Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. 1998 Aug 4; 95(16): 9596–9601.

Calcium dynamics in single spines during coincident pre- and postsynaptic activity depend on relative timing of back-propagating action potentials and subthreshold excitatory postsynaptic potentials


We compared the transient increase of Ca2+ in single spines on basal dendrites of rat neocortical layer 5 pyramidal neurons evoked by subthreshold excitatory postsynaptic potentials (EPSPs) and back-propagating action potentials (APs) by using calcium fluorescence imaging. AP-evoked Ca2+ transients were detected in both the spines and in the adjacent dendritic shaft, whereas Ca2+ transients evoked by single EPSPs were largely restricted to a single active spine head. Calcium transients elicited in the active spines by a single AP or EPSP, in spines up to 80 μm for the soma, were of comparable amplitude. The Ca2+ transient in an active spine evoked by pairing an EPSP and a back-propagating AP separated by a time interval of 50 ms was larger if the AP followed the EPSP than if it preceded it. This difference reflected supra- and sublinear summation of Ca2+ transients, respectively. A comparable dependence of spinous Ca2+ transients on relative timing was observed also when short bursts of APs and EPSPs were paired. These results indicate that the amplitude of the spinous Ca2+ transients during coincident pre- and postsynaptic activity depended critically on the relative order of subthreshold EPSPs and back-propagating APs. Thus, in neocortical neurons the amplitude of spinous Ca2+ transients could encode small time differences between pre- and postsynaptic activity.

Pre- and postsynaptic action potentials (APs) in synaptically connected neurons, when occurring within an interval of a few hundred milliseconds, can either increase or decrease synaptic efficacy of excitatory connections (1, 2). During subthreshold excitatory postsynaptic potentials (EPSPs), glutamate receptor channels are transiently opened and mediate a Ca2+ inflow (38), which may be restricted to single spines and adjacent portions of the dendritic shaft (3, 6, 8) or may extend also into larger parts of the dendritic shaft (5, 9). On the other hand, APs propagating back into the dendritic arbor open voltage-dependent Ca2+ channels (VDCCs) both in the dendritic shaft and spine heads (8, 10, 11). An elevation of the postsynaptic intracellular free Ca2+ concentration contributes to long-lasting changes in the efficacy of glutamatergic synapses (12, 13), but how the different sources of Ca2+ interact is unclear. We used multiphoton fluorescence microscopy (14) to measure how the Ca2+ signal in dendritic spines of pyramidal neurons in layer 5 of the rat neocortex depended on the exact timing of subthreshold EPSPs and back-propagating APs.


Multiphoton Microscopy.

A modified confocal microscope (TCS 4D, Leica Laser Technik, Heidelberg) was adapted to an upright microscope (BX50WI, Olympus Optical, Hamburg) equipped with a ×60 objective (LUMPlanFL 60×W0.9IR, Olympus Optical). For excitation we used short pulses at 76 MHz from a Ti:Sa-Laser (MIRA 900F, Coherent, Santa Clara, CA), either 90–110 fs at 840 nm or 150–170 fs at 890 nm. In line scan imaging both scan directions and external detection through the condenser were used for signal collection. This allowed us to use an average laser power (≤4–9 mW in the focal plane, depending on depth below slice surface) that was below the threshold (9–12 mW, measured in the system used) for photodamage (15). Transmission and epifluorescence signals were recorded by photomultiplier tubes (R6357, Hamamatsu Photonics, Herrsching, Germany) and summed off-line.

Slice Preparation.

Acute neocortical slices were prepared from 13–14-day-old Wistar rats as described (16). All experiments were performed on the spines of basal dendrites located within 20–80 μm of the soma. Bath temperature was 32–34°C. The bath solution contained 125 mM NaCl, 25 mM NaHCO3, 2.5 mM KCl, 1.25 mM NaH2PO4, 1 mM MgCl2, 25 mM glucose, and 2 mM CaCl2 (Biometra, Goettingen, Germany).


Patch pipettes were filled with 115 mM K-gluconate, 20 mM KCl, 10 mM Hepes, 4 mM ATP-Mg, 10 mM phosphocreatine, 0.3 mM GTP, and a calcium indicator, either Calcium Green-1 (CG-1, 100 μM, Kd = 190 nM, Molecular Probes) or Oregon Green 488 BAPTA-2 (OG-2, 500 μM, Kd = 580 nM, Molecular Probes). Cells were loaded with the indicator >15 min before the imaging experiments began. Electrical recordings were made with a patch-clamp amplifier (EPC-7, List Electronics, Darmstadt, Germany) operated in current clamp mode. To estimate the voltage dependence of Ca2+ transients mediated by the N-methyl-d-aspartate receptor (NMDAR) subtype of glutamate receptor channels, the voltage clamp mode was also used. In voltage- clamp experiments, Cs+ replaced K+ in the pipette solution. Initial access resistances were <8 MΩ. To evoke EPSPs by local extracellular stimulation, a theta glass pipette was placed near (≤10 μm) a secondary basal dendrite and a bipolar pulse (100 μs, 6–50 V) was applied. The stimulus amplitude was kept as low as possible and always below AP initiation threshold.

