Electrophysiology of Dendritic Spines: Information Processing, Dynamic Compartmentalization, and Synaptic Plasticity

Abstract

For many years, synaptic transmission was considered as information transfer between presynaptic neuron and postsynaptic cell. At the synaptic level, it was thought that dendritic arbors were only receiving and integrating all information flow sent along to the soma, while axons were primarily responsible for point-to-point information transfer. However, it is important to highlight that dendritic spines play a crucial role as postsynaptic components in central nervous system (CNS) synapses, not only integrating and filtering signals to the soma but also facilitating diverse connections with axons from many different sources. The majority of excitatory connections from presynaptic axonal terminals occurs on postsynaptic spines, although a subset of GABAergic synapses also targets spine heads. Several studies have shown the vast heterogeneous morphological, biochemical, and functional features of dendritic spines related to synaptic processing. In this chapter (adding to the relevant data on the biophysics of spines described in Chap. 1 of this book), we address the up-to-date functional dendritic characteristics assessed through electrophysiological approaches, including backpropagating action potentials (bAPs) and synaptic potentials mediated in dendritic and spine compartmentalization, as well as describing the temporal and spatial dynamics of glutamate receptors in the spines related to synaptic plasticity.

3.1 Dendritic Spine Electrophysiology

Dendritic spines, the subcellular formation characterized by tiny membrane protrusions (≤1 μm diameter) of dendrites, constitute the fundamental units of synaptic and information processing in the mammalian brain (Harris et al. 1992; Popovic et al. 2015b; Kwon et al. 2017). Most excitatory inputs on cortical pyramidal neurons are made on spines, rather than on dendritic shafts (Yuste 2010; Araya et al. 2014), but dendritic spines also modulate inhibitory synaptic inputs showing particularly high levels of plasticity (Chen et al. 2012; van Versendaal et al. 2012). Spines can individually detect temporal coincidence of synaptic activity, acting as basic functional units of neuronal integration (Yuste and Denk 1995; Sabatini et al. 2002; Noguchi et al. 2005; Yuste 2013). Additionally, spine shape and spine head size are correlated with changes in synaptic strength during synaptic plasticity (Yuste and Bonhoeffer 2001). While the complex and dynamic regional electrical properties of dendritic spines play an important role in determining the function and integration of neural circuits, the electrical behavior of spines is still not well understood and sometimes remains controversial.

For a long time, the limitations of available techniques prevented measuring voltage signals directly from the membrane of individual spines. Thus, the electrical properties of dendritic spines, including active conductances, were based on theoretical models (Harris and Stevens 1989; Koch and Zador 1993; Harris and Kater 1994; Johnston et al. 1996; Yuste and Tank 1996; Svoboda et al. 1996; Shepherd 1996; Yuste 2010) and based on channel densities and kinetics from recordings of dendritic patches. Whole-cell or cell-attached recordings of spine heads per se are still very difficult to perform in brain slices. However, advanced imaging techniques, including high voltage-sensitive dyes, low-affinity Ca²⁺ and Na⁺ indicators, two-photon Ca²⁺ imaging, and two-photon uncaging of glutamate, made it possible to detect membrane voltage signals from individual dendritic spines in brain slices (Holthoff et al. 2010; Popovic et al. 2015a, b; Miyazaki and Ross 2022). In addition, the intracellular voltage recordings from dendritic spines have been obtained by quantum-dot-coated nanopipettes in cells in culture and brain slices (Jayant et al. 2017), and patch-clamp recordings have been also performed in dissociated neurons and isolated spines (Priel et al. 2022). That is, the combination of advanced techniques in imaging and morphology with electrophysiology and spine models are providing valuable information about the spatiotemporal dynamics of biophysical properties of dendritic spines (Stuart et al. 1993; Stuart and Sakmann 1994; Denk et al. 1996; Mainen et al. 1999; Korngreen and Sakmann 2000; Oertner 2002; Nuriya et al. 2006; Araya et al. 2006).

3.1.1 Backpropagating Action Potentials and Synaptic Potentials in Dendritic Spine

Somatic action potentials (APs), after being generated in the axon initial segment, travel back through the dendritic tree as backpropagating action potentials (bAPs) (Häusser et al. 2000). Dendritic spines can regulate the synapses without attenuation of dendritic bAP (Popovic et al. 2015a; Cornejo et al. 2022), which can influence the processing of incoming synaptic inputs in the dendrites of most neurons (Stuart et al. 1997a, b; Xiong and Chen 2002). The coincidence and temporal sequence of bAP and excitatory postsynaptic potential (EPSP) within a dendritic spine determine the amount of Ca²⁺ entry through voltage-gated channels and/or ionotropic receptors, essential for synaptic plasticity (Yuste and Denk 1995; Schiller et al. 1998; Koester and Sakmann 1998; Yuste et al. 1999; Nevian and Sakmann 2004; Camiré and Topolnik 2014). The rapid time course of the bAP in spines may be a critical factor for the accurate regulation of spike timing-dependent synaptic plasticity within a very short time window, while the magnitude and time course of synaptically activated EPSP in dendritic spines determine how these synapses contribute to synaptic integration (Caporale and Dan 2008; Popovic et al. 2015a). Both synaptic integration and plasticity depend on spatiotemporal changes in the membrane voltage on the dendritic spine, and the ability to measure such modifications is improving.

Spine ionic conductance especially from voltage-gated sodium channels could play an important role in dynamically regulating bAP. Voltage-gated sodium channels have been found in spines (Caldwell et al. 2000; Araya et al. 2007; Bywalez et al. 2015; Miyazaki and Ross 2017). Calcium channel subunits (Caldwell et al. 2000; Bloodgood et al. 2009) and potassium channels (Kim et al. 2007; Kaufmann et al. 2013; Wang et al. 2014; Strobel et al. 2017; Tazerart et al. 2022) have also been localized in spines. These channels, together with functional N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) glutamate receptors (Yuste and Denk 1995; Schiller et al. 1998; Yuste et al. 1999; Kovalchuk et al. 2000; Makino and Malinow 2009; Park 2018), control synaptic potential. These relevant advances in detecting voltage changes in spine membrane are further discussed below.

Voltage-sensitive dye recordings based on wide-field laser-excitation fluorescence microscopy permit monitoring of subthreshold electrical signal integration between individual dendritic spines and adjacent dendritic shafts in acute brain slice preparations (Popovic et al. 2015b; Weng et al. 2023). It is, then, possible to measure membrane potential (V_m) transients from individual dendritic spines by using intracellular high-sensitivity voltage-sensing dyes and high-resolution wide-field epi-fluorescence microscopy applied at optical magnification 10 times higher than that used to record membrane voltage from other neuronal parts (Cohen and Salzberg 1978; Weng et al. 2023). These measurements have a sufficient spatial and temporal resolution (approximately 1 μm and submillisecond range) for recording neuronal AP and synaptic potential signals (Popovic et al. 2015a). With this approach, the AP at the soma can be recorded by patch-clamp technique, while voltage-imaging signals of bAP at dendritic spines can be recorded at a frame rate of 2 kHz at high optical magnification, with a satisfactory signal-to-noise ratio (S/N) and insignificant interference of the scattered light or the photodynamic damage (Holthoff et al. 2010) (Fig. 3.1). It is also possible to detect the bAP invading spines with a rapid time course critical for regulating plasticity and very similar to the spikes recorded in the parent dendrite.

Fig. 3.1

Optical recordings of backpropagating action potentials (bAPs) signals from individual dendritic spines of two neurons. (a, b) Upper micrographs show fluorescence images of dendritic branches with spines obtained with a CCD camera for voltage imaging. Lower micrographs show spiny dendrites reconstructed after deconvoluting stacks of spinning-disk confocal images. (c, d) Fluorescence intensity traces from locations 1 to 3 outlined in CCD images. Single-trial recordings and temporal averages of 4 (cell A) and 9 trials (cell B) are shown. Bottom traces are electrode recordings from the soma. The bAP signals, clearly recorded in spines, are absent from regions without spines (locations 2 and 3). (Reproduced from Holthoff et al. (2010), Figure 4, https://doi.org/10.1113/jphysiol.2009.184960, under CCC RightsLink^® license 5516071490678)

There is an important functional implication for these data. The magnitude of the synaptic depolarization (EPSP) at the dendritic spine, usually difficult to determine, could provide crucial information about the influence of synaptic inputs on local voltage-dependent processes and cellular responses (Acker et al. 2016). The EPSP signals present an amplitude fivefold to tenfold smaller than an AP. Therefore, the optical resolution used for monitoring the dendritic spines’ bAP is insufficient for recording unitary subthreshold synaptic responses at the spatial scale of individual spines, considering an adequate sensitivity and spatiotemporal resolution with a suitable S/N to allow a precise quantitative analysis.

