric proliferative AP divisions require a vertical cleavage plane, as this is the only way to equally divide, and distribute to the daughter APs, the full apicobasal cytoarchitecture (see above). Therefore, the change of BWCFF-S AP cleavage planes to a nearly randomized or majoritarian horizontal orientation is expected to prevent most APs from undergoing symmetric proliferative divisions to expand their pool size. Moreover, although both vertical and horizontal cleavage planes can be associated with asymmetric–self-renewing divisions, a shift away from vertical cleavages increases the probability that one or both daughter cells will delaminate from the VZ. In this way, the observed shift in cleavage plane orientation in BWCFF-S APs is likely to contribute to a reduction in the size of the VZ progenitor pool.

Next, we asked whether the shift in cleavage plane orientation in mitotic anaphase APs of BWCFF-S cerebral organoids results in an increased number of delaminated AP progeny cells, which in humans are predominantly BPs. To this end, we analyzed changes in BP numbers using immunofluorescence for the BP marker TBR2 (Appendix Fig. S6) and quantified the number of TBR2-positive BPs in the SVZ/NL. We found a significantly higher number of TBR2-positive BPs in the SVZ/NL of BWCFF-S cerebral organoids compared to control organoids (Fig. 5H). This finding is in line with our scRNA-seq data, which pointed to an elevated BP output (Fig. 3D). Taken together, these data indicate that the VZ progenitor pool in BWCFF-S cerebral organoids is diminished, likely due to the shift in cleavage plane orientation of mitotic anaphase APs, which increases the probability of progenitor delamination and thereby promotes BP generation at the expense of the primary progenitor type, the AP.

As actin is a major component of the cytoskeleton affecting multiple structural and morphological features of the cell, we next sought to analyze VZ progenitors of BWCFF-S organoids in detail for cytoskeletal and morphological changes using transmission electron microscopy. First, we examined the size and shape of VZ progenitors. We found that at the level of the apical adherens junction belt, the size of BWCFF-S VZ progenitors and the shape of their cell circumference appeared to be more variable in comparison to control (Fig. 6A). In addition, protrusions from VZ progenitors into neighboring cells were more frequently observed in BWCFF-S than control organoids (Fig. 6A; Movies EV1–EV6).

As these effects were particularly noticeable at the level of the apical adherens junction belt, we next examined this belt in greater detail. We found that a key feature of the apical adherens junction belt, that is, to connect a cell to its neighbors, appeared to be maintained in VZ progenitors of BWCFF-S cerebral organoids (Fig. 6B,C). In line with this observation, upon immunofluorescence for ZO-1—a tight junction protein that is also associated with adherens junctions—the apical adherens junction belt appeared as a thin line in the ventricle-like structures of control as well as BWCFF-S cerebral organoids. (Fig. EV4A). Any abnormal adherens junction belt morphology was observed to a similar degree in both control and BWCFF-S conditions (Fig. EV4B). However, in the BWCFF-S organoids, we observed a higher density of microtubule bundles at the apical cell cortex that appeared to be anchored to the apical adherens junction belt rather than being nucleated at the basal body of the primary cilium (Fig. 6D).

To further assess this observation, we performed tubulin immunohistochemistry. This indicated an increased concentration of both alpha- and beta-tubulin at the apical cell cortex of VZ progenitors in the BWCFF-S organoids (Fig. EV5). Taken together, using transmission electron microscopy, we identified several reproducible cytoskeletal and morphological irregularities in VZ progenitors of BWCFF-S cerebral organoids, notably in the apical region of these cells. While these findings do not establish a causal relationship, they provide reproducible observations (further images are available as source data in BioStudies) of altered cell morphology that may be relevant in the context of abnormal cleavage plane orientation of BWCFF-S mitotic APs (see “Discussion” for details).

