KLHL13 functional defects cause neurodevelopmental disorder in humans that can be rescued via inhibition of AURKB in cellular and animal models

Abstract

Purpose:

Neurodevelopmental disorders (NDDs) are characterized by limitations in brain development. This study aims to determine the genetic causes of NDD in humans.

Methods:

Exome sequencing was used to detect genetic variants in 4 families. In-silico protein modeling and overexpression in heterologous cells were used to determine the variants’ impact. klhl13 loss of function was modeled in zebrafish, followed by rescue studies using human KLHL13 mRNA and an AURKB inhibitor.

Results:

We found one frameshift and three missense hemizygous variants of KLHL13 in individuals exhibiting NDD characteristics, such as intellectual disability and macrocephaly. 3D protein modeling simulation predicted the alteration of the KLHL13 protein folding for missense variants. Overexpression of NDD-associated variants in HEK293T cells revealed a significant impact on KLHL13-mediated cell cycle regulation during mitosis, leading to genomic instability. Knocking down klhl13 in zebrafish resulted in developmental deficits, which were rescued by co-injection of human KLHL13^WT mRNA but not by transcripts encoding NDD variants. Treatment with AURKB selective inhibitor AZD1152-HQPA rescued genomic stability in heterologous cells and neurobehavioral deficits in zebrafish.

Conclusion:

Our results implicate KLHL13-mediated AURKB regulation as a significant contributor to NDD in humans. Inhibiting AURKB activity could serve as a potential therapy to improve brain development and cognitive function.

Introduction

Ubiquitination, a process that targets substrates for either proteasomal degradation or non-proteolytic processing, is one of the most well-known and extensively studied post-translational modification mechanisms.¹ The sequential reactions of three ubiquitin enzymes - ubiquitin recognizing and activating enzyme E1, ubiquitin-binding and conjugating enzyme E2, and ubiquitin ligase enzyme E3 – further catalyze this regulatory mechanism.² Unlike E1 and E2 ubiquitin enzymes, E3 ubiquitin ligase has a high level of substrate specificity and facilitates the ultimate transfer of the ubiquitin molecules onto the specific substrate.³ Cullin-RING E3 ligases (CRL3), the largest and highly conserved member of the E3 ubiquitin ligase family, are in charge of the ubiquitination of a target substrate.⁴ The CRL3 is made up of four components; a Cullin3 protein (HGNC: 2553) that acts as a scaffold, a kelch domains containing substrate recognition receptor, an adaptor protein with bric-a-brac tram track broad complex (BTB) domain to present the substrate to Cullin3, and a RING-Box 1 (RBX1; HGNC: 9928) protein for the recruitment of E2 and activation of the E3 ubiquitin ligase complex through binding with Cullin3.^{5; 6} In recent years, several Kelch-like protein family members have been defined as the substrate adaptor subunits of the Cullin3 and regulate diverse cellular signaling pathways.^7–10 As of May 2025, forty-two Kelch proteins have been identified, each playing an essential role in the ubiquitination of specific substrates in an E3 ubiquitin ligase-dependent manner.¹¹

Kelch proteins have five to six Kelch domains that recognize the substrate, a BTB domain for protein-protein interactions and homodimerization, which acts as an adaptor between the Kelch domains and Cullin3 to form the E3 ubiquitin ligase complex, and a BACK domain that stabilizes the overall protein structure.^{8; 12–14} Each member of the Kelch family recognizes and recruits a specific substrate that contributes to the diversity of the cellular pathways. KLHL13 (HGNC: 22931) encodes a kelch-like family member 13 (KLHL13) protein with six kelch domains for the target substrate Aurora kinase B (AURKB; HGNC: 11390) recognition, which is a member of the chromosomal passenger complex that aids chromosomal condensation and alignment at the equator plate during metaphase by attaching to the kinetochore.¹⁵ For normal segregation of the sister chromatids and completion of the anaphase, AURKB must migrate towards the central spindle during the transition from metaphase to anaphase, which requires a KLHL13-Cullin E3 ubiquitin ligase complex. CRL3 complex. Prior studies have shown that at the onset of anaphase, AURKB becomes a substrate of the anaphase-promoting complex/cyclosome (APC/C) ubiquitin ligase and undergoes proteasome-mediated degradation.^16–18 In the absence of a functional CRL3 complex, AURKB remains attached to the chromosomal arms, leading to aberrant cell division.¹⁹ Other than AURKB, Insulin Receptor Substrate 1 (IRS1; HGNC: 6125) has been recently identified as another substrate of the KLHL13-Cullin E3 ubiquitin ligase complex. As an adaptor protein in the insulin signaling pathway, IRS1 is responsible for sending signals from the activated insulin receptor to the downstream effectors, such as PI3K and AKT.²⁰

Here, we report hemizygous variants of KLHL13 causing mild to severe X-linked intellectual disability (XLID; HP:0001249), developmental (HP:0012758; HP:0002194), and speech delay (HP:0000750), macrocephaly (HP:0000256), hypotonia (HP:0001252), unsteady gait (HP:0002317), facial dysmorphism (HP:0001999), and behavioral problems (HP:0000708) in eight affected individuals from four human families of diverse ethnicities. Functional analyses of XLID-causing variants revealed abnormalities in mitosis, genomic content stability, and deficits in AURKB localization and ubiquitination. Our overexpression studies suggest that KLHL13 NDD-associated variants mimic loss-of-function phenotypes. Moreover, the knockdown of klhl13 in zebrafish resulted in developmental deficits and impaired locomotor activities. Intriguingly, treatment with AURKB pharmacological inhibitor AZD1152-HQPA (hydroxyquinazoline pyrazole anilide) rescued the cellular abnormalities in heterologous cells and the behavioral deficits in the zebrafish models of KLHL13. Our findings underscore the critical role of KLHL13 in neurodevelopment and establish the viability of additional testing and development of the AURKB inhibitor to treat patients with KLHL13-associated NDD.

Materials and Methods

Study Participants

This study was approved by the institutional review boards at the National Center of Excellence in Molecular Biology, University of the Punjab, Lahore, Pakistan, and the University of Maryland, Baltimore, MD, USA. The PKMR464 family was enrolled from Narowal city, Punjab, Pakistan. Three other families were enrolled in the collaborating institutes. Written informed consent to participate in the research study were obtained from all the unaffected individuals, while the consent for the affected individuals was obtained from the legal guardians/ parents. Affected individuals were examined by a neurologist and a geneticist for neurological, behavioral, morphological, auditory, ophthalmological, skeletal, and dermatological anomalies. MRI images from the medical records, and photographs of the face and any other visible abnormalities were also obtained. Peripheral blood samples were collected from the participating individuals, and genomic DNA was extracted following the standard protocol.²¹

Exome sequencing and bioinformatics analysis

Exome sequencing (ES) was performed on the DNA sample of individual II:6 of the PKMR464 family, and analyzed using the hg19 (GRCh37; GCF_000001405.13) human genome assembly as a reference. Sequencing data was filtered and prioritized using the criteria described previously.^{22; 23} Segregation analysis of the candidate variants was performed through Sanger sequencing. Families 2–4 were enrolled and exome sequenced at their participating institutes. Available photographs of the participating affected individuals were analyzed using the Face2Gene application.²⁴

The identified KLHL13 (NM_001168299.2) variants were analyzed for their predicted impact using various in-silico tools, including Clustal Omega,²⁵ MetaDome,²⁶ Missense Tolerance Ratio score.²⁷ Moreover, three-dimensional (3D) protein modeling was performed to determine the impact of variants on the encoded protein structure via the I-Tasser^{28; 29} database and PyMol software.³⁰

Generation of KLHL13 constructs

The wild-type (WT) human KLHL13 cDNA construct (Cat#HG23067-UT) was purchased from Sinobiological Inc. (Wayne, PA, USA) infused with pCMV3 untagged plasmid. Using this construct as a template, KLHL13^WT cDNA was PCR amplified using the cDNA-specific primers (Supplemental Table 3) and cloned into either the pEGFP-C3 (Cat#6082–1; Clontech, San Jose, CA, USA) or pcDNA3.1(+) plasmids via In-Fusion cloning. The NDD-associated variants were introduced into KLHL13 via site-directed mutagenesis (Supplemental Table 3).

