New models of Williams syndrome offer translational insights and future directions

Williams syndrome, a rare genetic condition that leaves children extraordinarily friendly yet struggling with anxiety, intellectual disability, and heart defects, has long defied attempts at curative treatment. A comprehensive review published in World Journal of Pediatrics now maps out how a new generation of laboratory models, from genetically engineered mice to human brain organoids grown in dishes, is transforming the field’s understanding of what happens when a chunk of chromosome 7 goes missing, and points toward the first mechanism-based therapies.

The condition arises from the heterozygous deletion of a 1.5 to 1.8 megabase region on chromosome 7q11.23, a stretch of DNA containing 26 to 28 genes. Occurring in roughly one in 7,500 live births, the deletion accounts for a strikingly complex clinical picture. Children with the classic 1.5 megabase deletion, which represents about 95 percent of cases, show a distinctive paradox: an intense drive toward social contact, often described as hypersociability, coupled with poor awareness of personal boundaries, social disinhibition, and difficulty sustaining friendships. Alongside this, they frequently experience attention deficit hyperactivity disorder, anxiety disorders, specific phobias, and heightened sensitivity to sound. Cognitive impairment is typically mild to moderate, with intelligence quotients averaging between 50 and 60, and visuospatial abilities are disproportionately affected, while fine motor skills are also commonly impaired.

Neuroimaging studies have begun to explain how this behavioral profile emerges from altered brain structure. Magnetic resonance imaging reveals total brain volume reductions of 10 to 15 percent in affected individuals, accompanied by diminished gray matter in the thalamus and the parieto-occipital regions that underpin the human visual-spatial system. Functional abnormalities in the hippocampal formation contribute to deficits in spatial navigation and long-term memory, while alterations in limbic circuitry, particularly the amygdala-prefrontal circuitry, are closely linked to hypersociability and increased anxiety. Intriguingly, cerebellar volumes relative to intracranial volume are increased, a finding that may relate to the motor deficits and poor short-term memory seen in patients.

The central challenge for researchers has been translating these clinical observations into testable biology. Because clinical cohorts are small, variable, and constrained by ethical considerations, animal and organoid models have become indispensable. The review, led by Ya-Yue Chen and colleagues at Zhejiang University School of Medicine, synthesizes two decades of progress across these platforms, with particular emphasis on how they illuminate neurodevelopmental impairments.

Mouse models have proven especially valuable because the genes in the Williams syndrome critical region are conserved between humans and mice, clustered on mouse chromosome 5G2. In 2009, Li and colleagues generated proximal and distal deletion strains using Cre-loxP technology, which when crossed produce double heterozygous mice mimicking the human deletion. These mice exhibit reduced brain volumes, altered neuronal distribution in the somatosensory cortex, heightened social interest, and impaired motor coordination. A landmark advance came in 2014, when the laboratory of Victoria Campuzano created a complete deletion model spanning Gtf2i to Fkbp6, avoiding the complicating homozygous loss of Limk1 present in earlier strains and more accurately replicating the human condition.

Complete deletion mice have since become the workhorse of the field. They display increased startle responses to acoustic stimuli, impaired fear memory, and hypersociability, mirroring the human phenotype. Studies of infant and adolescent mice have revealed reduced body growth, delayed sensory development, and altered patterns of ultrasonic vocalizations. Perhaps most importantly, these mice have become a platform for preclinical drug screening. Combined oral administration of verapamil, an L-type calcium channel blocker, and curcumin for five weeks significantly improved neurodevelopmental impairments in adolescent mice, apparently by modulating inflammasome-related, MAPK, and PI3K/AKT signaling pathways and reducing the number of activated microglia. Notably, neither drug worked alone. In a separate line of work, daily injections of JZL184, a selective monoacylglycerol lipase inhibitor, specifically normalized the social and cognitive phenotypes of mutant mice, suggesting that modulation of the endocannabinoid system represents a promising therapeutic avenue. By contrast, oxytocin, sometimes proposed as a treatment for social deficits, failed to attenuate fear memory impairments in these animals.

The review also dissects the contributions of individual genes within the deleted region, work that has been guided by patients carrying smaller, atypical deletions. When GTF2I and GTF2IRD1 are retained, visuospatial abilities and intellectual function are notably better preserved, and the intraparietal sulcus, a key node for visuospatial processing, remains structurally unaffected. GTF2I, which encodes a multifunctional transcription factor involved in embryonic development, cell cycle regulation, and epigenetic control, appears to account for roughly 10 to 20 percent of the transcriptional dysregulation seen in patient-derived cells, likely through disrupted interactions with the LSD1 repressive chromatin complex. In 2024, Adams and colleagues used CRISPR/Cas9 to generate GTF2I-deficient human cells and found increased proliferation in neural progenitor cells alongside increased cell death and synaptic dysregulation in resulting neurons and cortical organoids. In mice, heterozygous loss of Gtf2i produces hypersociability, reduced motor coordination, increased anxiety, and auditory hypersensitivity, and conditional deletion in forebrain excitatory neurons causes myelination defects that can be pharmacologically rescued by remyelinating drugs such as clemastine and 4-aminopyridine.