In light of the (unavoidable) facts (see “Methods”) that (i) the reprogramming of the control or BWCFF-S patient-derived fibroblasts had to be carried out at three different institutes, (ii) the culture conditions after reprogramming until the isolation of the iPSC clones were similar but not identical, and (iii) the genetic backgrounds of the healthy donor and the two BWCFF-S patients were (by definition) different, it was essential to demonstrate that the phenotypes observed in the BWCFF-S cerebral organoids as compared to control organoids were not due to these differences but indeed caused by the point mutations in ACTB or ACTG1. To this end, we subjected the control CRTDi011-A iPSCs to CRISPR/Cas9-mediated genome editing to introduce a single point mutation into ACTB that would selectively result in the Thr120Ile change. We then isolated two clones after this genome editing, referred to as CRTDi011-A–mutACTB-1 and CRTDi011-A–mutACTB-2, and used these to generate cerebral organoids, collectively referred to as cr. ACTB Thr120Ile organoids (see Fig. 7A for an overview of 30-day-old control and cr. ACTB Thr120Ile organoids). The two clones exhibited normal karyotype and expressed typical pluripotency markers (Appendix Fig. S12). Analyses in comparison to 30 days-old control CRTDi011-A iPSC–derived organoids of (i) the organoid area (Fig. 7B), (ii) the VZ radial thickness, perimeter, and area (Fig. 7C–E), (iii) the length of the apical surface (Fig. 6F) (iv) the number of VZ cells per ventricle-like structure (Fig. 6G), and (v) anaphase AP cleavage plane angles (Fig. 7H,I), revealed essentially the same phenotypes—with exception of the number of ventricle-like structures (Fig. 7J)—in the 30 days-old cr. ACTB Thr120Ile organoids as observed in the BWCFF-S ACTB Thr120Ile patient-derived organoids (see Figs. 2, 4, and 5). Moreover, scRNA-seq analysis of 30 and 50-day-old cr. ACTB Thr120Ile organoids revealed a cell type composition similar to that of BWCFF-S cerebral organoids (Figs. 3 and EV1; Appendix Fig. S11). These data indicate that the phenotypes observed in the BWCFF-S ACTB Thr120Ile patient-derived cerebral organoids were indeed caused by the point mutation in ACTB.

In conclusion, our results indicate that cerebral organoids expressing BWCFF-S-associated mutant ACTB or ACTG1 genes provide insight into the cellular mechanism underlying the cortical malformations seen in BWCFF-S patients. Thus, BWCFF-S organoids exhibit several cytoskeletal and morphological irregularities in VZ progenitors, notably in the apical region of these cells. Likely reflecting consequences of these irregularities, mitotic BWCFF-S APs show a striking deviation of the orientation of the mitotic spindle, which normally is parallel to the ventricular surface. This in turn is incompatible with AP proliferation increasing AP abundance, and likely results in an increased delamination of cells from the VZ due to the change in the orientation of the cleavage plane of mitotic APs from predominantly vertical to a nearly randomized or majoritarian horizontal orientation. This lack of AP proliferation and increase in cell delamination from the VZ, on the one hand, is associated with a decrease in the size of the VZ within the ventricle-like structures of the BWCFF-S cerebral organoids and, on the other hand, with an increased abundance of BPs in the SVZ/NL of BWCFF-S organoids. Since APs in the VZ constitute the primary progenitor type in the developing brain, a reduction of their pool size due to delamination at a stage when these cells normally still self-amplify or at least self-renew will, in the short run, be consistent with an increased BP abundance. In the long run, such a reduction in the AP pool could limit the source from which BPs are generated, potentially decreasing neuronal output—one mechanism that would be consistent with microcephaly.

After obtaining a positive statement (EK127032017 and EK44012019) from the ethics council of TU Dresden, fibroblasts were isolated, using a previously described procedure (Latham et al, 2018), from skin punch biopsies of a healthy adult female donor (control) and of two female BWCFF-S patients carrying either the NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile or the NM_001614.5↗ (ACTG1):c.608 C > T p.Thr203Met mutation. The presence of both BWCFF-S mutations was confirmed with locus-specific Sanger sequencing as previously described (Di Donato et al, 2014).

The three batches of isolated human fibroblasts were put into culture under identical conditions (in BIO-AMF™-2 medium (Biological Industries)), grown for 2–3 passages, and then subjected to reprogramming to generate iPSCs. Fibroblasts from the BWCFF-S mutant ACTB patient were frozen in liquid nitrogen before reprogramming. Due to differences in the availability of the healthy donor and the two BWCFF-S patients providing the skin punch biopsies, this reprogramming was performed correspondingly at three different locations. The human control fibroblasts were reprogrammed at the CRTD Stem Cell Engineering Facility at the Technische Universität Dresden. The fibroblasts from the BWCFF-S patient carrying the NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile mutation were reprogrammed at the Medizinische Hochschule Hannover. The fibroblasts from the BWCFF-S patient carrying the NM_001614.5↗ (ACTG1):c.608 C > T p.Thr203Met mutation were reprogrammed at the Medizinisch Theoretisches Zentrum Dresden. All reprogrammings were performed using the same type of kit, the CytoTune-iPS 2.0 Sendai Reprogramming Kit (Thermo Fisher Scientific).

