Specific dendritic spine modifications and dendritic transport: From in vitro to in vivo

Dendritic spines are small protrusions along dendrites that contain most of the excitatory synapses in principal neurons, playing a crucial role in neuronal function by creating a compartmentalized environment for signal transduction. The plasticity of spine morphologies provides a tunable handle to regulate calcium signal dynamics, allowing rapid regulation of protein expression necessary to establish and maintain synapses (Cornejo et al., 2022). If excitatory inputs were to be located primarily on dendritic shafts, dendrites would frequently short-circuit, preventing voltage signals from propagating (Cornejo et al., 2022). It is thus not surprising that the structural plasticity of dendritic spines is closely linked to synaptic plasticity and memory formation (Berry and Nedivi, 2017). While comprehensive in vitro studies have been conducted, in vivo studies that directly tackle the mechanism of dendritic transport and translation in regulating spine plasticity spatiotemporally are limited. To the best of our knowledge, there are only two studies prior to ours that demonstrated tracking of RNA binding protein or mRNA transport in vivo in vertebrate systems. Bestman and Cline electroporated Xenopus laevis tadpoles with fluorescence-tagged cytoplasmic polyadenylation element binding protein 1 (CPEB1) and tracked the transport of CPEB1 in optic tectum by two-photon microscopy (Bestman and Cline, 2009). However, neurons in Xenopus laevis tadpole optical tectum are not spiny. In another study, Nwokafor et al. (2019) adopted knock-in mice containing MS2 binding stem loop sequences inserted at the 3′ untranslated region of β-actin to track in vivo β-actin transcripts tagged with GFP in the Layer 2/3 neurons of visual cortex. This study aimed at demonstrating the methodology, and an in-depth investigation of dendritic spine plasticity regulation by β-actin transport and translation was not in its scope. Our recent studies thus aimed to address the in vivo mechanism of synaptic activity-dependent transport and translation at dendritic spines (Zhao et al., 2020; Fok et al., 2024). After demonstrating the functional specificity of kinesin 1 isoform, KIF5B, using in vitro primary hippocampal neurons, we generated a conditional knockout mouse model of kif5b to study the effects of impaired dendritic transport on the dendritic spine plasticity in vivo by two-photon imaging (Zhao et al., 2020). By using in utero electroporation, both Cre recombinases and Cre-dependent fluorescent proteins were simultaneously expressed in a kif5b-floxed transgenic mouse. This approach enables us to study the effects of postsynaptic knockout of KIF5B on dendritic spine plasticity (Fok et al., 2024). These experiments show that ablated dendritic KIF5B-mediated transport heightened basal dendritic spine turnover and impaired activity-dependent dendritic spine plasticity that would otherwise be necessary for mice to acquire freezing behaviors in auditory-cued fear conditioning and fear extinction (Fok et al., 2024). In addition to fear associative learning and extinction, mice with impaired KIF5B-mediated dendritic transport were also found to have other memory deficits, such as working memory, spatial memory, and social memory (Zhao et al., 2020). To visualize the impaired KIF5B-mediated dendritic transport, fluorescently tagged KIF5B cargoes, including FMRP, PSD95, and gephyrin, were exogenously expressed in Layer 2/3 pyramidal neurons in the frontal association cortex (Fok et al., 2024). The dynamics of FMRP and PSD95 dendritic transport were specifically implicated in KIF5B conditional knockout neurons and were correlated with spine instability and abnormal learning-dependent spine plasticity. To our knowledge, these studies are the first to provide in vivo evidence to support the important role of dendritic transport in regulating dendritic spine plasticity necessary for memory formation. While our study confirmed the specific need for KIF5B-dependent transport in dendrites to regulate plasticity-related proteins and RNA binding proteins, such as PSD95 and FMRP, and dendritic spine plasticity, it is noteworthy that the behavior deficits observed in Camk2a-Cre;Kif5bfl/fl conditional knockout mice could result from the combinatory deletion of Kif5b in both axons and dendrites involved in a memory circuit. Similar to CPEB1, FMRP is a well-studied RNA binding protein that assembles ribonucleoprotein granules. Therefore, these findings highlight the crucial role of localizing mRNA transcripts in dendrites and enabling their translation to facilitate activity-dependent spine plasticity, which is associated with the formation of memories. The