Data Analysis.

Line scan images were analyzed by using in-house software. Pixels of two lines between two positions enclosing the dendrite examined were averaged to obtain one time point. The temporal resolution of the scan was 4.5 ms. Stimulation protocols began 150 ms after the start of the line scan (512 scans of a single line every 2.273 ms). The first 100 ms of the fluorescence before a stimulus were averaged to obtain the basal fluorescence, F0. A region distant to any indicator-containing structure was chosen for subtraction of background fluorescence, FB. Relative fluorescence changes were calculated as ΔF/F(t) = (F(t) − F0)/(F0FB). (ΔF/F)max was the amplitude of the fluorescence transient. This value was obtained by fitting ΔF/F(t) with a single exponential by using a least-squares-fit routine (IGOR, WaveMetrics, Lake Oswego, OR). Offsets were caused by neutral density filters in only the initial experiments, which changed their optical density with increasing exposure time. Because correction for offsets did not change the results, offsets were neglected. All values are given as mean ± SD, unless stated otherwise.

Pharmacological Compounds.

Voltage-dependent Na+ channels were blocked with 1 μM tetrodotoxin (TTX, Research Biochemicals, Natick, MA), VDCCs with 100 μM Cd2+, NMDARs with 100 μM d(−)-2-amino-5-phosphonopentanoic acid (AP-5, Tocris Neuramin, Bristol, U.K.) and l-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptors (AMPARs) with 2 μM 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide (NBQX, Tocris Neuramin).

Indicator Saturation.

As a test for indicator saturation, we loaded neurons with CG-1 (100 μM) or OG-2 (500 μM) and determined the AP-frequency–ΔF/F relationship. It was compared with the linear relationship between the AP frequency and intracellular-free Ca2+ concentration (17). With CG-1, the AP frequency–ΔF/F relationship was sublinear at AP frequencies ≥10 Hz. At 10 Hz frequency (ΔF/F)max was about 1.4, indicating that for (ΔF/F)max ≥ 1.4 the fluorescence signal deviated from a linear response because of partial saturation of CG-1. The highest (ΔF/F)max that could be evoked was about 2–2.5 (e.g., by 40 Hz trains of APs). The fact that the (ΔF/F)max evoked by the EPSP–AP sequence had approximately this value supported the view that CG-1 was partially saturated in this experimental situation. Nevertheless, we used CG-1 in some experiments because it allowed us to observe differences in fluorescence changes with a better signal-to-noise ratio than with OG-2. In addition, the rise time, measured as the time between 20–80% of (ΔF/F)max could only be determined in CG-1 loaded neurons. With OG-2, the AP frequency–ΔF/F relationship was linear up to 40 Hz. With 40 Hz stimulation (ΔF/F)max was about 1.8, suggesting that up to this value an almost linear relationship between ΔF/F and free Ca2+ concentration can be assumed.


Ca2+ Transients in Dendritic Spines Evoked by EPSPs and Back-Propagating APs.

Fig. Fig.11A shows basal dendritic branches at low and high magnification. Single spines were readily identified (Right, white arrows). We compared the amplitude and decay time course of spinous Ca2+ transients evoked in response to local subthreshold synaptic stimulation and to suprathreshold electrical stimulation of the soma by using line scan imaging.

Figure 1
Imaging of Ca2+ transients in single spines at high spatial and temporal resolution. (A) Optical detection of dendritic spines in neocortical neurons. Highlight projection through a stack of 20 frames of a layer 5 pyramidal neuron at low and high ...

Brief, transient increases of fluorescence in spine heads were observed when APs were evoked by somatic current injection (Fig. (Fig.11B). The 20–80% rise time of the fluorescence transient was ≤4.5 ms (n = 6). The peak amplitudes (ΔF/F)max were 1.01 ± 0.31 (CG-1, n = 13) and 0.62 ± 0.25 (OG-2, n = 14), and the decay time constants were 203 ± 110 ms (CG-1) and 181 ± 188 ms (OG-2). Blocking Na+ channels or VDCCs abolished the fluorescence transients (Fig. (Fig.11D). Hence, the AP-evoked Ca2+ transients were dependent on voltage-activated channels that transiently opened during APs that propagated into the basal dendritic arbor (18).