Cell lines, culture maintenance, and transfection

HEK293 (ATCC CRL-1573) and COS-7 (ATCC CRL-1651) cells (ATCC, Wesel, Germany) were used for all the in vitro analyses. These cells were grown in Dulbecco modified Eagle’s medium (Cat#MT10013CV; Corning, New York, NY, USA) supplemented with 10% fetal bovine serum (Cat#100–106H-500; GeminiBio, West Sacramento, CA, USA) and penicillin/streptomycin (Cat#15140122; Gibco BRL, St. Louis, MO, USA) at 37°C temperature with 5% CO₂. Transfections of KLHL13^WT and variants cDNA constructs into HEK293 and COS-7 cells were performed at 70% cell confluency using Lipofectamine 2000 reagent (Cat#11668019; Invitrogen, St. Louis, MO, USA).

Immunolocalization analysis

Transfected COS-7 cells grown on glass coverslips were fixed with 1ml of 4% paraformaldehyde (PFA; Cat#50980487; Fisher Scientific, Pittsburgh, PA, USA) for 15 minutes at room temperature (RT), followed by washing three times with 1X PBS and permeabilization with 0.1% Triton-X100/ PBS for 10 minutes. Following permeabilization, cells were treated with 10% normal goat serum (NGS; 005000121; Jackson ImmunoResearch, West Grove, PA, USA), 0.1% Triton-X100 solution for 1 hr. at RT to minimize non-specific binding. Then, the cells were incubated with an anti-GFP monoclonal antibody (Cat#G6539–2ML, Sigma-Aldrich, St. Louis, MO, USA) and Tom20 polyclonal antibody (Cat#SC-11201, Santa Cruz Lab, Santa Cruz, CA, USA) diluted at a 1:500 ratio in 3% NGS/ 0.1% Triton-X100 overnight at 4°C. Next day, cells were washed two times with 1X PBS, followed by staining with DAPI, phalloidin, and Alexa Fluor-conjugated secondary antibodies at 1:500 dilution in 3% NGS/ 0.1% Tritox-X100 for 1 hr. at RT. After washing three times with 1X PBS, coverslips were mounted with Fluoro-Gel II (Cat#1798550; Electron Microscopy Sciences, Hatfield, PA, USA). Images were captured with a Nikon W-1 spinning disk fluorescent microscope and processed with Fiji ImageJ (Version 2.0).

Western blotting

Transfected HEK293 cells were collected in ice-cold 1X PBS and centrifuged at 1200 rpm for 5 minutes. The resulting cell pellets were treated with radioimmunoprecipitation (RIPA) cell lysis buffer (3ml 5M NaCl, 5ml 1M Tris-HCl (pH 8.0), 1ml NP40, 5ml 10% sodium deoxycholate, 1ml 10% SDS in 100ml ddH₂O), a protease inhibitor cocktail (Cat#78429; ThermoFisher Scientific, St. Louis, MO, USA), and DNase I (Cat#M0303S; New England Biolabs, Ipswich, MA, USA). Sample denaturing SDS Laemmli 2X concentrate buffer (Cat#S3401–1VL; Sigma-Aldrich) was added to each sample, and protein concentration was determined through Bradford assay. Twenty-five micrograms of protein samples were loaded onto a 4–20% Tris-Glycine SDS-PAGE gel (Cat#XP04202BOX; Invitrogen) and electrophoresed at 25 milliamperes for 2 hours. The proteins were then transferred to a 0.45 μm nitrocellulose membrane (Cat#88018; ThermoFisher Scientific) for 2 hours at 25 volts. After transfer, the membrane was immersed in blocking buffer (1X TBS, 5% w/v nonfat dry milk) for 1 hr. at RT, followed by incubation with anti-GFP, anti-AURKB (Cat# ab2254; Abcam, Waltham, MA, USA), Anti-KLHL13 (Cat#MA5–15658; Invitrogen) and anti-GAPDH (Cat#60004–1-Ig; Proteintech, Rosemont, IL, USA) antibodies at 1:500 dilution in 1X TBST/ 5% nonfat dry milk overnight at 4°C with continuous shaking. Next day, after washing, the membrane was incubated with fluorophore-conjugated IR-Dye 680RD (Cat#926–68070; LI-COR Biosciences, Lincoln, NE, USA) at a 1:5000 dilution in 1XTBST/ 5% non-fat dry milk for 1 hr. at RT. Finally, the membrane was imaged wet on the ChemiDOC MP imaging system (Cat#12003154; Bio-Rad, Hercules, CA, USA) using a fluorescent detection tool.

Cell cycle analysis

For immunofluorescence-based studies, twenty-four hours post-transfection, COS-7 cells were serum-starved for a day to halt the cell cycle at the G0/G1 phase. Next day, after washing with 1X PBS, cells were fixed with 4% PFA, permeabilized with 0.1% TritonX-100, and immune-stained with anti-GFP, anti-β-Tubulin, and anti-AURKB (Cat#MA5–27890; ThermoFisher Scientific) monoclonal antibodies overnight at 4°C. The following day, after 3X washes with 1XPBS, cells were stained with DAPI and Alexa flour-conjugated secondary antibodies for 1 hr. at RT. After washing, coverslips were mounted, and images were captured via the Nikon W-1 Spinning Disk microscope and processed with Fiji ImageJ (Version 2.0).

For flow cytometric studies, twenty-four hours post-transfection with KLHL13 constructs, HEK293 cells were incubated in 10μM (3.3μg/ml) of 5-bromo-2’-deoxyuridine (BrdU; Cat#423401; BioLegend, San Diego, CA, USA) for 2 hrs at 37°C temperature with 5% CO₂. After incubation, cells were harvested using the accutase cell dissociation reagent (Cat#A6964; Sigma-Aldrich) at 1200 rpm for 5 mins. The resulting pellets were washed twice with 1X PBS and fixed overnight with 70% ethanol at 4°C. The next day, cell cycle analysis was carried out following the previously published protocol.³¹ Briefly, 1×10⁷ cells were washed twice with 1XDPBS. After denaturing, cells were stained in the dark at RT for 1 hour. with a 1:30 dilution of anti-BrdU antibody (Cat#364113; BioLegend) in DPBS buffer supplemented with 5% FBS. Following staining with BrdU antibodies, cells were incubated in the dark in propidium iodide (PI; Cat#F10797; Invitrogen) diluted in the FACS staining buffer for 30 minutes at RT. A total of 1×10⁷ events were collected for cell cycle analysis per condition, and each experiment was performed in triplicate. The gating strategy involved setting markers based on DNA content, with G1 phase cells having 2n DNA content clustered in one group and G2/M phase cells having 4n DNA content clustered with much higher fluorescence intensity when samples were stained with only PI. S-phase cells were identified as Brdu-positive cells with intermediate DNA content. All samples were analyzed using BD LSR-II flow cytometer, and data were processed using FCS Express software.³²

Apoptotic assay using APC Annexin V and 7AAD staining

Twenty-four hours post-transfection with KLHL13^WT and variants constructs, HEK293 cells were harvested using the accutase cell dissociation reagent at 1200 rpm for 5 mins. The resulting pellets were washed twice with a cold FACS buffer at 2000 rpm for 5 minutes. The pellets were then resuspended in the Annexin V binding buffer (Cat#640930; BioLegend, San Diego, CA, USA) at a concentration of 5×10⁵ cells/100μl. After resuspension, 5μl of apoptotic APC Annexin V and 5μl of 7AAD viability staining markers were added to each sample tube, gently vortexed, and incubated for 15 minutes at room temperature in the dark. After incubation, 400μl Annexin V binding buffer was added to each sample and analyzed by flow cytometry to determine the percentage of viable, apoptotic, and necrotic cells treated with the AURKB inhibitor and untreated states.