The combination of high-sensitivity voltage-sensing dyes with the 2-photon uncaging of glutamate could be a suitable strategy for recording dendritic spines (Popovic et al. 2015b; Weng et al. 2023). The increment of the sensitivity can be achieved by increasing excitation intensity from a laser at a required wavelength and reducing the photodynamic damage (Popovic et al. 2015b; Weng et al. 2023). Then, a subthreshold uncaging EPSP signal (uEPSP) can be evoked by stimulating one synapse on a dendritic spine using 2-photon uncaging of glutamate. For example, this procedure could be done by simultaneously recording optical signals along a small segment of a basal dendrite from a cortical layer V pyramidal neuron at a frame rate of 2 kHz (Popovic et al. 2015a, b) (Fig. 3.2) and from cortical layers II/III, V, and VI pyramidal neurons at a frame rate of 5 kHz (Weng et al. 2023). The optical recordings of local uEPSP signals evoked by two-photon glutamate uncaging were followed by a bAP signal evoked by depolarizing pulse delivered by the somatic patch electrode (Popovic et al. 2015a, b; Weng et al. 2023).

Fig. 3.2

Recording subthreshold electrical events from individual dendritic spines. (a) Low-magnification fluorescence image of a basal dendrite labeled with a voltage-sensitive dye; z-stack of confocal images. Arrow: recorded spine. (b) High magnification confocal image of the same spine. Tip of iontophoretic electrode (labeled with the fluorescent dye) in the immediate vicinity of the spine head. (c) Single frame image of a spine in recording position obtained with a CCD camera for voltage imaging. (d) Traces shown on the left are evoked subthreshold excitatory postsynaptic potential (EPSP) recordings from the spine head (red) and its parent dendrite (green). Average of 16 trials. Bottom black traces are somatic electrode recording and the uncaging command pulse. Traces on the right are backpropagating action potentials (bAPs) signals from the same locations. Average of nine trials. (e) Left traces showing superimposed evoked subthreshold EPSP signals from the spine head and its parent dendrite calibrated in terms of membrane potential. Right traces represent bAP signals corrected for recording sensitivity difference. (f–j) Two-photon uncaging of glutamate. Same information as shown in (a–e). Red dot in h: position and approximate size of uncaging light spot. The evoked subthreshold EPSP and bAP recordings are average of 8 and 4 trials, respectively. (Adapted from Popovic et al. (2015b), Figure 5a–j, https://doi.org/10.1038/ncomms9436, under CC BY license, originally published by Springer Nature)

Also, the combined voltage imaging and glutamate uncaging using patterned illumination based on computer-generated holography allowed one-photon uncaging of glutamate on multiple spines to be carried out in parallel with voltage imaging from the parent dendrite and neighboring spines. In this procedure, somatosensory cortex in brain slices was used to perform electrical and optical recordings from cortical layer V pyramidal neurons filled with a voltage-sensitive dye. Subthreshold uEPSP and bAP signals were recorded at a frame rate of 2 or 5 kHz. Optical signals of a subthreshold uEPSP onto one spine and bAP evoked by depolarizing current pulse were able to be recorded from spine head and parent dendrite and also in two spines simultaneously, while somatic patch-clamp recordings of the subthreshold responses related to glutamate uncaging onto one and two spines were obtained (Tanese et al. 2017).

These approaches improved the accuracy required for spatiotemporal resolution to simultaneously record and reconstruct individual uEPSP and AP signals from the spine head and the parent dendrite at near physiological temperature (Popovic et al. 2015a, b; Tanese et al. 2017; Weng et al. 2023).

Recently, direct electrical recording from dendritic spines in culture and brain slices was achieved by using quantum-dot-coated nanopipettes (Jayant et al. 2017). The simultaneous electrical recordings on the soma and the spine of pyramidal cells from either primary hippocampal neuronal culture or neocortical slice preparations (Fig. 3.3) revealed that bAP invaded spines and was similar to a somatic AP. Also, the recorded spine bAP preceded the initial rise of EPSP measured in the soma (Fig. 3.4). Spontaneous EPSPs, with large amplitude (25–30 mV), fast rise time (1–2 ms), and half-width of approximately 5 ms were recorded as the nanopipette entered the spine head (Fig. 3.5) (Jayant et al. 2017). This possibility of direct measurement of the V_m at the spine level opens a way to explore the dynamics of local ionic conductances during synaptic integration and plasticity.

Fig. 3.3

Nanopipette dual somato-spine recordings. (a) Schematic describing the dual patch-nanopipette recording and associated signal transmission mechanisms to and from the spine. Action potentials (APs) originate in the axon hillock and propagate both forward (toward the postsynaptic cell) and backward into dendrites and spines. Backpropagating action potentials (bAPs) invade spines (inset) across the spine neck. Neurotransmitters released from the presynaptic axonal terminals activate receptors located on the spine head and cause spine head EPSPs (inset). Nanopipettes are labeled with QDs (CdSe/CdS/ZnS quantum dots) through adsorption (top right). (b) Schematic representation of dual somato-spine recordings in cultures and (c) slices. (d) Nanopipette “navigation” approach in slices toward a chosen target. (Adapted from Jayant et al. (2017), Figure 1a and Supplementary Figure 2 https://doi.org/10.1038/nnano.2016.268, under CCC RightsLink^® license 5516540783944)

Fig. 3.4

Nanopipette dual somato-spine recordings. (a, b) Dual somato-spine recordings in cultures (a, scale bar, 1.5 μm, panels (i)–(iii) and callout) and slices (b, scale bar, 20 μm left, 3 μm, inset middle). Dendritic and spine nanopipette recordings were restricted to within 100 μm from the soma to avoid space clamping and backpropagating action potential (bAP) attenuation. (c, f) Spine and presynaptic axon terminal recordings in culture (c, d) and slices (e, f). (c) bAPs invasion into spines. Somatic AP (bottom) and bAPs in the spine head (top) reveal that the nanopipette is capable of registering bAPs without the need for averaging. (d) Nanopipette recordings in the presynaptic axon terminal and postsynaptic soma. Note the characteristic 2–4 ms time delay Fig. 3.4 (continued) between the presynaptic AP peak and onset of the EPSP (inset: raw data, scale bar = 2 mV, 500 ms). (e) bAPs registration in spines probed in slices. (f) Putative presynaptic terminal recordings in slice preparations. Note the very short delay between the putative presynaptic AP peak and the EPSP rise measured at the soma. (g) Raw bAPs and somatic APs. Note the close fit in rise time and half-width indicating the bAPs invade the spine with nearly no loss in temporal structure. (Adapted from Jayant et al. (2017), Figure 5b–e, g–i https://doi.org/10.1038/nnano.2016.268, under CCC RightsLink^® license 5516540783944)

Fig. 3.5

Nanopipette synaptic potentials recorded in spine head. EPSPs in spines. Spontaneous break-in into the spine head (top) in cultures and corresponding somatic recordings (bottom). Note the large and fast-rising spontaneous EPSPs registered in the spine that are completely filtered from the soma. The gray-shaded region is zoomed in to show the typical rise and decay kinetics of the potential transient. (Adapted from Jayant et al. (2017), Figure 5j, https://doi.org/10.1038/nnano.2016.268, under CCC RightsLink^® license 5516540783944)

Furthermore, the changes in V_m of spines during EPSP are associated with variations of intracellular concentration of Na⁺ and Ca²⁺ caused by changes in channels and receptor conductances. While the influx of Na⁺ is important for driving the synaptic potential and bAP, the Ca²⁺ influx is essential for synaptic plasticity processing. Compared to Na⁺ channels, Ca²⁺ channels are well described in terms of type, distribution, and function in dendritic spines (Bloodgood and Sabatini 2005).

