ARP2/3 complex mediates the neuropathology of PTEN-deficient human neural cells downstream of mTORC1 and mTORC2 hyperactivation

Significance

Mutations in the phosphatase and tensin homolog (PTEN) gene cause neurodevelopmental disorders characterized by brain overgrowth, autism, and epilepsy. These conditions are associated with hyperactivation of mechanistic target of rapamycin (mTOR), a master regulator of cell growth, yet mTOR-targeted therapies have shown limited clinical benefit. Using human stem cell–derived neural cells, we found that excessive mTOR activity disrupts the actin cytoskeleton through the ARP2/3 complex. Inhibiting ARP2/3 reversed abnormal cell proliferation, hypertrophy, and electrical hyperactivity in PTEN-deficient neurons and precursors. These findings establish the PTEN–mTOR–ARP2/3 axis as a key driver of disease phenotypes and suggest ARP2/3 inhibition as a potential therapeutic strategy. This work uncovers a mechanism that may apply broadly to brain disorders involving mTOR hyperactivation.

Abstract

Mutations in the phosphatase and tensin homolog (PTEN) gene are linked to severe neurodevelopmental disorders. Loss of PTEN causes hyperactivation of both mechanistic target of rapamycin (mTOR) complexes, mTORC1 and mTORC2. Recent studies have shown that this dual hyperactivation is required for the neuropathology observed in PTEN-deficient human stem cell–derived neural cells. However, the molecular effectors that integrate these synergistic signals remain unknown. Here, we identify the actin-regulating ARP2/3 complex as a critical point of convergence downstream of mTORC1 and mTORC2. We show that concurrent hyperactivation of both complexes drives increased filamentous actin and elevated levels of the ARP2/3 complex subunits in PTEN-deficient human neural precursors (NPs) and neurons. Pharmacological or genetic inhibition of ARP2/3 is sufficient to rescue multiple disease-relevant phenotypes, including NP hyperproliferation, neuronal hypertrophy, and electrical hyperactivity, without affecting the upstream mTORC1 or mTORC2 hyperactivation. Together, these findings reveal the PTEN–mTOR–ARP2/3 signaling axis as a core mechanism of neuropathology and highlight ARP2/3 inhibition as a potential therapeutic strategy for PTEN-related neurodevelopmental disorders.

Germline heterozygous loss-of-function mutations in the phosphatase and tensin homolog (PTEN) gene cause a spectrum of clinical manifestations known as PTEN hamartoma tumor syndrome (PHTS), which includes a high prevalence of neurodevelopmental disorders (NDDs) such as autism spectrum disorder, intellectual disability, and epilepsy (1–3). A central mechanism underlying PHTS pathophysiology is the hyperactivation of the phosphoinositide 3-kinase (PI3K)-AKT-mTOR signaling pathway, a master regulator of cell growth, proliferation, and metabolism (4). Consequently, inhibitors of mechanistic target of rapamycin (mTOR) such as rapamycin have been considered as therapeutic avenues (5). However, clinical trials of mTOR inhibitors in patients with PTEN deficiency have yielded limited therapeutic benefits, likely due to the pathway’s broad physiological functions and the potential for both central and peripheral side effects (6, 7). These limitations underscore the need to identify more specific signaling effectors downstream of mTOR hyperactivation that directly contribute to neuropathology.

The mTOR kinase serves as the catalytic core of two distinct complexes, mTORC1 and mTORC2 (4, 8). While the roles of these separate complexes have been studied extensively, their interplay in the context of PTEN-related NDDs has been less clear. Using human pluripotent stem cells (hPSCs) derived neural models that recapitulate key aspects of the human disorder, we previously demonstrated that synergistic hyperactivation of both mTORC1 and mTORC2 is required to drive the neural phenotypes observed in PTEN-deficient human neural precursors (NPs) and neurons, including hyperproliferation, cellular hypertrophy, and electrical hyperactivity (9, 10). Importantly, selective normalization of either mTORC1 or mTORC2, via genetic disruption of RAPTOR or RICTOR, key components of each respective complex, was sufficient to prevent these abnormalities. These findings suggest that mTORC1 and mTORC2 converge on a common downstream effector or cellular process. Identifying this point of convergence is a critical next step toward developing more targeted therapeutics.