Human control fibroblasts were transduced according to the reprogramming kit manufacturer’s instructions. After 5 days following transduction, cells were plated onto irradiated CF1 Mouse Embryonic Fibroblasts (Thermo Fisher Scientific) and cultured in ReproTeSR (StemCell Technologies), After 10 days of culture, individual iPSC colonies were mechanically picked and expanded as clonal lines on Matrigel-coated cell culture dishes (Corning) in mTeSR™1 (StemCell Technologies). IPSC lines were passaged using ReLeSR (StemCell Technologies).

BWCFF-S ACTB Thr120Ile patient-derived fibroblasts were thawed in fibroblast medium I (DMEM (Gibco), supplemented with 10% FCS (PAA), 1% MEM non-essential amino acids (Gibco), and 1 mM L-glutamine (Gibco), cultured for 3 days, and then transduced according to the reprogramming kit manufacturer’s instructions. After 4 days following transduction, cells were plated on mitotically inactivated murine embryonic feeder cells and cultured in iPSC medium (knockout-DMEM (Gibco), supplemented with 20% knockout serum replacement (Gibco), 1 mM L-glutamine (Gibco), 0.1 mM 2-mercaptoethanol (Gibco), 1% MEM non-essential amino acids (Gibco), and 8 ng/ml bFGF-2 (PeproTech)). After 8 days of culture, individual iPSC colonies were mechanically picked and expanded as clonal lines on murine embryonic feeder cells in iPSC medium. IPSC lines were passaged using collagenase IV (Gibco).

BWCFF-S ACTG1 Thr203Met patient-derived fibroblasts were transduced according to the reprogramming kit manufacturer’s instructions. Directly after transduction, cells were cultured in fibroblast medium II (DMEM/F-12 (Gibco), supplemented with 1% GlutaMAX Supplement (Gibco), 20 ng/ml hbFGF (PeproTech), 1% 2-mercaptoethanol (Serva), 10% FBS (Sigma-Aldrich), and 1% MEM non-essential amino acids (Gibco). After 7 days of culture, the cells were passaged on Geltrex-coated cell culture dishes (Thermo Fisher Scientific) in fibroblast medium II. One day later, the medium was changed to Essential 8™ Medium (Gibco). Individual iPSC colonies were mechanically picked and expanded as clonal lines on Geltrex-coated cell culture dishes in Essential 8™ Medium, supplemented with 2 µM thiazovivin for 24 h (Sigma-Aldrich). IPSC lines were passaged using Versene solution (Thermo Fisher Scientific).

Although there were some differences in cell culture conditions after the reprogramming step, as can be tracked above, these are unlikely be responsible for the phenotypic differences observed with the various cerebral organoids described in the Results section. The same applies to the different genetic background of the healthy donor and the two BWCFF-S patients, in light of the following considerations. All iPSC clones obtained from the healthy donor- and the two BWCFF-S patient-derived fibroblasts were treated identically after the clonal expansion step (see below). All iPSC clones were tested for the removal of Sendai virus genomes and transgenes by reverse transcription-PCR analysis, and further checked for pluripotency marker expression, differentiation to three germ layers, and an intact karyotype by G-banding (Appendix Figs. S8, S9, and S13). One fully characterized clone of the control iPSC line (referred to as CRTDi011-A) and two clones per mutation (referred to as mutACTB-1 and mutACTB-2, and as mutACTG1-1 and mutACTG1-2) were used for cerebral organoid generation. Moreover, the presence of the NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile and NM_001614.5↗ (ACTG1):c.608 C > T p.Thr203Met mutation was additionally confirmed by locus-specific Sanger sequencing of the two clones each used in this study.

Furthermore, strong evidence that the phenotypic differences between the mutACTB-1-, mutACTB-2-, mutACTG1-1-, and mutACTG1-2-derived cerebral organoids and the control organoids are caused by the mutations in ACTB and ACTG1, respectively, is provided by our finding that introducing the Thr120Ile change into ACTB in the control iPSC clone CRTDi011-A using the CRISPR/Cas9 technology is sufficient to cause essentially the same cerebral organoid phenotype as observed in the BWCFF-S ACTB Thr120Ile patient-derived cerebral organoids (Fig. 7).