Imaging studies of AP and synaptically evoked Ca²⁺ transients have revealed multiple classes of voltage-sensitive Ca²⁺ channels that contribute to Ca²⁺ influx in dendritic spines (Bloodgood and Sabatini 2007a). Ca²⁺ ion influx has been studied in some detail in a series of experiments, using two-photon Ca²⁺ imaging and two-photon uncaging of glutamate. For example, through a combination of conventional electrophysiological techniques with two-photon excitation laser scanning microscopy, Ca²⁺ imaging, and optogenetic, it was possible to investigate local Ca²⁺ transients in dendrites and spines (Nakamura et al. 1999; Camiré and Topolnik 2018). The bAPs in dendrites opened voltage-gated Ca²⁺ channels (Jaffe et al. 1992; Popovic et al. 2012) with consequent Ca²⁺ transients in dendrites and spines (Jaffe et al. 1992; Popovic et al. 2012). Moreover, using combined electrophysiology, two-photon Ca²⁺ imaging, and two-photon glutamate uncaging in layer II pyramidal cells from the rat medial entorhinal cortex in acute brain slices, it was possible to detect dendritic spine bAP-evoked Ca²⁺ transients by R- and T-type voltage-gated Ca²⁺ channel activation (Theis et al. 2018). These bAP-evoked Ca²⁺ transients did not affect the EPSP measured at the soma but induced a downscaling of NMDA receptor currents (Theis et al. 2018). Na⁺ conductance on spines of mice neocortical pyramidal neurons has also been detected in experiments using two-photon uncaged glutamate associated with application of tetrodotoxin (TTX), a voltage-gated Na⁺ channel blocker that significantly decreased the spine potentials (Araya et al. 2007). Interestingly, the Na⁺ K⁺-ATPase α3 isoform was found to be expressed in spines with excitatory synapses (Blom et al. 2011), which suggests the active maintenance of ionic gradients by the dendritic spine at the postsynaptic cell.

Recently, the time courses of synaptically induced changes in Ca²⁺ and Na⁺ concentrations in the dendritic spines of hippocampal CA1 pyramidal cells were studied in brain slices of mice. Na⁺ and Ca²⁺ transients in a spine were simultaneously measured in response to a minimal electrical stimulation, in order to generate normal neurotransmitter release (smaller than uncaging glutamate), combined with low-affinity linear Na⁺ and Ca²⁺ indicators and single photon laser fluorescence stimulation with a high-speed, sensitive CCD camera (Miyazaki and Ross 2022). The intracellular Na⁺ concentration in the spine increased following a single synaptic stimulation. The kinetics of this synaptically activated increment of intracellular Na⁺ concentration in the spines of hippocampal CA1 pyramidal cells of mice was consistent with previous measurements in rat pyramidal neuron spines (Miyazaki and Ross 2017, 2022). The rise time of the synaptically activated increment of intracellular Na⁺ concentration was around 7.0 ms, about 2 ms slower than the evoked AP in the axon initial segment (Filipis and Canepari 2021; Miyazaki and Ross 2022). That is, a Na⁺ current contributes to the fast EPSP in the spine (Palmer and Stuart 2009; Popovic et al. 2015b), that is slower than the duration of the Na⁺ current. Almost all synaptically activated Na⁺ currents were mediated by AMPA receptors with little Na⁺ entry through either voltage-gated Na⁺ channels or NMDA receptors (Grunditz et al. 2008; Miyazaki and Ross 2017, 2022). In fact, using a highly sensitive electron microscopic immunogold technique, no evidence was found for Na⁺ channels in the dendritic spines of hippocampal CA1 pyramidal cells (Lorincz and Nusser 2010).

Similar to Na⁺, the intracellular Ca²⁺ concentration in a spine, also, increased following a single synaptic stimulation. The synaptically activated Ca²⁺ transients exhibited fast rise time and half-decay time kinetics (about 7.7 ms and 26 ms, respectively) (Kovalchuk et al. 2000; Sabatini et al. 2002; Bloodgood and Sabatini 2005; Enoki et al. 2009; Miyazaki and Ross 2022). Resembling the Na⁺ transient, most of the Ca²⁺ entry was driven by the AMPA receptor-mediated EPSP in the dendritic spine (Miyazaki and Ross 2022). Therefore, processes that modulate AMPA receptor activation, such as desensitization or potentiation, might have a direct effect on Ca²⁺ signaling and synaptic potential in dendritic spines. In contrast, two-photon uncaging experiments (Hestrin et al. 1990; Spruston et al. 1995) and uncaging evoked signals by repetitive stimulation (Polsky et al. 2009) have demonstrated much larger Ca²⁺ entry into spines through NMDA receptors. However, the larger Ca²⁺ current through NMDA receptors may be attributed to the intense two-photon uncaging activation compared with synaptically evoked responses by minimal electrical stimulation. In addition, bAP-induced spine depolarization opens NMDA receptors by lifting the Mg²⁺ block (Schiller et al. 1998; Yuste et al. 1999; Nevian and Sakmann 2004), further enhancing the spine V_m. This increase in spine V_m may be elicited either by additional activation of voltage-gated Na⁺ channels during the coincidence of bAP and incoming excitatory synaptic input (Stuart and Häusser 2001) or by previous depolarization caused by AMPA receptors in the spine head (Grunditz et al. 2008). For example, using whole-cell recording in acute brain slices of mice coupled with two-photon Ca²⁺ imaging, Ca²⁺ transients in spines were driven mostly by the activity of NMDA receptors in GABAergic intercalated neurons, located between the basolateral (BLA, from which they receive glutamatergic inputs) and the central amygdala (CeA) (Royer et al. 1999; Strobel et al. 2015). Both bAP and evoked spine Ca²⁺ transients in this cell type attenuated rapidly with distance, a finding caused by voltage-gated K⁺ channels present in the spine head (probably containing Kv4 channels) (Kaufmann et al. 2013; Strobel et al. 2017). Other techniques showed different results (Bloodgood and Sabatini 2005; Araya et al. 2007; Bywalez et al. 2015), which suggests that neurons and spines may exhibit distinct characteristics depending on the brain area studied, its connectivity and function, and the spine morphology (Bloodgood and Sabatini 2007b; Araya et al. 2007; Bywalez et al. 2015). Synaptically activated potentials can also involve distinct receptors and channel compositions that contribute to the variation in Ca²⁺ transients.

To date, there is no direct information on dendritic spine ionic conductance obtained from direct electrical activity measurement by whole-cell or cell-attached recordings of dendritic spines in slice preparation. However, electrical recordings of isolated and in situ dendritic spines in cultured hippocampal neurons of adult mice were performed using patch-clamping, detecting backpropagating dendritic electrical oscillations and NMDA-mediated electrical activity (Priel et al. 2022). First, dendritic spines were isolated from the dissected hippocampus and then reconstituted using a lipid bilayer membrane system. Ca²⁺-dependent spontaneous current oscillations, with peak frequency at 1–2 Hz, and NMDA receptor single-channel currents were recorded. NMDA and Ca²⁺-dependent oscillations at diverse frequencies and amplitudes were similarly recorded in the isolated dendritic spines using patch-clamp technique in both voltage and current-clamping configuration. Voltage-dependent conductances, consistent with both Na⁺ and K⁺ currents, and AP were also recorded in the isolated dendritic spine. The cell-attached patch-clamp recordings of in situ dendritic spines identified NMDA-dependent EPSP-like intrinsic oscillations at diverse frequencies and amplitudes. Double patch-clamp recordings of a dendritic spine and its parent dendrite detected a NMDA-dependent activity with diverse frequencies and waveforms coupled with and propagated to the parent dendrite. While synapses between dissociated cells do not necessarily exhibit the same properties as those in the intact brain tissue, they are more easily accessible for imaging. Valuable information can also be obtained about the biophysical properties of the dendritic spines and their interaction with the adjacent dendritic segments and the soma (Priel et al. 2022).

Although most spines receive an excitatory input, GABA-mediated inhibitory postsynaptic potentials (IPSPs) participate in the control of neuronal excitability. Experiments using cell type-specific optical stimulation in combination with two-photon Ca²⁺ imaging showed that the dendritic spines’ heads of cortical pyramidal cells are targeted by somatostatin-expressing interneurons, compartmentalize GABAergic inhibition, limit both bAP and synaptically evoked Ca²⁺ entry, and regulate NMDAR-dependent synaptic integration. These inhibitory synapses on spines can participate in the control of electrical and biochemical signaling along dendritic segments, including Ca²⁺ signaling and synaptic plasticity (Chiu et al. 2013).

The conductance of different channels and receptors and the specific changes in the V_m of dendritic spines are now being revealed by the use of combined new technical approaches, evidencing spatiotemporal patterns of induced or spontaneous depolarizations of dendrites and spines during synaptic integration and plasticity.

3.2 Dendritic Spines as Dynamic Compartments

Since their discovery by Ramón y Cajal, it has been hypothesized that dendritic spines may be involved in the compartmentalization of synaptic signals (Yuste 2015). Such a compartmentalized microenvironment could modulate the functional properties of different synapses (Calabrese et al. 2006; Araya et al. 2014; Tønnesen and Nägerl 2016; Vallés and Barrantes 2021) affecting synaptic strength, cell information processing, circuits’ organization, and behavior (Yuste and Denk 1995; Sabatini et al. 2002; Noguchi et al. 2005; Araya 2014).

