Maintaining the integrity of the proteome through effective protein turnover is essential for the proper functioning of all cells (Hinkson and Elias, 2011; Ross et al., 2021). Proteins that are unnecessary for current functions, misfolded, or damaged (due to oxidative stress, disease, or aging) must undergo degradation, with their amino acids recycled. This process is especially critical for cells with extended lifespans and intricate morphologies, such as neurons (Hanus and Schuman, 2013; Morozov and Karpov, 2018; Patrick et al., 2023). Beyond the shared processes common to all cells, protein degradation serves unique functions in neurons, such as shaping synapse formation and pruning during development and regulating excitatory and inhibitory transmission in the mature brain (Jarome and Helmstetter, 2013; Turker et al., 2021).

The ubiquitin-proteasome system is the primary mechanism responsible for the targeted degradation of aberrant and short-lived regulatory proteins (Nath and Shadan, 2009). This system involves a cascade of reactions successively catalyzed by three families of enzymes: ubiquitin-activating E1, ubiquitin-conjugating E2, and ubiquitin-ligase E3 enzymes. E3 ligases are noteworthy as they can directly bind to substrates and typically provide specificity for the substrate protein. The significance of E3 ligases is underscored by the many cellular processes they regulate and, correspondingly, the numerous diseases, including several neurological disorders, associated with their loss of function or mistargeting (George et al., 2018; Kawabe and Stegmuller, 2021). Two primary classes of E3 enzymes exist: HECT (homologous to E6-AP carboxyl-terminus) domain E3s and RING finger E3s. Due to its association with several neurological disorders, the ubiquitin protein ligase E3A (UBE3A, OMIM# 601623), the first E3 discovered in the HECT class (Figure 1A), holds particular significance in the central nervous system.

In the central nervous system, UBE3A, the gene that encodes UBE3A protein, undergoes imprinting, wherein the paternally-inherited allele is epigenetically silenced in neurons during development (Figure 1B) (Rougeulle et al., 1997; Judson et al., 2014). Mechanistically, paternal UBE3A silencing is disrupted by the transcription of an opposing long, non-coding RNA commonly referred to as UBE3A-ATS (Rougeulle et al., 1998; Hsiao et al., 2019). Consequently, only the maternal copy of UBE3A is expressed in neurons. Mutation or deletions of the maternally-inherited UBE3A allele leads to Angelman syndrome (AS, OMIM # 105830), a rare neurodevelopmental disorder characterized by developmental delay, severe intellectual disability, absence of speech, seizures, and ataxia (Buiting et al., 2016). Conversely, gain-of-function mutations in UBE3A, or the overexpression of UBE3A caused by extra copies of the 15q11-q13 chromosomal region in which UBE3A resides, are associated with autism spectrum disorder (Dup15q syndrome) (Hecht, 1975; Christian et al., 2008; Glessner et al., 2009; Xing et al., 2023).

Despite the identification of various putative targets of UBE3A protein (also known as E6-AP) (Sell and Margolis, 2015), the mechanisms governing the downstream effects of UBE3A gain or loss of function in neurodevelopmental disorders remain poorly understood, and effective treatments are lacking. As a result, current treatments for AS mainly focus on symptom management, utilizing anti-seizure medications and physical therapy approaches (Duis et al., 2022). Recent research by our lab and others has focused on reinstating the UBE3A protein, which can be achieved by reintroducing the UBE3A gene (Judson et al., 2021) or suppressing the UBE3A-ATS to reinstate expression of the paternal UBE3A allele (Huang et al., 2011; Meng et al., 2015; Wolter et al., 2020). While restoring UBE3A protein function holds potential therapeutic benefits, several obstacles must be addressed. Comprehensive understanding of UBE3A expression in a region- and cell-type-specific manner, both during early developmental stages and throughout adulthood, is a sine qua non for developing and evaluating potential therapeutic safety and effectiveness.

While human induced pluripotent stem cells provide some insights into UBE3A expression in humans (Chamberlain et al., 2010; Sirois et al., 2020), our knowledge of UBE3A expression within the intact human brain remains limited and has primarily taken advantage of western blotting (Williams et al., 2010; Daily et al., 2012). We previously used immunohistochemistry to explore UBE3A distribution in the human brain, but our focus was constrained to the adult temporal cortex due to the scarcity of high-quality human tissue samples (Burette et al., 2018). The assumption of ubiquitous UBE3A expression in the human brain and the broader idea of conserved UBE3A biology between humans and rodents stems primarily from research conducted on mice, but direct evidence is lacking. While there are optimistic grounds to believe in the validity of this assumption, caution is warranted.