Subthreshold EPSPs evoked spinous Ca2+ transients (Fig. (Fig.11D) that showed stochastic failures (<25%). The 20–80% rise time was 9–20 ms (n = 6) and, on average, had amplitudes (ΔF/F)max of 1.15 ± 0.41 (CG-1, n = 13) and 0.69 ± 0.32 (OG-2, n = 11). The respective decay time constants were 225 ± 116 ms (CG-1) and 325 ± 141 ms (OG-2). When the AMPAR subtype of glutamate receptor channels were blocked with NBQX, the somatic EPSPs were reduced to 21% of control, whereas spinous fluorescence transients were attenuated to only 76% of control (Fig. (Fig.11D). Selectively blocking NMDAR channels abolished the fluorescence transients (Fig. (Fig.11D), whereas the size of the somatic EPSP was reduced only slightly (16). The Ca2+ transients evoked by EPSPs are thus dependent predominantly on the activation of NMDARs. This view is supported further by experiments where the soma was voltage clamped to different membrane potentials. After synaptic stimulation, a maximal Ca2+ transient was seen at around −30 mV (n = 2; not shown).

Although somatic APs and EPSP differed (about 50-fold) in duration, the decay time constants of fluorescence transients did not show large differences, suggesting that the decay of Ca2+ transients is determined predominantly by the kinetics of Ca2+ buffering and Ca2+ clearance (17, 19, 20). Consistent with this idea transients recorded at the beginning of indicator loading revealed transients with decay time constants of <100 ms, whereas later they increased to 200–300 ms. Thus, both indicators, CG-1 and OG-2, acted as an added exogenous Ca2+ buffer that increased significantly the dendritic Ca2+ binding ratio.

Ca2+ Transients in Dendritic Spine and Shaft Evoked by Single EPSPs or Back-Propagating APs.

A significant difference between fluorescence transients evoked by APs and those evoked by EPSPs was found in their spatial distribution. In most experiments, the Ca2+ transients evoked by a single EPSP were largely restricted to a single, active (that is, showed a clear rise in fluorescence following an EPSP) spine (e.g., in Fig. Fig.22A the spine marked by an arrow) and extended only weakly into the adjacent dendritic shaft (Fig. (Fig.22B). In contrast, Ca2+ transients evoked by back-propagating APs were of comparable amplitude in spines and the adjacent dendritic shaft (Fig. (Fig.22 A and C). Conversely, a short train of 3–5 EPSPs elicited at 20 Hz evoked a fluorescence transient that extended for a small distance (5–10 μm) in the dendritic shaft with reduced amplitude (69 ± 20%, n = 6; not shown), but not into the neighboring spine heads examined.

Figure 2
Calcium transients in spine and dendritic shaft. (A) EPSP-evoked calcium fluorescence transients restricted to a single spine. The dendritic shaft and other spines near the active spine (arrow, lower right) were tested for fluorescence changes. In one ...

Spinous Ca2+ Transients Evoked by Coincident EPSPs and Back-Propagating APs.

To test for possible interactions between Ca2+ transients mediated by VDCCs and NMDARs in the spine during coincident APs and EPSPs, we quantified the amplitude of the spinous Ca2+ transient when the AP preceded the EPSP by 50 ms (Fig. (Fig.33 A and B) or vice versa (Fig. (Fig.33 C and D). The time interval of 50 ms was the optimal compromise between the time dendritic NMDAR channels remained activated after a short pulse of glutamate (21) and the detectability of the two responses in the fluorescence recording. With either sequence, the initial stimulus elicited a spinous fluorescence transient, which was increased in size by the second stimulus (Fig. (Fig.33 AD, lower traces). However, in each individual experiment, the amplitude of the fluorescence transient evoked by the sequence EPSP–AP was larger than that evoked by the AP–EPSP sequence in the same spine (Fig. (Fig.33 E and F). The differences between transients evoked by the EPSP–AP and AP–EPSP sequences were larger when observed with OG-2 (143%) than with CG-1 (121%), probably because of partial saturation of the indicator CG-1 (see Materials and Methods).

Figure 3
Summation of spinous Ca2+ transients depends on relative timing of back-propagating AP and EPSP. (A) EPSP preceded by AP (AP–EPSP) examined with CG-1 (100 μM). Upper trace shows whole-cell recording in current clamp mode of a single ...