3.2.1 Biochemical Compartmentalization

Dendritic spines, with a wide diversity of sizes and shapes, have organized membrane-bound enzymes, lipids, and proteins into delimited regions or subcompartments of varying extension and composition (Honigmann and Pralle 2016). Compartmentalization is generated by the spine neck, which relatively isolates the spine head from the parent dendrite (Yuste and Denk 1995; Nevian and Sakmann 2004; Honigmann and Pralle 2016).

A series of experiments, including those using two-photon excitation laser scanning microscopy in combination with Ca²⁺ imaging, has revealed that spines function as isolated biochemical compartments, not only in excitatory synapses, but also in GABAergic synapses that directly target spine heads (Chen et al. 2011; Chiu et al. 2013; Camiré and Topolnik 2018).

Spines act as biochemical and highly differentiated subcellular compartments. They have the ability to dynamically control the membrane area available for specific domains and clustering of neurotransmitter receptors as well as the diffusion of signaling and regulatory molecules, such as Ca²⁺, kinases, second messengers, enzymes, scaffolding proteins, and cytoskeletal elements within a very small volume (See Fig. 1.13, Chap. 1 of this book). Thus, biochemical compartmentalization restricts and regulates biochemical processes in individual synapses to smaller regions, modulating synaptic output, integration, and plasticity (Yuste 2011; Tønnesen and Nägerl 2016). Various neurotransmitter receptors, including the cholinergic nicotinic receptor, for example, have been located in the raft-type lipid domains of the spine membrane (Brusés et al. 2001; Allen et al. 2007; Egawa et al. 2016). Also, spine membrane domains show temporal modifications that affect the timing of biochemical cascades following synaptic activation. These changes induce dynamic alterations in the spine head and neck width associated with the degree of the synaptic signals compartmentalization, thereby biochemically isolating synaptic inputs and enabling input-specific plasticity (Magee and Johnston 1997; Lee et al. 2009; Tønnesen et al. 2014). In addition, the dynamic changes of the spine neck length have been associated with the bidirectional modulation of the synaptic and the parent dendrite voltage, acting as an electrical compartment and modifying synaptic integration (Vanderklish and Edelman 2002; Araya et al. 2014; Cornejo et al. 2022). This spatiotemporal organization involving dendritic spines is crucial to coordinate the synaptic function in a variety of neurons throughout the CNS.

3.2.2 Electrical Compartmentalization

Electrical compartmentalization relates to the ability of dendritic spines to process synaptic inputs, modulating the amplitude, kinetics, and integration of synaptic potentials into the spine head (Tsay and Yuste 2004). Spine heads exhibit voltage-gated ion channels and neurotransmitter receptors (Kim et al. 2007; Araya et al. 2007; Chung et al. 2009), which modulate the voltage inputs. In addition, the geometry of the spine neck influences voltage propagation into the adjacent dendritic shaft, acting as a high resistance pathway. The process of compartmentalization is influenced by membrane passive electrotonic and active properties (Koch and Segev 2000; Sjöström et al. 2008). Then, the geometry and passive properties of dendritic spines, such as axial resistance, membrane capacitance, and diameter, are crucial factors to understand how a spine modulates synaptic potentials and alters synaptic strength (Spruston et al. 1994), input integration (Yuste 2010), and, consequently, neural circuits (Yuste 2010; Kwon et al. 2017). As opposed to the well-established biochemical compartments, the electrical compartmentalization of dendritic spines is a matter of debate.

The spine electrical compartmentalization has been investigated with various experimental techniques in vitro and in vivo. For example, the strategy of combining a high-sensitivity organic voltage-sensitivity dye with two-photon uncaging of glutamate (Popovic et al. 2015b) detected subthreshold uEPSP in dendritic spines of layer V pyramidal neurons (Popovic et al. 2015a, b). The subthreshold uEPSP presented no significant attenuation along the spine neck, which exhibited low electrical resistance relative to the input impedance of the parent dendrite (mean value of 27 ± 6 MΩ) (Popovic et al. 2015b). These data suggested that spine synapses were not electrically isolated from the parent dendrites and may exhibit the same electrical behavior as those contacts made directly on dendrites (Popovic et al. 2014, 2015b). The low resistance of the spine neck relative to the impedance of the parent dendrite is similar to previous theoretical predictions and diffusional resistance measurements (Wilson 1984; Koch and Zador 1993; Svoboda et al. 1996; Bloodgood and Sabatini 2005; Tønnesen et al. 2014; Takasaki and Sabatini 2014).

Experiments using simultaneous Na⁺ and Ca²⁺ imaging with single-spine resolution in pyramidal neurons in rat hippocampal slices detected a rapid diffusion of Na⁺ out of the spine head through the spine neck to the parent dendrite, also suggesting that the neck offers low resistance for ionic diffusion (Miyazaki and Ross 2017). Moreover, the amplitude of EPSP and Ca²⁺ transients evoked by 2-photon glutamate uncaging also showed no correlation with the neck geometry, supporting the lack of a functional impact of spine shape on electrical compartmentalization (Takasaki and Sabatini 2014; Bywalez et al. 2015). With that assumption in mind, constricted spine neck diameter would rather serve to isolate metabolic events by reducing diffusion of activated molecules to neighboring synapses, without significantly influencing the transfer of synaptic charge to the postsynaptic dendrite (Harris and Stevens 1988, 1989).

Alternatively, other experiments have demonstrated that dendritic spines may constitute fundamental electric compartments (Tsay and Yuste 2004; Harnett et al. 2012; Yuste 2013; Acker et al. 2016; Jayant et al. 2017; Kwon et al. 2017; Strobel et al. 2017; Cornejo et al. 2022). For example, experiments using combined dendritic patch electrophysiology, Ca²⁺ imaging, and glutamate uncaging in spines at apical dendrites of rat hippocampal CA1 pyramidal neurons demonstrated that, when spine neck resistance was high (mean value of 514 ± 44 MΩ) (Harnett et al. 2012), passive amplification of the spine EPSP up to 50-fold would occur compared to an unitary dendritic EPSP. This amplification was very much dependent on the spine and neck structure in different neurons (Harnett et al. 2012).

Another study used two-photon voltage-sensitive organic dye recording with two-photon glutamate uncaging to measure the amplitude and duration of an uEPSP in single spines from the basal dendrites of layer V pyramidal neurons in acute brain slices of mice (Acker et al. 2016). Data from this study showed high spine neck resistance with a mean value of 179 ± 25 MΩ (ranging from 23 to 420 MΩ), suggesting the existence of a substantial electrical compartmentalization in spines.

Study using a genetically encoded voltage indicator and two-photon glutamate uncaging to identify electrical properties of dendritic spines in cultured hippocampal neurons showed that the invading bAP was unchanged. However, the spine neck resistance had values ranging from 15 to 332 MΩ (mean value: 101 ± 95 MΩ) reducing the synaptic and intrinsic voltage-gated currents that reach the adjacent dendritic shaft. Thus, the spines could compartmentalize voltage and attenuate the synaptic inputs by about twofold as they drive through the spine neck to the parent dendrite. Spines with longer neck attenuated uEPSP by fourfold as they propagate to the parent dendrites and exhibited an estimated neck resistance of 332 MΩ (Kwon et al. 2017).

Intracellular recordings of dendritic spines with long and narrow necks from neurons in brain slices using quantum-dot-coated glass nanopipettes revealed neck resistances ranging from 250 to 536 MΩ (mean: 425 ± 102 MΩ) (Jayant et al. 2017). The EPSPs recorded were large in the spine head (mean 26 mV), but strongly attenuated at the soma (0.5–1 mV), and the estimated neck resistance was large enough to generate significant local voltage compartmentalization in spines (Jayant et al. 2017) (Fig. 3.6).