Pharmacological treatment of heterologous cells

Post transfection and serum starvation, HEK293 and COS-7 cells were incubated in AURKB inhibitor AZD1152-HQPA (Cat#S1147; SelleckChem, Houston, TX, USA) at six different concentrations (5nM, 10nM, 20nM, 30nM, 50nM, and 100nM) for 24 hours at 37°C with 5% CO₂ for the dosage response studies. After treatment, COS-7 cells were fixed in 4%PFA and immunostained with anti-AURKB antibody, while the HEK293 cells were harvested, fixed, and stained with anti-BrdU antibody and PI for cell count at different phases of the cell cycle via flow cytometry analysis.

Immunoprecipitation

KLHL13^WT and pathogenic variants transfected HEK293 were either treated with 20 μM MG132 (Cat#HY-13259; MedChemExpress, Monmouth, NJ, USA) for 6 hours (to assess polyubiquitination pattern) or remained untreated to determine the monoubiquitination pattern of AURKB. The following day, both the MG132-treated and untreated cells were collected in ice-cold 1X PBS and centrifuged at 1200 rpm for 5 minutes, and the resulting cell pellets were treated with RIPA cell lysis buffer. After the protein lysate preparation, protein concentration was determined through the Bradford assay. Samples were incubated in the 1:500μl of anti-AURKB (Cat#MA5–27890; ThermoFisher Scientific, Waltham, MA, USA) monoclonal antibody and/or anti-IRS1 antibody (Cat#2382; Cell Signaling, Danvers, MA, USA) at 4°C overnight with continuous mixing. Next day, following the manufacturer’s immunoprecipitation protocol and using the Pierce protein A/G magnetic beads (0.25mg; Cat#88802; ThermoFisher Scientific), AURKB was immunoprecipitated. The resulting samples were mixed with the sample denaturing SDS Laemmli 2X concentrate buffer (Cat#S3401–1VL; Sigma-Aldrich) and subjected to western blotting using the Ubiquitin monoclonal antibody (Ubi-1; Cat#13–1600; ThermoFisher Scientific) to identify the ubiquitinated form of AURKB in KLHL13^WT and pathogenic variants.

Zebrafish maintenance

All the experiments performed in the zebrafish model were approved by the Institutional Animal Care and Use Committee at the University of Maryland, Baltimore, MD, USA (Protocol# 00000140, renewed on 8/22/2023). Zebrafish were maintained at 28°C under 14 hours light and 10 hours of dark cycle. All the experiments were performed either using the WT tubingen or the neurod1-egfp transgenic zebrafish line.³³ The developmental stages explained in this study were either hours post fertilization (hpf) or days post fertilization (dpf).

Morpholino (MO) mediated klhl13 knockdown

Following the breeding and fertilization, WT tubingen eggs were collected and microinjected with the klhl13 ATG translational blocking MO (ATG_MO), a splice site blocking MO (SS_MO), or with a standard control MO (GeneTools, Philomath, OR, USA) at 1 to 4ng/μl concentrations. All the embryos were maintained in an E3 embryo medium (5mM NaCl, 0.33mM CaCl₂, 0.17mM KCl, 0.33mM MgSO_4, and 0.1% methylene blue) at 28°C, and gross morphology was assessed at 3- or 5-dpf after anesthetization with tricaine. To rule out klhl13 ATG_MO cytotoxicity, additional studies were performed by co-injecting the embryos at the 1–4 cell stage with p53 and klhl13 MOs at 2ng/μl and 4ng/μl concentrations, respectively (Supplemental Table 4).

Zebrafish RNA extraction and qPCR

At 3dpf, 100 embryos per condition (uninjected control, ATG-MO, SS-MO) were collected and homogenized in 1 ml of Trizol reagent (Cat#15596018; ThermoFisher Scientific), and total RNA was extracted following a phase separation step with chloroform (Cat#C2432; Sigma Aldrich) and RNA precipitation with isopropanol (Cat#T036181000; ThermoFisher Scientific). RNA pellets were washed with 75% ethanol, air-dried, and resuspended in RNase-free water. RNA concentration and purity were assessed using a NanoDrop spectrophotometer (Thermo Fisher Scientific), and integrity was confirmed by agarose gel electrophoresis.

For each sample, 1 μg of total RNA was reverse transcribed using the PrimeScript RT-PCR Kit (Cat#RR014B; Takara). qPCR was performed using 100ng/ μL of cDNA with the SYBR Green qPCR master mix (Cat#330513; Qiagen, Venlo, The Netherlands) on a QuantStudio^™ 3 Real-Time PCR System. The graph was plotted for the relative klhl13 expression using ΔΔCt and the fold change method, normalized to gapdh (Supplemental Table 3: Primers). Statistical analysis was performed using GraphPad Prism (Version 10) with significance determined by a two-tailed unpaired Student’s t-test.

Human KLHL13 mRNA microinjections

For the rescue experiments, cDNAs of human WT and variants harboring KLHL13 were amplified (Supplemental Table 3) and cloned into the pT7TS vector. The resulting constructs were Sanger sequenced and verified. After verification, vector clones were linearized through EcoRI restriction enzyme digestion, and full-length human KLHL13 mRNA was transcribed using the HiScribe T7 Quick high-yield RNA synthesis kit (Cat#E2050S; NEB). Co-injection of klhl13 ATG_MO and human KLHL13 mRNAs (concentrations: 250pg/μl, 500pg/μl, 750pg/μl, and 1ng/μl) was performed in WT zebrafish embryos at the 1–4 cell stage. Moreover, to assess the cellular toxicity and loss of function impact of KLHL13 mRNAs, the WT and mutant mRNAs were injected individually at a 500pg concentration in WT zebrafish embryos at the 1–4 cell stage.

Behavior analysis of zebrafish embryos

Behavioral experiments were performed at 5dpf with the Viewpoint Zebrabox system (ViewPoint, Paris, France). Locomotor activity and visual motor startle responses (VMR) of the klhl13 morphants and control larvae were analyzed at the dark/light transition period. We also assessed the auditory startle responses after sound stimuli. Before the VMR experiment, larvae were acclimated to the room light (500–600 lux) for 20 minutes. After the acclimation period, larvae were kept in the dark for 10 minutes, followed by exposure to 3000 lux light for 10 minutes. Distance traveled (mm) from the last 30 seconds in the dark period till 30 seconds into the light phase was calculated for each larva and plotted. For auditory startle motor response (AMR) analysis, larvae were kept in a noise-free environment for 10 minutes, followed by exposure to a 96dB sound stimulus for 10 minutes. Distance traveled (mm) at the no sound-to-sound transition point was calculated for each larva and plotted.

Zebrafish brain imaging

For brain imaging and neuronal pattern analysis, neurod1-egfp zebrafish embryos at the 1–4 cell stage were microinjected with either klhl13 ATG_MO, control_MO, or human KLHL13^WT mRNA. After 24 hours, larvae were treated with 1X 1-phenyl-2-thiourea (PTU; Cat#P0237; TCI America, Portland, OR, USA) to prevent melanin synthesis. At 5 dpf, larvae were fixed with 4% PFA and embedded in 1% low-melting agarose, and brain morphology was analyzed. High-resolution images of the zebrafish brain were captured using the Nikon W1 Spinning disk fluorescent microscope. The optic tectum area was identified based on anatomical landmarks and selected as the region of interest (ROI) using the line selection tool in Fiji ImageJ software. The area marked with the red line (ROI) was calculated in square micrometers (μm²) using the measurement tools in Fiji ImageJ. Statistical analysis was performed to compare the optic tectum area between the control and experimental groups, with significance determined using a two-tailed unpaired Student’s t-test.

Pharmacological manipulations of zebrafish embryos

After 24 hours of klhl13 ATG_MO microinjections, tubingen WT, and neurod1-egfp morphants were treated with AZD1152-HQPA at 30nM final concentration. After treatment, larvae were analyzed for their developmental phenotypes, behavior, brain size, and neuronal pattern at the 3- or 5-dpf stage.

Quantification and statistical analysis

All experiments were performed at least three times (n=3) to ensure reproducibility and statistical validity. The number of cells counted for each experiment is mentioned in each experimental section. Quantification and statistical analyses were performed using GraphPad Prism (version 10) and Microsoft Excel. All the data is shown as the mean ± standard error of the mean (SEM). A comparison between different groups was carried out using the two-tailed unpaired Student’s t-test, with statistical significance set at p<0.05 (Source data).

