TAOK2 drives opposing cilia length deficits in 16p11.2 deletion and duplication carriers

Summary

Deletion and duplication of the 16p11.2 genomic locus are associated with opposing changes in brain size. To determine cellular mechanisms that underlie these opposing phenotypes, we performed quantitative phosphoproteomic analyses of induced pluripotent stem cells (iPSCs)-derived neural progenitor cells (NPCs) obtained from unaffected individuals, 16p11.2 deletion, and 16p11.2 duplication carriers. Differentially phosphorylated proteins were enriched in centrosomal and cilia proteins. Deletion NPCs showed longer primary cilium compared to unaffected individuals, while stunted cilia were observed in duplication NPCs. Through cellular screens in NPCs, we determined the contribution of genes within the 16p11.2 locus to cilium length. Protein kinase TAOK2 and phosphatase PPP4C were found to regulate primary cilia length. NPCs lacking TAOK2 exhibited elongated cilia, aberrant IFT88 and pericentrin (PCNT) accumulation, and were impaired in sonic hedgehog (SHH) signaling. These findings implicate aberrant cilia length in the pathophysiology of 16p11.2 copy number variation and establish TAOK2 kinase as a regulator of primary cilium length.

Graphical abstract

Highlights

• •16p11.2 deletion and duplication carriers show reciprocal deficits in cilium length
• •TAOK2 and PPP4C negatively regulate primary cilia elongation
• •SHH deficits with aberrant IFT and pericentrin distribution in TAOK2 knockout NPCs
• •Catalytic activity of TAOK2 negatively regulates primary cilium elongation

Byeon, Ferreccio et al. reveal opposing changes in primary cilium length in 16p11.2 deletion and duplication iPSC-derived neuronal progenitors. TAOK2 and PPP4C encoded within the 16p11.2 locus control cilia length in a dose dependent manner. TAOK2 knockout NPCs have elongated cilia with aberrant accumulation of IFT88 and pericentrin resulting in impaired SHH signaling.

Introduction

The 16p11.2 genomic region between breakpoint 4 and 5 contains 30 annotated genes and is flanked by segmental duplications, making it prone to copy number variation (CNV) such as deletion and duplication (Weiss et al., 2008; Zufferey et al., 2012). 16p11.2 CNVs are associated with autism spectrum disorder (Weiss et al., 2008), schizophrenia (McCarthy et al., 2009), motor and speech impairment (Steinman et al., 2016), body weight disturbances (Jacquemont et al., 2011; Maillard et al., 2015; Walters et al., 2010; Zufferey et al., 2012), and structural abnormalities of the brain (Maillard et al., 2015; Martin-Brevet et al., 2018; Qureshi et al., 2014). Carriers of 16p11.2 deletion and duplication exhibit common developmental, behavioral, and social impairments implying convergent mechanisms caused by reciprocal genetic changes. On the other hand, certain opposing clinical manifestations in 16p11.2 deletion and duplication carriers, such as alteration in brain size and body mass index (BMI), provide a unique opportunity to determine the individual contributions of genes in this locus to pathophysiology of 16p11.2 CNV. Carriers of 16p11.2 deletion present with increased brain size or macrocephaly, including larger cortex and thicker corpus callosum; in contrast, duplication is associated with microcephaly, thinner corpus callosum, and smaller cortex (Martin-Brevet et al., 2018; Owen et al., 2014; Qureshi et al., 2014; Steinman et al., 2016). Also, 16p11.2 deletion carriers are prone to obesity (defined as BMI ≥30 kg/m²) while duplication carriers have low BMI (defined as ≤18.5 kg/m²) (Arbogast et al., 2016; Jacquemont et al., 2011; Walters et al., 2010; Zufferey et al., 2012). Mechanisms underlying these gene dosage-dependent phenotypes are unknown, and systematic efforts to map the genetic basis of these opposing effects are lacking. Even though the syntenic locus in mice offers an excellent genetic model to study this chromosomal disorder, phenotypes of deletion and duplication mice are opposing to the human condition (Arbogast et al., 2016; Portmann et al., 2014). However, human iPSC-based neuronal and organoid models derived from 16p11.2 deletion and duplication carriers have provided important insights into the pathophysiology of these disorders (Connacher et al., 2022; Deshpande et al., 2017; Fetit et al., 2023; Roth et al., 2020; Urresti et al., 2021). For instance, excitatory neurons derived from 16p11.2 deletion carriers show hypertrophy, implicating increased neuronal soma and arbor size as contributors to the macrocephaly phenotype (Deshpande et al., 2017).

To identify mechanisms underlying opposing phenotypes in 16p11.2 CNV, we utilized an unbiased proteomic approach. Our phosphoproteomic analyses of neural progenitor cells (NPCs) differentiated from 16p11.2 deletion, duplication, and control iPSCs revealed differentially phosphorylated proteins enriched in cilia and centrosomal proteins. Primary cilia are microtubule-based membrane protrusions rich in receptors for signaling molecules, and are present in most cell types including neuronal progenitors, neurons, and astrocytes (Anvarian et al., 2019; Lepanto et al., 2016; Singla and Reiter, 2006). Defects in ciliogenesis, ciliary length, and trafficking cause several human diseases including neurodevelopmental disorders (Bettencourt-Dias et al., 2011; Ferkol and Leigh, 2012; Guemez-Gamboa et al., 2014; Han and Alvarez-Buylla, 2010; Reiter and Leroux, 2017). Our study demonstrates that primary cilia are impacted in a gene dosage-dependent manner in 16p11.2 CNV, with longer primary cilium in 16p11.2 deletion carrier iPSC-derived NPCs and shorter primary cilium in duplication carrier iPSC-derived NPCs. Further, we identify the genetic drivers of opposing changes in cilia length in 16p11.2 CNV, and find Thousand And One amino acid Kinase 2 (TAOK2) to be the most significant and dosage-dependent regulator of cilia length.