A promising candidate for this convergence is the actin cytoskeleton, a dynamic network of filaments crucial for cell morphology, migration, and synaptic function. Dysregulation of the actin cytoskeleton is recognized as a significant contributor to the pathophysiology of various NDDs (11). For instance, mutations in key actin-related genes, such as RAC1 and WASF1 (also known as WAVE1) are known to cause NDDs that share clinical manifestations with PTEN-related NDDs, including macrocephaly, intellectual disability, epilepsy, and autism spectrum disorders (12–14). Furthermore, altered neuronal actin content and dynamics have been mechanistically linked to other syndromic NDDs, including Fragile X syndrome, Tuberous Sclerosis, and Phelan-McDermid Syndrome (15–17). Both mTORC1 and mTORC2 have been independently shown to regulate actin dynamics. mTORC2, for example, is a well-established regulator of small GTPases such as RAC1, CDC24, and RHOA, which in turn control actin polymerization and cytoskeletal organization (18–20). mTORC1 has also been directly implicated in controlling cytoskeletal organization and filamentous actin (F-actin) content in neuronal contexts (20–22). Together, these insights raise the possibility that actin cytoskeleton may also contribute to the PTEN-deficient human neural cells, acting downstream of mTOR hyperactivation.

In this study, we used hPSC-derived neural models to investigate how mTOR hyperactivation drives neuropathology in PTEN-deficient cells. We found that dual hyperactivation of mTORC1 and mTORC2 increases levels of the actin-nucleating ARP2/3 complex and elevates F-actin content in both NPs and neurons. Inhibiting ARP2/3, either pharmacologically or genetically, rescued the core cellular and physiological deficits in PTEN-mutant human neural cells, including hyperproliferation, neuronal hypertrophy, and electrical hyperactivity. Together, our results establish a direct mechanistic link between mTOR and ARP2/3 signaling axis as the driving mechanism underlying PTEN-deficient neuropathology.

Results

PTEN Mutant Human NPs and Neurons Display Increased Actin Cytoskeleton Content Driven by mTORC1 and mTORC2 Hyperactivation.

Actin cytoskeleton dysregulation has been increasingly implicated in NDDs, but its role in PTEN-deficient human neuropathology remains poorly understood. To investigate this, we previously used CRISPR/Cas9-mediated gene editing to generate PTEN homozygous null mutant hPSCs from WIBR3, a well-characterized female embryonic stem cell line (10). Upon directed differentiation into cortical NPs, these PTEN mutant cells lacked PTEN protein and exhibited significantly elevated proliferation compared to isogenic controls, a disease-relevant phenotype that likely contributes to macrocephaly (9, 10). Given that the actin cytoskeleton regulates key cellular processes including cell proliferation and morphology, we first asked whether actin content is altered in PTEN mutant NPs. To visualize filamentous actin (F-actin), we performed phalloidin staining. When normalized to soma size, phalloidin intensity was significantly increased in PTEN mutant NPs compared to isogenic controls, indicating elevated F-actin content (Fig. 1 A and B).

Fig. 1.

Dual hyperactivation of mTORC1 and mTORC2 drives increased filamentous actin (F-actin) in phosphatase and tensin homolog (PTEN) mutant human neural precursors (NPs) and neurons. (A and B) Representative images (A) and quantification (B) of Phalloidin staining show that PTEN mutant WIBR3 human NPs have increased levels of F-actin. Both PTEN+RAPTOR and PTEN+RICTOR double mutants, which selectively normalize mTORC1 or mTORC2 signaling, respectively, rescue the F-actin levels. Lower panels are higher magnification views of the boxed areas in the corresponding Upper panels. (Scale bar, Upper panels, 20 µm and Lower panels, 5 μm.) Each data point represents a cell (n = 82 to 91 for each group from 4 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (C and D) Representative images (C) and quantification (D) of Phalloidin staining show that day 30 PTEN mutant WIBR3 human neurons have increased levels of F-actin. Both PTEN+RAPTOR and PTEN+RICTOR double mutants rescue the F-actin levels. Neurons are induced from WIBR3 NPs using Ngn2 induction. Lower panels are higher magnification views of the boxed areas in the corresponding Upper panels. (Scale bar, Upper panels, 20 μm and Lower panels, 5 μm.) Each data point represents a cell (n = 70 to 75 for each group from 4 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. Results are mean ± SEM. *P < 0.05, ****P < 0.0001.

We then asked whether the increased F-actin content in PTEN mutant NPs is directly caused by mTOR hyperactivation by examining the roles of mTORC1 and mTORC2 individually. Our previous work demonstrated that PTEN mutant human NPs display significantly elevated mTORC1 and mTORC2 activity, as measured by phosphorylation of S6 at Ser240/244 (mTORC1) and AKT at Ser473 (mTORC2), and that the synergistic hyperactivation of both pathways are required for the hyperproliferation phenotype (9). To dissect the contribution of these two pathways, we disrupted RAPTOR or RICTOR, key components of mTORC1 and mTORC2, respectively, to selectively rescue either pathway (9). In the present study, we tested whether both pathways are also required for actin cytoskeletal dysregulation. We analyzed four genotypes of human NPs: control, PTEN mutant, PTEN+RAPTOR, and PTEN+RICTOR double mutant. These double mutant NPs were differentiated from clonal WIBR3 hPSC lines carrying PTEN homozygous null mutations, along with compound heterozygous mutations in RAPTOR or RICTOR (consisting of one frameshift and one frame-preserving mutation) that reduce their protein levels by approximately 50% (9). Phalloidin staining showed that disruption of either RAPTOR or RICTOR restored F-actin levels to those observed in controls (Fig. 1 A and B). These findings demonstrate that the simultaneous hyperactivation of both mTORC1 and mTORC2 is required for elevated actin content in PTEN mutant human NPs.