For the target region NC_000007.14↗:g.5529165 G > A (NM_001101.5↗ (ACTB):c.359 C > T p.Thr120Ile), the sequence was analyzed for appropriate guide RNA binding sites using the Geneious software and the CRISPOR design webtool (http://crispor.tefor.net/↗). The selected target sequence plus PAM was: 5’-AGGTAGCGGGCCACTCACCTGGG-3’. Guide RNA was ordered as crRNA from IDT (Integrated DNA Technologies). Linear donor templates with 50-bp homology arms were purchased from IDT as PAGE Ultramer ssDNA oligonucleotides with three phosphorothioated DNA bases on each side. In order to generate a heterozygous iPSC line, two oligonucleotides were designed; one contained the point mutation c.359 C > T, whereas the other contained the sequence encoding the reference amino acid sequence, with both oligonucleotides containing silent mutations to prevent recutting of either targeted allele. The ss-oligodeoxynucleotide (ssODN) sequence for the c.359 C > T mutation was: 5′-AGGAGCACCCCGTGCTGCTGACCGAGGCCCCCCTGAACCCCAAGGCCAACAGGGAAAAAATGATACAGGTGAGTGGCCCGCTACCTCTTCTGGTGGCCGCCTCCCTCCTTCCTGG-3′. The ssODN sequence for the c.359 C > C reference maintenance was: 5’-AGGAGCACCCCGTGCTGCTGACCGAGGCCCCCCTGAACCCCAAGGCCAACAGGGAAAAAATGACACAGGTGAGTGGCCCGCTACCTCTTCTGGTGGCCGCCTCCCTCCTTCCTGG-3′ (underlined bases indicate match to guide RNA). CRTDi011-A cells were transfected with RNP (46 pmol guideRNA and 31 pmol Cas9) and 0.5 µl of each ssODN (100 µM stock, yielding 50 pmol final amount) using the NEON transfection system (settings: 1000 V, 20 ms, 1 pulse). Cells were single-cell sorted by FACS, and heterozygous single-cell clones were identified by Sanger Sequencing of the target region (primer_fwd: 5′-GTCACCAACTGGGACGACAT-3′, primer_rev: 5′-GCTAAGTGTGCTGGGGTCTT-3′). Two such clones, referred to as CRTDi011-A–mutACTB-1 and CRTDi011-A–mutACTB-2, were used for the generation of cerebral organoids, after successful validation of pluripotency marker expression (Appendix Fig. S12).

The clonal control iPSC lines SC102A-1 (System Biosciences) and CRTDi011-A (this study), the mutated CRTDi011-A-derived clones CRTDi011-A–mutACTB-1 and CRTDi011-A–mutACTB-2, and the BWCFF-S patient-derived iPSC lines BWCFF-S ACTB Thr120Ile (clones mutACTB-1 and mutACTB-2) and BWCFF-S ACTG1 Thr203Met (clones mutACTG1-1 and mutACTG1-2) (all this study) were subjected to an identical culture protocol under feeder-free conditions on Matrigel-coated plates (Corning) in mTeSR™1 (StemCell Technologies) at 37 °C, in a humidified atmosphere of 5% CO₂ and 95% air. Medium was changed daily, and cells were passaged every third day, using ReLeSR™ (StemCell Technologies).

Cerebral organoids were generated as previously described (Lancaster et al, 2013; Lancaster and Knoblich, 2014; Kanton et al, 2019; Mora-Bermudez et al, 2016) with minor modifications. Patient-derived organoids were generated from 2 clones each per mutation (clones mutACTB-1 and mutACTB-2, and clones mutACTG1-1 and mutACTG1-2), patient-like organoids were generated from the mutated CRTDi011-A-derived clones CRTDi011-A–mutACTB-1 and CRTDi011-A–mutACTB-2, control organoids were generated from two different iPSCs (line SC102A-1 and clone CRTDi011-A) using feeder-free conditions. Briefly, 9,000 cells/well were seeded into an ultra-low-attachment 96-well plate (Corning) in mTeSR™1 (StemCell Technologies) supplemented with 50 µM ROCK inhibitor Y-27632 (StemCell Technologies). After 48 h, the medium was changed to mTeSR™1 without ROCK inhibitor Y-27632. On day 5, mTeSR™1 was replaced by neural induction medium (DMEM/F12 (Gibco), 1% N₂ supplement (Gibco) 1% GlutaMAX supplement (Gibco), 1% MEM non-essential amino acids (Gibco), and 1 µg/ml heparin (Sigma-Aldrich)) and changed every other day. Five days later, embryoid bodies were embedded in Matrigel, medium was changed to differentiation medium (DMEM/F12 and Neurobasal (Gibco, ratio 1:1), 0.5% N₂ supplement (Gibco) 1% GlutaMAX supplement (Gibco), 0.5% MEM non-essential amino acids (Gibco), insulin (2.875 ng/ml final concentration Sigma-Aldrich), 1% B27 supplement (without vitamin A, Gibco), 0.00035% 2-mercaptoethanol (Serva), and 1% penicillin–streptomycin (Gibco)), and cerebral organoids were further cultivated on an orbital shaker. Medium was changed every other day. On day 16, medium was changed to differentiation medium containing 1% B27 supplement with vitamin A (Gibco), and organoids were cultivated under these conditions until day 30 and 50 after seeding.