Fig. 3.6

Nanopipette recordings in spines reveal electrical compartmentalization. (a) Electrical equivalent circuit of the passive dendritic spine. E_rest, leak reversal potential; R_m, spine head passive membrane resistance; C_m, spine head passive membrane capacitance; E_syn, synaptic reversal potential; g_syn, synaptic conductance; R_pore, pore resistance; R_seal, seal resistance; R_neck, neck resistance; R_e, pipette resistance; C_e, pipette capacitance; R_dendrite, dendritic resistance; R_m(d), dendrite passive membrane resistance; C_m(d), dendrite passive membrane capacitance; E_d, dendrite reversal potential. (b, c) Typical rise and decay kinetics for spontaneous PSPs recorded in the spine head. Notice the millisecond resolution rise and decay indicative of an extremely small RC time constant. The data shown in (b) and (c) are not deconvolved. (d) Raw EPSPs in spines from slices (marked stars) are larger than somatic EPSPs (∼0.5 mV flickers). Note the sudden registration of EPSPs in the soma (gray bar, bottom) that signify concomitant inputs from other spines. (e) Deconvolved R_neck (left), EPSP amplitudes (middle), and RMPs (right) in the spine head. The range of deconvolved EPSP amplitudes is in accordance with the spontaneous EPSP. The range for R_neck indicates that EPSPs that invade a low input resistance dendrite will be heavily attenuated. (Adapted from Jayant et al. (2017), (a–c) Supplementary Figure 12 and (d, e) Figure 5k, l, https://doi.org/10.1038/nnano.2016.268, under CCC RightsLink^® license 5516540783944)

Two-photon Ca²⁺ imaging coupled with whole-cell recording in amygdaloid GABAergic intercalated neurons in acute brain slices showed that Ca²⁺ enters the spine head mediated by the activity of NMDA receptors and with little spread of the Ca²⁺ signal. That finding also suggested that a dendritic spine head can act as an isolated electrical compartment (Strobel et al. 2017). High spine neck resistances, associated with longer and thinner spine necks, would be expected to increase the voltage in the spine head during a given synaptic input, reducing the driving force for synaptic current flow (Koch and Zador 1993).

Interestingly, the size of the spine head has also been correlated with the area of the postsynaptic density (PSD), a characteristic specialization of the postsynaptic cell membrane (Harris and Stevens 1989; Trommald and Hulleberg 1997; Arellano 2007), and with the amplitude of the generated EPSP (Matsuzaki et al. 2001; Noguchi et al. 2011). Varied EPSPs’ amplitudes have been observed in different experiments, which likely related to a spectrum of spine neck resistances correlated with distinct neck lengths, even in the same dendritic shaft and neuron type (Miyazaki and Ross 2022). Not only does EPSP modulation by spines have important repercussions for synaptic plasticity but also NMDA receptor activation and intra-spine Ca²⁺ levels activate a variety of signaling pathways that can modify actin filament assembly and contributing to activity-dependent alterations in spine morphology (Tada and Sheng 2006; Araya 2014; Yasuda 2017).

Finally, using in vivo two-photon microscopy and a genetically encoded voltage indicator, it was possible to measure the V_m of basal dendrites and their spines while performing simultaneous somatic whole-cell recordings in layers II/III pyramidal neurons of the somatosensory cortex of mice. Recordings were obtained during spontaneous activity and sensory stimulation (Cornejo et al. 2022) (Fig. 3.7). The estimated spine neck resistance was 226.6 ± 128.8 MΩ (mean ± SD), ranging from approximately 0 to 530.8 MΩ, similar to that seen in previous publications (Harnett et al. 2012; Tønnesen et al. 2014; Jayant et al. 2017). Spines and dendrites were depolarized together during AP, the dendritic potentials propagated into spines, and there was no attenuation of dendritic potentials or AP but synaptic potentials were significantly reduced. During subthreshold and resting potentials, spines could be activated independently, even in the absence of dendritic or somatic activity.

Fig. 3.7

Two-photon optogenetics and voltage imaging showing the voltage compartmentalization in dendritic spines in vivo. (a) Experimental design. (b) Construct and representative fluorescence changes in soma (light green, raw fluorescence; black, 10-Hz low-pass filtered) during 500-ms stimulation trials (red, 100 mW power). (c) Representative soma (top) and peak fluorescence response (bottom) during stimulation trials (×10, 500 ms, 100 mW). The dotted circle shows the stimulation area (scale bar, 10 μm). (d) Representative in vivo voltage-clamp recordings during optogenetic stimulation of proximal dendrites (100 mW, 100 ms) are shown on the left. Peak currents are shown on the right; −22.7 ± 11.3 pA (mean ± SD), 10 trials (n = 7 cells and 4 animals). (e) Representative peak fluorescence changes during optogenetic activation of dendritic shafts are shown on the left (stimulation ROIs are indicated by white dotted circles; 10 trials, 100 ms, 100 mW; color bar same as in (c); scale bar, 5 μm). Peak fluorescence changes in stimulated dendritic shaft (dendrite stim), adjacent dendritic spine (spine), and unstimulated dendritic shaft (dendrite no stim) are shown on the right (n = 34 dendrites and 9 animals). (f) Same as (e) during optogenetic activation of spines (n = 35 spines and 12 animals; scale bar, 5 μm). ****p < 0.0001. (g) Simplified electrical model. (h) Resistance ratio (R_n to R_den) of spine-only events during spontaneous (Spon, n = 116), sensory stimuli (Sen, n = 181), and optogenetic spine stimulation (Opto, n = 35). (i) Values of spine neck resistance for stimulated spines; median = 213.7 MΩ (n = 35 spines). In (d), (h), and (i), boxes and whiskers represent median (line), 25th to 75th percentiles (box), range (whiskers), and mean as a “+.” *postASAP: postsynaptic accelerated sensor of APs. (Reproduced from Cornejo et al. (2022), Figure 4, https://doi.org/10.1126/science.abg0501, under CCC RightsLink^® license 1338950-1)

These findings, including the spine-independent depolarization during spontaneous and evoked activity, contribute to our understanding of dendritic spine biophysics, showing that dendritic spines serve not only for biochemical but also for electrical compartmentalization of synaptic inputs essential for synaptic integration, information processing, and plasticity (London and Häusser 2005; Yuste 2011; Cornejo et al. 2022).

3.3 Spatiotemporal Dynamics of Dendritic Spines Related to Synaptic Plasticity

Dendritic spines are known to be the site of long-term synaptic plasticity (Matsuzaki et al. 2004; Lai and Ip 2013). In fact, the activity-dependent morphological changes in the head and/or neck of spines (Matsuzaki et al. 2004; Tønnesen et al. 2014; Araya et al. 2014) have been correlated with modifications in synaptic strength in cortical pyramidal neurons involving biochemical and electrical mechanisms (Araya et al. 2006; Araya 2014).

Studies using signal imaging techniques based on fluorescence resonance energy transfer (FRET) in combination with photostimulation of single synapses revealed the temporal sequence of biochemical signaling cascades and their spatial spreading from stimulated spines (Nishiyama and Yasuda 2015). Dendritic spine plasticity directly modifies the synaptic strength of neuronal networks through changes in protein synthesis, spine cytoskeleton structure, spine morphology, and density of receptors at the synaptic membrane (Nakahata and Yasuda 2018). These modifications guide rearrangements of connections between presynaptic and postsynaptic elements and consequently promote a shift of synaptic efficacy, either potentiating or depressing synaptic transmission. Over the long term, two processes involving various molecular mechanisms promote synaptic strength or weakness, respectively: long-term potentiation (LTP) and long-term depression (LTD) (Citri and Malenka 2008). Together, spine synaptic plasticity allows the establishment and reorganization of connectivity within neuronal circuits during normal brain development to adulthood and aging. It also may be involved in degenerative processes and in a large spectrum of brain disorders that involve deficits in information processing, such as schizophrenia, Alzheimer’s disease, autism spectrum disorder, and epilepsy (Penzes et al. 2011; Dickstein et al. 2013; Nishiyama 2019; Runge et al. 2020; Rossini et al. 2021; Zaccard et al. 2023).

Before describing the dynamics of receptors in the spines during LTP and LTD, it is important to consider a fundamental synaptic component that influences spine synaptic plasticity and the formation and stabilization of new dendritic spines, that is, the extracellular matrix.

The extracellular matrix, together with pre- and postsynaptic elements and glial processes, form the so-called tetrapartite synapse (Dityatev and Schachner 2003; Dityatev et al. 2006; Dityatev and Rusakov 2011; Song and Dityatev 2018). The fully developed extracellular matrix has the ability to stabilize neuronal connectivity, restraining morphological changes of dendritic spines and decreasing cortical plasticity in the adult brain. In fact, as the circuitry reaches its adult form, spine motility diminishes and connectivity stabilizes (Oray et al. 2004). A prominent component of the matrix, chondroitin sulfate proteoglycans (CSPGs), may play an important role in limiting cortical plasticity, restraining the structural remodeling of neurons, and inducing inhibition of synaptic transmission (Oray et al. 2004).

Degradation of CSPGs proved to affect ionic conductances, control receptors, and channel localization at the cell membrane (Gundelfinger et al. 2010; Dityatev et al. 2010), induce rapid structural changes of dendritic spines by increasing the size of the spine head, and enhance spine mobility thereby restoring plasticity (de Vivo et al. 2013). CSPGs modulate structural plasticity by preventing integrin activation and signaling at synaptic sites (Orlando et al. 2012).