differences observed between in vitro and in vivo studies of synapses indicate potential variations in the demands and strategies involved in dendritic transport and local translation, suggesting the intriguing possibility of a more complex transport mechanism in vivo. Cultured neurons have a higher percentage of excitatory synapses located on the dendritic shaft and demonstrate a different developmental pattern compared to age-matched in vivo neurons (Boyer et al., 1998). The speed of dendritic transport also differs. Kinesin-1 is known to be the common motor for β-actin mRNA, FMRP, and CPEB1. We did not observe apparent FMRP translocation in vivo within 1 hour (Fok et al., 2024), albeit in vitro consensus of translocation at around 1–1.3 m per second (Zhao et al., 2020). Similarly, β-actin mRNA and CPEB1 were also observed to exhibit slower motility in vivo compared to their in vitro counterpart (Bestman and Cline, 2009; Nwokafor et al., 2019). Furthermore, it is also noteworthy that in vivo studies so far have tracked dendritic transport in cortical neurons, while in vitro studies were predominantly conducted in hippocampal neurons. Although this may explain partially their discrepancies, it also suggests an interesting hypothesis whereas dendritic transport mechanisms may vary among neuronal types. Finally, the transportation of cargoes in vivo could be limited to specific time windows corresponding to animal experience and brain activity, which could easily elude detection without prior knowledge of such regulatory mechanisms in different animal experiences. Awake intravital imaging of dendritic transport, similar to how our recent study was conducted, would be preferable compared to the imaging of anesthetized animals to address this question. Synaptic activities upon artificial stimulations have been shown to induce dendritic transport and local translation at dendritic spines, which is necessary for structural plasticity. Several in vitro studies have elegantly adopted glutamate uncaging in live cells to demonstrate the recruitment of mRNAs to the stimulated spines that support the translation of new proteins and spine remodeling (Rangaraju et al., 2019). Sequestration of β-actin mRNA into condensates of membrane-less assemblies has also been shown to block β-actin translation at dendritic spines and induce spine shrinkage, as demonstrated by the optoMCP-FUS system in dissociated hippocampal neurons. This technique utilizes light to induce sequestration of MSB-tagged β-actin into small assemblies to disrupt translation (Lee et al., 2024). Interestingly, stimulation of spines does not always lead to an increase in local translation, but also leads to the reduction of translation through miRNA-mediated degradation (Sambandan et al., 2017). The enrichment of translation organelles near stimulated spines also supports this view. Studies on ribosomes, endoplasmic reticulum, and mitochondria have revealed that these organelles become less mobile and cluster near dendritic spines upon synaptic stimulation, creating an ideal environment to promote local translation. Depletion of such machineries is shown to impair spine modifications and protein synthesis (Rangaraju et al., 2019). While in vitro studies have made significant advancements, it remains unclear how these processes translate to in vivo conditions. Owing to technological advancement, similar approaches previously carried out in vitro could be attempted in vivo, including the use of in vivo glutamate uncaging. Nevertheless, the controlled stimulation of dendritic spines by in vivo glutamate uncaging is ultimately a mimicry of a subset of synaptic cues at spines. It still poses a challenge to translate these artificial spine stimulations into normal physiological conditions. Another approach would be to identify the stimulated spines retrospectively after the animal has gone through specific experience or learning task. Our study is an example of this approach, where the experience-associated spines related to fear conditioning and extinction are identified by the observed structural plasticity such as formation or elimination (Fok et al., 2024). By simultaneously tracking FMRP in the same dendrites, it is found that FMRP was transported to 2 μm proximity of the stimulated spines, suggesting the localization of mRNAs and translation machinery to support spine formation following stimuli in fear learning. Future studies can be carried out with the Targeted Recombination in Active Population (TRAP) transgenic mice to home in on the dendritic branches of activated neurons during memory formation (DeNardo et al., 2019). Furthermore, it is also possible to only label recently activated spines using a spine-directed fluorescent