TAOK2 is a protein kinase important for neuronal development, including dendritic spine development and synaptic function (Anda et al., 2012; Byeon and Yadav, 2024; Nourbakhsh and Yadav, 2021; Yadav et al., 2017). Mutations in TAOK2 are associated with autism spectrum disorder, and mutations impact both the catalytic and regulatory domains (Richter et al., 2018). The canonical isoform of TAOK2 encodes a transmembrane protein kinase localized to the endoplasmic reticulum (ER) and centrosomes, with ER-microtubule tethering property (Nourbakhsh et al., 2021). In this study, we find that TAOK2 knockout (KO) human iPSC-derived NPCs exhibit longer primary cilia, while TAOK2 overexpression leads to stunted cilia. Centrosomal accumulation of pericentrin (PCNT) and ciliary tip distribution of IFT88 are significantly increased in TAOK2 KO NPCs. Functionally, TAOK2 KO neuroprogenitors show deficits in sonic hedgehog (SHH) signaling. Finally, using pharmacological inhibition, we show that the catalytic activity of TAOK2 is essential for cilia length elongation. Our study reveals reciprocal alterations of cilia length in neural progenitors derived from 16p11.2 deletion and duplication carriers, elucidates the genetic basis for opposing changes in ciliary length, and reveals a hitherto unknown function of TAOK2 kinase in primary cilium regulation.

Results

Centrosomal and ciliary proteins are differentially phosphorylated in 16p11.2 CNV carrier-derived neural progenitors

Three of the 30 genes in the 16p11.2 genetic locus encode for proteins that impact protein phosphorylation: MAPK3 and TAOK2 are serine threonine kinases, and PPP4C is the catalytic subunit of a protein phosphatase (Figure 1A). Therefore, we hypothesized that the phosphoproteome in patients with duplication and deletion would be altered, and that identification of differentially phosphorylated proteins could reveal insight into the neuropathology of 16p11.2 CNV. IPSCs from fibroblasts of three deletion carriers, three duplication carriers, and three unaffected individuals were obtained (Figure S1A). Differentiation of iPSCs into dorsal forebrain NPCs was induced by dual SMAD inhibition using our previously published protocol (Figure S1B) (Deshpande et al., 2017). All NPCs expressed forebrain progenitor markers PAX6, NESTIN, SOX1, and SOX2 (Figures S1C and S1D). NPCs were grown to confluence, cell lysates were prepared, and equal amounts of protein were utilized for downstream proteomics. Each sample was chemically labeled using the 10plex tandem mass tag (TMT) reagent. Phosphopeptides were enriched using TiO₂ column, eluted and analyzed by mass spectrometry (Clauser et al., 1999; Guan et al., 2011) (Figure 1B). Mass spectrometry analyses detected 18,894 distinct phosphopeptides from 1,927 proteins, of which 43 phosphopeptides from 40 proteins were significantly increased (greater than 2-fold increase) in duplication NPCs, and 304 phosphopeptides from 213 proteins were significantly decreased (greater than 2-fold decrease) in deletion NPCs compared to control (Figures 1C and 1D; Table S1). Of the phosphoproteins downregulated in deletion carriers, 13.6% (29 out of 213) proteins had a known cilia and centrosomal function based on the Centrosome and Cilium Database (CCDB) (Gupta et al., 2015) (Figure 1D; Table S2).

Figure 1.

Quantitative phosphoproteomics of neural progenitors from 16p11.2 CNV carriers identifies ciliary proteins

(A) Schematic representations of annotated genes within the human 16p11.2 genomic locus (orange) that undergo heterozygous deletion or duplication. Genes encoding kinases MAPK3 and TAOK2 and phosphatase PPP4C are highlighted in red.

(B) Workflow for tandem mass tag (TMT) labeling of protein lysates isolated from dorsal neural progenitor cells (NPCs) differentiated from iPSCs derived from patients with 16p11.2 deletion and duplication as well as unaffected control individuals. Labeled peptides were enriched for phosphopeptides using TiO₂ affinity beads followed by identification through mass spectrometry. For proteomics experiments, one clone per patient sample with two to three patient samples per genotype were analyzed. Internal controls with clonal repeats were included. Created in BioRender. Byeon, S. (2025) https://BioRender.com/6qww2wz.

(C) Volcano plot shows phosphopeptides significantly downregulated (blue) in 16p11.2 deletion samples compared to controls.

(D) Phosphopeptides significantly downregulated in 16p11.2 deletion NPCs were enriched in centrosomal and ciliary proteins with 13.6% of proteins present in the Centrosome and Cilium DataBase (CCDB) and 60.5% proteins classified as cilia proteins in GeneCards.