We next investigated whether similar cytoskeletal abnormalities occur in PTEN mutant human cortical neurons. Using a doxycycline-inducible Ngn2 overexpression lentivirus, we transduced control, PTEN, PTEN+RAPTOR, and PTEN+RICTOR mutant WIBR3 hPSCs and differentiated them into neurons. Phalloidin staining revealed significantly increased F-actin content in day 30 neurons, even after normalizing to cell size (Fig. 1 C and D). This phenotype was rescued in both PTEN+RAPTOR and PTEN+RICTOR double mutant day 30 neurons, indicating that dual mTORC1 and mTORC2 hyperactivation is required for actin cytoskeleton dysregulation in PTEN-deficient neurons (Fig. 1 C and D), as in NPs.

We validated these findings in two independent genetic backgrounds. First, we used CRISPR/Cas9 to generate clonal PTEN homozygous null mutant hPSC lines in the PGPC17-iNgn2 system (SI Appendix, Fig. S1 B and C). PGPC17 is a well-characterized hPSC line reprogrammed from a healthy male donor (23), and the PGPC17-iNgn2 version contains a doxycycline-inducible Ngn2 expression cassette in the CLYBL safe-harbor locus (SI Appendix, Fig. S1A). Upon doxycycline treatment, isogenic control and PTEN mutant hPSCs were differentiated into cortical neurons. Phalloidin staining confirmed that PTEN mutant PGPC17 neurons also exhibit significantly increased F-actin content on day 30 (SI Appendix, Fig. S2 A and B). We further tested the phenotype using the dual-engineered iCas9-iNgn2 line in H1, a widely used male hPSC line. The iCas9-iNgn2 line harbors doxycycline-inducible Cas9 and Ngn2 expression cassettes inserted in each of the two AAVS1 safe-harbor loci (9, 24). Using this tool, we previously generated pooled, polyclonal PTEN mutant neurons to demonstrate morphological and electrophysiological phenotypes (9). In the current study, we validated that these PTEN mutant neurons, which showed an approximate 50% reduction in PTEN protein (SI Appendix, Fig. S3A), also displayed significantly elevated F-actin content at a population level on day 30 (SI Appendix, Fig. S3 C and D).

Synergistic mTORC1 and mTORC2 Hyperactivation Increases Levels of the ARP2/3 Complex In PTEN Mutant Human NPs and Neurons.

To investigate the molecular mechanisms of increased F-actin content in PTEN mutant WIBR3 human neural cells, we examined the protein levels of key regulators of actin dynamics. In both PTEN mutant NPs and neurons, we observed elevated levels of RAC1, a small GTPase known to be a key organizer of the actin cytoskeleton (). This finding is consistent with previous studies in Pten null mutant mouse fibroblasts (25). Given the broad role of RAC1 in regulating cytoskeleton dynamics, we sought to further specify this mechanism. RAC1 is known to regulate the ARP2/3 complex, a critical actin nucleator that generates branched actin networks and plays important roles in neurodevelopment (26, 27). We found that PTEN mutant NPs and neurons had significantly elevated levels of multiple ARP2/3 subunits, including ARP2, ARP3, ARPC1, and ARPC4 (Fig. 2 A–L).

Fig. 2.

Dual hyperactivation of mTORC1 and mTORC2 causes increased levels of the ARP2/3 complex in PTEN mutant human NPs and neurons. (A–F) Immunoblotting analysis shows that PTEN mutant WIBR3 human NPs have increased levels of RAC1 and key components of the ARP2/3 complex (ARP2, ARP3, ARPC1, and ARPC4). Both PTEN+RAPTOR and PTEN+RICTOR double mutants rescue these protein levels. Each data point represents an independent experiment (different hPSC clone or independent differentiation, n = 3 for each group). P values are calculated using post hoc multiple comparison after ANOVA. (G–L) Immunoblotting analysis shows that day 10 PTEN mutant WIBR3 human neurons have increased levels of RAC1 and key components of the ARP2/3 complex (ARP2, ARP3, ARPC1, and ARPC4), which are rescued in both PTEN+RAPTOR and PTEN+RICTOR double mutants. Neurons are induced from WIBR3 hPSCs using Ngn2 induction. Each data point represents an independent experiment (different hPSC clone or independent differentiation, n = 3 for each group). P values are calculated using post hoc multiple comparison after ANOVA. Results are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001.

