Introduction
In the brain, neurons are embedded in a dense environment, the extracellular matrix (ECM), which contains a complex array of directional cues. Neuronal development and migration are governed by molecular stimuli, acting over long distances, and by physical signals locally retrieved through contact guidance [
1,
2]. Here, contact sensing triggers complex intracellular signaling patterns that are integrated by cells to guide neuronal adhesion, migration, neurite wiring, and synaptic plasticity [
2,
3]. The outgrowth of neurites in neuronal cells is critically controlled by the development of focal adhesions (FAs), both in vitro [
4,
5] and in vivo [
6,
7]. FAs act as sensors by integrating signals from both the ECM and chemotactic factors [
3,
8] and mediate coordinated rearrangements of the cytoskeleton, essential for both neuronal growth and synaptic functionality [
7]. Neurons integrate information from multiple sources at the level of cytoskeletal signalling that ultimately modulates cell shaping, migration, and contractility [
9]. Thereafter, neuronal growth and guidance to the proper targets require concerted efforts from the cytoskeleton (actin and microtubule) and from adhesions [
10]. In this framework, the dilated tip of developing neurites, i.e., the growth cones (GCs), sense environmental cues leading the axons to their specific targets for precise neuronal wiring. The
actin cytoskeleton is the major component of the GC that powers its directional motility [
10]. However,
microtubules are also important in neuronal/growth cone guidance, because their polarized invasion into the peripheral domain on one side of the GC is essential for it to turn [
11].
Deficits in neuronal micro-connectivity leading to functional connectivity deficits are recently emerging as crucial in many cognitive disorders (e.g., autism spectrum disorders, schizophrenia) [
12]. However, the role of neuronal sensing mechanisms during development and migration is under-investigated in pathological conditions. Recently, a pivotal role of ubiquitination (i.e., at the level of E3 ligases) emerged in the processes which orchestrate adhesion and cytoskeletal signaling pathways [
13]. Among these, the ubiquitin protein ligase E3A (UBE3A) has a key role in neurodevelopment, in particular at early neurodevelopmental stages [
14]. Importantly, the exact level of UBE3A in the brain is crucial: its lack leads to Angelman syndrome (AS; OMIN#105830) [
15], while its increase can cause non-syndromic autism spectrum disorder (ASD) and Dup15q syndrome (OMIN#608636) [
16‐
18]. AS and Dup15q show phenotypic overlap characterized by autistic features, intellectual disability, motor deficits, speech absence/delay, and epileptic seizures [
19,
20]. A strong correlation between AS-associated deficits and the loss of UBE3A ligase activity has been reported [
21], as well as between an autism UBE3A-linked mutation and the hyper-activation of UBE3A [
22]. Importantly, the
Ube3a gene is transcribed to form distinct splice variants encoding two UBE3A protein isoforms (designated as isoforms 2 and 3, in mice). Recent studies showed that these isoforms have different cellular localization and likely different function [
23‐
25].
Ube3a-deficient mouse models showed abnormalities at the level of dendritic arborization and spine development (both in vitro and in vivo), even if with overall conflicting outcomes [
23,
24,
26‐
30], as well as at the level of spine actin reorganization [
31,
32] and dendritic polarity in vivo [
23]. In addition, the increase of UBE3A leads to a reduction in dendritic arborization and synapse formation in cortical neurons [
30]. Finally, UBE3A has been reported to be highly present at GCs [
26,
33]. Although several UBE3A targets were described [
34,
35], still little is known about the role of UBE3As in neuron morphogenesis and the pathogenesis of UBE3A-associated disorders. However, it is noteworthy that UBE3A gene reinstatement in adult mice rescued electrophysiological properties but did not result in a behavioral rescue, suggesting that UBE3A could be involved in neuronal wiring [
14,
36].
Recent developments in biomaterial science allow direct investigation of the processes that control cell contact sensing by using nano-textured substrates [
37]. Nano/microstructured surfaces are capable of tuning the properties of the cell surroundings at the nanoscale and to test cell response to physical stimuli in vitro [
38,
39]. Microgratings (GRs; anisotropic topographies composed of alternating lines of grooves and ridges with sub-micrometer lateral dimensions) could effectively apply directional topographical stimuli via pure contact interaction, and tune neuronal differentiation, polarization, and neurite orientation [
40‐
43]. We previously demonstrated, in differentiating PC12 neuronal cells on GRs, that contact guidance requires Rho-associated protein kinase (ROCK)-myosin-II-activated contractility [
4,
44] and dynamically responds to topographical noise [
45]. Here, nocodazole treatment was demonstrated to improve the neurite alignment to the GR pattern when the topographical stimulus was disturbed [
45]. Importantly, thanks to GRs, we showed for the first time the aberrant morphological phenotype of AS neurons in vitro: neurite contact guidance is defective in
Ube3a-deficient (i.e., model of AS) hippocampal neurons at early stages of development (DIV1-3), and this phenotype is linked to an impaired activation of the FA signaling pathway (at the level of FAK phosphorylation) [
5].