reporter expressed downstream of an immediate early gene promoter such as the e-GRASP technique (Lee et al., 2023). Combined with in vivo tracking of mRNAs (e.g., MS2-based mRNA reporting system), nascent protein synthesis (e.g., FRAP and FLIP microscopy), and plasticity-related protein transport/ turnover (e.g., PSD95-HaloTag mice) (Bulovaite et al., 2022), the relationship between spine plasticity, dendritic transport, and local translation can be analyzed retrospectively. It is also noteworthy that advances in correlated light and electron microscopy and high throughput electron microscopy technologies such as automated tape-collecting ultramicrotome combined with scanning electron microscopy allow a post hoc ultrastructural analysis of activity-related dendritic spines in a scale much higher than before to complement in vivo imaging of spine plasticity (Sohn et al., 2022). It is thus possible to validate local translation machinery increase in activity-related spines by electron microscopy analysis of polyribosomes, compared to non-related spines. Moreover, there is an emerging perspective of an increasingly complex environment where dendritic local translation takes place. In the past two decades, the field has advanced a lot in the understanding of the decentralized capabilities of neurons to regulate translation machinery and synthesize proteins. Nonetheless, mRNA transcripts and ribosomes are believed to be originated from neuronal soma and are moved to distal dendrites through motor protein-dependent transport. Interestingly, emerging studies are introducing extracellular sources of mRNA and translation machinery such as ribosome and miRNA into the picture. Through the release and fusion of extracellular vesicles, inter-neuronal or glia-to-neuron transfer of mRNAs, miRNAs, and synaptic proteins were reported to impact dendritic signaling cascades, transport, local translation, and spine plasticity (see review such as Akbari-Gharalari et al., 2024). That being said, it remains to be answered about the prevalence and the extent neurons tap into these extracellular sources of translation materials. However, as the majority of studies heretofore about dendritic transport and local translation have come from cultured hippocampal neurons, it is expected to garner new insights when more in vivo studies are conducted in the future. For example, with the supply of extracellular sources of mRNAs and translation machinery, in vivo dendritic transport dynamics could simply rely less on long-distance transport from the soma. In our study, the effect of dendritic Kif5b knockout was not compensated by the endogenous presence of glial cells. However, we reported that the effect was not in the reduction of FMRP abundance in dendrites, but rather the fine localization of FMRP. As a proxy of mRNA presence in dendrites, our findings about KIF5B-dependent FMRP transport cannot rule out the role of glial cells as an external source of translation machinery. In addition, local translation regulation would also consider the states of the animal, which correlate with the activity of glia (Akbari-Gharalari et al., 2024). These are areas that can only be addressed through in vivo studies of dendritic transport and local translation. Finally, dendritic spines are heterogeneous in their morphologies, molecular composition, and plasticity, which would pose different demands on dendritic transport and translation mechanisms (Berry and Nedivi, 2017). Spines have long been studied as the structural correlate for synapses, and their classification has been limited to the morphological perspective, such as mushroom, thin, and stubby spines. Mushroom spines are believed to be the memory spines since they are the most stable and likely to contain the strongest synapses, while thin spines are believed to be the learning spines due to their plasticity (Berry and Nedivi, 2017). This forced morphological classification of dendritic spines is increasingly challenged as analyses reveal the continuum nature of spine morphology. Furthermore, in vivo visualization of synaptic markers such as PSD95 and gephyrin concurrently with spine structures has demonstrated another perspective to classifying spines, which is the molecular composition (Berry and Nedivi, 2017; Fok et al., 2024). Since spines contain most excitatory synapses, studies about the excitatory synaptome suggest that spines of the same morphological category still vary in their molecular composition and lifetime in different neuronal types and brain regions (Bulovaite et al., 2022). Why and how such heterogeneity of dendritic spines arises will then require further investigation in the future. Just like how synaptome complements our understanding of the connectome, the