To determine whether RAC1 and ARP2/3 upregulation depends on mTORC1 and mTORC2 hyperactivation, we analyzed protein levels in PTEN+RAPTOR and PTEN+RICTOR double mutant WIBR3 human NPs and neurons. In both genotypes, elevated levels of ARP2/3 subunits and RAC1 were rescued to control levels, indicating that both pathways are required to drive this upregulation (Fig. 2 A–L). We further validated these findings in two additional genetic backgrounds. First, we showed that ARP2/3 complex components are also significantly elevated in clonal PTEN homozygous null mutant neurons derived from the PGPC17 hPSC line (SI Appendix, Fig. S2 C and D). Second, we analyzed polyclonal PTEN mutant neurons from the H1-iCas9-iNgn2 hPSC line and again observed increased ARP2/3 levels (SI Appendix, Fig. S3 E–I). Importantly, this increase was also rescued in both PTEN+RAPTOR and PTEN+RICTOR double mutants (SI Appendix, Fig. S3 E–I). Together, these data establish that synergistic mTORC1 and mTORC2 hyperactivation is necessary for ARP2/3 upregulation in PTEN-deficient human neurons.

Pharmacological Inhibition of the ARP2/3 Complex Rescues PTEN Mutant NPs and Neurons.

We next asked whether the ARP2/3 complex is required for the disease-relevant molecular, cellular, and physiological phenotypes of PTEN mutant human NPs and neurons. We treated control and PTEN mutant NPs with the ARP2/3 inhibitor CK-666 (28, 29) (100 µM) for 24 h and observed significantly reduced F-actin content as measured by phalloidin staining (Fig. 3 A and B). Using KI67 immunostaining and an EdU incorporation assay, we showed that PTEN mutant human NPs display hyperproliferation, consistent with our previous reports (9, 10). The ARP2/3 inhibitor CK-666 was effective in reducing the number of proliferating cells to control level (Fig. 3 C and D).

Fig. 3.

The ARP2/3 complex inhibitor CK-666 restores F-actin levels and rescues hyperproliferation in PTEN mutant human NPs. (A and B) Representative images (A) and quantification (B) show that chronic CK-666 treatment reduces the increased levels of F-actin in PTEN mutant WIBR3 human NPs. (Scale bar, 20 μm.) Each data point represents a cell (n = 115 to 131 for each group from four independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (C and D) Representative images (C) and quantification (D) show that chronic CK-666 treatment reduces the overproliferation of PTEN mutant WIBR3 human NPs, as measured by Ki67 immuno-staining and EdU incorporation assay. (Scale bar, 50 μm.) Each data point represents an independent experiment (different hPSC clone or independent differentiation, n = 4 for each group). P values are calculated using post hoc multiple comparison after ANOVA. Results are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001.

Next, we treated differentiating control and PTEN mutant PGPC17 neurons with the CK-666 inhibitor (100 µM) starting on day 1 of Ngn2 induction and continuously throughout. Phalloidin staining showed that chronic CK-666 treatment significantly reduced the F-actin content in PTEN mutant neurons on day 14 (Fig. 4 A and B). While PTEN mutant neurons showed significant increases in soma and nucleus sizes as well as dendritic arborization, CK-666 treatment rescued these phenotypes to control level (Fig. 4 C–G). We recorded neuronal electrical activity using a multielectrode array (MEA) platform over the span of 30 d. PTEN mutant neurons were significantly more active and synchronized in their firing pattern compared to control neurons, consistent with our previous report (9). Chronic CK-666 treatment prevented the development of these hyperactive electrical phenotypes (Fig. 4 H–J and Movie S1). We further validated these findings in the polyclonal PTEN mutant neurons derived from H1-iCas9-iNgn2 hPSCs (SI Appendix, Fig. S4 and Video S3). Together, these findings demonstrate that the ARP2/3 complex mediates the disease-relevant morphological and physiological phenotypes in PTEN mutant human NPs and neurons.

Fig. 4.