Results

Identification and clinical manifestations associated with rare variants in KLHL13

Individuals with NDD were referred by consultants specializing in disability and genetic services. Following the approval of study protocols by the Institutional Review Boards of all participating institutions, written informed consent was obtained from all adult participants and the guardians of affected individuals. Genomic DNA was extracted from the venous blood of participating individuals and was analyzed by exome sequencing at their ascertaining institutes. After identifying the potentially pathogenic variants in KLHL13, the participating investigators were connected through the GeneMatcher website.³⁴ The pedigrees of the families included in this study and available images of affected individuals are shown in Figure 1A–B, while the clinical summaries for each patient are provided in Supplemental Table 1.

Figure 1. Hemizygous rare variants in KLHL13 are associated with intellectual disability, developmental delay, hypotonia, macrocephaly, and dysmorphic facial features.

A Pedigrees of families from different ethnicities segregating NDD-associated phenotypes with maternally inherited variants in the X-linked KLHL13 gene. Filled symbols represent affected individuals, and genotypes for the identified variants are given below the symbols of the participating individuals. The black arrow represents the proband. B Photographs of affected individuals (I:1, II:2, and II:6) of PKMR464 and Family 3 (I:2, and II:2) revealed dysmorphic facial features, including broad nasal bridge, short philtrum, triangular mouth, highly arched eyebrows, hypertrichosis of the eyebrow, broad nasal tip, protruding low-set ear, broad forehead, hypertelorism, hypoplasia of maxilla, facial hypotonia, and down slanting palpebral fissures, and thin upper lips. C Percentages and prevalence of different NDD-associated clinical features that are variably expressed in individuals harboring KLHL13 variants. ID: intellectual disability; DD: developmental delay; MC: macrocephaly; BP: behavioral problems; SD: speech delay; FD: facial dysmorphism; UG: unsteady gait; RF: rare clinical features; HT: hypotonia. D Brain MRI of the normal (II:3) and affected individual (II:6) of family PKMR464. The red arrows represent a larger head-to-neck area, while the yellow arrow shows mucus retention cysts in the bilateral maxillary sinuses. E Structural representation of the human KLHL13 gene and encoded protein, along with four variants identified in our cohort.

Family PKMR464, which included four individuals with NDD and mild ID (Supplemental Table 1), was identified in the Narowal city of Punjab, Pakistan. Exome sequencing of proband II:6, followed by Sanger sequencing of participating family members DNA samples, revealed a rare (gnomAD: 0) transition variant c.953C>T (g.117043686G>A) in the KLHL13 gene co-segregating with NDD phenotypes in an X-linked dominant pattern in the family (Figure 1A). The resulting missense variant p.(Ala318Val) in the encoded protein due to c.953C>T transition was not found in 200 ethnically matched Pakistani control samples and was predicted pathogenic by various in silico algorithms (Supplemental Table 2).

Through the GeneMatcher program,³⁴ we identified three additional unrelated families with variants in the KLHL13 gene (Figure 1A). Family 2 of European American origin had an affected male with c.179A>G p.(His60Arg) variant (g.117079467T>C), while the NDD phenotype in Family 3 with Hispanic ethnicity was co-segregating with dominantly inherited c.557G>A p.(Cys186Tyr) variant (g.117053506C>T) of KLHL13 (Figure 1A). Finally, the affected male of German Family 4 had a frameshift c.1781_1782del p.(Val594Glyfs*6) variant (g.117033068_117033069del) of KLHL13, inheriting in an X-linked recessive pattern from a phenotypically normal mother (Figure 1A). Besides the KLHL13 variant, we also identified a rare de novo missense variant in the BTBD10 gene (HGNC: 21445) in the affected individual of Family 4 (See Supplemental notes). In contrast to heterozygous affected females of family PKMR464 and family 3, the I:1 female of family 4 displays no symptoms of the NDD phenotype, which could be attributed to the potential influence of X-inactivation patterns, reduced penetrance, or genetic factors impacting expressivity. All four KLHL13 variants were not found in the gnomAD database³⁵ and were predicted as damaging by various prediction programs (Supplemental Table 2).

Next, we comprehend clinical phenotype information from the medical records of the available affected individuals harboring rare pathogenic variants in KLHL13 (Supplemental Table 1). All the affected individuals had mild to severe ID (HP:0001249), macrocephaly (HP:0000256), developmental (HP:0001263; HP:0012758) and speech delay (HP:0000750), hypotonia (HP:0001252), dysmorphic facial features (HP:0001999) as predicted by the Face2Gene application (Figure 1B) as well as behavioral abnormalities (HP:0000708; HP:0000718), and unsteady gait (HP:0002317) along with chewing and swallowing problems (Supplemental Table 1; Figure 1C). The affected individual of Family 2 also had progressive leg numbness (HP:0031959), paresthesias (HP:0003401), unsteady gait, urinary issues, constipation (HP:0002019), tinnitus (HP:0000360), and bilateral papilledema, while the affected individual of Family 3 experienced bilateral cryptorchidism (HP:0008689; Supplemental Table 1). Unlike the neurotypical sibling (II:3), brain magnetic resonance imaging (MRI) of the affected individual II:6 from family PKMR464 only revealed mucus retention cysts in the bilateral maxillary sinuses without any apparent degeneration or malformation of brain tissue (Figure 1D).

In silico analysis of KLHL13 NDD-associated variants

KLHL13, encoded by eight coding exons, is a relatively midsize protein with 658 amino acids and consists of three structurally and functionally important domains, including an N-terminus bric-a-brac, tram track, broad complex (BTB) domain, C-terminus six Kelch repeats, and a BACK domain (Figure 1E). The BTB region is critical for the protein-protein interaction, while Kelch motifs are essential for substrate recognition and binding, and the BACK domain facilitates the interaction between BTB and Kelch domains, along with stabilizing the overall protein structure.¹⁵ The identified NDD-associated variants in KLHL13 are located within the structured regions of the protein (Figure 1E). The MetaDome tolerance landscape and mutation tolerance ratio score predicted the damaging impact of all the variants due to their positions in the relatively intolerant regions of the encoded protein. (Supplemental Figure 1A). Moreover, all three identified missense variants were predicted to replace the amino acid residues that are highly evolutionary conserved (Supplemental Figure 1B).

To gain further insights into the structural impact of the identified missense variants, we performed three-dimensional (3D) molecular modeling of KLHL13 using I-Tasser and PyMol (Supplemental Figure 1C). The p.His60Arg variant is predicted to cause a bump into the secondary structure because of the larger size of the arginine and repulsion between residues due to the positive charge (Supplemental Figure 1C). In contrast, the p.Cys186Tyr is predicted to cause a loss of hydrophobic interactions at the core or surface of the protein (Supplemental Figure 1C). Finally, the p.Ala318Val variant might increase the interactions with other amino acids due to the larger size of the mutant residue (Supplemental Figure 1C).³⁶

NDD-associated KLHL13 variants impact mitotic progression and completion of cell division in heterologous cells

Next, to functionally evaluate the impact of identified KLHL13 variants, with or without EGFP-tagged WT, p.His60Arg, p.Cys186Tyr, p.Ala318Val, and p.Val594Glyfs*6 variants harboring human KLHL13 cDNAs were transiently transfected in HEK293 and COS-7 cells. Forty-eight hours post-transfection, cells were harvested and analyzed for protein levels through immunoblotting. We did not observe any significant difference in the expression of EGFP-tagged vs untagged KLHL13 cDNA constructs (Supplementary Figure 2). When compared with full-length WT, KLHL13 variants (p.His60Arg and p.Val594Glyfs*6) exhibited significantly reduced protein steady-state levels, while p.Cys186Tyr and p.Ala318Val were expressed at comparable levels to the WT protein (Supplemental Figure 2).

