Hippocampal Vs Neocortical Spines: The clash of the Titans:
Hippocampal Vs Neocortical Spines: The clash of the Titans:
Dendritic spines are the protrusions emanating from the dendritic shaft and they act as a specialized compartment for receiving inputs from the boutons (1). They relay the information of individual inputs impinging on them by forming synapses so that they can pass the information or synaptic potentials to the soma for the generation of action potential. A synapse is formed between the apposition of pre-synaptic active zones filled with the neurotransmitter and the spines having the postsynaptic density holding the receptors such as NMDAR, AMPAR etc. (2). Spines in the dendrites make synapses with as many axonic boutons as possible rendering the neurons to receive maximum input. With such a structured pattern all around the dendritic trees, spine enables the increase in connectivity, integrating the inputs linearly and regulating the plasticity of specific inputs in a distributed circuit model (3). Spines are highly plastic structures, new spines can be formed either de novo or from the existing shaft synapse. Moreover the spines vary from a range of volume 0.001µm3 to at the most 2µm3. Spines are rich with actin filaments and actin play a prominent role in the growth and the shrinkage of spines by remodeling the cytoskeleton (4) Molecules such as Ca2+ regulate the actin dynamics in spines during synaptic activity via opening of NMDA receptors thereby in fluxing Ca2+ in the dendritic spines (5). For instance, blocking NMDA receptors causes pruning of the spines however tetanic activation of neurons with glutamate causes the removal of spines pointing to the direction that spines are not static and they undergo pruning as well as spine formation, so an estimation of spine numbers is always variable. Interestingly the time window for the formation of de novo spines is large and it is a slow process induced by tetanic stimulation for inducing long-term potentiation (LTP).
In LTP inducing experiments, ensued by the electrophysiological recording from this neurons shows an increase in 50 – 100% of synaptic transmission than the basal synaptic transmission, which is not co-related with the increase in the formation of 2 – 3 spines per 100µm (6) of dendritic segment. Despite that, there is a consensus that a part of the population of spines in vivo after experience dependent structural plasticity, novel experiences, (7) are stable throughout the age of the mouse suggesting that weight of the inputs are stable lifelong arguing in favor of the idea that stable spines stores life – long memories. Interestingly the in vitro data on experiments evoking tetanic stimulations and by long –term potential evoking protocols, which causes an increase in the spine head diameter, indicates that synaptic proteins are very labile with a half-life or turn over at the maximum of two days (8). Considering that the spines are stable for life long and the synaptic proteins have a turnover of only days, it is still puzzling how the spines are maintained. Work by Gray NW et al, postulated that larger spines, which are considered as the one which stores memory retain PSD-95 in vivo (Postsynaptic density) for longer duration of time than the smaller one by interacting with PSD thus by keeping the spine size for days however a paper by Woods GF et al, 2011 showed that in a subset of neurons during spine retraction there is no concomitant decrease in PSD-95 though majority of spines lost PSD-95 during spine shrinkage indicating that PSD -95 loss is not a pre-requisite during the retraction of the spines (9). This data is very interesting especially in light of the fact that stronger synapses should be associated with larger PSD so that it can occupy more number of AMPA type glutamate receptors (turning the silent synapses in to very active one), conversely weaker synapses possess less enrichment of PSD-95 and concomitantly less number of AMPA –type glutamate receptors.
The model for the changes in spines during experience dependent structural plasticity states that during basal condition some of the spines are stable, while there is also a formation of some new and transient spines, though majority are short lived however during a novel experience such as change in sensory experience creates a structural remodeling of synapses. Sensory experience also causes a loss of previously persistent spines and an increase in the size and strength of synapses. Spines are broadly classified in to stubby, thin and mushroom depending on their morphology. Stubby spines are the one that has a head, but no defined spine neck, while thin spines are more of filiopodia like protrusions, however mushroom spines are distinguished by large mushroom like head and they have the biggest spine head with a large volume compared to other spine classes. It has been widely speculated that the variance in the size and the volume can be positively correlated with the strength of the synapses though this has not been still resolved very clearly. It has been postulated that spine neck plays a prominent role in compartmentalizing the signaling molecules and the ions by dynamically regulating their diffusion (10). LTP (long term potentiation) causes the enlargement of spines, while LTD (long term depression) causes the shrinkage of spines. In spite of that, even the thin protrusions make synapses and there is no absolute co-relation between the stability of the spines and the size. In vivo experiments have shown that thin or small spines can persist, while larger spines get disappeared (7). Though there has been not universal consensus about morphology of spines and the strength of synapses, none of the key questions in the research in neuroscience is whether spines are important in the formation of memories.