Primary cilia in 16p11.2 deletion and duplication patient-derived neural progenitor cells exhibit opposing defects in length

To test whether cilia were perturbed in 16p11.2 deletion and duplication carriers, we immunostained NPCs with antibodies against the cilia membrane protein ARL13B and NPC marker PAX6. We found that compared to NPCs from unaffected individuals (mean = 2.36-2.55 μm, n = 60 each line), cilia in NPCs from patients with deletion (mean = 3.65–4.79 μm, n = 60 each line) were significantly longer , while cilia in NPCs from patients with duplication were shorter (mean = 1.34–1.83 μm, n = 60 each line) (Figures 2A and 2B). Cilia length is coupled to the cell cycle (Kasahara and Inagaki, 2021). Cilia elongate throughout the G1-S phase, disassemble in M phase, and regrow in G1. NPCs were grown in serum free media for 24 h to synchronize the cell cycle. We found that the reciprocal change in cilia length in NPCs from deletion and duplication carriers persisted, even when NPCs were grown in serum free media for 24 h (Figures 2A–2C). NPCs responded to serum starvation by increasing their ciliary length, with cilia in deletion NPC (mean = 4.16–4.97 μm, n > 33 each line) significantly longer, and cilia in duplication NPCs (mean = 2.01–2.39 μm, n > 45 each line) shorter compared to control (mean = 2.96–3.40 μm, n > 38 each line). This suggests that cilia length deficits observed in 16p11.2 CNV carriers arise through a cell intrinsic mechanism independent of cell cycle phase.

Figure 2.

Opposing cilia length alterations in neural progenitors derived from 16p11.2 deletion and duplication carriers

(A) Representative confocal images of human iPSC-derived NPCs from unaffected individuals (RMK0119, RMK0162, and RMK0165), 16p11.2 duplication carriers (16pDup03, 16pDup04, and 16pDup05), and 16p11.2 deletion carriers (16pDel06, 16pDel07, and 16pDel08) immunostained for dorsal forebrain progenitor marker PAX6 (red) and primary cilium marker ARL13B (green). Nuclei were stained with DAPI. Scale bar represent 3 μm.

(B) Length of primary cilia in human iPSC-derived NPCs from unaffected control, and 16p11.2 deletion and duplication carriers, growing in nutrient rich conditions is plotted (n = 60 cilia from 3 experiments for each NPC line, ^∗∗∗∗p < 0.0001 and n.s. not significant, by ordinary one-way ANOVA with Dunnett’s multiple comparisons).

(C) Length of primary cilia in NPCs from unaffected control, and 16p11.2 deletion and duplication carriers, grown in starvation conditions for 24 h before fixation is plotted (n > 33 cilia from 3 experiments for each NPC line, ^∗∗∗∗p < 0.0001, ^∗∗∗p < 0.001, and n.s. not significant, by ordinary one-way ANOVA with Dunnett’s multiple comparisons). For primary cilia length, two clones per line were used and data were pooled from three experimental replicates. Three patient samples per genotype were used, and each genotype is plotted separately. All error bars are standard error of the mean.

Cellular screens identify genetic basis of altered cilium length

To uncover mechanisms underlying altered ciliary length caused by opposing gene dosage changes in the 16p11.2 locus, we performed complementary shRNA and overexpression cellular screens. First, we employed an shRNA screen in NPCs differentiated from the human iPSC line WTC11 (Coriell Institute) and RMK0119b line (UCSF, Weiss Lab) using commercially validated shRNA targeting 23 annotated genes in the 16p11.2 locus. NPCs were transfected with GFP-tagged shRNA, starved for 24 h to induce ciliogenesis, fixed and then immunostained with ARL13B antibody 48 h after transfection (Figure 3A). We found that shRNA against TAOK2 (mean = 5.43 μm, n = 26 NPCs, SEM = 0.32) and PPP4C (mean = 4.35 μm, n = 27 NPCs, SEM = 0.27) significantly increased cilia length compared to control shRNA (mean = 3.43 μm, n = 29 NPCs, SEM = 0.15) in WTC11-derived NPCs (Figures 3B and 3C). Similarly, in the RMK0119b-derived NPCs, shRNA against TAOK2 (mean = 4.82 μm, n = 24 NPCs, SEM = 0.25) and PPP4C (mean = 4.60 μm, n = 16 NPCs, SEM = 0.43) significantly increased the cilia length compared to control shRNA (mean = 3.21 μm, n = 19 NPCs, SEM = 0.19) (Figure 3D). Next, we tested in an overexpression screen, the impact of increased gene expression on cilia length. Briefly, we generated mammalian expression vector with a BFP-tag inserted at the N-terminal of TAOK2 and PPP4C along with six other genes in the 16p11.2 locus. NPCs derived from WTC11 line were transfected with either control-BFP construct or constructs expressing the 16p11.2 genes, starved for 24 h, and immunostained with ARL13B antibody 48 h after transfection (Figure 3E). Overexpression of TAOK2 had the strongest effect in decreasing cilia length (mean = 1.66 μm, n = 25 NPCs, SEM = 0.08), followed by PPP4C (mean = 2.39 μm, n = 19 NPCs, SEM = 0.15), while control (mean = 3.19 μm, n = 22 NPCs, SEM = 0.16) and none of the other six 16p11.2 genes had a significant effect on cilia length (Figure 3F). Therefore, gene dosage of TAOK2, and to a lesser extent PPP4C, drives opposing cilia length changes in NPCs derived from 16p11.2 CNV carriers.

Figure 3.

Cellular screens identify genetic drivers of altered ciliary length in 16p11.2 CNV

(A) Workflow for shRNA screen involves transfection of WTC11 or RMK0119 iPSC-derived NPCs with shRNA expression constructs (GFP), followed by growth under starvation conditions for 24 h before fixation. NPCs are immunostained for ARL13B to mark cilia and 60× images were acquired through confocal microscopy.

(B) Representative images of WTC11-derived NPCs transfected with control shRNA (green) and TAOK2-shRNA (green) stained for cilium marker ARL13B (magenta). Scale bar represents 2 μm.