Several other molecules of the extracellular matrix have been involved in the regulation of synaptic plasticity. For example, hyaluronan, the backbone of extracellular matrix, was found to regulate hippocampal synaptic plasticity by modulating postsynaptic L-type Ca²⁺ channels (Kochlamazashvili et al. 2010). Another important glycoprotein, Reelin, is secreted into the extracellular matrix by neurons, such as GABAergic cells. Reelin regulates the composition of NMDA receptors, the increment of AMPA responses by increasing the number of AMPA receptors on the postsynaptic membrane, and robustly enhances LTP (Qiu et al. 2006; Campo et al. 2009; Hellwig et al. 2011). Reelin strongly induces spine remodeling, regulating spine size and stability, and the number of synaptic contacts per spine (Beffert et al. 2006; Pujadas et al. 2010). Reelin interacts with integrins to regulate AMPA receptor internalization and cell membrane trafficking of NMDA receptor subunits (Dityatev et al. 2010; McGeachie et al. 2011), important for receptor dynamics during synaptic plasticity.

Furthermore, in mature neurons, integrins are involved in homeostatic synaptic scaling and interact with matrix metalloproteinases, promoting spine volume changes induced by long-term plasticity (Nagy et al. 2006; Wang et al. 2008). Cell adhesion molecules (CAMs) mediate extracellular matrix and PSD signaling that may influence the actin cytoskeleton network of dendritic spines, and consequently, spine shape (Huntley et al. 2002; Washbourne et al. 2004; Koleske 2013). Spine remodeling is, also, controlled by autocrine signaling of brain-derived neurotrophic factor (BDNF), crucial for LTP (Harward et al. 2016; Hedrick et al. 2016) and for proteolytic cleavage of extracellular matrix (Nakahata and Yasuda 2018). The increased intracellular Ca²⁺ concentration, mediated by NMDA receptors or by voltage-gated Ca²⁺ channels, induces lysosomal fusion with the plasma membrane, releasing the protease Cathepsin B to the extracellular matrix, which cleaves and inhibits the endogenous inhibitor of the matrix metalloprotease 9. The disinhibited matrix metalloprotease 9 cleaves the CAMs and the extracellular matrix, facilitating the structural remodeling of spines (Nakahata and Yasuda 2018). Thus, the extracellular matrix that fills perisynaptic spaces surrounding dendritic spines plays a crucial role in remodeling neural circuits, affecting long-term synaptic plasticity and receptor trafficking (Dityatev et al. 2010; Orlando et al. 2012).

3.3.1 Dynamics of Receptors in Spines During Long-Term Potentiation

The most studied form of spine structural plasticity is LTP (Nakahata and Yasuda 2018). The typical protocol for inducing LTP involves brief and high-frequency stimulation (for review, see Nicoll 2017), and paired pre- and postsynaptic stimulation, known as spike timing-dependent plasticity (STDP) (Markram et al. 1997; Debanne et al. 1998; Bi and Poo 1998; Sjöström et al. 2008; Feldman 2012). However, there are protocols applying lower-frequency stimulations to induce LTP mediated by metabotropic glutamate receptor (mGluR) and protein kinase A activation (Lanté et al. 2006). Importantly, LTP is input-specific at the level of individual synapses, that is, only stimulated spines may trigger LTP-induced synaptic modifications (Magee and Johnston 1997; Lee et al. 2009; Tønnesen et al. 2014). Input-specific LTP induction can be performed by two-photon glutamate uncaging. For example, using an uncaged glutamate-LTP protocol to study the visual cortex of anesthetized rats in vivo, it was possible to demonstrate the induction of individual spine enlargements (Noguchi et al. 2019) reinforcing the LTP-synaptic specificity (Harvey and Svoboda 2007). Although LTP can also be mediated by presynaptic mechanisms, here we focus on the NMDA receptor-dependent form of LTP (Luscher and Malenka 2012).

Multiple molecular and cellular mechanisms contribute to induction, expression, and maintenance of NMDA-dependent form of LTP. Ca²⁺ represents a key factor for long-term synaptic plasticity (Mateos-Aparicio and Rodríguez-Moreno 2020). During postsynaptic depolarization, Ca²⁺ entry through activated NMDA receptors results in a local increase in dendritic excitability and an increase in synaptic strength (Luscher and Malenka 2012). Activation of NMDA receptors with subsequent increase of Ca²⁺ influx is a well-known mechanism involved in the early phase of LTP, which lasts approximately 60 min (Becker and Tetzlaff 2021). Experimental approaches using chemical and selective activation of NMDA receptors were able to induce synaptic potentiation, that was blocked by a competitive antagonist, D-2-amino-5-phosphonopentanoate (AP5) (Lu et al. 2001). Following NMDA receptors agonist washout of cultured hippocampal neurons, synaptic potentiation was still enhanced, revealing the contribution of AMPA receptors in the maintenance of postsynaptic excitatory response potentiation (Lu et al. 2001). Additionally, NMDA-dependent LTP could be blocked by MK-801, a competitive NMDAR antagonist, and BAPTA, a Ca²⁺ chelator, highly supporting the involvement of NMDA and AMPA receptor activation and Ca²⁺ influx in LTP expression (Lu et al. 2001) (Fig. 3.8).

Fig. 3.8

LTP induction. (a) Up: The illustration demonstrates that brief stimulation of postsynaptic NMDA receptors with glycine may induce LTP. Bottom: Representative traces of mEPSC. Glycine-induced LTD increases mEPSC amplitude and frequency measured by whole-cell patch clamp in hippocampal neurons of primary dissociated cultures. (b) Averaged mEPSCs demonstrating before, during, and following the application of glycine. (c) Pretreatment with AP5, MK-801, and BAPTA blocked LTP, decreasing both mEPSC amplitude and frequency. (Adapted from Lu et al. (2001), https://doi.org/10.1016/S0896-6273(01)00194-5, (a) Figure 1A,D,F,G and (b, c) Figure 2E,F, under CCC RightsLink^® license 5516571126317)

LTP induced by STDP at CA3-CA1 hippocampal synapses and some synapses of neocortical layers II/III pyramidal cells are dependent on NMDA receptors (Nishiyama et al. 2000; Froemke et al. 2005). NMDA receptor-mediated Ca²⁺ accumulation in a spine is also associated with actin polymerization-dependent shrinkage of the activated spine neck and an increase in synaptic strength, as demonstrated during the timing-dependent LTP triggered by two-photon glutamate uncaging at the presynaptic site paired with postsynaptic spikes in layer V pyramidal neurons from juvenile mice (Tazerart et al. 2020) (Fig. 3.9a, c).

Fig. 3.9

Induction of t-LTP and t-LTD in single dendritic spines. (a) Experimental protocol for t-LTP induction in single dendritic spines (sp). (b) Representative experiment where a spine was activated with t-LTP pre- and post-pairing protocol of +13 ms. Traces correspond to average of ten uncaging EPSPs (uEPSPs) recorded in the soma and generated by 2P uncaging before (control, black trace) and after t-LTP induction (red trace) over the indicated spine (red dot). (c) Experimental protocol for t-LTD induction in single dendritic spines. (d) Representative experiment where a spine was activated with t-LTD post–pre pairing of −15 ms. Traces correspond to average of 10 uEPSPs recorded at the soma and generated by 2P uncaging before (control, black trace) and after Fig. 3.9 (continued) t-LTD induction (red trace) over indicated spine (red dot). (e) Biochemical signaling pathways of the STDP induced by NMDA receptors. N and A, NMDA, and AMPA receptors, respectively. (Figure and legend adapted from (a–d) Tazerart et al. (2020), https://doi.org/10.1038/s41467-020-17861-7, Figures 1a,b and 2a,b, respectively, under CC BY license, originally published by Springer Nature; (e) Feldman (2012), https://doi.org/10.1016/j.neuron.2012.08.001, Figure 4A, under CCC RightsLink^® license 5516580280377)

Furthermore, during synaptic potentiation, increased Ca²⁺ influx leads to activation of biochemical cascades through Ca²⁺-calmodulin-dependent protein kinase II (CaMKII), followed by phosphorylation of AMPA receptors and insertion of new receptors in the postsynaptic membrane sites. CaMKII antagonists block the synaptic potentiation, preventing exocytosis of new receptors at the postsynaptic membrane (Matsuzaki et al. 2004; Lee et al. 2009; Makino and Malinow 2009; Lisman et al. 2012; Jaafari et al. 2013; Penn et al. 2017). In vivo studies also revealed that CaMKII is crucial for LTP, learning, memory, and long-lasting spine enlargements (Miller et al. 2002; Matsuzaki et al. 2004; Buard et al. 2010). Indeed, the enlargement of dendritic spines associated with increase in AMPA receptor-mediated currents (Matsuzaki et al. 2004) and the expression of new AMPA receptors in the spine membrane at the stimulated synapse (Tanaka and Hirano 2012; Vitureira and Goda 2013) contribute to the enhancement of synaptic transmission during LTP (Park 2018) and are associated with larger PSDs (Chater and Goda 2014). Those changes were dependent on NMDA receptors, Ca²⁺ entry, CaMKII, and actin polymerization (Matsuzaki et al. 2004; Lee et al. 2009; Makino and Malinow 2009; Lisman et al. 2012; Jaafari et al. 2013; Penn et al. 2017). Also, the enlargement of individual spines precedes the increase in AMPA receptors at the postsynaptic site at spines (Chater and Goda 2014).

