Fast Synaptically Activated Calcium and Sodium Kinetics in Hippocampal Pyramidal Neuron Dendritic Spines

Identification of single spine sodium and calcium signals

In a typical experiment, we simultaneously measured the spine sodium and calcium transients in response to electrical stimulation. For the experiment shown in Figure 1, we first elicited a bAP in a pyramidal neuron with a 1-ms pulse in the soma. This pulse was followed by two synaptic stimuli separated by 50 ms. Although the stimulation pulses were equal, the second response reached threshold for generating an action potential. This protocol was then repeated in the presence of 100 μm APV, which blocked NMDA receptors. This kind of experiment was repeated many times in different cells. The responses in this cell were particularly clean, providing a strong template for further analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

[Ca²⁺]_i and [Na⁺]_i changes in a dendritic spine evoked by a bAP and by synaptic stimulation. A, Top, Fluorescence image of a dendritic segment with an ROI and the position of the stimulating electrode marked. Middle, Difference image (between the times at the ends of the double arrow in B) of the [Ca²⁺]_i change (ΔF) evoked by synaptic stimulation (first arrow at the bottom of B). The ROI was positioned over the location of the [Ca²⁺]_i change. Bottom, Difference image of the [Na⁺]_i change. The location of the change is the same as the location of the [Ca²⁺]_i change. The length of both the calcium and sodium spots was 0.8 μm (FWHM). The color bars on the side show the range of values in the image from maximum to minimum. These color bars also apply to Figures 2 and 7. B, [Ca²⁺]_i and [Na⁺]_i changes at the ROI in response to an intrasomatically evoked bAP and two synaptic stimuli; the second synaptic response was suprathreshold. [Na⁺]_i increases are shown downward because SBFI fluorescence decreases with increasing [Na⁺]_i. C, Responses to the same stimuli with 100 μm APV added to the ACSF. The [Na⁺]_i changes were almost unaffected but most of the [Ca²⁺]_i increase was blocked, leaving only a small signal (red arrow).

We first made a difference image of the fluorescence changes corresponding to the [Ca²⁺]_i and [Na⁺]_i changes elicited by synaptic stimulation (start and stop times indicated by double headed arrow over the synaptic optical response in Fig. 1B). These revealed locations of [Ca²⁺]_i and [Na⁺]_i changes on the dendrite, as shown in the two lower images in Figure 1A. For both sodium and calcium signals these locations were small spots about the size of a dendritic spine (average size ∼0.7 ± 0.2 μm, FWHM, measured with no spatial filter; n = 18). Using the camera as a detector allowed us to search for the activated spines after the trial was recorded and could determine, in most cases, if more than a single spine was activated in the illuminated region on the dendrite. In the example shown in Figure 1, only one spine in the image field (32 × 32 μm²) was activated, although from the amplitude of the EPSP in the soma (∼5 mV) we know that multiple spines in the dendrites must have been activated (Magee and Cook, 2000). In this dendrite, we could not see the activated spine, possibly because it was projecting above or below the dendrite. But its size, and the isolated response, strongly suggests that only one spine on this dendritic segment was activated. We then placed a small analysis box (ROI) over the spot and measured the dynamic [Ca²⁺]_i and [Na⁺]_i changes from that location. In this way we analyzed calcium signals from 66 isolated spines [OGB-5N (51), Fluo-5F (15), most from mice (35; OGB-5N (20), Fluo-5F (15)]; the remainder were from rats. Most of these responses also had localized sodium signals. Ten responses [OGB-5N (9), Fluo-5F(1)] were further identified as signals from single spines because we could clearly see the spine where the signal came from. This was probably because the spine projected to the side of the dendrite and the dendrite was close to the surface of the slice, making for a sharper image with the camera.