(C) Quantification of cilium length in WTC11-derived NPCs transfected with the indicated shRNA against genes in the 16p11.2 genomic locus (n > 20 NPCs from 3 experiments for each shRNA, ^∗∗∗∗p < 0.0001, ^∗∗p < 0.005, and ordinary one-way ANOVA with Dunnett’s multiple comparisons).

(D) Quantification of cilium length in RMK0119-derived NPCs transfected with the indicated shRNA against genes in the 16p11.2 genomic locus (n > 15 NPCs from 3 experiments for each shRNA, ^∗∗∗∗p < 0.0001, ^∗∗p < 0.005, and ordinary one-way ANOVA with Dunnett’s multiple comparisons).

(E) Representative images of WTC11-derived NPCs transfected with BFP empty vector (blue) and BFP-TAOK2 (blue) stained for cilium marker ARL13B (magenta). Scale bar represents 3 μm.

(F) Quantification of cilium length in WTC11-derived NPCs transfected with BFP-tagged protein expression constructs for the indicated genes in the 16p11.2 locus (n > 12 NPCs from 3 experiments for each indicated expression construct, ^∗∗∗∗p < 0.0001, ^∗p < 0.05, and ordinary one-way ANOVA with Dunnett’s multiple comparisons). All error bars are standard error of the mean.

Cilium length deficits in TAOK2 knockout neural progenitors

TAOK2 is a serine threonine kinase localized to the ER (Nourbakhsh et al., 2021) that is important for neuronal development (Anda et al., 2012; Byeon and Yadav, 2024; Yadav et al., 2017). In mitotic cells, TAOK2 localizes strongly to spindle poles (Koo et al., 2017; Nourbakhsh et al., 2021). We used an antibody against TAOK2 to determine its distribution in ciliated NPCs. TAOK2 was present on the ER membrane in a punctate manner as expected, but was also strongly concentrated at the centrosomes and the base of cilia marked by ARL13B antibody (Figure 4A). To further test the role of TAOK2 in regulating primary cilium length, we generated TAOK2 KO iPSC lines using gene editing. We used the AICS0032 stem cell line (Allen Institute), where the centrosomal protein centrin-2 (CETN2) is tagged with RFP at its endogenous locus, allowing exquisite visualization of centrosomes (Figures S2A and S2B). AICS0032 iPSCs were genetically edited using Cas9 along with guide RNA targeting exon4 to generate TAOK2 KO cell lines 4.4 and 4.9 (Figure S2C). Successful editing was confirmed by sequencing and KO was validated by western blot (Figures S2D and S2E). Unedited wild-type (WT) AICS0032 iPSCs were used as a control. All three lines were karyotypically normal (Figure S2F), and iPSCs retained pluripotency as evident by expression of NANOG and OCT4 (Figure S2G). Dorsal forebrain NPCs were obtained from iPSCs through dual SMAD inhibition using our published protocol (Deshpande et al., 2017). TAOK2 KO and WT control NPCs expressed neuronal progenitor markers PAX6, NESTIN (Figure S3A), as well as FOXG1 and SOX2 (Figure S3B). Loss of TAOK2 protein in NPCs was confirmed by western blot (Figure 4B).

Figure 4.

TAOK2 kinase localizes to the base of the primary cilium and restricts cilium elongation

(A) Human iPSC-derived NPCs stained for endogenous TAOK2 (magenta) and cilia marker ARL13B (green) show ER localization of TAOK2 as well as its localization at the base of the primary cilia. Nuclei were stained with DAPI. Scale bar represents 5 μm. Magnified inset is shown on the right with individual frames of the z stack acquired at 0.3 μm. Scale bar represents 1 μm.

(B) Validation of TAOK2 KO in NPCs by western blot analyses.

(C) Neural rosettes from control WT and TAOK2 KO NPCs expressing RFP tagged centrin-2 (green) were stained for apical polarity marker ZO1 (white) and NPC marker PAX6 (red). Scale bar represents 10 μm.

(D) NPCs from control WT and TAOK2 KO NPCs expressing RFP-centrin-2 (red) were stained for ARL13B (white). Nuclei were stained with DAPI (cyan). Scale bar represents 3 μm.

(E) Quantification of cilium length in control WT (n = 260 NPCs from 3 experiments), TAOK2 KO 4.4 (n = 151 NPCs from 3 experiments, ^∗∗∗∗p < 0.0001) and KO 4.9 (n = 170 NPCs from 3 experiments, ^∗∗∗∗p < 0.0001, and one-way ANOVA with Dunnett’s multiple comparison) NPCs immunostained with ARL13B. All error bars are standard error of the mean.

Neural progenitors from control and TAOK2 KO iPSCs self-assembled into rosettes and maintained cell polarity, with enrichment of ZO1 and N-cadherin on the apical end of the rosettes (Figures 4C and S3C), and primary cilia projecting toward the apical side of the rosette (Figure 4D). To test whether primary cilium length is altered in TAOK2 KO NPCs, we immunostained NPCs with ARL13B (Figures 4D and 4E). We found a significant increase in cilia length in TAOK2 KO4.4 (mean = 3.52 μm, n = 151 NPCs, SEM = 0.08) and TAOK2 KO4.9 NPCs (mean = 3.35 μm, n = 170 NPCs, SEM = 0.06), compared to WT (mean = 2.14 μm, n = 260 NPCs, SEM = 0.04) (Figure 4E).