The ARP2/3 complex inhibitor CK-666 restores F-actin levels and rescues morphological and electrophysiological phenotypes in PTEN mutant human neurons. (A and B) Representative images (A) and quantification (B) show that chronic CK-666 treatment reduces the increased levels of F-actin in PTEN mutant PGPC17 human neurons on day 14. (Scale bar, 20 μm.) Each data point represents a cell (n = 101 to 105 for each group from 3 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (C and D) Representative images (C) and quantification (D) show that chronic CK-666 treatment reduces the soma and nucleus size of PTEN mutant human neurons on day 14. (Scale bar, 10 μm.) Each data point represents a cell (n = 57 to 61 for each group from three independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (E) Representative images of the traces of individual GFP-labeled neurons showing their dendritic morphology on day 14. (Scale bar, 50 μm.) (F and G) Sholl analyses reveal increased dendritic complexity (F) and total dendritic length (G) in PTEN mutant human neurons on day 14, which are rescued by chronic CK-666 treatment. ANOVA found significant effects of genotype and treatment (F, black asterisks marked on the right side of the graph). The post hoc multiple comparison test revealed differences at specific distance from cell soma, in which the black asterisk represents comparison between control and PTEN mutant neurons, and the red asterisk represents comparison between PTEN mutant neurons treated with vehicle or CK-666 (F). Each data point represents a cell (n = 20 for each group from 3 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (H–J). Multielectrode array analysis shows that chronic CK-666 treatment rescues the electrical hyperactivity (I, mean firing rate) and hypersynchronization (J) phenotypes of PTEN mutant human neurons. Raster plots (H) show spike distribution on day 30. The purple rectangle highlights the incidence of network burst. ANOVA found significant effects of genotype and CK-666 treatment (black asterisks marked on the right side of the graphs). The post hoc multiple comparison test revealed differences on individual days of recording, in which the black asterisk represents a comparison between vehicle-treated control and PTEN mutant neurons, and the red asterisk represents a comparison between vehicle- and CK-666-treated PTEN mutant neurons. N = 6 for each group from 6 independent experiments. Results are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001.

Genetic Inhibition of the ARP2/3 Complex Rescues PTEN Mutant NPs and Neurons.

We sought to further validate the role of the ARP2/3 complex using a CRISPR-based genetic approach to disrupt ARPC4, an essential subunit of the ARP2/3 complex. Previous mouse studies have shown that knockout or knockdown of Arpc4 in various tissues (including the epidermis, intestinal epithelium, and pancreatic cancer models) leads to destabilization of the Arp2/3 complex. This results in a marked reduction in the protein levels of most other Arp2/3 subunits and a significant decrease in F-actin content (29–31). Notably, patients with ARPC4 germline loss-of-function mutations develop a rare NDD called DEVLO (Developmental Delay, Language Impairment, and Ocular Abnormalities) (32), which is often characterized by microcephaly, a phenotype opposite of the macrocephaly found in PTEN-related NDDs (2), and consistent with the microcephalic phenotypes of a patient with PTEN duplication (33). We generated PTEN+ARPC4 double mutant WIBR3 NPs in a pooled manner by transducing PTEN homozygous null mutant WIBR3 NPs with lentiviruses expressing Cas9 and sgRNA against the exon 2 of ARPC4. This polyclonal population of transduced cells was selected with puromycin and assessed via Sanger sequencing to confirm partial ARPC4 genetic disruption (SI Appendix, Fig. S5). We further demonstrated that ARPC4 protein levels were significantly reduced compared to PTEN single mutants, returning to levels comparable to wild-type controls (Fig. 5 A and B). We also observed that other ARP2/3 complex subunits were significantly reduced to control levels, consistent with the knowledge that reduction of ARPC4 disrupts ARP2/3 complex formation (29–31) (Fig. 5C). Phalloidin staining demonstrated that PTEN+ARPC4 mutant human NPs displayed rescued levels of F-actin content (Fig. 5 D and E). We performed KI67 immunostaining and EdU incorporation assay to show that PTEN+ARPC4 mutant human NPs displayed rescued proliferation levels comparable to isogenic control NPs (Fig. 5 F–H).

Fig. 5.

Genetic inhibition of the ARP2/3 complex rescues PTEN mutant human NPs. (A–C) Immunoblotting analysis shows reduced ARPC4 protein levels in the PTEN+ARPC4 double-mutant WIBR3 human NPs. Note the concomitant reduction in other key components of the ARP2/3 complex (ARP2, ARP3, and ARPC1) in the PTEN+ARPC4 double-mutant WIBR3 human NPs. Each data point represents an independent experiment (different hPSC clone, independent transduction, or ARPC4-sgRNA, n = 3 to 6 for each group). P values are calculated using post hoc multiple comparison after ANOVA. (D and E) Representative images (D) and quantification (E) show that PTEN+ARPC4 double-mutants rescue F-actin levels in human NPs. (Scale bar, 20 μm.) Each data point represents a cell (n = 153 to 168 for each group from 3 to 6 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (F–H) Representative images (F) and quantification (G and H) show that PTEN+ARPC4 double-mutants rescue the overproliferation phenotypes of PTEN mutant human NPs, as measured by KI67 immunostaining and EdU incorporation assay. (Scale bar, 50 μm.) Each data point represents an independent experiment (different hPSC clone, independent transduction, or ARPC4-sgRNA, n = 4 to 8 for each group). P values are calculated using post hoc multiple comparison after ANOVA. Results are mean ± SEM. *P < 0.05, **P <0.01, ***P < 0.005, ****P < 0.0001.