In addition, we measured the fluorescence transients in single spines evoked by short bursts consisting of two or three EPSPs (at 20 Hz) and two or three APs (at 20 Hz), but shifted by a time interval of 15 ms. Again the peak amplitude of the fluorescence transient depended on the relative timing (Fig. (Fig.33F, open squares and circles). The largest differences were seen when three EPSPs (at 20 Hz) were either preceded or followed by a single AP, separated by 15 ms (open triangles, n = 2). When NMDAR channels were blocked with AP-5, this dependence of the peak Ca2+ transient on the sequence of AP and EPSP was not observed (n = 3).

Interaction Between EPSPs and Back-Propagating APs in Evoking Ca2+ Transients.

We compared the fluorescence transients evoked by a single EPSP or a single AP (Fig. (Fig.44A, upper two traces) to transients evoked by an EPSP in the sequence AP–EPSP or an AP in the sequence EPSP–AP (Fig. (Fig.44A, lower two traces). An EPSP-evoked transient was strongly reduced by a preceding AP (Fig. (Fig.44B, AP–EPSP), and the reduction increased with the amplitude of the AP-evoked fluorescence transient (correlation coefficient r = 0.49). Also, an AP-evoked transient was reduced in amplitude by a preceding AP (Fig. (Fig.44B, AP–AP). This reduction was not caused by saturation of the indicator as confirmed by measurements on the dendritic shaft with a ratiometric dye (fura-2, 250 μM), using a CCD camera for fluorescence detection. In contrast, the fluorescence transient evoked by an AP that followed an EPSP (Fig. (Fig.44A, lowermost trace) was increased compared with the fluorescence change evoked by an AP alone (Fig. (Fig.44B, EPSP–AP). Thus, pairing AP and EPSP evoked a supralinear response when the AP followed EPSP, but a sublinear response in the reverse case. The differences were qualitatively the same with CG-1 (100 μM), but supralinear summation was not observed, probably because of partial saturation of the indicator.

Figure 4
Back-propagating APs reduce or increase Ca2+ inflow through NMDAR channels. (A) Calcium fluorescence transients evoked by single and paired APs and EPSPs. Fluorescence transients were measured with OG-2 (500 μM) as indicator. Transients ...


These experiments demonstrated a large difference in the spatial spread of Ca2+ transients evoked by back-propagating APs and by subthreshold EPSPs, in agreement with previous reports (3, 6, 8). The EPSP-evoked transient was restricted, in most experiments, to a single spine head as expected from buffering and diffusion in a spine neck (22, 23). In contrast, the AP-evoked Ca2+ signal was observed in all spines examined and was of almost equal magnitude in spine heads and dendritic shafts. The spinous Ca2+ transients were comparable in size and time course when evoked by APs or by EPSPs, although they were mediated by different ion channels, VDCCs and NMDARs, respectively. The fast rise times of the Ca2+ transients suggest that they are colocalized in the spine (8, 23). Previously, supralinear summation of Ca2+ transients evoked by coincident trains of EPSPs and APs was reported for the spines of hippocampal pyramidal neurons (8). Our results indicate, however, that pairing of AP and EPSP does not necessarily cause supralinear summation of Ca2+ transients. Rather, whether summation is sub- or supralinear depends on the relative order of EPSPs and APs.

Because the amplitude of spinous Ca2+ transients evoked by coincident back-propagating APs and subthreshold EPSPs depends on their relative timing, back-propagating APs may represent a global dendritic signal that either reduces or increases spinous Ca2+ inflow in active synaptic contacts mediated mostly by NMDARs. The mechanism underlying the reduction of synaptic Ca2+ inflow by an AP could be an inhibition of NMDAR channels by a brief, localized rise in Ca2+ concentration (24, 25). On the other hand, a brief glutamate pulse produces a long-lasting activation of dendritic NMDAR channels (21), which do not conduct current because they are blocked by extracellular Mg2+. The increased Ca2+ influx when the AP followed the EPSP could be caused by a transient removal of this Mg2+ block of NMDARs (26, 27) resulting in an amplification of Ca2+ influx through NMDAR channels (28).

Differences in summation of spinous Ca2+ signals could represent one mechanism for the detection of delays between back-propagating APs and subthreshold EPSPs. This mechanism would be sensitive to time intervals in pre- and postsynaptic activity as short as a few tens of milliseconds. The fact that AP-evoked Ca2+ transients summate sublinearly and that EPSP amplitudes in connections between layer 5 pyramidal neurons show frequency-dependent depression (29) suggest that only a repetitive, burst-like coincident activity should be effective in differentially increasing spinous Ca2+ concentrations. The “read out” of coincident activity would be encoded by different amplitudes of spinous Ca2+ transients. Whether and how this is related to the long-term changes in synaptic efficacy induced by synchronous activity of synaptically connected neurons remains to be elucidated.