TAOK2 knockout neural progenitors exhibit ciliary trafficking deficits

Dedicated trafficking machinery comprising intraflagellar transport (IFT) proteins mediates anterograde and retrograde motor-based transport of cargo in and out of the primary cilium (Ishikawa and Marshall, 2011; Pan and Snell, 2014). We first tested whether the distribution of the centrosomal scaffold protein PCNT, important for trafficking of IFT bound cargo (Jurczyk et al., 2004; Salisbury, 2004) and cilium elongation (Nakazato et al., 2023), was disrupted in TAOK2 KO NPCs. WT and TAOK2 KO NPCs were fixed and immunostained for endogenous PCNT and ciliary membrane marker ARL13B (Figure 5A). Mean fluorescence intensity of PCNT within a 5 μm radius of the CETN2-RFP tagged centrosomes was significantly increased in TAOK2 KO4.4 (mean = 336.4, 220 NPCs from n = 3 experiments, SEM = 23.22) and TAOK2 KO4.9 NPCs (mean = 354.4, 220 NPCs from n = 3 experiments, SEM = 29.60) compared to WT (mean = 204, 220 NPCs from n = 3 experiments, SEM = 8.10) cells (Figure 5B). Next, we tested whether IFT is maintained in TAOK2 KO cells by assessing the distribution of anterograde IFT-B component IFT88. NPCs were fixed and immunostained for endogenous IFT88 and ciliary membrane marker ARL13B (Figure 5C). IFT88 particles are present at the base of the cilia and distributed along the length of the cilia, indicating normal transport. Accumulation at the cilia tip typically occurs in IFT-A mutants and indicates defective balance between retrograde and anterograde transport (Qin et al., 2011; Weijman et al., 2024). In WT condition, 9.76% and 16.37% of NPCs had IFT88 enriched at the ciliary base and tip, respectively, while in a majority (73.87%) of NPCs, IFT88 was uniformly distributed along the ciliary length (482 NPCs from n = 3 experiments). In contrast, IFT88 was enriched in the cilia tip in 61.95% of TAOK2 KO4.4 and 52.8% of KO4.9 NPCs, while it was uniformly distributed in 33.44% of TAOK2 KO4.4 and 35.43% of KO4.9 NPCs (460 and 595 NPCs, respectively from n = 3 experiments) (Figure 5D). These data show that TAOK2 regulates PCNT and IFT88 trafficking in the primary cilium, with loss of TAOK2 resulting in increased centrosomal PCNT and IFT88 accumulation at the primary cilium tip. We also tested whether IFT88 trafficking is impacted in 16p11.2 deletion carriers by immunostaining control and deletion NPCs for IFT88 and ARL13B (Figure 5E). We found IFT88 was enriched in the cilia tip in 57.90% of deletion and 27.57% of control NPCs, while it was uniformly distributed in 28.03% of deletion and 58.36% of control NPCs (679 and 639 NPCs, respectively from n = 3 experiments) (Figure 5F).

Figure 5.

TAOK2 KO and 16p11.2 deletion NPCs show aberrant IFT88 accumulation at ciliary tip

(A) NPCs derived from control WT, TAOK2 KO 4.4, and 4.9 iPSC tagged with RFP-centrin-2 (blue) were immunostained for pericentrin (PCNT) (yellow) and cilia marker ARL13B (magenta). Bottom row shows PCNT immunostaining in inverted grayscale. Scale bar represents 5 μm.

(B) Fluorescence intensity of PCNT staining within a 5 μm radius around the centrosome was measured in control WT and TAOK2 KO 4.4 and 4.9 NPCs (220 NPCs each genotype from n = 3 experiments, ^∗∗p < 0.005, and repeated measures one-way ANOVA with Dunnett’s multiple comparisons).

(C) Control and TAOK2 KO4.4 and 4.9 NPCs tagged with RFP-centrin-2 (blue) were immunostained for IFT88 (yellow) and cilia marker ARL13B (magenta). Bottom row shows merged images of IFT88 (yellow) with endogenous centrin-2 (blue) that marks the base of the primary cilium. Scale bar represents 5 μm.

(D) Percentage of control WT NPCs (482 NPCs from n = 3 experiments), TAOK2 KO 4.4 NPCs (460 NPCs from n = 3 experiments), and TAOK2 KO4.9 NPCs (595 NPCs from n = 3 experiments, ^∗∗∗∗p < 0.0001, and two-way ANOVA with Tukey’s multiple comparisons) with IFT88 enrichment at the cilium base, cilium tip, or uniform distribution along the length of the cilia are plotted.

(E) Control NPCs (RMK0119, RMK0162, and RMK0165) and 16p11.2 deletion carrier NPCs (16pDel06, 16pDel07, and 16pDel08) were immunostained for IFT88 (yellow), cilia marker ARL13B (magenta), and PCNT (magenta).

(F) Percentage of control NPCs (639 NPCs from n = 3 experiments) and deletion carrier NPCs (679 NPCs from n = 3 experiments, ∗∗∗∗p < 0.0001, and two-way ANOVA with Sidak's multiple comparisons) with IFT88 enrichment at the cilium base, cilium tip, or uniform distribution along the length of the cilia are plotted. Error bars are standard error of the mean.