bAP-evoked signals

To establish a reference for synaptic responses we first measured the [Ca²⁺]_i and [Na⁺]_i changes from bAPs using low-affinity indicators. In the experiment shown in Figure 1, the bAP elicited a rapid calcium transient, which rose to peak <2 ms, the effective frame interval for most calcium measurements. In eight cells, the average rise time was also <2 ms, as in previous experiments with the same time resolution (Sabatini et al., 2002). To examine the rise time of these signals more accurately we made a series of measurements at 4× higher speed (2000 Hz) without simultaneous sodium imaging. With this higher precision the 10–90% rise times were 1.58 ± 0.14 ms (n = 4) in the proximal dendrites, consistent with previous measurements (Jaafari et al., 2014). The same values were measured in the distal, and oblique dendrites. The half decay time of the bAP signal in the dendrites measured at this speed was 20.5 ± 5.2 ms (n = 4). When measured at 500 Hz the half recovery time in the spine was 12.0 ± 5.2 ms (n = 13) and 17.6 ± 3.6 ms (n = 9) in the nearby dendrites. These decay times, measured with 150 μm OGB-5N, are comparable to the half decay times measured previously (Sabatini et al., 2002) with very low concentrations of higher affinity indicators (OGB-1, fluo-4, Mg-Green). This concentration of OGB-5N was not buffering the calcium in the spine since similar half decay times in the dendrites were measured with 600 μm OGB-5N (18.6 ± 2.4 ms; n = 3). This fact, together with the observation that the magnitudes of all evoked calcium transients (see below) are far below the K_d of OGB-5N (∼40 μm), means that the fluorescence decay times measured in the experiments in this paper are an accurate reflection of the underlying calcium dynamics and are not sensitive to the level of indicator filling in the dendrites and spines.

A sodium response in the spine to a single bAP was not detected using SBFI (n > 20). Since the [Ca²⁺]_i change from the bAP in the spine in Figure 1 (and all other spines selected for analysis) was clear we know the spike reached the spine. In three additional cells we averaged the response to trains of 40 bAPs and looked for spine sodium signals; none were detected. Figure 2A shows that, with the same apparatus, we easily detected an action potential-evoked [Na⁺]_i increase from an area of 0.4 × 0.4 μm² (comparable to spine dimensions) in the axon initial segment, which is known to have a higher density of Na⁺ channels (Kole et al., 2008; Fleidervish et al., 2010) than dendrites (Stuart and Sakmann, 1994; Magee and Johnston, 1995). Previous experiments on rats showed that small [Na⁺]_i increases could be detected from large dendritic segments from trains of bAPs with signal averaging (Rose et al., 1999; using SBFI; Miyazaki and Ross, 2017; using ING-2), This result is not a difference between rats and mice since additional experiments using ING-2 (data not shown) detected small [Na⁺]_i changes in large segments of murine dendrites with averaging of multiple bAP signals.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Sodium transients in dendritic spines and the axon initial segment. A, SBFI fluorescence transient detected by a single pixel (red box) covering an area of 0.4 × 0.4 μm² in the axon initial segment (AIS) in response to an intrasomatically evoked AP (blue trace). The rise time of the signals was <2 ms (the frame interval of the camera) and was easily detected without averaging or filtering. B, An example of two spine signal with fast (4.1 ± 1.2 ms) and moderate (31.7 ± 17.2 ms) Na⁺ half recovery times activated by the same extracellular stimulus (calculated from exponential fits to the transients). Fluorescence signals from the two ROIs on a mouse pyramidal neuron and the corresponding somatic electrical recording are shown. C, Distribution of half decay times of spine Na⁺ transients in response to single synaptic stimuli. Measurements were made from 20 mouse spines (57 traces) using SBFI, 20 rat spines (43 traces) using SBFI, and 14 rat spines (36 traces) using ING-2. Half decay times were binned in 10-ms groups and normalized to the peak number for each protocol. The distributions were similar among the three groupings. Recordings with ING-2 from Miyazaki and Ross (2017).