This study aims to fill existing gaps in knowledge by exploring the temporal and spatial expression patterns of UBE3A and UBE3A-ATS in the developing and adult brains of macaque monkeys. Macaques are a compelling model for the study of neurodevelopmental disorders (Aida and Feng, 2020). Macaque and humans share numerous brain regions, circuits, and cell types critical for higher cognitive functions. Additionally, macaques exhibit advanced cognitive and social communication abilities, including face recognition, eye gazing, vocalizations, and complex behavioral patterns that are challenging to model in rodents.

We first mapped UBE3A protein expression in the postnatal macaque brain using immunohistochemistry (Figures 2– 4). The regional distribution of UBE3A in the infant macaque brain (2 weeks, 1 month, and 3 months of age) closely resembled that of the adolescent brain (5.5-year-old, Figure 2). Immunoperoxidase-reacted sections exhibited staining for UBE3A throughout the post-natal macaque brain, concentrating in the grey matter (Figures 2, 3). The hippocampus, cerebellum, and neocortex exhibited the highest UBE3A staining. The striatum, thalamus, and brainstem displayed comparatively weaker staining. The white matter exhibited the least intense staining.

UBE3A was found in all layers of the cerebral cortex (Figure 4A). Staining was weaker in layer 1 (Figure 4B) and the white matter (Figure 4C), reflecting lower cell density. At the subcellular level, the staining concentrated in nuclei, with comparatively weaker staining detected in somata and dendrites (Figure 4D). In the hippocampus, UBE3A staining ranged from moderate to intense in layers abundant in somata, encompassing the pyramidal cell layer of the hippocampus and the granule cell layer of the dentate gyrus (Figures 2, 3, 4E). Staining was comparatively lower in the paucicellular stratum radiatum and oriens. The subcellular distribution mirrored that of the neocortex, with prominent nuclear staining and weaker staining in the neuropil. In the basal ganglia, strongly stained nuclei were dispersed throughout the gray matter, while the fascicles of myelinated fibers exhibited minimal staining (Figure 4F). Positive nuclei were also present in the diencephalon (Figure 4G) and rhombencephalon (Figures 4H–I). In the cerebellar cortex, UBE3A staining concentrated in the nuclei of Purkinje cells, and in small nuclei scattered in the molecular layer (Figure 4J). In contrast, staining was lower in the nuclei of granule cells.

Next, we explored UBE3A expression in neurons and glial cells. Simultaneous labeling for UBE3A and the neuronal marker NeuN showed that nearly all NeuN-positive cells were also positive for UBE3A (Figure 5). However, the nuclear expression of UBE3A in neurons exhibited variability, as evidenced by heightened staining in larger nuclei and diminished levels in smaller nuclei (Figures 5B,C). Given the established connection between deficits in GABAergic circuitries and AS pathogenesis (Judson et al., 2016; Gu et al., 2019), we examined UBE3A expression in glutamatergic and GABAergic neurons. Multiplex ISH experiments indicated that the majority of VGLUT1-positive (glutamatergic) and VGAT-positive (GABAergic) neurons coexpressed UBE3A transcript (Figures 5D,G), and suggest lower UBE3A levels in VGAT-positive neurons (Figures 5E,F). Not all UBE3A-positive cells were identified as neurons based on NeuN, VGLUT, or VGAT expression, suggesting that a significant portion comprises glial cells. To confirm UBE3A expression in glial cells, we used multiple labeling and multiplex ISH for UBE3A and SOX9 (Figure 6). We chose to examine SOX9 over GFAP because SOX9 is known to be expressed in all astrocytes (Sun et al., 2017). Immunohistochemistry revealed UBE3A protein in all SOX9-positive nuclei, albeit at a lower level than neurons (Figures 6A,B). Complementary examination using in situ hybridization (ISH) shows astrocytes displayed reduced UBE3A mRNA labeling compared to neurons (Figures 6C,D). We also investigated UBE3A expression in oligodendrocytes, finding that, like astrocytes, this cell type appeared to have lower UBE3A protein levels than neurons (Figures 7A,B).