Matsuzaki et al, provided evidence that spines obey hebbian plasticity. At the level of single synapses, input specific strong activation such as LTP creates rapid enlargement of small-stimulated spines that persist for longer time while the large ones such as mushroom spines, the enlargement is transient. Moreover the large spines such as mushroom spines are refractory to induce LTP, while the small-stimulated spines shows persistent increase in the glutamate evoked currents and a concomitant increase in the volume of the spine head. Matsuzaki et al, found within 2 – 4 minutes of LTP stimulation, an increase in the head volume of three fold, which drops to 25 – 30% increase in volume after 20 – 40 minutes of stimulation. They found that the molecular machinery responsible for this increase in the glutamate evoked AMPA receptors current are the activation of NMDA receptors, dynamic changes in the actin polymerization and plasticity regulating kinase such as Ca2+ / calmodulin – dependent protein kinase II. They postulated that small spines are the silent synapses which have the potential to become larger spines as LTP induced increase in the current and the head volume preferentially acts on small spines, which are considered to be the memory storage unit. The large spines, which are refractory to LTP, are the one that stores the trace of already formed memory (11) Interestingly Kopec et al, 2006 has shown using chemical LTP protocol that there is a translocation of AMPA receptors with a persistent increase in spine volume. This increase in the volume of spine is not dependent on the initial size of the spine, which is in contrast with the studies of Matsuzaki et al, 2004. The LTP induced addition of new AMPA receptors to the PSD in the dendritic spines is regulated by molecules such as syntaxin, myosin Vb which causes the mobilization of recycling endosomes, which carries the AMPA receptors.
Lang et al addressed the key question of whether changes in the morphology of spines are a pre-requisite for the induction of synaptic plasticity. It has been shown that the LTP inducing stimulus causes only a transient increase in the spine head volume in hundreds of spines imaged. Moreover electrophysiological recordings in this slices has shown that there is an induction of LTP, however the spines which are potentiated couldn’t be identified (12) indicating that changes in the morphology of spines doesn’t need measuring the induction or the expression of synaptic plasticity. Time lapse experiments has given valuable information about the formation of new spines either by inducing activity or during learning however the number and the time course of the formation of new spines varies with different studies. Despite this, there is a consensus in the field that spines undergo structural changes during learning, novel experiences, encountering enriched environments and by plasticity – inducing stimulus. LTP induction causes a parallel increase in the size of the spine head conversely LTD cause spine shrinkage, but still one of the unanswered question remains is what’s the real consequence of change in the morphing of the spines and their physiological effects across different regions of the brain.