We then analyzed the effect of the variants on the subcellular localization of KLHL13 constructs in COS-7 cells and observed no apparent differences compared to the wild-type protein (data not shown). AURKB is a known substrate for KLHL13, which aids in chromosomal condensation and alignment at the equator plate during metaphase^{9; 15; 19} regulating the cell division process; therefore, we further investigated whether the KLHL13 variants impact the cell division and the cell cycle regulation. For these studies, transiently transfected COS-7 cells with KLHL13^WT and variants encoding GFP-tagged constructs were synchronized at the G0/G1 phase using a serum starvation method.^{37; 38} After harvesting, cells were immunolabeled (Figure 2A) with anti-β tubulin (red) antibody, and nuclear content was marked with DAPI (blue). COS-7 cells overexpressing KLHL13^WT had no cell division abnormalities from interphase through cytokinesis (Figure 2A). In contrast, cells overexpressing NDD-associated variants harboring KLHL13 showed abnormal and irregulated cell division, including chromosomal misalignment, missegregation, lagging, and bridging of the sister chromatids (Figure 2A). Moreover, abnormal cytokinesis with the formation of giant multinucleated cells was observed in cells overexpressing NDD-associated KLHL13 alleles (Figure 2B; split channels with the individual images of each construct are shown in Supplemental Figures 3–7). For the quantification of GFP-positive cells, at least 120 cells were analyzed per condition in each experiment. Our results indicated that the overexpression of KLHL13 variants led to phenotypes consistent with loss-of-function effects. We observed chromosomal misalignment at metaphase, missegregation and lagging at anaphase, and bridging at telophase, along with genomic instability in cells overexpressing the KLHL13 variants (Figure 2C, D). These findings support the deleterious nature of NDD-associated variants in KLHL13.

Figure 2. KLHL13 NDD-causing variants impact cell division and cycle regulation at the M/G1 phase in heterologous cells.

A Confocal image of COS-7 cells overexpressing GFP-tagged KLHL13^WT or NDD-associated variants synchronized at the G0/G1M phase of the cell cycle. Immunolabeling with anti-GFP (green) and anti-β-tubulin (red) antibodies revealed chromosomal misalignment at metaphase (green arrows) in p.His60Arg and p.Val594Glyfs*6, while all four variants caused abnormal segregation and lagging of sister chromatids (white arrows) at anaphase. Chromosomal bridging (yellow arrows) at telophase was also observed in cells overexpressing KLHL13 with p.Ala318Val, p.His60Arg, and p.Val594Glyfs*6 variants. Moreover, all four variants impacted the completion of cytokinesis via the formation of giant multinucleated cells (red arrows). Nuclear contents were labeled with DAPI (blue). Scale bar: 10 μm. Note: Figure. 2A shows merged images, while Supplemental Figures 3–7 display individual channels for each stage of mitosis. B Graphical representation of mitotic abnormalities observed in cells overexpressing KLHL13 constructs. All four NDD-associated variants showed incomplete cytokinesis with significantly high (*p<0.02, **p<0.001, ***p<0.0001) formation of bi/multinucleated cells when compared with KLHL13^WT expressing cells. At least 120 GFP-positive cells were imaged for each condition, with 40 cells quantified per experiment over 3 independent experiments (n=3). C Bar graph representing the percentage of KLHL13^WT and variants overexpressing cells at metaphase, containing aligned vs. misaligned chromosomes. There was a significantly higher number of cells impacting chromosomal alignment during metaphase in variants expressing cells compared to KLHL13^WT (**p<0.001, ***p<0.0001). A total of 90 GFP-positive cells were imaged at metaphase, with 30 cells quantified per experiment over 3 independent experiments (n=3). D The bar graph represents the percentage of cells observed with lagging and bridging chromosomes showing an incomplete segregation of the sister chromatids. The variants overexpressing cells were found to have a significantly higher number of cells observed with the lagging and bridging phenotypes during anaphase compared to the KLHL13^WT overexpressing cells (*p<0.01, **p<0.001). A total of 60 GFP-positive cells were imaged at anaphase, with 20 cells quantified per experiment over 3 independent experiments (n=3). E Graphical representation of the percentages of KLHL13^WT and NDD variants overexpressing HEK293 cells at different phases of the cell cycle. In contrast to WT-expressing cells, KLHL13 NDD-associated variants-expressing cells had significantly higher (*p<0.01, **p<0.001) cell cycle abnormalities, leading to genomic instability. A total of 2×10⁵ events were collected for cell cycle analysis per condition, and each experiment was performed in triplicate (n=3).

To further confirm the dysregulation of mitosis in cells overexpressing KLHL13 variants, we analyzed the BrdU and PI content of HEK293 cells through flow cytometry. BrdU labels the proliferating cells as it incorporates into the newly synthesized DNA during the S phase of the cell cycle, while PI determines the cell viability and overall DNA content in all phases of the cell cycle.³⁹ Consistent with our immunostaining results, we observed significantly high genomic content at the M/G1 phase transition, leading to polyploidy in cells overexpressing NDD-associated KLHL13 variants (Figure 2E; Supplemental Figure 8A, B). Moreover, compared with KLHL13^WT, all four variants resulted in abnormal genomic content at each phase of the cell cycle (Figure 2E). Taken together, our ex vivo studies in heterologous cells confirmed the pathogenic nature of all four NDD-associated variants and impaired function of KLHL13.

As previously stated, KLHL13 is a specific substrate adaptor for AURKB within the BCR-E3 ubiquitin ligase complex, facilitating the ubiquitination of AURKB, which leads to its removal from chromosomal arms to the central spindle at anaphase.¹⁹ Translocation of AURKB to the central spindle facilitates accurate segregation of sister chromatids at anaphase and successful completion of cytokinesis.¹⁹ We reasoned that KLHL13 variant-associated mitotic abnormalities might have stemmed from retention of AURKB at the chromosomal arms due to impaired ubiquitination. Therefore, we analyzed AURKB expression, ubiquitination, and localization during anaphase in cells transfected with KLHL13 constructs. As anticipated, in cells overexpressing KLHL13 NDD-associated variants, AURKB had significantly high expression and reduced the monoubiquitination (Supplemental Figure 9A–C) and polyubiquitination compared to cells expressing the KLHL13^WT construct (Figure 3A–D). Furthermore, in contrast to AURKB localization at the central spindles in cells expressing KLHL13^WT, in all the NDD-variants expressing cells, AURKB remained accumulated at the chromosomal arms (Supplemental Figures 9D–E and 10), further supporting impaired ubiquitination. However, all the NDD-variants, except p.His60Arg, which showed mild deficits, did not significantly impact the ubiquitination levels of IRS1, another known substrate of KLHL13,²⁰ in heterologous cells (Supplemental Figure 11), which might indicate substrate-specific functional impact of these variants.

Figure 3. KLHL13 variants lead to increased expression and reduced ubiquitination level of AURKB, resulting in abnormal cell cycle regulation, which can be rescued by AZD1152-HQPA treatment.

A Immunoblot showing the expression of AURKB in KLHL13^WT and variants overexpressing cells. B Quantitative representation of AURKB in KLHL13^WT and variants overexpressing cells revealed a significant increase in the AURKB levels in the variants (p.His60Arg, **p<0.0117; p.Cys186Tyr, *p<0.0193; p.Val594Glyfs*6, *p<0.0419) compared to KLHL13^WT. C Immunoprecipitation of AURKB in KLHL13^WT and variants overexpressing cells, followed by western blot analysis using ubiquitin and AURKB antibodies. The yellow dotted line box represents the area that was quantified for ubi-AURKB expression. D Graph representing a significantly reduced AURKB ubiquitination level in KLHL13 variants (p.His60Arg, **p<0.0037; p.Cys186Tyr, **p<0.0046; p.Ala318Val, *p<0.0118; p.Val594Glyfs*6, *p<0.0215) overexpressing cells as compared to KLHL13^WT. E Graphical representation of KLHL13 constructs overexpressing cells at different cell cycle phases. In contrast to untreated cells, AZD1152-HQPA (30nM) treatment significantly (**p<0.001) improved mitosis and genomic stability of cells overexpressing variants harboring KLHL13 constructs. F Confocal images of COS-7 cells at anaphase overexpressing KLHL13 NDD-associated variants, with or without AZD1152-HQPA (30nM) treatment, and immunolabelled with anti-AURKB (orange), anti-GFP (green), and anti-β-tubulin (red) antibodies, while DNA content was marked with DAPI (blue). After treatment, AURKB localization at the central spindle was restored for cells expressing KLHL13 pathogenic variants. Scale bar: 20μm. G Quantitative representation of cells with AURKB localization at the central spindle vs. chromosome arms. AZD1152-HQPA treatment significantly (**p<0.001, ***p<0.0007) improved AURKB localization at the central spindles when compared with untreated cells. A total of 90 cells for each condition were imaged and plotted, with 30 cells quantified per experiment over 3 independent experiments.