Na⁺ signals from synaptic stimulation

In contrast to the bAP result, the middle panel of Figure 1 shows that a spine [Na⁺]_i increase was clearly detected following single synaptic stimulation. In general, while [Na⁺]_i increases in spines from single bAPs were never detected, synaptic responses were often detected. This result was found from spines within 150 μm from the soma, where bAP amplitude is expected to be large (Spruston et al., 1995b), and from spines in the most distal dendrites, where the bAP amplitude is smaller and the EPSP is bigger (Magee and Cook, 2000).

The kinetics of the synaptically activated [Na⁺]_i increases on mouse spines was consistent with previous measurements on rat pyramidal neuron spines (Miyazaki and Ross, 2017). For these new mouse experiments, we measured the spine [Na⁺]_i changes with SBFI at higher time resolution (500 Hz) and improved S/N. The 10–90% rise time of the [Na⁺]_i increase was 7.0 ± 4.5 ms (n = 13). This rise time is slower than the <2-ms rise time of the AP-evoked signal in the axon initial segment (n = 6; data not shown; see also Filipis and Canepari, 2021), showing that the slower rise time is not because of technical factors like camera response time or indicator rate constants. It likely reflects the integral of the sodium current underlying the fast EPSP in the spine (Palmer and Stuart, 2009; Popovic et al., 2015), which is slower than the duration of the current. The [Na⁺]_i change in response to the second synaptic stimulus was almost identical, although the response was suprathreshold, evoking a bAP. As with the response to the bAP alone, there was no obvious contribution of the synaptically evoked bAP to the [Na⁺]_i increase, in contrast to previously reported results (Rose and Konnerth, 2001). The larger spine [Ca²⁺]_i increase, when the spike was evoked, confirmed that the bAP reached this location. These results indicate that activation of voltage-gated sodium channels (VGSCs) make little contribution to the synaptically evoked [Na⁺]_i increase in spines (see also Miyazaki and Ross, 2017). The kinetics of the sodium signal were almost unaffected by 100 μm APV (rise time reduced from 6.6 ± 3.6 to 6.0 ± 2.7 ms; p = 0.5; n = 4; half decay time reduced from 22.8 ± 4.5 to 19.7 ± 1.8 ms; p = 0.4; n = 4). The amplitude (%ΔF/F), while hard to measure because of the background fluorescence, was reduced from 8.8 ± 3.0 to 7.6 ± 2.4 (p = 0.4; n = 4). Since almost all the [Na⁺]_i transient remains in APV, most of the influx underlying the signal does not come though NMDA receptors, although there must be some contribution through this pathway since >93% of the current through NMDA receptors is carried by sodium (Schneggenburger et al., 1993). Most likely the influx is through AMPA receptors, since most of it is blocked by NBQX (Fig. 5). This conclusion is consistent with the observation that the EPSP, as recorded in the soma, was almost unaffected by APV (Fig. 1), an observation reported by many groups (Feldmeyer et al., 2002).