Here, we selectively study axon and dendrite topographical guidance of wild-type, UBE3A-deficient, and UBE3A-overexpressing neurons on micro-grooved substrates. The aim is to unravel the role of UBE3A in neuronal contact guidance and explore neuronal morphological aspects relevant to AS and ASD/Dup15q syndrome. We exploit GRs (with a pattern of 1-μm ridge, 1-μm groove, and 500-nm depth), which transfer an optimal directional impulse to the cells. We further investigate the FA molecular pathway and the link between FA and actin fibers, at the level of α-actinin organization. Finally, we test the potential of either UBE3A reinstatement or pharmacological treatment acting on cytoskeletal contractility for restoring correct topographical guidance in UBE3A-deficient neurons.
Discussion
Here, we investigated the role of UBE3A in contact guidance, with the aim to identify morphological and molecular aspects relevant for neuronal development in AS and ASD/Dup15q. We found that AS HNs have a specific deficit in axonal topographical guidance in response to GR directional stimuli in vitro, and show in parallel increased axonal branching, with more frequent main secondary axonal segments. This deficit is specific to loss of UBE3A, as overexpression of UBE3A has no influence on neuronal guidance. We further demonstrate that AS neurons present impairments in the FA pathway and that their defective axonal guidance can be rescued by low doses of the cytoskeletal drug nocodazole, while only partially by UBE3A protein reinstatement.
Axon formation is the initial step in establishing neuronal polarity [
50] and our in vitro results evidence the presence of a specific deficit in the axonal contact guidance of AS HNs in front of a topographical directional stimulus. In line with our present results, Miao et al. [
23] demonstrated in vivo that shRNA-mediated downregulation of UBE3A selectively inhibited apical dendrite outgrowth and resulted in impaired polarity both in pyramidal and CA1 neurons (p7). Miao defined
normal dendrite polarity as a single dendrite within the ± 15° range of orientation toward the pial surface, consistent with previous studies in vivo [
55], and similar to in vitro tests [
44]. Interestingly the mean axonal alignment of WT HNs on GRs was right around this value (i.e., 22°), while for AS was almost double. However, we did not find a major difference in the dendrite network orientation along the GRs. This might stem from the fact that axonal processes primarily drive cell directional polarization [
53] and account for more than 50% of the total neurite length, in vitro. Overall, AS neurons also show an increased number of axonal secondary branches on GRs. It almost seems that AS axons do not know or understand which way they should go, while WT do. Here, Noco normalizes both axonal main directional growth and also its branching. In vivo complex signalling pathways that are activated by extracellular cues regulate the growth and guidance of axon branches but the ultimate target of all these signal transduction pathways is the cytoskeleton (both actin fibers and MTs), which can reorganize by changes in dynamics and polymerization/depolymerization (reviewed in [
56]). Our results suggest the presence of defects in cytoskeleton dynamics/ regulation in AS neurons.
Miao et al. [
23] further demonstrated that the UBE3A loss–mediated impairment of apical dendrite polarity could be counteracted by coexpressing UBE3A isoform 2, but not isoform 3. However, recently Avagliano Trezza et al. [
25] demonstrated the prominent role of UBE3A nuclear isoform 3 in the behavioral and electrophysiological phenotypes of AS mice. For these reasons and for the strong correlation between AS–associated mutations and the loss of Ube3a E3 ligase activity [
21], we reinstated both the cytosolic isoform 2 and the nuclear isoform 3 (both catalytically active) to achieve the most promising strategy to rescue the directional guidance of AS neurons. Importantly, reinstatement of UBE3A isoforms 2 and 3 in AS neurons at an early stage of development (DIV2) only partially ameliorated their defective axonal guidance on GRs, once already started (Fig.