One of the recent issues of Nature, Attardo et al (13); from Mark Schnitzer’s lab has shown that in vivo, the neocortex possess more permanent spines (> 50%), while the spines on the basal dendrites in the CA1 region of the hippocampus possess a single population of unstable spines, i.e. nearly full erasure of their synaptic connectivity in the hippocampus. However when they measured the turnover dynamics of the spines, they found that transient CA1 spines stay twice longer than that of the neocortical spines (10 days compared to 5 days). Moreover they had discovered that NMDA receptor activation stabilizes CA1 spines, while they prune cortical spines pointing an opposite role in synapse maintenance. Interestingly, when they exposed the animals to enriched environment, they found neither change in the number of impermanent spines nor the turn-over dynamics of the CA1 spines on the basal dendrites, but a small increase in the spine volume which scales back to its original size after a few days. One of the hallmarks of the study is the use of two-photon microendoscopy to image the deep regions in the brain. Combining with STED microscopy they had elegantly shown that spines that are present nearby are merged in two photon imaging however they are actually individual spines. Their study also questioned the validity of time-lapse experiments researchers had performed in the past on measuring the spine dynamics and turnover kinetics in vivo in densely packed spines on CA1 pyramidal neurons. They had beautifully had shown that many spines which appeared to be stable in conventional imaging can be merged two spines because of the diffraction limit of the microscope and hence been erroneously measured as stable spines. Interestingly this work had provoked some interesting questions and suggestions that I would like to enumerate below:
- It will be interesting to see whether all the spines in the basal dendrites are impermanent. The authors pointed out that there might exist a small population of permanent spines, with their elegant experimental data had clearly shown the presence of mostly transient spines. Given that spines can be classified into stubby, thin and mushroom, will it be possible to further classify whether the majority of the transient spines broadly fall in to stubby and thin, while the minor population that persist permanently are mushroom? The rationale for this claim is that apparently it has been considered that mushroom spines are larger, stable and store traces of memory, while stubby and thin are learning spines.
- Will it be the case during hippocampal learning during pattern completion (auto-associated CA3 network and the CA3 to CA1 network), some of the transient spines change to permanent ones? Will there be a concomitant increase also in the different types of spines, such as more mushroom spines during conditioned learning paradigms compared to the non – learners? It had been shown that upon uncaging glutamate on thin and stubby spines, they could be converted to mushroom, which becomes refractory to any structural changes in vitro in slices. It will be interesting to test whether such changes happens during specific hippocampal learning paradigms.
- One of the approach i) will be combining the methodology of in vivo two photon endoscopy for time lapse imaging in vivo with TRAP (Targeted recombination in active populations, (13)) so that one can permanently label the active neurons genetically which are exposed to learning paradigms. In this way, one can measure the dynamics, turnover kinetics as well as the other changes in spines such as the length, volume, width of the spine neck in those activated neurons. ii) Will these activated spines stay longer than the observed values? What will be the fate of these activated spines on the basal dendrites? One can also identify memory engram neurons in the hippocampus by a similar approach of activity dependent labeling regulated by drug such as doxycycline (14). The transgenic mouse with tTA under the control of immediate activity dependent gene can be crossed with Thy-1: GFP to label the active neurons in red and the excitatory neurons in green respectively. A doxycycline diet during learning paradigm to mark the activated neurons combined with two-photon endoscopy will be an alternative approach to address the above-mentioned question.
4. It will be interesting to decode the activity changes in single spines in vivo using a genetically encoded calcium indicator or by bulk loading of fluorescent dye into the spines to measure the calcium changes during learning paradigm (15) This might give glimpse of idea of i) whether there are hot spots of active individual spines on the dendrites and ii) whether these active spines will have a different turn over dynamics, lifetime and kinetics compared to the non-active spines? Given that, not all the geometrically apposed synapses are functional, it will be worth to check by calcium imaging experiments whether all the transient spines are functionally active
5. Does the presence of a single population of impermanent spines on the dendrites is specific for basal dendrites or it also holds true for the spines on the apical dendrites in the CA1 region of hippocampus? The basal dendrites receive most of the connections from the CA3 neurons that are close to the CA1 region. It has been suggested that the attenuation of EPSP (Excitatory synaptic potential) is steeper in basal dendrites than in the apical dendrites when they propagate to the soma (16). Besides that, the basal dendrites display significantly attenuated back propagating APs (Action potential) which forms the basis of the conclusion that basal dendrites integrated information in a sub-threshold way. The mode of input summation in basal dendrites is location – dependent in contrast to the spatial amplification of temporally clustered inputs processed by apical or distal dendrites. Moreover, the efficacy of generation of APs from the dendritic spikes of the basal dendrites differs from that of the apical dendrites as the former are more involved in the temporal coding (Ariav et al., 2003) and hence It will be worth to measure the spine turn over and kinetics of spines on the apical dendrites.