Cerebral organoids were fixed after 30 and 50 days in culture in 4% paraformaldehyde (PFA) in 120 mM phosphate buffer (pH 7.4) for 1 h at room temperature and then washed in PBS for at least 12 h. Organoids were then incubated overnight in phosphate-buffered saline (PBS) containing 30% sucrose at 4 °C, embedded in Tissue-Tek OCT (Sakura, Netherlands), and frozen on dry ice. Cryosections of 12 µm thickness were cut on a cryostat (Microm HM 560, Thermo Fisher Scientific) and stored at –20 °C until further use.

Fetal human brain samples were collected in accordance with the Helsinki Declaration 2000 and approval of the Ethical Committees of the School of Medicine, University of Zagreb, and University Hospital Centre Zagreb (641-01/19-02/01; 8.1-7/170-02; 02/21AG). The brain specimens were obtained from fetuses following spontaneous abortions (with 12 h postmortem delay, or less). The deaths were attributed to respiratory distress syndrome or non-fetal causes. Two human fetuses at the age of 12 pcw and two human fetuses at the age of 16 pcw were clinically evaluated by a gynecologist and a pathologist and included in this study. The fetuses showed no evidence of growth restriction or developmental malformations; the fetal age was estimated based on the crown-rump length, pregnancy records, and assessment of histological sections. Whole fetal brains were fixed in 4% PFA in 0.1 M PBS pH 7.4 for 48 to 96 h, depending on the brain size, and afterwards embedded in paraffin according to standard procedures. The brain tissue was cut into 20-µm-thick sections, mount on glass slides, deparaffinized, and rehydrated prior to immunohistochemistry as described below.

Fetal human brain tissue was obtained from the Klinik und Poliklinik für Frauenheilkunde und Geburtshilfe, Universitätsklinikum Carl Gustav Carus of the Technische Universität Dresden with maternal agreement and permission of the Ethics Committee at the TU Dresden. After pregnancy termination, the tissue was directly transferred to the lab, and fragments of the cerebral cortex were identified by morphology and fixed in 4% PFA in 120 mM phosphate buffer pH 7.4 for 3 h at room temperature, followed by 24 h at 4 °C. The specimen was then incubated overnight in PBS containing 30% sucrose at 4 °C and processed for cryosectioning as described above for cerebral organoids.

For immunohistochemical detection of βCYA and γCYA, cerebral organoids and fetal brain sections were washed twice with PBS and permeabilized with ice-cold methanol (–20 °C) for 5 min. After two washing steps in PBS, samples were incubated in blocking buffer (2% BSA/PBS) for 1 h. Primary antibodies were diluted in blocking buffer and incubated for 1 h, followed by a 1 h incubation with secondary antibodies also diluted in blocking buffer (Latham et al, 2018). Immunostained tissue sections were counterstained with DAPI (1:1000 in PBS) and were mounted in Mowiol (Millipore). Other immunohistochemical stainings were performed as previously described (Kanton et al, 2019) using antigen retrieval in 0.01 M sodium citrate buffer (pH 6.0) for 1 h at 70 °C. The following primary antibodies were used: mouse monoclonal IgG₁ anti-βCYA (Bio-Rad, clone 4C2, RRID:AB_2571580), mouse monoclonal IgG_2b anti-γCYA (Bio-Rad, clone 2A3, RRID:AB_2571583), mouse anti-Caspase3 (abcam, #ab208161), CREST anti-centromere (kinetochore) (Antibodies Incorporated, #15-235, RRID:AB_2939059), rat anti-Ctip2 (abcam, #ab18465, RRID:AB_2064130), rabbit anti-Ki-67 (abcam, #ab15580, RRID:AB_443209), rabbit anti-Nestin (Sigma-Aldrich, #N5413, RRID:AB 1841032), mouse anti-pan-actin (Novus Biologicals, #NB600-535, RRID:AB_2222881), mouse anti-pan-cadherin (Sigma-Aldrich, #C1821, RRID:AB_476826), rat anti-PH3 (abcam, #ab10543, RRID:AB_2295065), goat anti-SOX2 (R + D Systems, #AF2018, RRID:AB_355110), rabbit anti-TBR2 (abcam, #ab23345, RRID:AB_778267), mouse anti-alpha-tubulin (Sigma-Aldrich, #T6199, RRID:AB_477583), rabbit anti-beta-tubulin (abcam, #ab179513, RRID:AB_3073861), rabbit anti-tubulin (Sigma-Aldrich, #T3526, RRID:AB_261659), mouse anti-Tuj1 (BioLegend, #801201, RRID:AB_2313773) and rabbit anti-ZO-1 (Invitrogen, #61-7300, RRID:AB_138452). All fluorescent-conjugated secondary antibodies were purchased from Thermo Fisher Scientific.