field would benefit from a more systemic and holistic description of spine heterogeneity across neuronal types and brain regions, which we term “spinome” in this perspective. Since spine plasticity (both formation and elimination) is a consistent indication of memory formation, it is intriguing to hypothesize memory-specific modifications to the “spinome.” While most in vivo studies of dendritic spines randomly sample dendritic branches from one cortical area, future research that probes into these questions will require transgenic mouse line that labels endogenous synaptic markers to sort dendritic spine into subtypes, machine learning solutions to automatically score fluorescent puncta in a large scale, and super-resolution imaging by two-photon in vivo nanoscopy (e.g., STED) in multiple brain regions. With these technological advances, it will be possible to conduct high throughput imaging, for example, the dendritic spines in the somatosensory cortex and the thalamus in a sensory discrimination task with morphological and molecular composition details, to characterize the “spinome” in this behavior context. In conclusion, dendritic spine plasticity is an important outcome of dendritic transport and local translation, as the constant fine-tuning of spine structures would require these processes to be well-regulated (Figure 1). By far, in vivo findings displayed different dendritic transport speeds and different sources of dendritic mRNAs and translation machinery. A global description of molecular heterogeneous spines would also provide a novel dimension, which we term “spinome,” to understand memory formation and storage in addition to synaptome and connectome. How dendritic transport and local translation support the plasticity of molecularly heterogeneous spines in a memory context will be of great interest to the field in the future.Figure 1: Summary schematic displaying the unique considerations for in vivo studies of dendritic transport and dendritic spine plasticity.(1) In vivo studies of dendritic transport thus far demonstrated significantly different transport dynamics of either mRNA or RNA-binding protein than in vitro findings. We speculate more complex mechanisms are involved in vivo to regulate motor protein (such as kinesin 1, light blue dimeric icon) speed on microtubules (green) such as synergistic movement of multiple motors for a cargo complex (yellow), coordination of multiple motor types attached to a cargo complex and stimulation dependent transport that is challenging to detect with anesthesia paradigms. (2) Studies of spine plasticity in vivo considering the behavior contexts. Mechanisms of dendritic transport and translation that support spine plasticity in one behavior context might be distinct from the other, given different synaptic cues, plasticity forms, signaling pathways involved. In the schematic, different behavior contexts (such as auditory-cued fear conditioning and whisker stimulation by air puffs) are designated to modulate different groups of spines under different mechanisms (colored in red and pink). Behavior related spine plasticity is the structural basis for memory formation. It is thus imperative to classify these spines a priori or a posteriori to investigate the underlying mechanisms of dendritic transport and translation with advanced genetic or analytic methodologies. (3) In vivo scenarios are further complicated by recent findings about inter-neuronal and glia-to-neuron transfer of mRNA and translational machinery (glia in purple, neuron in grey, extracellular vesicles in dark blue). The demand for long distance dendritic transport is speculated to be drastically different from in vitro scenarios as a result. Since glial secretory functions are closely tied to the behavior and internal states of the animals, mechanisms of dendritic transport, translation, and spine plasticity will likely depend on different behavior contexts or types of memories. These considerations add to the value of conducting in vivo studies of dendritic transport and translation. The models of neurons and glia are originally sketched by authors. Other graphics such as mice and icons of kinesin are adapted from BioRender.com.This work was supported by the National Natural Science Foundation of China (NSFC/RGC/JRF N_HKU735/21); Research Grant Council of Hong Kong, China (17102120, 17108821, 17103922, C1024-22GF, C7074-21G); Health and Medical Research Fund (HMRF 09200966) (to CSWL); and FRQS Postdoctoral Fellowship (to AHKF). Open peer reviewer:Jun Noguchi, Kokuritsu Seishin Shinkei Center, Japan. Additional file:Open peer review report 1.P-Reviewer: Noguchi J; C-Editors: Zhao M, Liu WJ, Qiu Y; T-Editor: Jia Y

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.