- Spines are involved in the integration of inputs linearly or in the summation of inputs either sub-linearly or supra-linearly. Information is propagated from CA1 to neocortex for storing long-term memories and that this study has shown that neocortical spine lifetime and dynamics are in contrast to the dynamics exhibited by the CA1 spines, i) it will be interesting to learn, what are the learning rules implemented by the CA1 spines to neocortical spines? ii) What is the functional importance of each input, especially with respect to the input – specific plasticity in individual synapses.
7.This study also has shown that the spines on the basal dendrites in the CA1 of the hippocampus are transient single population with a mean life time of 10 – 15 days ensued by a full erasure of their synaptic connectivity. i) It will be worth to check the axon arborisation connectivity matrix of those axonal contacts with the transient spines? ii) Does the axon connectivity matrix also changes or disappears with the transient synapses? Iii) What is the fate of pre-synaptic boutons, are they also transient or stay longer even after the disappearance of the spines? As Thy-1: GFP also labels the boutons; it will be interesting to see the structural rearrangements in the pre-synaptic part.
8. NMDA receptor activation in neocortical spines causes elimination, while their activation in hippocampal spines causes stabilization of spines shown by this study. Given NMDA receptor activation causes LTP by recruiting more AMPA receptors to exocytose on the dendritic spines there by increasing synaptic transmission and concomitantly increase in the growth of the spine. What different signaling cascades are activated by NMDA receptors in different context?
Despite the discrepancies from data from different labs, there is a consensus in the field that spines undergo structural changes during learning, novel experiences, encountering enriched environments and by plasticity – inducing stimulus. Future studies addressing the functionality of such synapses, strength of the spines and the mechanism affecting such changes will further prove the causality of the observed morphological changes in the neocortex and in the hippocampus.
- Veronica AA, and Sabatini BL. “Anatomical and physiological plasticity of dendritic spines.” Annual Review of Neuroscience (2007)
- Harris KM, and Kater SB. “Dendritic spines: cellular specializations imparting both stability and flexibility to synaptic function.” Annual Review of Neuroscience (1994)
- Yuste R and Denk W. “Dendritic spines as basic functional units of neuronal integration.” Nature (1995).
- Matus A. “Actin-based plasticity in dendritic spines.” Science (2000).
- Sabatini BL, Maravall M, and Svoboda K. “Ca(2+) signaling in dendritic spines.” Current Opinion in Neurobiology (2001).
- Engert F, and Bonhoeffer T. “Dendritic spine changes associated with hippocampal long-term synaptic plasticity.” Nature (1999).
- Holtmaat A, et al., “Experience-dependent and cell-type-specific spine growth in the neocortex.” Nature (2006).
- Ehlers MD, “Activity level controls postsynaptic composition and signaling via the ubiquitin-proteasome system.” Nature Neuroscience (2003).
- Woods GF et al., “Loss of PSD-95 enrichment is not a prerequisite for spine retraction.” The Journal of Neuroscience (2011).
- Bloodgood BL, and Sabatini BL. “Neuronal activity regulates diffusion across the neck of dendritic spines.” Science (2005).
- Matsuzaki M et al., “Structural basis of long-term potentiation in single dendritic spines.” Nature (2004).
- Lang C et al., “Transient expansion of synaptically connected dendritic spines upon induction of hippocampal long-term potentiation.” PNAS (2004).
- Guenthener CJ et al., “Permanent genetic access to transiently active neurons via TRAP: targeted recombination in active populations. Neuron (2013).
- Liu X et al., “Identification and manipulation of memory engram cells”. CSHL Quant Biol, (2014).
- Chen X et al., “Functional mapping of single spines in cortical neurons in vivo.” Nature (2011).
- Inoue, M et al., “Dendritic attenuation of synaptic potentials in the CA1 region of rat hippocampal slices detected with an optical method.” The European Journal of Neuroscience 13 (2001).
- Ariav G et al., “Sub millisecond precision of the input-output transformation function mediated by fast sodium dendritic spikes in basal dendrites of CA1 pyramidal neurons.” The Journal of Neuroscience (2003)