Next, we performed a similar genetic inhibition experiment on human neurons. We transduced PTEN homozygous null mutant PGPC17 hPSC lines with sgRNA targeting ARPC4 to generate a pooled population of PTEN+ARPC4 mutant hPSCs. Sanger sequencing and immunoblotting confirmed ARPC4 genetic disruption (SI Appendix, Fig. S6) and protein reduction in ARPC4 as well as other ARP2/3 complex subunits (Fig. 6 A–C). Phalloidin staining showed that PTEN+ARPC4 mutant neurons displayed rescued F-actin content (Fig. 6 D and E), soma and neurite morphology (Fig. 6 F–K) and electrical activity (Fig. 6 L–N and Movie S2). We further validated these findings in the H1 genetic background. We transduced the iCas9-iNgn2 hPSCs with control sgRNA, sgRNA targeting PTEN alone, or dual-sgRNA targeting both PTEN and ARPC4, and confirmed ARPC4 disruption at the genetic and protein levels (SI Appendix, Figs. S7 and S8 A–C). ARPC4 reduction rescued the PTEN mutant phenotypes in F-actin content, neuronal morphology, and electrical activity (SI Appendix, Fig. S8 D–K and Movie S4).

Fig. 6.

Genetic inhibition of the ARP2/3 complex rescues PTEN mutant human neurons. (A–C) Immunoblotting analysis shows reduced ARPC4, as well as other key components of the ARP2/3 complex (ARP2, ARP3, and ARPC1) in the PTEN+ARPC4 double-mutant PGPC17 human neurons on day 10. Neurons are induced from PGPC17 hPSCs using Ngn2 induction. Each data point represents an independent experiment (different hPSC clone, independent transduction, or ARPC4-sgRNA, n = 3 to 6 for each group). P values are calculated using post hoc multiple comparison after ANOVA. (D and E) Representative images (D) and quantification (E) show that PTEN+ARPC4 double-mutants rescues F-actin levels in day 30 human neurons. (Scale bar, 20 μm.) Each data point represents a cell (n = 101 to 106 for each group from 3 to 6 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (F–H) Representative images (F) and quantification (G and H) show that PTEN+ARPC4 double-mutants rescue the soma and nucleus size of PTEN mutant human neurons on day 30. (Scale bar, 10 μm.) Each data point represents a cell (n = 127 to 130 for each group from 3 to 6 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (I–K) Sholl analysis shows that PTEN+ARPC4 double-mutants rescue the increased dendritic complexity and total dendritic length in PTEN mutant human neurons on day 30. ANOVA found significant effects of genotype and treatment (J, black asterisks marked on the right side of the graph). The post hoc multiple comparison test revealed differences at specific distance from cell soma, in which the black asterisk represents comparison between control and PTEN mutant neurons, and the red asterisk represents comparison between PTEN and PTEN+ARPC4 mutant neurons (K). Each data point represents a cell (n = 20 for each group from 3 to 6 independent experiments). P values are calculated using post hoc multiple comparison after ANOVA. (L–N). Multielectrode array analysis shows that PTEN+ARPC4 double-mutants rescue the electrical hyperactivity (M) and hypersynchronization (N) phenotypes of PTEN mutant human neurons. Raster plots (L) show spike distribution on day 30. The purple rectangle highlights the incidence of network burst. ANOVA found significant effects of genotype (black asterisks marked on the right side of the graphs). The post hoc multiple comparison test revealed differences on individual days of recording, in which the black asterisk represents a comparison between control and PTEN mutant neurons, and the red asterisk represents a comparison between PTEN and PTEN+ARPC4 mutant neurons. N = 4 for each group from 4 independent experiments. Results are mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001.

Together, these results corroborate our findings using ARP2/3 inhibitor and demonstrate that ARP2/3 is necessary for the development of cytoskeletal and functional abnormalities in PTEN mutant human NPs and neurons.

ARP2/3 Inhibition Rescues PTEN Mutant Neural Phenotypes Without Normalizing mTORC1 and mTORC2 Hyperactivation.

We showed that the increased ARP2/3 levels required the concomitant hyperactivation of mTORC1 and mTORC2. To place the ARP2/3 complex within this signaling cascade, we investigated whether ARP2/3 inhibition affects the upstream hyperactivation of mTORC1 and mTORC2. Immunoblotting for phosphorylation of S6 (Ser240/244) and AKT (Ser473) showed that CK-666 treatment did not rescue the hyperactivation of mTORC1 and mTORC2 in PTEN mutant WIBR3 NPs (Fig. 7 A–C), PGPC17 neurons (Fig. 7 G–I), and H1 neurons (SI Appendix, Fig. S9). Similarly, hyperactivation in these pathways persisted in PTEN+ARPC4 double mutant WIBR3 NPs and PGPC17 neurons (Fig. 7 D–F and J–L). Taken together, these findings support the model that ARP2/3 is a convergent downstream effector of mTORC1 and mTORC2 pathways (Fig. 7M).