Dysregulated SHH signaling in TAOK2 knockout neuroprogenitor cells

IFT dysfunction is associated with impaired SHH signaling as IFT proteins play key roles in cilia structure and ciliary trafficking (Huangfu et al., 2003; Liem et al., 2012; Tran et al., 2008). SHH is a key signaling pathway important for brain development including the forebrain (Altaba et al., 2002). Upon SHH ligand binding, the repressor protein patched1 (PTCH1) is removed from cilia, resulting in ciliary activation of smoothened (SMO), which turns on transcription of SHH target genes such glioma-associated oncogene (GLI) (Rohatgi et al., 2007). First, we tested if ciliary accumulation of SMO was affected by absence of TAOK2. WT and TAOK2 KO NPCs were fixed and immunostained with antibody against SMO and ARL13B, and proportion of NPCs that had SMO present in the primary cilia was quantified (Figure 6A). We found a significant increase in percentage of NPCs with ciliary SMO distribution, with 34.69% of TAOK2 KO 4.4 NPCs (1,033 NPCs from n = 3 experiments, SEM = 0.77) and 39.22% of TAOK2 KO 4.9 NPCs (1,202 NPCs from n = 3 experiments, SEM = 3.39) showing ciliary SMO compared to 17.58% (1,147 NPCs from n = 3 experiments, SEM = 0.83) WT NPCs (Figure 6B). Since presence of SMO within the cilia impacts downstream SHH signaling, we next measured the transcriptomic outcome of SHH activation by small molecule SMO agonist SAG that activates SHH signaling independently of PTCH protein. We performed quantitative PCR to measure mRNA levels of SHH target genes in WT and TAOK2 KO NPCs upon 24 h treatment with 0.5 μM SMO agonist SAG. We found a significant upregulation of GLI1 expression in TAOK2 KO4.4 and KO4.9 NPCs compared to WT NPCs (n = 7–9), whereas there was no significant change in PTCH1 mRNA level (Figures 6C and 6D). Therefore, consistent with the upregulation of SMO within the cilia in TAOK2 KO NPCs, there was increased activation of SHH dependent GLI1 transcription compared to control NPCs.

Figure 6.

TAOK2 is important for SHH signaling

(A) Control WT and TAOK2 KO 4.4 and 4.9 NPCs expressing RFP-centrin-2 (blue) were immunostained for SMO (yellow) and ARL13B (magenta). Scale bar represents 2 μm.

(B) Quantification of primary cilia positive for SMO in control WT NPCs (1,147 NPCs from n = 3 experiments), TAOK2 KO 4.4 (1,033 NPCs from n = 3 experiments, ^∗∗p = 0.002), and TAOK2 KO 4.9 (1,202 NPCs from n = 3 experiments, ^∗∗∗p = 0.0006, one-way ANOVA with Dunnett’s multiple comparisons) NPCs.

(C and D) Bar graph plots of individual values and geometric means of mRNA levels of GLI1 and PTCH1 in SAG treated TAOK2 KO 4.4 and 4.9 NPCs normalized to WT NPCs (^∗∗∗∗p < 0.0001, mixed effects analyses with Geisser-Greenhouse correction). Error bars are standard error of the mean.

TAOK2 catalytic activity is required for its role in cilium length control

To determine whether TAOK2 controls cilium length through phosphorylation, we utilized a small molecule TAO kinase inhibitor compound 43 (CP43) with an IC₅₀ of 15 nM (Giacomini et al., 2018; Koo et al., 2017). WT NPCs were treated with control vehicle DMSO or CP43 at 10 μM for 3 h. After drug treatment, cells were fixed and immunostained for cilia marker ARL13B. We found that NPCs treated with CP43 showed a robust increase in cilia length (mean = 3.13 μm, n = 100, SEM = 0.06) compared to DMSO treated controls (mean = 2.29 μm, n = 100, SEM = 0.07) indicating that TAOK2 kinase activity is important for its role in cilia length control (Figures 7A and 7B). These data reveal that TAOK2 catalytic activity controls cilium length, and its acute inhibition leads to increased cilia length, phenocopying the effect of TAOK2 KO.

Figure 7.

TAOK2 catalytic activity is important for its role in ciliary growth restriction

(A) Control WT NPCs endogenously tagged with RFP-centrin-2 (magenta) were treated with DMSO or TAO kinase inhibitor CP43 (10 μM for 3 h), fixed and immunostained for ARL13B (yellow) and nuclear dye DAPI (blue). Scale bars represent 5 μm. Inset on right shows magnified representative cilia treated with DMSO or CP43. Scale bar represents 2 μm.

(B) Schematic shows the chemical structure and IC₅₀ of CP43. Cilia length in NPCs treated with vehicle DMSO or CP43 is plotted (n = 100 NPCs each condition from 3 experiments, ^∗∗∗∗p < 0.0001, and Student’s t test with Welch correction). Error bars are standard error of the mean.

Discussion

Human diseases caused by mutations in genes affecting centrosomes and primary cilia, called ciliopathies, are commonly associated with brain development deficits (Guemez-Gamboa et al., 2014; Han and Alvarez-Buylla, 2010). Also, disruption in cilia length has been observed in several neurological disorders. Cells derived from patients with Down syndrome exhibit shorter cilia, aberrant centrosomal accumulation of PCNT and impaired trafficking to the cilium (Galati et al., 2018). Aberrant ciliogenesis is seen in patients with focal malformations of cortical development caused by somatic activating mutations in mTOR (Park et al., 2018). Seckel syndrome, caused by mutations in the centrosomal-P4.1-associated protein (CPAP) is associated with microcephaly. NPCs derived from patients with Seckel syndrome exhibit extremely long cilia and undergo premature neuronal differentiation (Gabriel et al., 2016). At the outset, several lines of evidence pointed to a potential role of ciliary dysfunction in 16p11.2 CNV. First, transcriptomic profiles of individuals carrying reciprocal 16p11.2 CNV were enriched for genes associated with ciliopathies (Migliavacca et al., 2015). Second, radial glial cells in cerebral organoids generated from 16p11.2 CNV carriers show differentially expressed cilium assembly genes (Urresti et al., 2021), and lastly, patients with 16p11.2 CNV exhibit BMI and metabolic imbalances (Jacquemont et al., 2011; Maillard et al., 2016; Walters et al., 2010; Zufferey et al., 2012), phenotypes often found in patients with ciliopathy (Oh et al., 2015; Reiter and Leroux, 2017; Ritter et al., 2018). Reciprocal changes in cilia length in NPCs derived from 16p11.2 deletion and duplication carriers were identified in our study through unbiased proteomics. Given the incomplete penetrance of 16p11.2 clinical symptoms (Arbogast et al., 2016; Weiss et al., 2008), if such an association of cilia length is made with brain size, cilium length alteration could serve as a valuable biomarker for 16p11.2 CNV.