Images were acquired as Z-stacks of thirty 0.3-µm optical sections in the case of βCYA and γCYA immunostainings or of twelve 1-µm optical sections in the case of all other immunostainings using a Zeiss LSM 880 microscope with a 63×/1.3 LCI Plan-Neofluar, W/Glyc, DIC (Zeiss) objective. Prior to the image analysis, 10 adjacent slices in the center of the stack of the thirty 0.3-µm optical sections of the βCYA and γCYA immunostainings, and three adjacent slices in the center of the stack of the twelve 1-µM slices of all other immunostainings, were averaged. Overview images of the cerebral organoids were acquired using a Keyence’s bench-top BZ-X810 all-in-one fluorescence microscope with 2× Plan-Apochromat, air, and 10x Plan-Fluor, air objectives (Keyence). Haze reduction function (Keyence) for the elimination of fluorescence blurring was used for the overview images of DAPI-stained cerebral organoids. Images to detect astral microtubules were acquired with an LSM 780 NLO laser scanning microscope, using a 63× Plan-Apochromat 1.4 N.A. oil objective (Carl Zeiss). The stacks acquired were of 512 × 512 pixels × 15–20 optical sections (xyz sampling: 0.09 × 0.09 × 0.75 μm). Images were analyzed with ImageJ (http://imagej.nih.gov/ij/↗) and Zen software (Carl Zeiss). The brightness and contrast of images were recorded and adjusted linearly.

Analysis of the βCYA and γCYA distribution was performed with custom Jython scripts written for Fiji (Schindelin et al, 2012). The input image stacks containing the γCYA isoform images in the magenta channel, the βCYA isoform images in the green channel, and an additional DAPI channel were first corrected for shifts caused by chromatic aberrations. Then, background intensity was subtracted per channel, and incompletely acquired images at the beginning and end of the stack were removed. These preprocessed image stacks were re-sliced along the z axis for visualization purposes. To obtain the βCYA and γCYA distributions in the x–y plane, 10 adjacent slices in the center of the stack were averaged. The gray-scale images to visualize the difference in localization pattern of γCYA and βCYA were obtained by first normalizing the intensities of each channel (histogram normalization). Then, the green channel signals were subtracted from the magenta channel signals, and a gray lookup table was applied. Thus, bright regions corresponded to areas in which γCYA was localized relatively strongly (compared to the localization distribution of βCYA). Conversely, in dark regions, βCYA was predominantly localized. Note that the generated images visualize relative localization patterns and do not quantify absolute concentrations of γCYA or βCYA.

In a second analysis step, the differences between the γCYA and βCYA localization patterns were quantified. Within each of the various zones (AJ, VZ, iSVZ, oSVZ, IZ, CP, MZ) of fetal human neocortical tissue, a region of interest (ROI) was defined for which the relative localization of βCYA and γCYA was compared. To this end, the ratio between the mean signal intensities of the respective ROI was calculated for each channel separately. Differences in the value of the ratio for the two channels then indicated a different localization preference of γCYA versus βCYA. For statistical analysis, intensity ratios of multiple images were obtained and compared with GraphPad Prism. The relative intensities of γCYA and βCYA were normalized to the mean fluorescence intensity of all positions.

Electron microscope analysis was performed as previously described (Dubreuil et al, 2007; Meinhardt et al, 2014; Camp et al, 2015). Cerebral organoids were shortly pre-fixed in 0.2% glutaraldehyde in culture medium, before being fixed in 1% glutaraldehyde, 2% PFA in 0.1 M Pipes pH 7.4, 0.09 mM CaCl₂ for 1 h at room temperature. After washing steps, the samples were transferred into 4% low-melting agarose in PBS. Sections of 200 µm thickness through the cerebral organoids were cut on a vibratome (Leica VT1200S). Sections were post-fixed with 2% osmium tetroxide (OsO₄), 1.5% potassium ferrocyanide for 30 min at room temperature. Contrast was enhanced by subsequent incubation (without washing) with 1% OsO₄ for 1 h. After washing, the samples were stained en bloc with 0.5% uranyl acetate overnight at 4 °C. Samples were dehydrated by an ascending series of ethanol (15 min for each step) and three times in pure ethanol. After gradual infiltration with resin, the samples were flat-embedded in Epon replacement (Embed-812, Science Services) and polymerized at 60 °C.