Other genes add further layers of complexity. LIMK1, which regulates the actin cytoskeleton by phosphorylating cofilin, has been implicated in impaired visuospatial constructive cognition, and mouse studies reveal effects on dendritic spines, hippocampal long-term potentiation, and long-term memory that appear to operate through the transcription factor CREB rather than cofilin itself. CLIP2, a microtubule-binding protein, contributes to hippocampus-dependent memory and motor coordination when haploinsufficient. Additional genes, including Ncf1, Eif4h, Stx1a/b, Dnajc30, Fzd9, Nsun5, Hip1, and Cldn3, each produce distinctive phenotypes ranging from mitochondrial dysfunction to impaired oligodendrocyte proliferation and blood-cerebrospinal fluid barrier disruption. Yet the picture is not uniform: Fkbp6, for instance, produces no phenotype unless both alleles are knocked out, a reminder that some symptoms likely emerge from interactions among multiple genes or from physiological differences between mice and humans.

Human-specific insights have come from induced pluripotent stem cell and organoid technologies, which preserve the pathogenic mutations of individual patients while recapitulating molecular signaling pathways of human neurodevelopment. Early work showed that cortical neurons derived from patient cells exhibit prolonged action potential repolarization and deficits in voltage-activated potassium currents, alongside morphological alterations consistent with postmortem findings. More recently, forebrain organoids derived from patients have revealed abnormal proliferation and differentiation of neural progenitor cells and aberrant expression of neurodevelopmental genes. A particularly elegant strategy employs CRISPR/Cas9 editing to create isogenic cell lines that differ only at the 7q11.23 region, allowing researchers to study dosage effects against an identical genetic background. Using this approach in 2024, researchers demonstrated that ribosome biogenesis plays a key role in the neurodevelopmental disorder, a finding with potential therapeutic implications. Gene-edited organoids carrying deletions spanning NSUN5 to GTF2IRD2 have similarly revealed significant downregulation of synaptic genes and pathways relevant to GABAergic neurons.

The authors are candid about limitations on both sides of the modeling divide. The Williams syndrome critical region is inverted between human chromosome 7q11.23 and mouse chromosome 5G2, and broader differences in brain structure and developmental trajectories limit translation; cerebellar volumes, for example, are enlarged in patients but unaffected in some mouse models. Murine microglia differ substantially from their human counterparts, complicating the study of neuroinflammation. Organoids, meanwhile, lack vascularization and immune components, which restricts their use in modeling cardiovascular phenotypes and can lead to hypoxia and limited tissue maturation in long-term cultures. The rarity of clinical samples and the reliance on single-sex iPSC lines in some studies further constrain generalizability.

Future directions outlined in the review include coculturing brain organoids with blood vessel organoids to form neural-specific vascular networks, introducing microglia or microglial progenitors to build immune-competent models, and deploying microfluidic organ-on-a-chip systems that mimic the blood-brain barrier and fluid shear stress. The authors also propose expanding beyond mice to rat, domestic dog, and non-human primate models. Domestic dogs are particularly intriguing because their hypersociability relative to wolves likely stems from structural variations in GTF2I and GTF2IRD1, the very genes implicated in Williams syndrome. Non-human primates offer comparable cognitive skills and social complexity that would allow assessment of fine motor deficits, rhythmic abilities, and intelligence quotients that are difficult to measure in rodents. Emerging technologies, including single-cell and spatial multiomics and optogenetics, are expected to further enhance model fidelity, while advanced neuroimaging in animal models could bridge preclinical findings to clinically useful biomarkers.

The overarching message is one of cautious optimism. For a disorder in which current treatment remains limited to symptomatic management with antidepressants, anxiolytics, stimulants, surgery for vascular anomalies, and educational support, the convergence of complete deletion mice, human forebrain organoids, and precision gene editing is steadily converting descriptive observation into mechanistic understanding. The path forward, the authors argue, lies in integrating these complementary systems into clinical trials designed specifically for Williams syndrome, using biomarkers to stratify patients and developmentally appropriate endpoints to measure success. Such an approach holds promise for moving beyond symptom management toward treatments that target the biological roots of the disorder.