Pharmacological inhibition of AURKB restored normal cycle and genomic stability in cells expressing KLHL13 variants

Next, we reasoned that the inhibition of AURKB would restore an accurate mitosis process and genomic stability in cells expressing KLHL13 NDD-associated variants. The pharmacological agent AZD1152-HQPA has been investigated as a selective AURKB inhibitor in several studies, including phase I/II clinical trials in participants with cancers.⁴⁰ Therefore, we used a dose escalation (5–100nM) paradigm of AZD1152-HQPA inhibitor in HEK293 cells overexpressing KLHL13 variants (Supplemental Figure 12). Interestingly, we found significant restoration of the genomic stability at each phase of the cell cycle in KLHL13 variants expressing cells after treatment with AZD1152-HQPA inhibitor (Supplemental Figure 12). Using the optimum AZD1152-HQPA dose (30nM), we found that polyploidy content at the M/G1 transition was reduced, with the apoptosis [Annexin-V and 7-amino actinomycin D (7-AAD) positive cells]⁴¹ of the abnormally dividing multinucleated cells (Figure 3E; Supplemental Figures 13–14) after the treatment. In conjunction, the treatment also reduced the percentage of viable KLHL13 variants expressing cells, however, the genomic content of these cells was stable (Figure 3E; Supplemental Figure 13). Concordantly, immunolabeling of the viable cells expressing KLHL13 NDD-variants after AZD1152-HQPA inhibitor (30nM) revealed normal localization of AURKB at the central spindles as compared to untreated cells (Figure 3F–3G; Supplemental Figure 15). Finally, AZD1152-HQPA treatment significantly increases the polyubiquitination levels of AURKB in p.Cys186Tyr and p.Val594Glyfs*6 variants expressing cells as compared to the untreated cells (Supplemental Figure 16). Taken together, these findings support the notion that the NDD-associated KLHL13 variants impact cell division and genomic stability via deficits in AURKB ubiquitination, and treatment with the AZD1152-HQPA inhibitor significantly rescued their pathological impact in heterologous cells.

klhl13 knockdown in zebrafish impaired development and brain function

Next, to gain insight into the role of KLHL13 in brain development and function and to study the pathogenic mechanism of NDD-associated variants in vivo, we used a zebrafish model. The zebrafish genome contains multiple KLH protein orthologs, including KLHL13. The klhl13 in zebrafish is located on chromosome 5 and is orthologous to the human KLHL13.⁴² Zebrafish klhl13 ortholog has ~84% identity and 87% amino acid similarity with human KLHL13. Through microinjection of translational blocking antisense morpholino (ATG_MO; Supplemental Table 4) and splice site blocking morpholino (SS_MO; Supplemental Table 4) in 1 to 4-cell stage embryos, we knocked down the klhl13 expression in AB/tubingen zebrafish strain (Figure 4A–C; Supplemental Figure 17A). First, we performed a dose escalation study using the ATG_MO to investigate the impact on the overall zebrafish development (Supplemental Figure 17A) and based on the results, selected the 4ng dose for further morphological and behavioral studies. In contrast to control_MO (Supplemental Table 4) injected larvae, the klhl13 ATG_MO injected morphants depicted several morphological and developmental deficits, which we grouped into four categories (Figure 4A, B; Supplemental Figure 17); Normal: no apparent developmental abnormality; mild: pericardial edema, absence of swim bladder, and skeletal deficits, including scoliosis, and protruded lower jaw; moderate: besides features noted in mild class, these larvae also had yolk sac edema, malformed tail, and delayed development; and severe: pericardial and yolk sac edema, no swim bladder, skeletal deficits, malformed curled tail, and shorter body length. To rule out cellular toxicity and off-target effects of klhl13 MO, we then co-injected a p53 morpholino to block the p53 pathway.⁴³ The absence of rescue by p53 morpholino indicated that the observed phenotypes were very likely due to klhl13 knockdown and not due to morpholino toxicity (Supplemental Figure 17B). To further confirm gene-specific phenotype, we used a second morpholino (SS_MO) to block the klhl13 exon 4 splice donor site (Supplemental Figure 18). Similar phenotypic abnormalities were observed in larvae injected with the klhl13 SS_MO (Supplemental Figure 18A, B). As anticipated, the SS_MO injection resulted in aberrant splicing and premature truncation of klhl13 mRNA, which was likely subjected to degradation by the NMD machinery (Figure 4C; Supplemental Figure 18C–E). Taken together, these results support the klhl13 function in regulating the overall development of zebrafish.

Figure 4. Morpholino-mediated klhl13 knockdown caused developmental delay and macrocephaly in zebrafish.

A Schematic representation of the zebrafish klhl13 gene (top), along with the location of ATG-blocking morpholino (MO). Shown also are representative images (bottom) of resulting phenotypes in ATG_MO injected zebrafish larvae at 3 dpf. Based on the phenotypic severity, larvae were categorized into four classes; normal: no apparent developmental abnormality; mild: pericardial edema, absence of swim bladder, and skeletal deficits, including scoliosis and protruding lower jaw; moderate: besides features noted in the mild class, these larvae also had yolk sac edema, malformed tail, and delayed development; and severe: pericardial and yolk sac edema, no swim bladder, skeletal deficits, malformed curled tail, and shorter body length. SC: scoliosis; PM: populated head melanocytes; PLJ: protruding lower jaw; PCE: pericardial edema; YSE: yolk sac edema; MT/CT: malformed curled tail. White star marked absence of the swim bladder. Scale bar: 10 μm. B The graph represents percentages of controls (uninjected and control MO_injected larvae) along with klhl13 ATG_MO. A significantly reduced (***p<0.0001) number of ATG_MO-injected morphants was observed in the normal morphological class of larvae when compared to control groups. C Graph representing the relative klhl13 expression in the ATG_MO and SS_MOs compared to the uninjected control, normalized to gapdh. SS_MOs showed a significant reduction in the relative klhl13 expression (**p<0.0044), whereas ATG_MO revealed a comparable klhl13 expression (ns: p<0.0955) to uninjected control. D Visual motor startle behavior analysis demonstrated delayed motor reflexes during the dark-to-light cycle transition stage in ATG_MO injected morphants compared to control injected larvae (*p<0.01, **p<0.001, ***p<0.0001, ****p<0.00001). E Representative confocal images of 5 dpf neurod1:EGFP transgenic zebrafish larvae. Compared to the control groups (uninjected and control_MO-injected), larvae injected with ATG_MO exhibited a significantly enlarged brain (***p<0.00001; quantified in panel F), particularly in the optic tectum (highlighted by red lines). Additionally, the midbrain-hindbrain boundary appeared disrupted in ATG_MO-injected larvae (indicated by red boxes). Anatomical regions are labeled: OE: olfactory epithelium; OB: olfactory bulb; RGC: retinal ganglion cells; TL: torus longitudinalis; OT: optic tectum; CR: cerebellum. Scale bar: 200 μm. F Co-injection of human KLHL13^WT mRNA with ATG_MO significantly rescued the brain phenotype, restoring brain size and the integrity of the midbrain-hindbrain boundary (****p<0.0001 compared to ATG_MO alone). G Visual motor startle behavior analysis showed a significant improvement in the overall locomotor activity in the human KLHL13^WT mRNA and ATG_MO co-injected larvae in the light/dark transition period compared to the ATG_MO injected larvae (**p<0.001, ***p<0.0001, ****p<0.00001). H The graph represents the number of ATG_MO injected morphants as well as human KLHL13^WT and NDD-associated variants mRNA coinjected morphants in various morphological categories. Microinjections of human KLHL13^WT encoding mRNA, along with ATG_MOs, significantly rescued the number of larvae in the normal developmental class as compared to ATG_MO alone (***p<0.0002). However, NDD-associated KLHL13 hemizygous variants (c.179A>G, c.557G>A, c.953C>T, and c.1781_1782del) encoding mRNAs did not rescue the phenotype when compared to the normal class of KLHL13^WT encoding mRNA (***p<0.0001), supporting their pathogenic nature. Moreover, significantly more larvae in the severe class were found when ATG_MO was co-injected with human KLHL13 mRNAs harboring c.179A>G (**p<0.0012) and c.1781_1782del variants (**p<0.0052).