Selected regions were re-mounted on resin stubs, and 70 nm-thick sections were cut on an ultramicrotome (Leica UCT). Sections were post-stained with uranyl acetate and lead citrate according to standard protocols. Images were acquired on a Tecnai 12 Biotwin transmission electron microscope (Thermo Fisher Scientific/FEI) at 100 kV with a TVIPS F416 CMOS camera and SerialEM software (Mastronarde, 2005). Image processing was performed in Fiji (Schindelin et al, 2012).

For electron tomography, 300 nm-thick sections were cut on an ultramicrotome (Leica UCT) and post-stained as described above. Both sides of the section were coated with 15 nm colloidal gold beads (BBI solutions). Tilt series at +/− 64° with a 1° increment were acquired on a Tecnai F30 transmission electron microscope (Thermo Fisher Scientific/FEI) at 300 kV with a Gatan OneView camera and SerialEM software. Second axis tilt series were acquired after a 90° sample rotation. Electron tomograms were reconstructed using the IMOD software package. The arrangement of the adherens junctions (and centrioles/basal bodies) in the reconstructed tomograms was visualized by 3D-rendering with inverted grey values in ORS Dragonfly software (Comet Technologies Canada Inc., Montreal, Canada; software available at https://www.theobjects.com/dragonfly↗).

Quantification of the EM data shown in Fig. 6 was not performed due to technical limitations. The apical-most neuroepithelium in both wild-type and mutant organoids exhibits a highly complex three-dimensional architecture, making accurate measurements of cell circumference or cytoskeletal attachments at adherens junctions unreliable in single 2D sections. The orientation of individual sections (e.g., diagonal vs. transverse cuts) and potential membrane folds could not be determined with certainty. Furthermore, the resolution of the tomograms was in some cases insufficient to unambiguously identify individual actin filaments (~5 nm) or microtubules (~20 nm) within 300-nm sections. Reliable quantification of these elements would require volumetric imaging.

Organoid protein lysates were obtained from organoid cell cultures after 30 days of culture using the RIPA lysis buffer system (Santa Cruz). Samples with a protein amount of 4 µg were prepared using 4× Protein Sample Loading Buffer (Li-Cor), denatured for 10 min at 70 °C, and loaded onto NuPAGE® Novex 4–12% Bis-Tris gels (Thermo Fisher Scientific) for separation. Separation was performed by running the gels for ~2.5 h at 90 V in 1× NuPAGE™ MOPS SDS running buffer (lower buffer chamber) and 1× NuPAGE™ MOPS SDS running buffer supplemented with NuPAGE® Antioxidant (upper buffer chamber) (Thermo Fisher Scientific). Proteins were transferred to iBlot™ PVDF Transfer Stacks (Thermo Fisher Scientific) for 6 min at 25 V. After blotting, membranes were dried at 37 °C for 10 min and rehydrated in TBS. For total protein detection, membranes were stained using Revert™ 700 Total Protein Stain (Li-Cor). After destaining, membranes were blocked in 5% milk (Roth) in TBST (1× TBS, 0.1% Tween-20) and incubated overnight at 4 °C while rotating. After three washes with 1× TBST, membranes were incubated with primary antibodies (mouse anti-human actin beta antibody, clone 4C2, or mouse anti-human actin gamma antibody, clone 2A3, Bio-Rad) at a 1:7500 dilution for 2 h at room temperature while rotating. Following three additional washes with 1× TBST, membranes were incubated in the dark for 1 h with a goat anti-mouse IgG H&L (IRDye® 800CW) secondary antibody (Li-Cor) at a 1:15000 dilution at room temperature while rotating. Washing with 1× TBST and 1× TBS was repeated. For visualization, a Li-Cor Odyssey® Imager was used. Image Lab (Bio-Rad) was used to analyze the protein bands.

For scRNA-seq of cerebral organoids, 3–6 organoids (depending on size) were used per iPSC line or clone, derived from one to three independent batches, at 30 and 50 days of culture. For dissociation into single cells, the organoids were cut with a scalpel and washed three times with HBSS (Thermo Fisher Scientific) to remove apoptotic cells from the core. Subsequently, the organoids were incubated with Solution A (50 μL Enzyme P + 1900 μL Buffer X) from the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec) for 15 min at 37 °C in a water bath. After the addition of Solution B (10 μL Enzyme A + 20 μL Buffer Y), the organoids were dissociated using a 1000 μL pipette tip and incubated for 10 min at 37 °C. The mechanical dissociation and incubation steps were repeated four times before filtering the cell suspension through a 30-μm filter and washing it three times with HBSS.

