Dendritic spines: it takes two to make an impression

One of the longstanding challenges in neurophysiology is to understand the integration by a neurone of multiple synaptic inputs in space and time, and how it translates into action potentials and phenomena like learning and memory. In order to address these questions, this Journal Club article discusses a theoretical study of spino-dendritic interactions that was recently published in The Journal of Physiology by Schmidt et al. (2007). Information processing by neurones is classically viewed as the spatiotemporal summation of postsynaptic potentials on a millisecond time scale, followed by the generation of action potentials if the membrane potential reaches threshold. On longer time scales, it is rather Ca2+ transients in dendrites and spines that mediate neuronal information processing through the regulation of signalling pathways. Indeed, these spikes of intracellular Ca2+ concentration ([Ca2+]i) regulate several neuronal properties and functions including excitability, gene expression and synaptic plasticity. This versatility of Ca2+ signalling is allowed by considerable diversity in timing, amplitude and spatial localization of the Ca2+ transients. In addition, the spatiotemporal extent of Ca2+ transients is shaped by a complex interaction between Ca2+ extrusion mechanisms, Ca2+-induced Ca2+ release from internal stores, diffusion and binding to endogenous buffers. Finally, intracellular Ca2+ regulatory function is accomplished when Ca2+ ions come across Ca2+ binding effector proteins that are involved in the regulation of various protein targets and signalling pathways. In recent years, Ca2+ microdomains have been described in several cellular types, including cardiac myocytes and neurones. A microdomain of Ca2+ is defined by the sum of localized elementary events that form during the opening of Ca2+ channels in the plasma membrane or intracellular stores (each elementary event, i.e. the opening of a single Ca2+ channel, is defined as a nanodomain of Ca2+ in neurones). The spatiotemporal dynamics of these Ca2+-related events occur within a limited domain of intracellular space. Microdomains of Ca2+ thus greatly enhance the information processing capacity of individual neurones, by allowing each synapse to be an autonomous signalling unit capable of generating localized Ca2+ signals independently of their neighbouring synapses. Although dendritic spines are isolated biochemical compartments specialized in Ca2+ signalling, microdomains of Ca2+ do not require spines, since they exist in aspiny neurones as well (Goldberg et al. 2003). Signal transmission from spines to dendrites occurs on a regular basis in active synapses by the propagation of postsynaptic potentials. In a previous study, Schmidt et al. (2003) reported another form of ‘spino-dendritic talk’ that occurred through mobile Ca2+ binding proteins (CaBPs). Briefly, they carried out high-resolution imaging of Ca2+ transients elicited in spiny dendrites of cerebellar Purkinje neurones by stimulation of the climbing fibre afferents. The experiments were performed on acute cerebellar brain slices of wild-type and knockout mice for two endogenous Ca2+ binding proteins, calbindin D28k and parvalbumin. From a total of 252 spine and 131 dendrite recordings in whole-cell patch-clamp configuration, Schmidt et al. (2003) quantified the average number of Ca2+ ions entering the spine and the dendrite during a Ca2+ transient (4700 and 35 000, respectively). Results indicated a significant contribution of CaBPs to the kinetics of synaptically evoked Ca2+ transients. The paper by Schmidt et al. (2007) is a theoretical study of signal integration in spiny dendrites of cerebellar Purkinje neurones. The authors used data that were obtained in previous experimental studies to build a biologically realistic 4-D model (3-D space + time). The temporal dynamics of Ca2+ transients were studied in a model of spiny dendrite, with the aim to elucidate the consequences of spino-dendritic coupling on signalling pathways in dendritic shafts. The spatially resolved model was comprised of a Ca2+ extrusion mechanism and reaction–diffusion processes for Ca2+ and CaBPs (calbindin, parvalbumin, calmodulin). Before adding calmodulin to the model, the authors measured its previously unknown diffusion coefficient by using two-photon fluorescence recovery after photobleaching (FRAP). First, the model was tested for its ability to reproduce previous experimental results (Schmidt et al. 2003). For this purpose, several mechanisms specific to Ca2+ imaging were included, comprising the endogenous CaBPs washout known to occur during prolonged whole-cell recordings, and Ca2+ indicator dye-related effects. Robust concordance with Ca2+ imaging data was obtained. The Ca2+ imaging-related mechanisms were then removed to allow the study of spino-dendritic coupling under unperturbed conditions. Two distinct types of synaptic activity were simulated: the activation of (i) voltage-operated