A spectrum of synaptically activated [Na⁺]_i half decay times was measured in murine pyramidal neurons with 2 mm SBFI. An example of transients in two spines on the same dendrite with fast and moderate Na⁺ recovery times, activated by the same stimulus, is shown in Figure 2B. The half recovery times ranged from 10–60 ms (mean 21.7 ± 9.6 ms; n = 57; Fig. 2C), with approximately the same distribution measured in rat pyramidal neurons with SBFI (mean 23.9 ± 12.2; n = 43) and with ING-2 (mean 20.0 ± 15.7; n = 33; Miyazaki and Ross, 2017). Two cells (data not shown) had very long Na⁺ transients, whose decay times could not be measured. The similarity of decay times measured with the two indicators is consistent with their being weak buffers with high K_d values for the sodium:indicator reaction (Miyazaki and Ross, 2017). If the indicators did buffer sodium, the recovery times of the true sodium transients would be even faster. Similar decay time distributions were measured from both rat and mouse spines (Fig. 2C), suggesting that these are general properties, at least among rodent spines. The rapid decay times are because of Na⁺ diffusion from the spine to the dendrite as shown previously for spines in rat pyramidal neurons (Miyazaki and Ross, 2017). Similar fast removal times were measured from other localized sites of Na⁺ entry in the axon initial segment and node of Ranvier (Fleidervish et al., 2010). In those axon experiments, the fast recovery times were not affected by ouabain, a blocker of Na/K-ATPase, showing that for short times the removal was not because of a pump. In other conditions, when there is no significant [Na⁺]_i gradient, e.g., in the dendrites following a train of bAPs, Na⁺ is removed slowly (seconds) via a temperature-sensitive membrane pump (Rose et al., 1999). In this situation, removal by diffusion is less significant, since there is expected to be little [Na⁺]_i gradient in the dendrites following bAPs. The fast removal time of Na⁺ by diffusion from spines is coincidently similar to the removal time of synaptically evoked Ca²⁺ (Figs. 3, 4B) but through an entirely different mechanism. Ca²⁺ is removed through a combination of pumps and buffers (Scheuss et al., 2006) and not by diffusion to the dendrites (Sabatini et al., 2002).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The effect of APV (100 μm) on synaptically evoked Ca²⁺ transients in dendritic spines. APV reduced the magnitude to 0.33 ± 0.09% of control (p = 0.00,006) and half decay times from 30.2 ± 8.6 to 18.7 ± 9.2 ms (p = 0.003) but had no significant effect on the rise times (91% of control; p = 0.5). Changes for each of ten spines are shown.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Magnitudes and time courses of synaptically evoked Ca²⁺ transients. A, Histogram of the number of traces with different values for the ratio of the synaptic [Ca²⁺]_i increase to the [Ca²⁺]_i increase evoked by a bAP at the same spine location. Events were binned in five groups (n = 40 trials from 24 spines). B, Distributions of the ΔF/F magnitudes, 10–90% rise times, and half decay times of synaptically evoked Ca²⁺ transients in spines as a function of the somatically recorded EPSP amplitude. All parameters were measured using 150 μm OGB-5N. No correction was made for background fluorescence (n = 58 total trials from 20 spines). C, Distribution of signals where only one active spine was visible in the camera field (16 trials from single activated spines in the field of view from 10 spines). D, Distribution of signals from the subset of responses from clearly visible spines (16 trials from 8 spines from 8 cells). The dashed lines in each panel show the best linear fits to the data. In most of these cells, the amplitude of the EPSP suggests that multiple axons were stimulated. The responses in one cell where only a single presynaptic fiber was activated is shown in Extended Data Figure 4-1.

The variations in Na⁺ decay times were greater than the error in measuring the times. In two spines where we were able to measure the signal many times, we found that the half decay times were 22.2 ± 5.4 (n = 7) and 28.1 ± 7.3 (n = 7) ms; the errors are much less than the range of half decay times. Since the somatic EPSP and the spine [Na⁺]_i increase were almost unaffected by APV the spine EPSP also should be unchanged in APV. Therefore, these different decay times reflect real physiological variation. Most likely they are because of variations in spine properties, like neck length, but some part of the slower signals may be because of a contribution from a dendrite included in the same ROI where diffusion from the spine slows the response. Without this component (unknown) the recovery times would be even faster.