ScRNA-seq was then performed based on the 10x Genomics Single-cell transcriptome workflow (Zheng et al, 2017). Specifically, accurate cell numbers were obtained using a Neubauer Hemocytometer. Cell samples were carefully mixed with Reverse Transcription Reagent (Chromium Single Cell NextGem 3′ Library Kit v3.1) and loaded onto a Chromium Single Cell G Chip to reach a recovery of up to 10,000 cells per sample. The samples were processed further following the guidelines of the 10x Genomics user manual for single-cell 3′ RNA-seq v3.1. In short, the droplets were directly subjected to reverse transcription, the emulsion was broken, and cDNA was purified using silane beads (Chromium Single Cell 3′ Gel Bead Kit v2). After amplification of cDNA with 11 cycles, samples were purified with 0.6x volume of SPRI select beads to deplete DNA fragments smaller than 400 bp, and cDNA quality was monitored using the Agilent FragmentAnalyzer 5200 (NGS Fragment Kit). In total, 10 μL of the resulting cDNA was used to prepare scRNA-seq libraries, involving fragmentation, dA-tailing, adapter ligation, and 10–13 cycles of indexing PCR, following the manufacturer’s guidelines. After quantification, libraries were sequenced on multiple Illumina NovaSeq 6000 S4 flow cells in 100 bp paired-end mode, aiming for 30,000 fragments per cell. The initial sequencing data were processed using the count command of the CellRanger software (v6.1.2) (https://support.10xgenomics.com/single-cell-gene-expression/software↗). To build the reference, the human genome (hg38) and gene annotation (Ensembl 104) were used. The annotation was filtered using the mkgtf command of the CellRanger tool, as suggested by 10x Genomics, to include only genes with the types ‘protein_coding’, ‘lincRNA’, ‘antisense’, ‘IG_LV_gene’, ‘IG_V_gene’, ‘IG_V_pseudogene’, ‘IG_D_gene’, ‘IG_J_gene’, ‘IG_J_pseudogene’, ‘IG_C_gene’, ‘IG_C_pseudogene’, ‘TR_V_gene’, ‘TR_V_pseudogene’, ‘TR_D_gene’, ‘TR_J_gene’, ‘TR_J_pseudogene’, ‘TR_C_gene’. The human genome and the filtered annotation were then used as input for the CellRanger mkref command to build a CellRanger reference.

The resulting expression tables were then analyzed using the Scanpy software (v1.9.1) following best-practice guidelines (Wolf et al, 2018; Luecken and Theis, 2019). For each 10x barcode, a Doublet Score was calculated using the Scrublet tool (Wolock et al, 2019). Quality control was performed, and only cells with at least 5000 reads, at least 2500 detected genes, a maximum of 30% of reads in the 50 most highly expressed genes, a maximum of 15% of reads in mitochondrial genes, and a Doublet Score of at most 0.25 were included in the analysis. This resulted in a dataset with 260,880 cells. Only genes expressed in at least 10 cells were used for further analysis.

All functions mentioned below are from Scanpy. Unless otherwise specified, standard parameters were used. CPM normalization and log-transformation were performed using the normalize_total and log1p functions (base=10). The 5000 most variable genes were selected using highly_variable_genes. A principal component analysis (PCA) was performed with the function pca. Sample integration was carried out with harmony_integrate (Korsunsky et al, 2019). On the Harmony-corrected principal components, a k-nearest neighbors (k-NN) network was computed with neighbors. This network was used for UMAP calculation with umap. Clustering was performed using Leiden (resolution=0.4), which resulted in 17 clusters. Marker genes for each cluster were computed using rank_genes_groups. The source code for the analysis is available upon request. The complete software stack is available as a Singularity container (https://gitlab.hrz.tu-chemnitz.de/dcgc-bfx/singularity/singularity-single-cell/container_registry/↗, v1.1.0).

GraphPad Prism version 9.3.1 for Windows (GraphPad Software, La Jolla California USA, http://www.graphpad.com/↗) was used for graphical illustration and statistical analysis of all data. All data were tested for normal distribution (Shapiro–Wilk test) and further analyzed with unpaired t test, Mann–Whitney U test, one-way or two-way ANOVA, or Kruskal–Wallis test (*P < .0.05; **P < .0.01; ***P < 0.001; ****P < 0.0001).

The authors declare no competing interests.

There are no restrictions on the published data. All data are accessible in this article or in the supplementary material. The single-cell RNA sequencing data from this publication have been deposited in the German Human Genome-Phenome Archive (https://data.ghga.de/browse↗) and assigned the identifier GHGAS92787677252496.

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44319-025-00647-7↗.