To further understand UBE3A expression variations among cell types, we compared UBE3A levels in neurons, astrocytes, and oligodendrocytes (Figures 7C,D). We selected the visual cortex for study because its structure and size facilitate the collection of small yet uniform pieces, thereby conserving the limited macaque tissue available. Moreover, changes in dendritic morphology have been observed within visual cortex in one of the only human AS anatomical studies to date (Jay et al., 1991), suggesting this is a potential site of pathology in Angelman and Dup15q syndromes. Cells were identified based on specific markers: NeuN for neurons, SOX9 for astrocytes, and Olig2 for oligodendrocytes. Neurons showed higher UBE3A expression compared to astrocytes and oligodendrocytes in agreement with our prior findings in mouse and human studies (Burette et al., 2017, 2018). Intriguingly, our data also suggest the existence of two distinct astrocyte populations based on UBE3A expression: a predominant group constituting 80% of all astrocytes and a smaller group (20%) characterized by approximately half the UBE3A concentration.

To gain insight into paternal UBE3A silencing, we performed ISH for UBE3A-ATS (Figure 8). In contrast to mRNAs, whose signals are scattered across the cytoplasm, the UBE3A-ATS signal was confined to nuclei and observed as small, bright focal puncta, often surrounded by several tiny speckles (Figure 8). A close 3D examination of the z-stack dataset showed a single UBE3A-ATS punctum per nucleus (Figure 8A). These observations are consistent with previous findings supporting that UBE3A-ATS expression is mono-allelic with a very short (~4 h) half-life (Meng et al., 2012).

Multiplex experiments show the presence of UBE3A and UBE3A-ATS signals throughout the post-natal macaque brain (Figures 8B,C). However, not all UBE3A-expressing cells also exhibited a UBE3A-ATS signal. In areas characterized by well-organized anatomical structures, cells that lack UBE3A-ATS can be identified as non-neuronal based on their distinctive size, shape, and arrangement. For example, in the cerebellar cortex, UBE3A-ATS positive Purkinje cells nested between UBE3A-ATS negative Bergmann glia (Figure 8D). Of note, the subgranular zone of the hippocampal dentate gyrus, a prominent neurogenic niche in the adult brain, exhibited very little UBE3A-ATS signal (Figure 8E). Validation of the absence of UBE3A-ATS in glial cells was confirmed through multiplex ISH for UBE3A and SOX9 (Figure 9).

Collectively, these findings delineate a UBE3A expression pattern in the postnatal macaque brain closely resembling that observed in mice, characterized by robust nuclear expression in neurons across the whole brain and comparatively weaker expression in glial cells. Notably, the expression profile of UBE3A-ATS suggests that paternal UBE3A is silenced in neurons but not in glial cells, mirroring observations in mice.

To investigate UBE3A and UBE3A-ATS in the prenatal macaque brain, we focused on the neocortex. The neocortex has undergone significant expansion during primate evolution, lending to markedly greater complexity of this structure in the macaque compared to mice. Additionally, the well-studied ‘inside-out’ development of the neocortex provides the anatomical advantage of observing multiple developmental stages simultaneously across the cortical thickness. We studied three-time points: GD 48 during the neurogenesis of deep cortical layers, GD 100 during the neurogenesis of upper layers and the start of neuronal maturation in deep layers, and GD 151 toward the end of cortical synaptogenesis in all cortical layers (Wallace and Pollen, 2024). At GD 48, UBE3A staining was widespread across the cortical wall but most prominent at the interface of the marginal zone and the cortical plate and in the ventricular zone. Lower UBE3A intensity was evident in the subventricular zones (Figures 10A–C). At the subcellular level, UBE3A staining appeared diffuse, filling the thin layer of cytoplasm surrounding largely unstained nuclei and extending into cellular processes. In contrast to UBE3A’s widespread distribution, UBE3A-ATS signal was seldom observed at GD 48 (Figure 11).

At GD 100, there were significant changes in UBE3A subcellular localization (Figures 10D–G). In the superficial layers, UBE3A staining remained predominantly extra-nuclear, like at GD 48. However, in the deeper, more developed layers, UBE3A was primarily located inside the nucleus, as it is after birth. This developmental shift in UBE3A subcellular distribution was accompanied by changes in UBE3A-ATS expression, with cells in the more mature layer 6 exhibiting a notable increase in UBE3A-ATS compared to their younger counterparts near the overlying marginal zone (Figures 12A–D). At GD 151, UBE3A staining was present throughout the developing cerebral cortex, resembling the postnatal distribution (Figures 10H–K). UBE3A staining concentrated in cell nuclei across all layers, with additional staining in the cytoplasm, cellular processes, and neuropil, albeit at reduced levels. UBE3A-ATS signal was found in numerous cells, many displaying sufficient maturity to be identified as neurons when counterstained with NeuroTrace (Figures 12E–H).