A protocol of glycine-induced LTP in dissociated cortical and hippocampal neuronal cultures also showed an increase of AMPA receptors expression in the spine membrane by exocytosis from intracellular pools. It begins within a few minutes after the LTP-inducing stimulation and is associated with a long-lasting potentiation of AMPA receptor-mediated excitatory postsynaptic current (EPSC) (Lu et al. 2001; Jaafari et al. 2013; Vitureira and Goda 2013; Henley and Wilkinson 2016). These new AMPA receptors permeable to Ca²⁺ (CP-AMPA receptors) are GluA2-lacking receptors and most of them contain the GluA1 subunit (Lu et al. 2001; Clem and Barth 2006; Jaafari et al. 2013), whose reversible phosphorylation plays an important role in long-term plasticity (Lee et al. 2010). The reduction of AMPA receptors diffusion in the spine membrane directly impairs hippocampal-dependent fear learning (Penn et al. 2017). There is strong evidence that synaptic CP-AMPA receptors have a crucial role in vivo and in vitro, tuning neuronal excitability to induce plasticity. However, how and to what extent CP-AMPA receptors are incorporated during LTP remains an area of research (Henley and Wilkinson 2016) (Fig. 3.10a, b).

Fig. 3.10

AMPA receptors trafficking in synaptic plasticity. (a) Under basal conditions, synaptic AMPA receptors (AMPAR) are predominantly calcium impermeable and are anchored to postsynaptic density protein 95 (PSD95) and the actin cytoskeleton. They undergo both lateral diffusion in the membrane and rapid constitutive recycling. Furthermore, under certain conditions, Ca²⁺-permeable AMPARs (CP-AMPARs) can be observed extrasynaptically. (b) During long-term potentiation (LTP) induction, activation of protein kinase A (PKA), which is anchored at synapses by the scaffold protein AKAP150, leads to phosphorylation (indicated by the P) of GluA1 at S845 in both spines and dendrites, promoting its surface expression and leading to the appearance of synaptic CP-AMPAR. In addition, enhanced binding of protein interacting with C kinase 1 (PICK1) to GluA2-containing Ca²⁺-impermeable AMPARs (CI-AMPAR) in intracellular vesicles restricts their recycling to the cell surface, thus further enhancing the ratio of CP-AMPAR to CI-AMPAR. (c) During activity deprivation to induce synaptic homeostatic scaling, calcineurin (CaN) activity is reduced, favoring PKA-mediated phosphorylation of GluA1 subunits at S845, and promoting surface expression of GluA1-containing AMPAR and their synaptic incorporation. Furthermore, induction of homeostatic scaling via NMDA receptor (NMDAR) blockade promotes the local translation of GluA1 in dendrites, and an increase in surface expression of GluA1-containing CP-AMPAR. By contrast, (d) long-term depression (LTD) is associated with dephosphorylation of GluA1 at S845, reducing the extrasynaptic pool of GluA1-containing CP-AMPARs, which then undergo lysosomal degradation. (Reproduced from Henley and Wilkinson (2016), https://doi.org/10.1038/nrn.2016.37, Figure 4, under CCC RightsLink^® license 5516911470380)

Besides AMPA receptors, NMDA receptors also exhibit gating and trafficking during LTP. This includes the rapid delivery of functional receptors and an increase in receptor clusters on the membrane of dendrites and dendritic spines thereby modulating neuronal excitability (Lan et al. 2001; Grosshans et al. 2002; Lau and Zukin 2007; Lau et al. 2010). The modifications in NMDA receptors are induced by protein kinase C (PKC) and tyrosine kinase Src (Lan et al. 2001; Grosshans et al. 2002; Lau and Zukin 2007; Lau et al. 2010) (Fig. 3.11a). Therefore, both AMPA and NMDA receptors contribute to synaptic potentiation. Increase of both components of EPSP (AMPA and NMDA) with a prominent potentiation of AMPA over the NMDA was reported during early and late LTP (Dozmorov et al. 2006), as well as, for the delayed NMDA-dependent LTP (Watt et al. 2004) (Fig. 3.12).

Fig. 3.11

NMDA receptors trafficking in synaptic plasticity. (a) In the hippocampal CA1 of adult animals, high-frequency stimulation (HFS) and long-term potentiation (LTP) promote rapid insertion of NR2A-containing NMDA receptors and an increase in NMDA field excitatory postsynaptic potentials (fEPSPs) at CA1 synapses through a protein kinase C (PKC)- and Src-dependent pathway (blue box). HFS-induced activation of mGluR5 can also induce LTP with NMDA receptors trafficking (red box). (b) Long-term depression (LTD) of synaptic NMDA EPSCs (LTD_NMDAR) occurs by retrieval of NMDA receptors from synaptic sites. Low-frequency stimulation (LFS) of Schaffer collaterals can induce metabotropic glutamate receptors (mGluR)-dependent LTD_NMDAR or NMDA-dependent LTD. (mGluR)-dependent LTD is associated with enhanced internalization of NMDA receptors (red box). NMDA-dependent LTD promotes actin depolymerization and lateral diffusion of NMDA receptors away from the synapse site (blue boxes). LFS of connected pairs of CA3 pyramidal neurons induces LTD_NMDAR by stimulating dynamin-dependent internalization of NMDA receptors (green box). (Adapted from Lau and Zukin (2007), https://doi.org/10.1038/nrn2153, Figure 5, under CCC RightsLink^® license 5516591094919)

Fig. 3.12

AMPA and NMDA components of the LTP response. (a) Representative traces of mEPSCs before (control), immediately after LTP induction (early LTP), and >2 h after LTP induction (late LTP). At early LTP, mEPSC amplitude and frequency increase significantly. At times >2 h after LTP induction (late LTP), mEPSC frequency returned to control levels. (b) Average mEPSC waveforms from control (gray) and early LTP (black) showing that the rapid AMPA component of the mEPSC was potentiated (AMPA-LTP), whereas the slowly decaying NMDA component was unaffected. (c) Average mEPSC waveforms from early LTP (gray) and late LTP (black). During late LTP, potentiation of the NMDA component (NMDA-LTP) was now also evident. (d) AMPA and NMDA components of mEPSCs at early and late LTP. (Up) Increased AMPA component of mEPSCs at both periods, and a delayed potentiation of NMDA component of mEPSCs only at late LTP. (Bottom) Time course of AMPA (open boxes) and NMDA (filled triangles) mEPSC amplitudes, with a prominent contribution of AMPA component. (Adapted from Watt et al. (2004), https://doi.org/10.1038/nn1220, (a–c) Figures 1 and (d) 6a,b under CCC RightsLink^® license 5516591392907)

At later stages of LTP, structural modifications of spines are triggered to support synaptic strength enhancement. Spine structural modifications are essential to synaptic plasticity (Fig. 3.13) (Harvey and Svoboda 2007; Noguchi et al. 2019), since it is highly correlated with the molecular reorganization and functional expression of receptors, such as AMPA and NMDA in excitatory synapses (Cao and Harris 2012; Baltaci et al. 2019). Increased NMDA-mediated synaptic transmission not only can modify dendritic spine morphology but also can induce new spine formation, stabilizing recently formed spines and enlarging already-existing spines (Lai and Ip 2013). Indeed, an increase in synaptic strength by LTP triggers the pruning and replacement of nonactivated spines by new ones together with a selective stabilization of activated synapses (De Roo et al. 2008) and the rapid formation of new spines that can be blocked by NMDA receptors antagonists (Toni et al. 1999; Ruddy et al. 2015).