2). These “partial” rescue results are in agreement with the demonstrated essential role of UBE3A in the early stages of neurodevelopment [
14]. These previous findings were obtained in vivo by using an AS model, which allows for temporally controlled Cre-dependent reinstatement of the maternal
Ube3a allele [
14], and demonstrated that there are distinct temporal windows during which UBE3A restoration can rescue different AS-relevant phenotypes. In fact, most of the AS phenotypes (i.e., epilepsy, autism- and anxiety-related features) appear to be established early and are only rescued when
Ube3a is reinstated during prenatal or early postnatal development [
14]. Likewise, deletion of UBE3A in the mature brain was shown to have little effect [
57] emphasizing that UBE3A plays its critical role in early development. Hence, it is possible that the absence of UBE3A during early neuronal development leads to irreversible morphological and wiring defects that later on contribute to the neurological deficits in AS. In line with this scenario, aberrant brain connectivity was revealed at the level of white matter architecture by diffusion tensor imaging in AS patients, pointing out a potential underlying error with axon guidance mechanisms. Microcephaly has been reported in an AS patient [
58] and in AS mouse model [
15] and a quantitative MRI analysis with voxel-based morphometry recently showed cortical/subcortical grey matter volume loss in AS patients [
59], strongly suggesting the presence of developmental abnormalities in AS.
Several data suggest that the proper level of UBE3A is critically important for the normal differentiation of neurons. On our GRs, the defective axonal alignment is due selectively to UBE3A loss, while UBE3A increase (WT+UBE3A) does not impair axonal guidance on GRs. These results are in accordance with previous in vivo reports showing that UBE3A isoform 2 overexpression does not cause any effect on apical dendrite polarity in pyramidal neurons [
23]. However, different results have been also reported. The inhibition of terminal dendritic arborization has been reported in the sensory neurons of both Drosophila
Ube3a-null and
Ube3a-overexpressing mutants [
29]. Moreover, the overexpression of UBE3A induced the loss of dendritic arborization both in vitro in hippocampal neurons (DIV12) and in vivo in the cortex [
30]. Overall, UBE3A dosage seems to be crucial during neuronal development, but only its loss impacts negatively on neuronal topographical guidance.
Immature hippocampal neurons (within DIV5) exhibit high levels of UBE3A in GCs [
23,
25], and GCs have a primary role in neurite guidance. Our results show that both WT and AS GCs could read and follow the GR topography, even if with different sensitivity and reinforcement at the level of actin fibers organization (Fig.
3). We know that filopodia located at GC tips that are aligned along micropatterned lines are more stable and rich in actin fibers compared with misaligned protrusions [
39,
60]. Here, we show that the guidance process is mediated by an anisotropic actin fiber organization at the GC level in WT. Aligned GCs (≤ 15° vs. GRs), indeed, are richer in actin fibers, a condition that may favor (and explain) neuronal growth in the GR direction. This is not the case for AS GCs that show, in particular, a lower ratio of actin fiber content in aligned vs. non-aligned GCs, finally suggesting the presence of deficits in actin fiber formation/stability. Interestingly, Drosophila Ube3a mutants have significantly less filamentous actin than WT larvae, consistent with the identification of actin targets regulated by UBE3A in Drosophila [
32]. These defects in GC actin cytoskeletal reorganization likely contribute to the AS impaired guidance on GRs, when GCs are actively in charge of exploring and responding to topographical stimuli, and again point to defects in cytoskeleton responses in AS. In line with our findings, AS mice also showed impaired theta burst stimulation (TBS)–induced actin polymerization, leading to defective synaptic plasticity [
31].
Finally, we tested a pharmacological rescue strategy. HNs were cultured on GRs in the presence of a low dose of nocodazole, a microtubule (MT) depolymerizing agent that activates the RhoA-ROCK-MLC pathway, leading to an increase of actin cell contractility [
49] and focal adhesion maturation [
45]. We previously showed in the PC12 neuronal model that Noco, administered at nanomolar concentration and once that the neurite growth was started (i.e., allowing the MTs stabilization of a major neurite), showed an ameliorative effect on neurite contact guidance response when the GR topographical stimulus is disturbed [
45]. Consistently here, the axonal alignment along GRs in AS neurons is improved by a low dose of Noco (i.e., 40 nM), thus restoring the correct directional development at WT levels. Noco is also able to reduce the upregulated secondary axonal branching present in AS neurons. The amelioration of the guidance defects in AS HNs can stem from a Noco-induced increase of actin contractility, in line with other defects found in AS models [
31], or of MTs depolymerization. In this framework, Bleb, a myosin-II-contractility-inhibiting drug which was shown to impair mechanotransduction in PC12 neuronal cells [
4,
45], had no effect on both WT and AS HNs guidance (Fig.
4), suggesting the presence of additional factors to actin dynamics in this process. We report just a trophic effect on dendritic network length growth upon Bleb treatment, which is in agreement with a previous study [
61].
In agreement with previous results showing that FA maturation was favored by Noco treatments [
45,
62], we show that Noco rescues some FAs’ molecular effectors that are affected in AS HNs, in particular, at the level of FAK activation (i.e., phosphorylation), talin and, at a little extent, zyxin (Fig.