Our findings indicate a dynamic shift in UBE3A subcellular localization between GD 48 and GD 151, with UBE3A progressively becoming more nuclear. Interestingly, this shift coincides with the increase in UBE3A-ATS expression in mature neurons.

While the overall trends in the spatial and temporal gene expression patterns are generally similar between humans and macaques, variations exist. These differences are particularly evident in genes related to synapse formation, neuronal development, and myelination (Zhu et al., 2018). The limited availability of high-quality human tissue samples has hindered our ability to perform detailed histological analyses. Therefore, to determine the applicability of our findings to humans, we analyzed publicly available human RNA sequencing data (Figure 13). UBE3A expression peaked during the first trimester in humans, after which it gradually decreased until birth. Following a brief increase in the early postnatal period, the expression levels of UBE3A stabilize after three years of age, aligning with levels seen at birth. In contrast, the expression of UBE3A-ATS is very low during the first trimester but steadily rises until adolescence. When examined in specific brain regions, these expression patterns show similar trajectories, with minor variations in timing that may be attributed to differences in developmental progression. Notwithstanding the developmental asynchrony observed between macaques and humans, these data suggest that the dynamic expression profiles of UBE3A and UBE3A-ATS in humans closely agree with our histological observations in macaques.

Mice have been the foundation for understanding UBE3A biology. Mouse models have played a critical role in characterizing AS and Dup15q syndrome pathology and are extensively used to design and screen therapeutic strategies. For example, AS mouse models successfully recapitulate various aspects of AS, including seizures and motor deficits. However, these mouse models cannot replicate the full spectrum of phenotypes, especially those related to severe intellectual disability, language impairment, and complex social behaviors. Therefore, as with other disorders affecting core human traits, a key challenge is determining the extent to which findings in murine models can be effectively translated to individuals with AS and Dup15q syndromes. This challenge arises from the substantial evolutionary divergence (~80 million years) between mice and humans, leading to significant discrepancies across various biological domains. These discrepancies range from the genomic and cellular level to neurocircuitry and complex behaviors (Izpisua Belmonte et al., 2015; Kaiser and Feng, 2015; Jennings et al., 2016). Transcriptomic analyses highlight these limitations by revealing substantial differences in molecular signatures between mice and primates; notably, genes implicated in neurological disorders display these variations (Parikshak et al., 2015; Pembroke et al., 2021). Studying UBE3A in the macaque brain thus provides a crucial complement to mouse studies and helps to bridge the gap between rodent and human research. Here, we provide the first detailed account of UBE3A and UBE3A-ATS expression in developing primate brains.

Despite species-specific variations in developmental timing and brain anatomy, our findings reveal remarkable conservation in the developmental trajectory of UBE3A expression between macaques and mice. This strengthens the validity of mouse models for investigating UBE3A-related disorders. Key similarities include a broad regional distribution across all ages examined, expression in most brain cell types, with consistently higher levels in neurons (both excitatory and inhibitory) compared to non-neuronal glial cells, and, most obvious in neurons, a developmental shift from cytoplasmic to predominantly nuclear subcellular localization. Importantly, the expression pattern of UBE3A-ATS in macaques suggests, akin to mice, that paternal UBE3A is developmentally silent in neurons and not in glial cells. The conservation of critical UBE3A biological features between the two species suggests the preservation of functionality. For example, the subcellular location of UBE3A likely influences its interaction with different substrates and, consequently, its functional activity. In particular, the noted similarity in UBE3A’s developmental transition from the cytoplasm to the nucleus in both mice and macaques suggests that UBE3A’s cellular functions are highly conserved during neuronal maturation.

Translating developmental data from animal models to humans presents challenges due to complex timing differences across species (Wallace and Pollen, 2024). Determining exact equivalencies is difficult, as factors like cerebral growth, neurogenesis, and synaptogenesis vary. This complexity is further compounded by regional differences in brain development timing. Despite this, our developmental data offer valuable insights, especially when focusing on neocortical development (Figure 14).