Ca²⁺ signals from synaptic stimulation

The calcium response to synaptic stimulation was superficially similar to the bAP signal (Fig. 1A). However, careful examination showed that the 10–90% rise time was slower (7.9 ± 3.3 ms; n = 20 spines) and the half decay time (25.4 ± 11.6 ms; range 15–50 ms; n = 18 spines) was slower than the bAP signal. As was shown many times previously (Yuste et al., 1999), most of the [Ca²⁺]_i increase was because of entry through NMDA receptor channels, since the addition of 100 μm APV reduced the Ca²⁺ signal to undetectable levels in many cells when OGB-5N was used (n = 8). However, it is possible that some of these nonresponses were because of failure of transmitter release. In nine cells where a clear signal in APV was detected (Fig. 3), the remaining signal amplitude was ∼32% of the control signal and is likely to reflect Ca²⁺ entry through VGCCs (Bloodgood et al., 2009). One example is shown in Figure 1, arrow. The half decay times of these remaining transients in APV (18.7 ± 9.2 ms) were faster than the synaptic response without APV (30.2 ± 8.6) and were comparable to the half decay times of the voltage-evoked bAP signals in spines (12.0 ± 5.2 ms). The fast rise times were unaffected by APV. This result suggests that both the NMDA receptor-mediated current and the remaining VGCC-mediated current are both driven by the same sharp voltage rise from the AMPA-mediated EPSP. This means that the small [Ca²⁺]_i increase in APV is about the magnitude of the VGCC signal in physiological conditions and is only a small fraction of the total synaptic [Ca²⁺]_i increase. When the synaptic response was suprathreshold for generating a bAP (Fig. 1) the [Ca²⁺]_i increase always was larger, reflecting the additional Ca²⁺ entry through VGCCs (n > 10). This was confirmed by the response to suprathreshold stimulation in APV. In this case, the time course of the Ca²⁺ transient was almost identical to the response to the bAP alone (Fig. 1C).

Magnitude of synaptic [Ca²⁺]_i increase

The amplitude measurements cannot easily be calibrated in terms of the [Ca²⁺]_i change since there was significant background fluorescence. However, these amplitudes can be directly compared with the amplitudes of the bAP-evoked ΔF/F values in the same spine; the background levels are the same for both measurements. The distribution of these ratios is shown in the histogram in Fig. 4A; the average is 1.2 ± 0.5 (n = 40 trials from 24 cells). Significant variation is expected since spines are morphologically heterogeneous, the number of released vesicles varies from trial to trial, and the amplitude of the bAP varies in different dendritic regions. Previously, Sabatini et al. (2002), using two-photon microscopy, estimated the amplitude of bAP-evoked [Ca²⁺]_i increases, in the limit of zero added indicator, as 1.7 ± 0.6 μm. These measurements were not contaminated by background fluorescence because of the high z-axis resolution of the two-photon process. Using their value as a reference suggests that the average synaptically evoked [Ca²⁺]_i increase was 2.0 ± 0.8 μm (1.7 × 1.2). This is the expected value without indicator since 150 μm OGB-5N is not a significant buffer.

One potential correction to this calculation for some events is that the ROI for the spine might include some underlying or overlying dendrite. In this case, the dendritic fluorescence would be background for the synaptic signal but would be an active signal source for the bAP signal. To try to correct for this error we selected spines that were clearly visible to the side of the dendrite (Fig. 2B). For these events the dendrite should make little contribution to a signal with an ROI placed over the spine. For five cells that met this criterion we found a ratio of 1.1 ± 0.5, leading to an average peak [Ca²⁺]_i change of 1.9 ± 0.9 μm, about the same as determined from the full dataset.

Possible effects of other spines

Since almost all the synaptically activated [Na⁺]_i increase is because of Na⁺ entry through AMPA receptors on the activated spine, which has little voltage dependence, and there is little voltage-gated Na⁺ entry, it is unlikely that activation of other spines by the stimulation current affects the localized Na⁺ signal in the spine, although some of the voltage from those other spines invades the measured spine. However, the contribution of VGCCs to the [Ca²⁺]_i increase in the spine, suggests that the voltage spread from other spines, could affect the magnitude and time course of the spine calcium signal (Palmer and Stuart, 2009). To investigate this possibility, we plotted the rise time, decay time, and amplitude of the [Ca²⁺]_i change, measured with 150 μm OGB-5N, as a function of the somatically recorded EPSP (Fig. 4B). We reasoned that although there is not a direct relationship between the somatic EPSP and the EPSP in the examined spine, there would be some correlation if spine EPSP amplitude was important in determining these signal parameters. There was a weak correlation of ΔF/F amplitude with EPSP amplitude. However, it is not clear whether the correlation is meaningful since the ΔF/F values were not corrected for background fluorescence and may be different for each spine. There was no correlation of rise time with EPSP size. There was a correlation with the half decay time of the synaptic Ca²⁺ transient, which became more rapid at higher EPSP amplitudes. This trend might reflect the greater contribution of VGCC-mediated Ca²⁺ entry at higher EPSP amplitudes. Increased Ca²⁺-activated K⁺ conductance might also affect the time course of the calcium signal.