During development of the mammalian neocortex, polarized radial glial cells are organized with their apical plasma membranes forming the luminal side of the neural tube, such that the primary cilia extend into the ventricular cerebrospinal fluid rich in signaling molecules (Willaredt et al., 2008). Enriched within the primary cilium are numerous signaling receptors that sense growth factors and ligands important for brain development such as insulin-like growth factors, bone morphogenic proteins, SHH, retinoic acid, and lysophosphatic acid (Han and Alvarez-Buylla, 2010; Reiter and Leroux, 2017). Cilia length dynamics are observed during neuronal differentiation, with the circadian rhythm and in response to growth factors. Through its critical signaling functions and dynamic regulation of length that dictates receptor availability, the primary cilium serves as key regulator of neural stem cell proliferation, neuronal differentiation, migration as well as site of specialized axon-cilium synapse (Ding et al., 2019; Han and Alvarez-Buylla, 2010; Han et al., 2009; Phua et al., 2017). Changes in cilia length are often associated with microcephaly. CPAP mutations lead to elongated cilia and microcephaly through premature neurogenesis (Gabriel et al., 2016). Mutations in CENPF, a centrosomal protein, also lead to microcephaly but result in shortened cilia (Cappuccio et al., 2022). Interestingly, mutations in EML1, a microtubule-associated protein, lead to shorter cilia and are associated with macrocephaly (Bizzotto et al., 2017; Oegema et al., 2019; Uzquiano et al., 2019). Whether and how primary cilium length deficits contribute to opposing brain size changes observed in 16p11.2 CNV remain to be investigated. Multiple studies have focused on the impact on NPC proliferation in 16p11.2 CNV using stem cell models with variable results. Studies have reported no differences in NPC proliferation in 16p11.2 CNV carriers using 2D cultured NPCs (Deshpande et al., 2017; Roth et al., 2020), hyperproliferation in deletion NPCs (Connacher et al., 2022), prolonged cell cycle in deletion 3D organoid models (Fetit et al., 2023), or alterations in the ratio of neurons to progenitors in 3D organoids during early neurogenesis (Urresti et al., 2021). Heterogeneity in the model systems and methods of assessing proliferation could contribute to these disparate findings.

Our results show that TAOK2 gene dosage regulates cilia length and that NPCs lacking TAOK2 have elongated cilia, increased centrosomal PCNT, and accumulation of IFT88 at distal cilia tip. Moreover, SHH signaling is dysregulated in TAOK2 KO NPCs. A previous study found that in neuroepithelial cells SHH signaling upregulates PCNT expression, which accumulates at the centrosome and recruits the protein kinase A (PKA) to centrosomes, thereby exerting a positive-feedback onto SHH signaling (Saade et al., 2017). Therefore, it is possible that increased PCNT that we observe in TAOK2 KO NPCs is contributing to the increase in GLI1 activity downstream of SMO. Upregulation of SHH signaling has been noted in Nestin-Gpr161 conditional KO mice that leads to increased progenitor proliferation causing ventriculomegaly (Shimada et al., 2019). Other outcomes of high SHH signaling include dysregulated neuroepithelial cell to radial glia transition and aberrant cell cycle exit in neuroprogenitors, each of which could lead to an increase in brain size (Chenn and Walsh, 2002; Sahara and O’Leary, 2009; Shimada et al., 2019). Notably, TAOK2 is the only gene within the 16p11.2 locus, single nucleotide missense mutations in which have been associated with both obesity (Agrawal et al., 2021; Elkhateeb et al., 2024; Loid et al., 2022) and macrocephaly (Elkhateeb et al., 2024). Whether and how the ciliary role of TAOK2 contributes to obesity and macrocephaly is an exciting direction for future investigation.

Limitations of the study

Whether ciliary deficits exist broadly in other cell types within the brain and in other tissues in 16p11.2 CNV carriers needs to be examined. This study was based on samples from three distinct 16p11.2 deletion and three duplication carriers, with heterogeneous age and sex distribution, and distinct genetic background. Future studies that employ cilia length and function analyses in larger number of 16p11.2 CNV carriers or genetically edited iPSC lines harboring 16p11.2 BP4-5 CNV would be insightful.