In the neocortex, neuron formation progresses from the inside out. Neurons destined for the outer layers migrate past those forming the deeper layers. Consequently, neurons in the deeper layers mature days before those in the upper layers. In mice, neurogenesis begins in the deeper layers around E11 and finishes in the upper layers by E18 (van den Ameele et al., 2014; Kroon et al., 2019). This is followed by a rapid maturation period for principal neurons during the first two postnatal weeks. The cytoplasmic-to-nuclear shift of UBE3A also occurs during this maturation phase (Judson et al., 2014). The precise profile of Ube3a-ATS expression in mice is unclear. Although Ube3a-ATS can be detected as early as E10 in mice (Meng et al., 2013), analyses of cortical lysates via Western blot revealed that the paternal Ube3a allele is fully silenced in neurons soon after birth (Sonzogni et al., 2020), following the completion of neurogenesis, within the neuronal maturation period. This developmental epoch in mice is very brief, making extrapolation to human timelines difficult. Macaque brain development, similar to human, is much more protracted, leading to a greater temporal resolution of key aspects of UBE3A and UBE3A-ATS biology. We show that in macaques, the increase in UBE3A-ATS expression and the shift of UBE3A to the nucleus happen after neurogenesis, during the initial phases of neuron maturation, and before synapse formation (Rakic, 1975; Wallace and Pollen, 2024). Specifically, in the deeper layers of the cortex, the increase in UBE3A-ATS expression begins between GD 48 and GD 100, while in the outer layers, this increase starts around 50 days later, between GD 100 and GD 151. Therefore, our data support a conserved timeline between mice and macaques where maternal UBE3A silencing and nuclear localization happen post-neurogenesis, coinciding with neuronal maturation.

Fewer data are available for the developing human brain. Our RNA-Seq analysis indicates that in the prefrontal cortex, UBE3A-ATS expression begins in the second trimester and peaks in the third, suggesting that, as in mice and macaques, the maternal allele silencing of UBE3A in the human brain also takes place after the completion of neurogenesis and neuronal migration, as principal neurons begin to mature. Understanding the timeframe when UBE3A-ATS expression begins to silence paternal UBE3A is crucial. Before this period, expression of UBE3A from the paternal allele may partially offset the maternal allele loss in neurons. Once the paternal allele is fully silenced, however, maternal UBE3A loss is likely to be much more impactful. Thus, restoring UBE3A function might yield the greatest long-term benefits if done before closure of this window, which our data indicate lies between the late second and mid-third trimesters. This proposed timing is supported by mouse studies, where the critical window for reversing most behavioral phenotypes spans from birth to P21, the period when neuronal maturation occurs (Silva-Santos et al., 2015; Sonzogni et al., 2020; Milazzo et al., 2021). After P21 in mice, UBE3A functional reinstatement can still mitigate specific conditions such as susceptibility to seizures and alterations in sleep patterns, suggesting that even late reinstatement of UBE3A expression can provide clinical benefits despite having a narrower impact (Rotaru et al., 2018; Gu et al., 2019; Spencer et al., 2022; Lee et al., 2023).

Collectively, our findings support the validity of mouse models for developing and testing AS therapies. Furthermore, they suggest the middle third of pregnancy as an ideal window in humans, during which restoring UBE3A function could yield the greatest benefits. While these findings are promising, they highlight rather than detract from the need for an AS model in macaques. The spatiotemporal expression of UBE3A is crucial for its function, yet this is just one element influencing UBE3A’s biology. Such a model would not only enhance our comprehension of AS, but would also improve the evaluation of both the effectiveness and, critically, the safety of potential treatments during early development, offering a model with greater predictive accuracy for human outcomes.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

All macaque monkey experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of California, Davis. The study was conducted in accordance with the local legislation and institutional requirements.

CGR: Investigation, Methodology, Writing – review & editing. SGS: Investigation, Methodology, Writing – review & editing. RKRP: Investigation, Methodology, Writing – review & editing. SC: Investigation, Methodology, Writing – review & editing. NWM: Methodology, Writing – review & editing. LMJ: Methodology, Writing – review & editing. NWR: Methodology, Writing – review & editing. JMS: Formal analysis, Methodology, Writing – review & editing. JB: Resources, Writing – review & editing. DGA: Methodology, Resources, Writing – review & editing. ACB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Supervision, Visualization, Writing – original draft, Writing – review & editing. BDP: Funding acquisition, Writing – review & editing, Conceptualization, Methodology, Supervision.