It is possible that some of these conclusions depend on whether the spines were isolated in the camera field or had close neighbors that were activated at the same time. Also, the results might differ if the spines were visible at the side of the dendrite or were not clear because they were above or below the dendrite. To check on these possibilities we measured the results from trials with isolated spines and visible spines and plotted them separately. The patterns in each of the panels in Figure 4C,D show that there were no obvious differences correlated with how the events were selected.

The highest EPSP, ∼12 mV, probably reflects activation of at least 15 synaptic sites (Magee and Cook, 2000). The weak correlation of these parameters with the somatic EPSP amplitude suggests that the EPSPs generated in other spines had little influence on the EPSP in the analyzed spine, although we cannot be precise in this estimate. These other spines must have acted at some distance since there was either only one activated spine or widely separated spines in the camera field for most the data points in Figure 4C. Since the EPSP did spread to the soma, this result suggests that the spine EPSP amplitude is much higher than the somatic EPSP amplitude. In other related experiments, Carter et al. (2007) found only small interactions among nearby spines in striatal medium spiny neurons, and Bloodgood et al. (2009) found little interaction among pyramidal neuron spines. In contrast, Harnett et al. (2012) and Weber et al. (2016) reported significant interactions among spines.

A more direct way of eliminating the influence of other spines on the Ca²⁺ signal of the examined spine, is to stimulate only a single axon, which is hard to do. After some effort we had two successes; the best one is shown in Extended Data Figure 4-1. We found that the optical time constants evoked by this single axonal fiber (rise time: 11.4 ± 2.5 ms; half decay time: 31.4 ± 10.9 ms) were like those measured with activation of multiple spines (Fig. 4B) and the EPSP amplitude (0.85 ± 0.24 mV) was similar to those measured with single fiber activation in purely electrical experiments (Allen and Stevens, 1994).