Fig. 7.

ARP2/3 complex inhibitions rescue PTEN mutant neural phenotypes without normalizing mTORC1 and mTORC2 hyperactivation. (A–F) Immunoblotting analysis shows chronic CK-666 treatment (A–C) and PTEN+ARPC4 double-mutants (D–F) did not reduce the hyperactivation of mTORC1 (pS6 at Ser240/244) and mTORC2 (pAKT at Ser473) in PTEN mutant WIBR3 human NPs. Each data point represents an independent experiment (different hPSC clone, independent differentiation, or ARPC4-sgRNA, n = 4 for each group of CK-666 inhibition, n = 3 to 6 for each group of genetic inhibition). P values are calculated using post hoc multiple comparison after ANOVA. (G–L) Immunoblotting analysis shows chronic CK-666 treatment (G–I) and PTEN+ARPC4 double-mutants (J–L) did not reduce the hyperactivation of mTORC1 (pS6 at Ser240/244) and mTORC2 (pAKT at Ser473) in PTEN mutant PGPC17 human neurons on day 10. Neurons are induced from PGPC17 hPSCs using Ngn2 induction. Each data point represents an independent experiment (different hPSC clone, independent differentiation, or ARPC4-sgRNA, n = 4 for each group of CK-666 inhibition, n = 3 to 6 for each group of genetic inhibition). P values are calculated using post hoc multiple comparison after ANOVA. (M) Schematic diagram of the PTEN–mTOR–ARP2/3 signaling pathway. Results are mean ± SEM. **P < 0.01, ***P < 0.005, ****P < 0.0001.

Discussion

PTEN deficiency disorder belongs to a wide spectrum of NDDs known as mTORopathies, which are characterized by mTOR hyperactivation and overlapping clinical features, including macrocephaly, epilepsy, and autism (34, 35). Although mTOR hyperactivation is a hallmark of PTEN-associated NDDs, the downstream mechanisms linking this signaling abnormality to neural dysfunction remain poorly defined. Our previous work demonstrated that synergistic activation of both mTORC1 and mTORC2 is required to drive the morphological and electrophysiological phenotypes of PTEN-deficient human neural cells (9). In the current study, we identify the ARP2/3 complex as a downstream effector of mTORC1 and mTORC2 hyperactivation. Using isogenic hPSC models, we show that PTEN loss leads to a marked increase in F-actin content in both NPs and neurons, accompanied by elevated protein levels of RAC1 and multiple ARP2/3 subunits. Both pharmacological inhibition and genetic disruption of ARP2/3 rescued F-actin accumulation, precursor hyperproliferation, and neuronal hypertrophy and prevented the emergence of electrical hyperactivity in neurons. Hyperactivation of both mTORC1 and mTORC2 is required to drive these phenotypes, and ARP2/3 inhibition rescued them without affecting the elevated mTORC1 or mTORC2 activity, positioning ARP2/3 as a downstream effector of this dual pathway activation.

Our findings contribute to a growing recognition that cytoskeletal dysregulation is a key mechanistic driver of NDDs (11, 36). Previous studies in human patients have shown that loss-of-function mutations in ARP2/3 complex components, such as ARPC4, cause severe neurodevelopmental abnormalities, including microcephaly and developmental delays (32). Similarly, genetic ablation of ARP2/3 components in conditional knockout mouse models leads to profound defects in neuronal migration, dendritic branching, and synaptic development (37–39). These loss-of-function phenotypes emphasize the essential roles of ARP2/3 during key early stages of neurodevelopment. In contrast to this established literature, our work highlights the detrimental impact of increased ARP2/3 activity in human neural cells. Our study demonstrates that elevated ARP2/3 levels can contribute to neurodevelopmental pathology. This finding positions the ARP2/3 complex as a dosage-sensitive causal mechanism for NDDs, similar to other genes such as PTEN, MECP2, SHANK3, SCN2A, UBE3A, and RAC1. Supporting this concept, activating mutations in RAC1, a key regulator of ARP2/3, cause macrocephaly in human patients, likely by driving excessive ARP2/3 activation (12). Furthermore, increased gene dosage of CYFIP1, a regulator of RAC1, is thought to contribute to the macrocephaly, autism, and epilepsy phenotypes in 15q11-13 duplication syndromes (40). It is possible that elevated ARP2/3 activity enhances aberrant actin remodeling and neurite growth, leading to the electrical hyperactivity and increased network synchrony observed in PTEN mutant cultures. Future studies to systematically investigate the pathological impact of excessive ARP2/3 complex levels, alone or in the context of other NDDs, could yield further mechanistic insights and therapeutic opportunities.