Next, to determine the brain function of klhl13 morphants, we assessed their behavior in response to light and sound stimuli. For these studies, we only used normal and mild class morphants. As compared to controls, the ATG_MO injected klhl13 morphants displayed significant delays in their locomotor activity both after light and sound stimuli (Figure 4D; Supplemental Figure 17C), supporting impaired neuronal activity. Among the salient clinical phenotypes of participants harboring variants in KLHL13 was macrocephaly (Figure 1C); therefore, we also measured the head size to body length ratio of control- and ATG_MO larvae (Supplemental Figure 17D, E). As compared to control_MO injected larvae, the klhl13 ATG_MO injected larvae had a significantly larger head size to body length ratio, supporting macrocephaly (Supplemental Figure 17E). To further investigate the impact of klhl13 knockdown on brain development, we microinjected ATG_MO in the neurod1-eGFP transgenic zebrafish, which express GFP in the developing neurons.³³ When analyzed at 5 dpf, a significant increase in the brain size of klhl13 morphants was observed (Figure 4E, F). Particularly, the optic tectum (OT) area in the midbrain, equivalent to the mammalian superior colliculus that processes visual sensory information and generates motor reflexes,⁴⁴ was abnormally large (Figure 4F). Moreover, pyramidal and projection neurons within the optic tectum were accumulated towards the tegument and torus longitudinalis, and the midbrain-hindbrain boundary (MHB) that connects the optic tectum and cerebellum to coordinate overall skillful body movements and generate motor responses⁴⁵ was also not intact in klhl13 morphants (Figure 4E). Finally, the number of parallel fibers that join the two halves of the cerebellum in the hindbrain was also reduced (Figure 4E). Importantly, co-injection of human KLHL13^WT mRNA with ATG_MO significantly rescued the brain size, the neuronal pattern deficits in the optic tectum and cerebellum (Figure 4E, F), as well as overall locomotor activity, particularly at light/dark transition points (Figure 4G). Taken together, these results suggest that the abnormal locomotor activity observed in klhl13 morphants likely stemmed from structural deficits in the optic tectum and cerebellum regions.

Next, to study the in vivo impact of NDD-associated variants, we co-injected 1–2 cell stage embryos with klhl13 ATG_MO along with either WT or variants harboring human KLHL13 mRNAs (Figure 4H). Similar to the rescued phenotypes observed in the brain size and function (Figure 4F, G), we also observed significantly higher klhl13 morphants in the normal class after co-injection with human KLHL13^WT mRNA (Figure 4H). In contrast, mRNAs harboring NDD-associated variants were not able to rescue the phenotypic abnormalities (Figure 4H), thus confirming their pathogenic nature. In summary, these observations revealed compelling evidence of the critical role of KLHL13 in neurodevelopment and brain activity.

Our studies in heterologous cells revealed impaired ubiquitination and expression of AURKB substrate due to KLHL13 NDD-associated variants (Figure 3); thus, we wondered if klhl13 knockdown also dysregulates aurkb in zebrafish. Using the anti-aurkb antibody,⁴⁶ we performed western blot analyses on whole protein lysates collected from the control, ATG_MO injected, as well as human KLHL13^WT mRNA+ATG_MO co-injected zebrafish larvae. In contrast to controls, significantly high expression of aurkb was observed in klhl13 morphants, which was significantly reduced in KLHL13^WT mRNA+ATG_MO co-injected morphants (Figure 5A, B). These results suggest that similar to higher vertebrates, zebrafish aurkb is also a substrate for klhl13-mediated ubiquitination, and thus treatment with AURKB inhibitory drug AZD1152-HQPA might also rescue the klhl13 morphant phenotype and neuronal function.

Figure 5. AZD1152-HQPA treatment rescued developmental and behavioral deficits in klhl13 zebrafish morphants.

A Endogenous zebrafish aurkb expression level at 3dpf in the uninjected control, KLHL13^WT mRNA + ATG_MO co-injected, and in the ATG_MOs alone injected larvae. B Quantification of the aurkb expression. The klhl13 morphants (ATG_MOs) revealed significantly high aurkb levels (**p>0.0048) as compared to uninjected controls. Intriguingly, co-injections with the human KLHL13^WT mRNA significantly (*p>0.037) reduced the aurkb levels in the klhl13 morphants. C Treatment of klhl13 ATG_MO injected zebrafish larvae within 24 hpf with AZD1152-HQPA inhibitor (30nM) significantly (**p<0.001, ***p<0002) rescued the number of larvae in the normal developmental class. D Visual motor startle behavior analysis demonstrated a significantly impaired locomotor activity in ATG_MO injected larvae during the dark-to-light transition period in terms of distance (mm) traveled. Intriguingly, AZD1152-HQPA inhibitor (30nM) treatment significantly improved the locomotor activity when compared to the untreated ATG_MO alone injected group (*p<0.01, **p<0.001, ***p<0.0001, and ****p<0.00001). E Brain morphology of neurod1-eGFP zebrafish at 5 dpf. ATG_MO injected larvae after treatment with AZD1152-HQPA (30nM) had a significant improvement in the overall normal brain size (comparable to controls), particularly the optic tectum, when compared to the untreated ATG_MO, quantified in panel F. Moreover, the midbrain-hindbrain boundary was also intact in the AZD1152-HQPA treated ATG_MO injected larvae *p<0.0419, ***p<0.0001, ****p<0.00001). Scale bar: 200 μm. G Schematic of known kelch-dependent E3 ubiquitin ligase-mediated ubiquitination of the AURKB pathway. H Based on the data presented in this study, we hypothesize that the impairment of KLHL13 function in the recognition and presentation of AURKB to the ubiquitin (UB) molecules for the ubiquitin-dependent proteasomal degradation caused abnormal cell cycle regulation. I AZD1152-HQPA-mediated inhibition of AURKB can restore genomic stability and AURKB levels at the central spindle in cells expressing KLHL13 pathogenic variants. Thus, the AZD1152-HQPA inhibitor could be further developed and evaluated as a potential treatment option for brain disorders stemming from KLHL13 dysfunction.

AZD1152-HQPA treatment partially rescued the neurobehavioral phenotypes of klhl13 zebrafish morphants

To evaluate the impact of drug treatment on klhl13 zebrafish morphants’ neurodevelopment and behavioral function, 1dpf ATG_MO injected larvae were treated with AZD1152-HQPA inhibitor (30nM) for 48 hours, followed by phenotypic analysis. Intriguingly, we found a significant rescue of the normal phenotypic class of zebrafish in the treated morphants as compared to untreated ATG_MO-only morphants (Figure 5C). The treated morphants also showed a significant improvement in the overall locomotor activity when compared to the untreated ATG_MO alone at light/dark and sound/no sound transition points (Figure 5D; Supplemental Figure 19). Furthermore, the optic tectum area of AZD1152-HQPA inhibitor-treated ATG_MOs was significantly smaller than that of ATG_MO alone-injected morphants, along with a decrease in the accumulation of the pyramidal and projection neurons in OT (Figure 5E, F). Finally, the MHB was found intact, and the number of parallel fibers in the cerebellum was restored in the treated morphants (Figure 5E, F). In summary, the AZD1152-HQPA inhibitor improved the overall development, brain size, neuronal patterning, and behavioral function of klhl13 morphants.