We thank Drs. E. Neher (Goettingen), L. Wollmuth, M. Larkum, G. Borst, and A. Korngreen (Heidelberg) for reading the manuscript and M. Kaiser for technical assistance. We also thank Dr. K. Kaiser (Heidelberg) for making fluorescence measurements on dendritic shafts with a CCD camera; Dr. K. Schaller, C. Brugger, R. Hoche, and Dr. R. Uhl (Munich) for supporting us with optical engineering; and Dr. W. Zinth and collaborators (Munich) for helping us set up the laser system.


l-α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptor
action potential
d-(−)-2-amino-5-phosphonopentanoic acid
excitatory postsynaptic potential
voltage-dependent Ca2+ channel
N-methyl-d-aspartate receptor
amplitude of calcium fluorescence transient


Parts of this work were published in abstract form (30).


1. Magee J C, Johnston D. Science. 1997;275:209–213. [PubMed]
2. Markram H, Lübke J, Frotscher M, Sakmann B. Science. 1997;275:213–215. [PubMed]
3. Müller W, Connor J A. Nature (London) 1991;354:73–76. [PubMed]
4. Alford S, Frenguelli B G, Schofield J G, Collingridge G L. J Physiol (London) 1993;469:693–716. [PMC free article] [PubMed]
5. Murphy T H, Baraban J M, Wier W G, Blatter L A. Science. 1994;263:529–532. [PubMed]
6. Denk W, Sugimori M, Llinas R. Proc Natl Acad Sci USA. 1995;92:8279–8282. [PMC free article] [PubMed]
7. Petrozzino J J, Miller L D P, Connor J A. Neuron. 1995;14:1223–1231. [PubMed]
8. Yuste R, Denk W. Nature (London) 1995;375:682–684. [PubMed]
9. Eilers J, Augustine G J, Konnerth A. Nature (London) 1995;373:155–158. [PubMed]
10. Jaffe D B, Fisher S A, Brown T H. J Neurobiol. 1994;25:220–233. [PubMed]
11. Jaffe D B, Brown T H. J Neurophysiol. 1997;78:10–18. [PubMed]
12. Malenka R C, Kauer J A, Zucker R S, Nicoll R A. Science. 1988;242:81–84. [PubMed]
13. Bliss T V P, Collingridge G L. Nature (London) 1993;361:31–39. [PubMed]
14. Denk W, Yuste R, Svoboda K, Tank D W. Curr Opin Neurobiol. 1996;6:372–378. [PubMed]
15. Sako Y, Sekihata A, Yanagisawa Y, Yamamoto M, Shimada Y, Ozaki K, Kusumi A. J Microsc (Oxford) 1997;185:9–20. [PubMed]
16. Markram H, Lübke J, Frotscher M, Roth A, Sakmann B. J Physiol (London) 1997;500:409–440. [PMC free article] [PubMed]
17. Helmchen F, Imoto K, Sakmann B. Biophys J. 1996;70:1069–1081. [PMC free article] [PubMed]
18. Schiller J, Helmchen F, Sakmann B. J Physiol (London) 1995;487:583–600. [PMC free article] [PubMed]
19. Neher E, Augustine G J. J Physiol (London) 1992;450:273–301. [PMC free article] [PubMed]
20. Markram H, Helm P J, Sakmann B. J Physiol (London) 1995;485:1–20. [PMC free article] [PubMed]
21. Spruston N, Jonas P, Sakmann B. J Physiol (London) 1995;482:325–352. [PMC free article] [PubMed]
22. Gamble E, Koch C H. Science. 1987;236:1311–1315. [PubMed]
23. Svoboda K, Tank D W, Denk W. Science. 1996;272:716–719. [PubMed]
24. Legendre P, Rosenmund C H, Westbrook G L. J Neurosci. 1993;13:674–684. [PubMed]
25. Medina I, Filippova N, Bakhramov A, Bregestovski P. J Physiol (London) 1996;594:411–427. [PMC free article] [PubMed]
26. Nowak L, Bregestovski P, Ascher P, Herbert A, Prochiantz A. Nature (London) 1984;307:462–465. [PubMed]
27. Mayer M L, Westbrook G L, Guthrie P B. Nature (London) 1984;309:261–263. [PubMed]
28. Schiller J, Schiller Y, Clapham D E. Nat Neurosci. 1998;1:114–118. [PubMed]
29. Markram H, Tsodyks M. Nature (London) 1996;382:807–810. [PubMed]
30. Koester H J, Sakmann B. J Physiol (London) 1998;509P:55P. (abstr.).

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...