Analysis of the components of the calcium signals

The rise times and decay times of the synaptically evoked calcium transients are faster than many previous measurements of these times. Nevian and Sakmann (2004) and Enoki et al. (2009) measured fast rise times although they used high-affinity indicators, which may have distorted the responses. In our experiments using OGB-5N, most rise times were <10 ms, which is much faster than the duration of the NMDA receptor conductance when there is no Mg²⁺ block (Spruston et al., 1995a) This conclusion suggests that a large part of the calcium signal is driven by the fast EPSP in the spine, which briefly relieves the Mg²⁺ block. This idea, which has been suggested previously (Bloodgood et al., 2009; Holbro et al., 2010), does not reveal which fraction of the [Ca²⁺]_i increase results from this mechanism, and which part results from Ca²⁺ entering through the NMDA receptor after the EPSP is finished and the membrane potential returns to resting level (Kovalchuk et al., 2000). One way to separate these two contributions is to deconvolve the synaptic calcium signal into temporal components, using the [Ca²⁺]_i transient from a bAP or other brief voltage-gated Ca²⁺ signal as a template. Bloodgood et al. (2009) used this approach to analyze the components of signals evoked by two-photon glutamate uncaging. They found that an early phase of Ca²⁺ entry was driven by the spine AMPA-mediated EPSP since it was blocked by the AMPA receptor antagonist NBQX. However, most of the signal was because of the second slow component resulting from Ca²⁺ entry through the NMDA receptor at resting potential. We used this method to analyze the synaptically evoked Ca²⁺ transients. We used the Ca²⁺ signal from the bAP as the impulse current (kernel) and deconvolved the synaptic signal in experiments using different calcium indicators [150 μm OGB-5N, 300 μm fluo-5F (to match conditions in previous published experiments), and 50 μm OGB1]. The bAP signal and the synaptic signal in normal ACSF and with 10 μm NBQX are shown in Figure 5A. NBQX blocks the fast component of the synaptic signal and almost all the somatically recorded EPSP. The deconvolution shows that most of the calcium influx occurred in the first 20 ms, and this component was almost eliminated in NBQX. Figure 5B shows that in experiments using fluo-5F the slow Ca²⁺ current (at 20 and 40 ms) comprised <10% of the early signal. Similar results were found with the other indicators (Fig. 5C). The results were qualitatively the same as in Bloodgood et al. (2009), although we found that the early synaptic component was significantly larger than the fast component recorded in their two-photon uncaging experiments. The difference was not because of different deconvolution techniques since we reproduced the deconvolution records of Bloodgood et al. (2009) when we used their recorded calcium transients in our deconvolution calculations.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Synaptic Ca²⁺ currents analyzed by deconvolution of the Ca²⁺ transients using three indicators of different K_d values. A, Left, bAP-evoked Ca²⁺ transient detected with fluo-5F used as the kernel in the deconvolution. Center, Responses with and without NBQX of synaptically evoked Ca²⁺ transients, using fluo-5F. In NBQX the response is smaller and slower. Right, Deconvolution of the synaptic responses using the bAP-evoked kernel. The largest response is in the first 10 ms after synaptic stimulation, and this component is blocked by NBQX. Arrows indicate 20 and 40 ms after the peak of the calcium current. B, Normalized histogram of the Ca²⁺ currents in the two conditions binned in 20-ms groups. Most of the effect of NBQX is in the first bin (0–10 ms; n = 5). C, Histogram of the effect of NBQX on the first 20 ms of the synaptic response, analyzed with the three indicators. The effects were similar (OGB-5N, n = 7; fluo-5F, n = 8; OGB-1, n = 2).

These experiments show that most of the synaptic signal recorded using fluo-5F, especially the fast component, was blocked by NBQX. Similar results were found in spiny stellate cells (Nevian and Sakmann, 2004). However, some experiments (Kovalchuk et al., 2000) suggest that similar AMPA receptor blockers like CNQX have little effect on the subthreshold synaptic calcium signal when 100 μm OGB-1 is used as the calcium indicator. One possible explanation for the difference is that synaptic responses using high affinity indicators like OGB-1, might saturate the indicator since the synaptic [Ca²⁺]_i increase was typically ∼2–3 μm (see above), while the K_d for OGB-1 is only 0.2 μm. Recordings with OGB-5N (K_d ∼40 μm) would not saturate (Higley and Sabatini, 2008). Another possible explanation is that Kovalchuk et al. (2000) used multiple stimuli to evoke a synaptic calcium response, which often evoke regenerative NMDA receptor responses. To examine these possibilities, we tested the effect of NBQX on the synaptic signals generated by single shocks using three different indicators with different K_d values. Figure 6A shows examples of these experiments. The recovery times of the synaptic calcium signals are fastest with OGB-5N, consistent with it being the least buffering indicator. More to the point, NBQX blocked a much higher fraction of the synaptic signal when OGB-5N was used. Summary statistics for this experiment are shown in Figure 6B. This result suggests that experiments using OGB-1, and to some extent fluo-5F, are missing the peak of the spine synaptic transient. When the less buffering indicator, OGB-5N, was used, it becomes clearer that most of the transient is blocked by NBQX.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

The effect of NBQX on synaptically evoked Ca²⁺ transients detected with indicators of different K_d values. A, Examples of responses with and without NBQX. The effect of NBQX is greatest when OGB-5N was used. All records are single trials except the recording of NBQX on the OGB-5N cell (average of 3 traces). The effect of NBQX on the somatically recorded EPSP is shown below. B, Histogram of the effect of NBQX on the amplitudes of the synaptic transients detected with the three indicators. The fractional reduction is greater with OGB-5N than with fluo-5F (p = 0.004) or with OGB-1 (p = 0.001; OGB-5N, 150 μm, n = 5; fluo-5F, 300 μm, n = 5; OGB-1, 50 μm, n = 5).