5). These results confirm the involvement of FAs in AS impaired contact guidance response. Because FA maturation and physical anchorage to cytoskeleton fibers are impaired in AS HNs, our results indicate a possible lower cross-talk between FAs and cytoskeleton fibers in AS neurons. This picture seems to be confirmed by the reduced periodicity of α-actinin bundles organization in AS HNs: in fact, α-actinin exploits an important structural and regulatory roles in cytoskeleton organization and FA maturation [
63]. In line with our view, the existence of two distinct pathways for the upregulation of traction forces after Noco-induced MTs depolymerization have been proposed: [
1] a Rho-myosin-II-dependent and FAK-independent mechanism, [
2] a FAK-dependent and myosin-II-independent pathway [
64]. Taken together, the ameliorative effects of Noco on axonal directional growth in AS may stem from its ability to modulate cell traction forces both by increasing actin fiber contractility and by reinforcing FAs, thus activating a feedback crosslink to the actin cytoskeleton.
However, Noco acts primarily on MTs and, in addition to the well-established function of the actin cytoskeleton, local MTs stabilization is a physiological signal specifying axonal/neuronal polarization [
65,
66]. In neurons, the loss of polarity correlates with characteristic changes in MTs turnover and consistently, modulation of the MTs stability is sufficient to alter neuronal polarization [
65]. Importantly, dendrites differ from axons in patterns of microtubule stability and polymerization during neuronal development [
61]. Therefore, our Noco experiments on GRs seem to point out the following: [
1] a co-primary role of MTs in AS axonal guidance, [
2] a different MTs cytoskeleton stability and reactivity of AS neurons to Noco. Interestingly it has been reported that in neurons of the UBE3A-autism mouse model, the overexpression of UBE3A induces the cleavage of MTs, thus leading to local degeneration and retraction of dendritic branches. Even more interesting, the treatment with 5 nM Taxol, a drug that stabilizes MTs (i.e., therefore with a mechanism opposite to Noco), prevented this UBE3A overexpression-induced morphological changes [
30]. These findings indicate that dysregulation in neuronal structural stability is a cellular hallmark in UBE3A-overexpressing autism and suggest a remarkable mirrored situation for AS neurons. We can envision that UBE3A deficiency in AS may cause an excessive stabilization of MTs, leading to contact guidance deficit that can be in turn rescued by low doses of Noco (i.e., which destabilizes MTs at a little extend). In agreement with this, medial ganglionic eminence interneurons exposed to a low (100 nM) concentration of Noco were showed to modify their direction of migration [
67]. We also register reduced axonal straightness in AS neurons, likely linked to their lower topographical guidance. However, this aspect is not fully rescued by Noco. On the other side, in WT HNs, the perturbation of the MTs dynamics by Noco is sufficient to alter their axonal growth along GRs, thus resulting in a lower axonal alignment along GRs (i.e., higher alignment angle) and in reduced straightness.
E3 ligases have emerged as key cell-intrinsic regulators of diverse aspects of neuronal morphogenesis and connectivity at distinct temporal phases [
13,
68]; however, the specific role of UBE3A in the brain is still unclear. In our view, several UBE3A targets converge on the regulation of cytoskeleton pathways [
34]. We already hypothesized, in [
5], that the deficits in contact guidance of AS HNs may be linked, directly or indirectly, to deregulations at the level of cytoskeleton dynamics. In our hypothesis, UBE3A could directly regulate upstream one or more mediators of cytoskeleton signaling and its loss (with the consequent accumulation of UBE3A-targets) may lead to impaired axonal topographical guidance during the early phases of neuronal development. The protein p27
Kip1 (an UBE3A target substrate) [
69] stands out now as an interesting candidate. In fact, p27
Kip1 plays roles in the regulation of the actin and MT cytoskeletons: its loss (i.e., the opposite situation of AS, where p27 should be accumulated) leads to increased RhoA/ROCK-myosin contractility activity, with increased neurite branching and reduced EB3 comets transport and consequently reduced MTs polymerization [
70]. Recently, it has been shown that p27
Kip1 controls the axonal transport and, at the molecular level, does this through the acetylation of MTs [
71]. Additionally or alternatively, the impaired contact guidance of AS HNs could also originate from the deregulation of gene expression at the nuclear level. In fact, it has just emerged the important role of the localization of UBE3A [
25]: importantly, the clinical AS deficits are mainly due to the loss of the UBE3A isoform 3, with its nuclear localization.
All together, our results indicate the presence of altered cytoskeleton dynamics, at actin and/or MTs levels, in AS HNs in response to topographical stimuli that can be relevant during early neuronal development.