Methods

Induced pluripotent stem cell culture and neuronal differentiation

All iPSC lines were differentiated into NPCs as reported previously (Deshpande et al., 2017). Stem cells were treated with 1 U/mL dispase, lifted using a cell scraper, resuspended in mTeSR media (STEMCELL Technologies) containing 10 μM ROCK Inhibitor (Selleck Chemical), and then transferred to uncoated T25 flasks for embryoid body formation. After 24–48 h (day 0), EBs were gravity settled and transferred to new uncoated T25 flasks in neural media (DMEM/F12 (Gibco), 1% N-2 supplement (Gibco), 1% non-essential amino acids (Gibco), 2 μg/mL Heparin, and 1% penicillin/streptomycin [Gibco]) containing small molecule TGF-β/SMAD inhibitors, SB431542 (5 μM; PeproTech), and LDN193189 (0.25 μM; PeproTech). On day 3, EBs were transferred to Matrigel (1:50 dilution; Corning)-coated 6-well plates in fresh neural media with SB and LDN. Media was changed with neural media without SB and LDN every other day or as needed until day 11 for formation of neural rosettes. On day 11, neuroepithelia were lifted using cell scraper and transferred to uncoated T25 flasks in neural media. Neurospheres were fed with neural media every other day, or as needed, until day 25. On day 25, neurospheres were plated onto Matrigel-coated plates in STEMDiff Neural Progenitor Media and NPCs were obtained.

Proteomic sample preparation

Confluent 75mL flasks of NPCs were washed with PBS and then homogenized in 1 mL 8 M urea, using a probe sonicator (Thermo Fisher Scientific). Protein content in extracts was determined using Micro BCA Protein Assay Kit (Thermo Fisher Scientific). Aliquots containing 500 μg protein in 590 μL 8 M urea were added with 236 μL of 200 mM ammonium bicarbonate buffer and phosphatase inhibitors (Sigma Phosphatase Inhibitor Cocktails 1 and 3), then treated with 5 mM DTT at 56°C for 15 min, followed by a 30 min incubation at room temperature in dark with 7.5 mM iodoacetamide. For tryptic digestion, the samples were then diluted 4-fold with 100 mM ammonium bicarbonate to reduce urea to 2 M and then 5% (W/W) trypsin (Promega) was added. The pH was adjusted to 8.0 with 250 mM ammonium bicarbonate, and samples were incubated 12 h at 37°C. Another aliquot of trypsin was added (2% W/W) and digested for additional 6 h. The digests were desalted using a MAX-RP SepPak C18 cartridge (Waters) following manufacturer’s protocol. Sep Pak eluates were dried-evaporated in preparation for labeling with TMT10plex reagents. TMT labeling, phosphopeptide enrichment, mass spectrometry, and peptide identification are described in the supplemental information.

Generation of TAOK2 knockout stem cell line

Two independent TAOK2 KO lines were generated using CRISPR-Cas9 genome editing in an AICS0032 stem cell line. Guide RNA was designed using Synthego design tool (https://www.synthego.com) to target coding exon4. Guide was cloned into CrisprV2pSpCas9(BB)-2A-Puro (PX459) V2.0 (Addgene no. 62988). Cells were passaged in single cell suspension and plated at 50% confluence. Cultures were then transfected Lipofectamine 2000 Transfection reagent (Invitrogen) and 2 mg of guide RNA used per KO line. Cells were then selected with Puromycin for 48 h to select for transfected cells. Edited cells were expanded into single cell colonies. Genomic DNA was extracted and the region around the cutting site was PCR amplified to send for sequencing. Successful TAOK2 KO clones 4.4 and 4.9 were confirmed by Sanger sequencing analysis, and absence of encoded protein was validated by western blot (described in supplemental information). Absence of large chromosomal alterations was confirmed by karyotyping.

Confocal microscopy

Imaging was performed on a Nikon Ti2 Eclipse-CSU-X1 confocal spinning disk microscope equipped with four laser lines (405, 488, 561, and 670 nm) and an sCMOS Andor camera for image acquisition. The microscope was caged within the OkoLab environmental control setup enabling temperature and CO₂ control during live imaging. Imaging was performed using Nikon 1.49 100× Apo 60× or 40× oil objective lenses. All image analyses and quantification of fluorescence intensity was performed using Fiji ImageJ.

Statistics

Statistics were performed in GraphPad software Prism 10.0. Multiple groups were analyzed using ANOVA, while two group comparisons were made using unpaired t test unless otherwise stated. Statistically p value less than 0.05 was considered significant. All experiments were done in triplicate, and experimental sample size and p values are indicated with the corresponding figures.

Resource availability

Lead contact

Requests for further information and resources should be directed to the lead contact, Smita Yadav (smitay@uw.edu).

Materials availability

Stem cell lines used are available from SFARI iPSC collection. Stable reagents generated in this study are available from the lead contact with a completed materials transfer agreement. Key resources are listed in Table S3.

Data and code availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD066614.

Acknowledgments

We are grateful for funding provided by the National Institutes of Health (R01MH121674) to S.Y. and the Jaconette Tietze Award to S.Y. We extend our gratitude to the patients and families of 16p11.2 CNV carriers who have shared their skin samples to support research and funding from SFARI (345471) to L.A.W. for generation of the iPSC lines. Mass spectrometry was provided by the MS Resource at UCSF (A.L.B., Director) supported by the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation and the UCSF Program for Breakthrough Biomedical Research.

Author contributions

S.B. characterized the TAOK2 KO iPSC lines, differentiated them into NPCs and performed all experiments on the TAOK2 KO lines with the help of S.M. and A.B. A.F. designed and performed shRNA and overexpression screening experiments and generated the TAOK2 KO iPSC lines. Cilia database analyses of proteomic data and molecular biology experiments were performed by M.C. Proteomics experiments were performed and analyzed by J.A.O-P. under the supervision of A.L.B. Neuronal differentiation of 16p11.2 CNV and control iPSC lines were performed by A.D. under the supervision of L.A.W. The manuscript was written by S.B. and S.Y. All aspects of work, with exception of the proteomics, were supervised by S.Y.

Declaration of interests

The authors declare no competing interests.

Published: August 28, 2025