Comparison between synaptic and glutamate uncaging-evoked calcium transients

All the above experiments analyzing calcium transients used electrically evoked synaptic responses and OGB-5N to measure the calcium signals. But most recent experiments (Bloodgood et al., 2009; Holbro et al., 2010; Beaulieu-Laroche and Harnett, 2018) used glutamate uncaging to evoke the synaptic responses and relatively higher affinity indicators to measure the calcium transients. The responses we recorded were clearly faster than those recorded in most published uncaging experiments and the slow calcium current (20–40 ms) was less than slow current produced by two photon glutamate uncaging. But it is not clear how much the difference is because of the indicator and how much to the method of stimulation. To examine this issue, we did some basic glutamate uncaging experiments, following the protocols of other investigators, with the exception that we used OGB-5N to detect the Ca²⁺ transients.

Figure 7A shows a typical experiment. We first confirmed that when properly aligned the pulsed laser activates a small spot. Figure 7B shows the fluorescence of a 4-μm bead detected at 500–600 nm when excited at 800 nm. The response, detected by the camera, had a FWHM of 0.7 μm). We then positioned a spine just next to the laser spot. In our experience, uncaging directly on the spine often generated fluorescence artifacts and bleached the indicator. Using a flash of 1.0 ms in duration, we generated a small uEPSP (1.1 ± 0.4 mV) and a calcium transient (Fig. 7A). The [Ca²⁺]_i change was confined to the region of the spine. The calcium signal had a fast rise time (∼10 ms) and a slower half decay time (∼50 ms). We repeated this experiment for many other spines, measuring their rise times and half decay times. For quantitative analysis we rejected signals that were not confined to identified spines, and which had strong stimulus artifacts. In some cases, the shape of the response was sensitive to the position of the ROI. For those responses we chose the location with the fastest kinetics. Figure 7C,D shows these times on a cumulative probability plot and compares them with similar times measured from synaptically activated responses (Fig. 4B). The rise times of synaptically evoked (7.9 ± 3.3 ms; n = 63) and uncaging evoked (10.8 ± 3.9 ms; n = 16) transients are close, although the p-value for a t test of the difference was 0.02. In contrast, as shown in Figure 7D, the half decay times were clearly different. The synaptically evoked half decay time (26.0 ± 11.5 ms; n = 55) was almost twice as fast as the uncaging evoked half decay time (49.4 ± 15.6 ms; n = 16). The p-value for the difference was 1.9 × 10⁻⁵. One issue of concern is that Figure 4B shows that the events with large EPSPs have the fastest decay times, which might be because of a strong contribution from VGCCs in addition to NMDA receptors. Even if we eliminate all events with EPSPs larger than 5 mV, the average half decay time for the synaptically activated events is 29.0 ± 14.3 ms (n = 26). This still differs from the uncaging half decay time with a p-value of 0.004. In summary, both synaptically evoked and uncaging evoked calcium transients are fast, but the half decay times of synaptically evoked transients are ∼25 ms faster.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Time courses of Ca²⁺ transients evoked by uncaging MNI-glutamate on a dendritic spine detected with 150 μm OGB-5N. A, Image of a dendritic segment with a targeted spine indicated with a red arrow. A 1-ms flash at 720 nm generated a small EPSP and a calcium transient from the red ROI. The image in the inset shows the same segment without the ROI to make the spine clearer. The pseudocolor difference image (between the times at the ends of the black arrow) shows that the [Ca²⁺]_i increase was only over the spine. B, Enlargement of a part of an image of a fluorescent bead excited by an 800-nm two-photon flash. Each pixel is 0.2 μm. The FWHM of a profile through the image (data not shown) was 0.7 μm. C, Cumulative frequency of rise times from many synaptically activated (n = 63) and uncaging (n = 16) calcium transients. D, Similar cumulative frequencies of half decay times. There is a clear difference between the two profiles.