We focused our mechanistic analysis of cytoskeleton abnormalities on the RAC1–ARP2/3 axis based on prior evidence showing that Pten loss in mouse fibroblasts leads to increased cell motility, driven in part by elevated Rac1 level (25). Additionally, a recent study in a Kras-driven pancreatic cancer mouse model showed that elevated Rac1–Arp2/3 activity downstream of mTOR hyperactivation plays a functional role in promoting tumor progression (29). However, our findings do not exclude the involvement of other cytoskeleton regulators in PTEN deficiency. For instance, Cdc42, another key Rho-GTPase important for filopodia formation and actin remodeling, has also been reported to be elevated in Pten knockout mouse fibroblasts (25), suggesting possible parallel mechanism of cytoskeletal dysregulation. Future studies can expand upon our current approach to investigate the contributions of these additional pathways.

The current study establishes a causal link from mTOR hyperactivation to elevated F-actin content and ARP2/3 levels in human neural cells. We showed that genetic inhibition of either mTORC1 or mTORC2, via the disruption of RAPTOR or RICTOR, respectively, is sufficient to rescue the F-actin content and ARP2/3 subunit levels. Previous work has reported that F-actin content is pathologically elevated in the aging Drosophila brain, a phenotype that can be rescued by the mTOR inhibitor rapamycin (22). Using an in vitro model of PTEN deficiency disorder, our current study now established hyperactivation of the mTOR–ARP2/3 axis as a pathogenic driver during human neurodevelopment. Given that mutations in other mTORopathy genes such as PIK3CA, AKT1/3, and MTOR also can activate both mTORC1 and mTORC2, it is possible that ARP2/3 dysfunction represents a shared downstream mechanism contributing to neuropathology across multiple mTOR-driven NDDs.

While mTOR hyperactivation is recognized as a key driver of PTEN-deficient neuropathology, mTOR inhibitors have shown limited therapeutic benefits for patients with PTEN deficiency, particularly in addressing neurocognitive symptoms (6, 7). This likely reflects the broad physiological roles of mTOR in both central and peripheral systems, as well as the high potential for dose-limiting side effects. Our study was motivated by the hypothesis that dissecting downstream effectors of mTOR signaling could yield more specific mechanistic insights and therapeutic targets. We demonstrate that inhibition of the ARP2/3 complex can ameliorate disease-relevant phenotypes in an in vitro model of PTEN deficiency NDD. This aligns with previous work in the aging Drosophila brain, where reducing elevated F-actin levels via siRNA slowed aging and extended lifespan (22). Although ARP2/3 inhibitors such as CK-666 (used in the current study) have been extensively tested and shown to suppress tumor cell migration and invasiveness in preclinical cancer models (29, 41–43), clinical translation of these inhibitors has yet to occur. Given the importance of ARP2/3 in fundamental cellular processes, including neuronal morphogenesis and synaptic function, direct therapeutic targeting of this complex requires careful evaluation. Moreover, it is likely that ARP2/3 dysfunction contributes to a subset, but not the entirety, of the neural phenotypes associated with PTEN deficiency. Therefore, its inhibition may offer only partial correction of the broad pathological spectrum. A cautious yet plausible direction for future research could involve evaluating combination therapies of inhibitors against mTOR, ARP2/3, or other related downstream effectors. Finally, while our study was conducted in vitro, future research testing ARP2/3 inhibition in vivo, including its ability to rescue seizures and neurocognitive phenotypes in Pten mutant animal models, will be essential to fully evaluate the therapeutic potential of this strategy. Taken together, our findings support the broader therapeutic concept that targeting downstream effectors of mTOR signaling, such as ARP2/3, may offer a more specific and effective strategy for PTEN-related and other mTOR-driven NDDs.

Materials and Methods

For a complete description of the materials and methods used in this study, please refer to SI Appendix, Materials and Methods. The human iPSC line PGPC17 and ESC lines WIBR3 and H1 were previously described (9, 23, 44). Clonal PTEN mutant, PTEN+RAPTOR double mutant, and PTEN+RICTOR double mutant WIBR3 ESC lines were generated in earlier studies (9, 10). In this study, CRISPR/Cas9-mediated gene editing was used to generate clonal PTEN homozygous mutant hPSCs in the PGPC17-iNgn2 background. Neuronal differentiation was performed through doxycycline-inducible expression of Ngn2.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

MEA movie, control and PTEN mutant PGPC17 human neurons treated with 100uM CK-666, day 30 post DOX induction.

Movie S2.

MEA movie, control, PTEN mutant, and PTEN+ARPC4 double mutant PGPC17 human neurons, day 30 post DOX induction.

Movie S3.

MEA movie, control and PTEN mutant H1 human neurons treated with 100uM CK-666, day 30 post DOX induction.

Movie S4.

MEA movie, control, PTEN mutant, and PTEN+ARPC4 double mutant H1 human neurons, day 30 post DOX induction.

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information.