Ca2+ channels and/or ionotropic glutamate receptor channels (short Ca2+ transients), or (ii) metabotropic glutamate receptors (long-lasting Ca2+ transients), in dendritic spines. In the case of short spineous Ca2+ transients, ∼3/4 of the Ca2+ influx exited the spine via buffered diffusion of CaBPs. However, both single spine activity and neighbouring spines co-activation failed to elicit a significant change of [Ca2+]i in the dendrite. Free Ca2+ was confined within the spine, except for a negligible leak towards the dendritic shaft. In the case of long-lasting Ca2+ transients in a single spine, massive spino-dendritic Ca2+ efflux occurred. However, these Ca2+ ions diluted in the far greater volume of the dendrite, and thus failed to elicit local [Ca2+]i elevations in the dendrite. On the other hand, neighbouring co-active spines elicited a significant increase of Ca2+ concentration in the dendrite, suggesting a spatial summation of Ca2+ signals in the dendritic shaft. Calmodulin (CaM) was then included in the model to examine how the spatial summation of Ca2+ signals that seems to take place in the dendrite might affect the local concentration of active calmodulin (CaM*) and the cellular processes it regulates. Short Ca2+ transients in single spines or in neighbouring, co-active spines failed to elicit a sizeable increase of [CaM*] in the dendrite or in the spine itself. Long-lasting Ca2+ transients, on the other hand, increased the [CaM*] in the spine for single spine stimulation, while neighbouring spine co-activation caused a significant increase of spineous as well as dendritic [CaM*]. The latter was rather due to Ca2+ buffered diffusion than to CaM* diffusion from spine to dendrite. This result indicates that the spatial summation of Ca2+ signals can result in functional consequences through the activation of dendritic CaM. Although microdomains of Ca2+ were described in aspiny neurones and non-neuronal cells, dendritic spines seem to be specialized structures favouring Ca2+ microdomains. The experimental and computational results of Schmidt et al. (2007) confirm that microdomains of Ca2+ are present in most forms of synaptic activity in the dendritic spines of cerebellar Purkinje neurones. However, they also show that a small amount of ‘spino-dendritic crosstalk’ occurs during synaptic activity, through CaBPs diffusion. In the particular case of long-lasting Ca2+ transients occurring simultaneously in neighbouring spines, this diffusion of CaBPs results in a Ca2+ microdomains overlap, thus providing a coincidence detection mechanism for metabotropic glutamate receptor-mediated activity. From an information theoretical viewpoint, dendritic computation by spatial summation of Ca2+ transients most certainly increases the information processing power (D. H. Johnson, personal communication). Potential roles include the generation of local dendritic Ca2+ signals, and dendritic Ca2+ waves which propagation is dependent of Ca2+-induced Ca2+ release (Augustine et al. 2003). In cerebellar Purkinje neurones, these mechanisms may contribute to the spread of certain forms of long-term synaptic depression to nearby or distant synapses (Wang et al. 2000). The results by Schmidt et al. (2007) suggest that such heterosynaptic plasticity may be facilitated by ‘spino-dendritic crosstalk’ due to mobile CaBPs, in part because it can modulate inositol-1,4,5-triphosphate-mediated Ca2+ release in the dendrite. In theoretical studies, a usual criticism is expressed in doubts on the physiological signification of the model. Indeed, a good fit with experimental data is not equivalent to a validation of the hypotheses expressed in the model's equations. It is, however, important to bear in mind that a simple linear relation between two variables, rarely questioned on the same grounds, is also a model. A strong aspect of the study by Schmidt et al. (2007) is that most parameters values used in the model are measured experimentally (many by Schmidt and collaborators) and not the result of optimization algorithms. One limitation to the study is that these parameters were measured ex vivo, on acute cerebellar brain slices. Neurones may in such conditions display characteristics that might not be present in vivo since they are partly deprived of their usual environment. In summary, Schmidt et al. (2007) introduce a novel mechanism that may play a role in information processing by cerebellar Purkinje neurones. Co-activation of neighbouring spines mediated by metabotropic glutamate receptors is hypothesized to induce ‘spino-dendritic crosstalk’ via mobile Ca2+ buffers, thus causing a significant activation of dendritic calmodulin. Further experimentation is necessary to validate results that were obtained by use of a combination of ex vivo imaging recordings and in silico methods. In particular, the physiological conditions under which microdomains of Ca2+ might overlap and possibly act synergistically need to be characterized.

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.