Discussion

We identified KLHL13 variants in seven individuals who exhibited mild to severe intellectual disability (ID), developmental and speech delays, macrocephaly, hypotonia, dysmorphic facial features, chewing and swallowing difficulties, unsteady gait, and various behavioral abnormalities (Supplemental Table 1). The identified NDD-associated KLHL13 variants significantly impacted the regulation of cell division and genomic stability in heterologous cells. Protein expression, ubiquitination, and localization of the KLHL13 substrate, AURKB, at the central spindle during anaphase was significantly altered due to the identified variants. AURKB was accumulated at the chromosomal arms instead of at the central spindle, which consequences in the abnormal segregation of sister chromatids with the formation of bridging, leading to abnormal cytokinesis. The phenotypes observed in our study are consistent with loss-of-function effects. Similar findings have been reported in other studies, where the overexpression of mutant genes resulted in phenotypes mimicking loss of function due to interference with normal protein function.⁴⁷ Together, our in vitro findings implicated that NDD-associated KLHL13 variants likely impacted the formation of the E3 ubiquitin ligase complex for the ubiquitination of the target substrate AURKB, leading to impaired regulation of the normal cell cycle. While our functional studies in heterologous cells provide valuable mechanistic insights, validation in patient-derived cells would certainly enhance the clinical relevance of our findings, especially for the c.1781_1782del variant, which lies in the last exon of KLHL13 and may escape nonsense-mediated decay.⁴⁸

To study the KLHL13 function in brain neuronal cells, we modeled KLHL13 dysfunction in zebrafish, which also resulted in severe neurodevelopmental and behavioral comorbidities comparable to human participants. The klhl13 morphants exhibited macrocephaly, neuronal accumulation in the optic tectum, and an incomplete midbrain-hindbrain boundary, resulting in sensorineural delay and reduced locomotor activity. Co-injection of human KLHL13^WT mRNA, but not NDD-associated variants harboring mRNAs, was able to rescue the morphological, neuronal, and behavioral deficits in the klhl13 morphant zebrafish. The clinical phenotype of macrocephaly in individuals with KLHL13 variants is consistent with observations in other neurodevelopmental disorders, where impaired cell cycle regulation and cell proliferation lead to brain enlargement. For example, pathogenic variants in genes involved in cell cycle regulation, such as PTEN,^{49; 50} are associated with macrocephaly due to increased cell size, altered dysregulation of neuronal differentiation, and gliosis. Furthermore, studies in other animal models would help in deciphering the primary cause of macrocephaly and NDD due to impaired KLHL13 function. Collectively, our in vivo functional studies demonstrated the critical role of KLHL13 in neurodevelopment within both zebrafish and humans.

The KELCH proteins play an essential role in the recognition and ubiquitin-mediated proteasomal degradation of their target substrates.⁸ To date, 42 KLHL proteins have been identified in the human genome. However, little to no knowledge is available of the functions of most of these proteins. Some KELCH proteins have been implicated in essential cellular mechanisms, including transcription, cell signaling, cell division, apoptosis, response to oxidative stress, autophagy, and cytoskeletal remodeling. Consistent with their diverse activity, variants in a few KELCH proteins have been associated with inherited as well as somatic disorders.^{15; 51} For instance, haploinsufficiency or loss of KLHL19 has been associated with various types of cancers.⁵² Additionally, variants in KLHL3 (HGNC: 6354), KLHL7 (HGNC: 15646), KLHL9 (HGNC: 18732), KLHL12 (HGNC: 19360), and KLHL16 (HGNC: 4137) have been implicated in various inherited human disorders, including familial hyperkalemic hypertension (MIM: 614495),⁵³ autosomal dominant retinitis pigmentosa (MIM: 613731),⁵⁴ distal myopathy (MIM: 160500),⁵⁵ Sjogren’s syndrome (MIM: 270150),⁵⁶ and giant axonal neuropathy (MIM: 256850), which affects both the central and peripheral nervous systems with the predominantly dysfunctional axons. Under normal conditions, KLHL37 (HGNC: 3345) is expressed in the neurons and not the astrocytes. However, the brain tumor cells of astrocytic origin express the KLHL37, supporting the role of KLHL37 in growth and cytoskeletal organization.⁵⁷ Our study demonstrates the essential role of KLHL13 in neurodevelopment and expands our understanding of the genetic basis and the association of KLHL13 variants with neurodevelopment disorders and intellectual disability in humans.

Our studies in heterologous cells and the zebrafish model suggest that the phenotypes observed in our participants are likely stemming from the disruption of the ubiquitination process mediated by the KLHL13-CUL3-RBX1 complex. Prior studies have reported that variants in CUL3 also cause neurodevelopmental disorders in humans, with clinical features including autism spectrum disorder, developmental delay, intellectual disability, and attention deficit hyperactivity disorder.⁵⁸ CUL3 is a core component of the ubiquitin E3 ligase complex, and the human disease-associated variants impair the protein degradation process, leading to cellular dysfunction.^59–61 The overlapping clinical presentation of individuals harboring variants in KLHL13 or CUL3 genes suggests a shared molecular mechanism within the CUL3-KLHL13 ubiquitination axis.^{19; 62} Apparently, through the ubiquitination and proteasomal degradation of AURKB in neurons, KLHL13 regulates brain development and cognitive function. Intriguingly, treatment with the pharmacological antineoplastic AZD1152-HQPA inhibitor of AURKB rescued normal development, brain size, and behavioral function of klhl13 morphant zebrafish. Considering the overlapping clinical presentation and likely shared molecular mechanism, it might be worth studying the impact of AZD1152-HQPA inhibitor on the cellular and/or animal models of CUL3.

Aurora kinases (AURKA, B, and C) play a critical role in the regulation of cell division, and their functional deficits contribute to various pathologies, including tumorigenesis.⁶³. The Overexpression of AURKB has been documented in multiple anomalies, including acute myeloid leukemia (AML; MIM: 601626),⁶⁴ anaplastic thyroid cancer (MIM: 188550),⁶⁵ and colorectal carcinoma (MIM: 120470).⁶⁶ Thankfully, there are a number of AURKB kinase inhibitors that have been developed and evaluated as anti-cancer drugs.^{67; 68} Preclinical trials have determined that inhibitors that target Aurora kinases can lead to growth arrest in the cell lines,⁶⁹ and prevent tumor expansion in the xenograft studies.⁷⁰ Among the known inhibitors, AZD1152-HQPA (hydroxyquinazoline pyrazole anilide; a.k.a. Barasertib) has shown the most promising response rate for AML in several studies.^71–76 AZD1152-HQPA has 1000-fold more selectivity for the AURKB protein⁷⁷ as compared to the other Aurora kinases (AURKA, AURKC), along with an IC₅₀ (half maximal inhibitory concentration) value of about 0.37nM.^{78; 79} In the AML cell lines, AZD1152-HQPA has been described to inhibit cell proliferation, leading to polyploidy and subsequent elevation in apoptosis.⁶⁹ Moreover, in the small cell lung carcinoma with the overexpression of MYC family proteins, AZD1152-HQPA inhibited the tumor proliferation in vivo.⁸⁰ Further, inhibiting AURKB led to the growth inhibition of human, colon, hematological, and lung malignancy xenografts in immunodeficient mice.^{40; 70; 71} Phase I/II clinical studies showed a well tolerability for AZD1152-HQPA in AML patients.^{72; 73} Thus, considering the safety profile and significant rescue of normal brain size and activity, AZD1152-HQPA could be repurposed to treat individuals harboring KLHL13 variant-linked NDD.

In summary, our in vitro and in vivo findings suggest that targeting AURKB with its selective inhibitor AZD1152-HQPA can effectively ameliorate the cellular and developmental phenotypes associated with KLHL13 variants, thereby rescuing the downstream cellular and developmental abnormalities. (Figure 5G–I). These findings are consistent with previous studies demonstrating the efficacy of AZD1152-HQPA in correcting mitotic defects in various disease models.^{69; 71} Given the promising results from preclinical and clinical studies, AZD1152-HQPA holds potential as a therapeutic agent for neurodevelopmental disorders involving AURKB dysregulation along with KLHL13 variant-associated syndrome.

Supplementary Material

The online version of this article contains supplemental material, which is available to authorized users.

Data availability

The data supporting the findings of this study are available from the corresponding author upon request. Source data are provided in this paper.