Fig. 3.13

In vitro and in vivo enlargement of spines under LTP. (a) (Up) LTP induction in vitro leads to rapid expansion of spine heads. Images before (23 min) and after (25 min) LTP induction. At time = 0, the LTP protocol (30 uncaging pulses at 0.5 Hz, 4-ms pulse duration, postsynaptic potential 0 mV) was applied to the spine marked by a circle (LTP spine). A triangle marks a tested nearby spine. (Middle) uEPSCs, averaged across all cells, in response to test stimuli before (23 min; gray) and after (40 min; black) the LTP protocol. (Bottom) Time course of the changes in uEPSC amplitude and spine volume at the LTP spine (filled circles; n 5 7) and at nearby spines (open triangles; uEPSC, n 5 7; Vol, n 5 31). The arrow marks the LTP protocol. (b) (Up) Induction of spine enlargement in the visual cortex in vivo using two-photon glutamate uncaging. Time-lapse images show an expansion of stimulated spines. (Bottom) Spine’s volume increased after LTP stimulation. Spine (i) in cyan; and spine (ii) in magenta. (Adapted from (a) Harvey and Svoboda (2007), https://doi.org/10.1038/nature06416, Figure 1a,b under CCC RightsLink^® license 5517380739551; and (b) from Noguchi et al. (2019), https://doi.org/10.1038/s41598-019-50445-0, Figure 1b,c under CC BY license, originally published by Springer Nature)

NMDA receptor-dependent enlargement of spines can be reversed by low-frequency stimulation (LFS) via a phosphatase-dependent mechanism, a well-known mechanism of LTD (Luscher and Malenka 2012).

Therefore, spine dynamics may constitute the basis of long-term molecular reorganization correlated with long-term memory and cognition. Strengthening synaptic connections is believed to be a common mechanism of both processes, and abnormal spine profiles can be correlated with altered synaptic transmission in various neurological and neuropsychiatric conditions (Kasai et al. 2010; Lai and Ip 2013).

3.3.2 Dynamics of Receptors in Spines During Long-Term Depression

LTD has been considered as a parallel and opposite process of LTP, associated with a reduction of functional synapses and, consequently, reduction in reactivity to afferent stimulation (Luscher and Malenka 2012). There are several protocols for LTD induction, including a low-frequency stimulation (LFS: 0.5–3 Hz), chemical induction (He et al. 2011) and paired pre- and postsynaptic stimulation (STDP) (Markram et al. 1997; Debanne et al. 1998; Bi and Poo 1998; Sjöström et al. 2001; Feldman 2012) or mediated by metabotropic glutamate receptors (mGluR) (Hasegawa et al. 2015).

The NMDA receptor-mediated LTD involves the activation of NMDA receptors followed by a small but lasting elevation in postsynaptic Ca²⁺ concentration, and activation of a serine-threonine protein phosphatase cascade that act to modulate the spine actin cytoskeleton (Stein and Zito 2019). For example, NMDA receptor-mediated moderated intracellular Ca²⁺ signals in spines and parent dendrites were observed in the timing-dependent LTD triggered by two-photon glutamate uncaging using the post- and pre-protocol in a single spine of layer V pyramidal neurons in juvenile mice (Tazerart et al. 2020) (Fig. 3.9b, c). The LTD induced by STDP at CA3-CA1 hippocampal synapses and some synapses on neocortical layers II/III pyramidal cells is dependent on NMDA receptors (Nishiyama et al. 2000; Froemke et al. 2005; Feldman 2012) (Nishiyama et al. 2000; Froemke et al. 2005). The LTD induced by STDP at several synapses in layers II/III and V of somatosensory and visual cortex and in cortical synapses onto striatal medium spiny neurons is dependent on mGluR- and endocannabinoids receptors (CB1) (Nevian and Sakmann 2006; Feldman 2012).

Following Ca²⁺ entry through activated NMDA receptors, LTD requires the activation of protein phosphatase 1 (PP1), which is essential for the functional process of LTD but not for spine structural modifications. Cofilin, a family of actin-binding proteins that trigger a rapid depolymerization of actin microfilaments, mediates spine structural modifications, that is, spine shrinkage induced by LFS (Zhou et al. 2004). In addition, the activation of NMDA receptors with Ca²⁺ entry and subsequent activation of calcineurin and AP2 are required for both functional and structural processes of LTD (Zhou et al. 2004) and for triggering AMPA receptors endocytosis at the dendritic spine (Vitureira and Goda 2013). The NMDA receptor-induced internalization of AMPA receptors can be inhibited by preventing the increment of intracellular Ca²⁺ concentration or by specific calcineurin inhibitors (Carroll et al. 2001). The removal of extrasynaptic CP-AMPA receptors and replacement by Ca²⁺-impermeable AMPA receptors (CI-AMPA receptors) is associated with LTD also in mGluR-induced LTD (Bellone and Lüscher 2005; Henley and Wilkinson 2016). During another form of plasticity, in which the neuron self-tunes its synapses, called the homeostatic synaptic scaling (Fig. 3.10a, c) (Turrigiano 2008), the recruitment of CP-AMPA receptors remains controversial and it may be dependent on the synapse, developmental stage and mode of induction. Overall, the activity-dependent removal of CP-AMPA receptors can define a time frame in which memories can be erased (Henley and Wilkinson 2016) (Fig. 3.10a, d).

NMDAR trafficking is also implicated in LTD following NMDA receptor activation (Carroll et al. 2001; Lau and Zukin 2007) (Fig 3.11b). In the hippocampus, for example, NMDAR trafficking requires actin depolymerization (Morishita et al. 2005) for the lateral diffusion of NMDA receptors between synaptic and extrasynaptic sites or dynamin-dependent internalization (Montgomery et al. 2005). Notably, activity-dependent alterations in NMDA synaptic strength influence metaplasticity, experience-dependent plasticity, and structural remodeling of local circuitry (Philpot et al. 2003; Lau and Zukin 2007).

Live imaging of dendritic spines after stimulation by low-frequency uncaging glutamate demonstrates that the stimulus of LTD leads to spine shrinkage specifically of the stimulated spine but not of neighboring spines. Like LTP-induced spine enlargement, spine shrinkage induced by LTD is also synapse-specific (Oh et al. 2013). Volume reduction was observed in both large and small spines after pharmacological manipulations in combination with two-photon glutamate uncaging. Shrinkage of small dendritic spines required activation of NMDA receptors, whereas shrinkage of large spines required signaling through both NMDA and metabotropic glutamate receptors mGluR and the second messenger inositol 1,4,5-trisphosphate receptors (IP3Rs) (Oh et al. 2013). LTD induction in dendritic spines followed or not by spine shrinkage is highly correlated with synaptic weakness (Fig. 3.14). However, these mechanisms also play a vital role in the neural circuit plasticity that underlies learning. Indeed, the formation and stabilization of new dendritic spines as new circuits are formed during learning is only possible due to the elimination of preexisting spines (Yang et al. 2008; De Roo et al. 2008) (see also Chap. 7 in this book).

Fig. 3.14

LTD induction and structural plasticity of spines in vitro and in vivo. (a) Low-frequency stimulation (LFS) induces long-term depression. (b) (Up) In vitro shrinkage of spines at different time points before and after stimulation. (Bottom) Quantitative measure of spine head diameter demonstrating a decrease of spine volume after LTD induction. (c) (Up) In vivo shrinkage of spines after LTD induction, showing a decrease of spine volume in the stimulated spine. (Bottom) The magenta, yellow, and white circles indicate spine S1, n1, and n2 traces, respectively. (Adapted from (a) and (b): Zhou et al. (2004), https://doi.org/10.1016/j.neuron.2004.11.011, (a, b Bottom) Figure 2A,B and (b Up) Figure 1C,D, under CCC RightsLink^® license 5518550626563; (c) from Noguchi et al. (2019), https://doi.org/10.1038/s41598-019-50445-0, Figure 3A,B under CC BY license, originally published by Springer Nature)

3.4 Concluding Remarks

For many years, the electrical properties of dendritic spines, such as membrane voltage and ionic conductances, were not able to be directly measured due to technical limitations usually difficult to overcome. However, several important biophysical properties of spines have been unveiled recently. Advancements result from the ongoing development of novel and integrated technologies, as described in this chapter.

We highlighted some important achievements for detecting the electrical properties of dendritic spines. First, we discussed the direct measurement of the V_m (bAPs and synaptic potentials) at the spine head, composed by a rich variety of ion channels and receptors and whose dynamics and trafficking are influenced by intracellular signaling and extracellular matrix during synaptic activation and plasticity. By comparing the modifications of the V_m of spines to the parent dendrites and soma, we can now have a better understanding about signal processing and integration. Second, we described the possibility to measure the spine neck resistance, which is associated with the electrical compartmentalization of spines. Spine neck resistance is crucial for tuning the amplitude and kinetics of synaptic signals along the dendrite and controls the spatiotemporal dynamics of information during the synaptic integration and plasticity. Although the electrical role of dendritic spines remains an open issue, these constant advances in technology are providing valuable evidence on the intrinsic properties of dendritic spines for the biochemical and electrical compartmentalization and associated molecular and structural modifications during synaptic plasticity that contribute actively to neuronal behavior.