Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 Feb 17;89(4):756-69.
doi: 10.1016/j.neuron.2016.01.010. Epub 2016 Feb 4.

Inhibitory Synapses Are Repeatedly Assembled and Removed at Persistent Sites In Vivo

Free PMC article

Inhibitory Synapses Are Repeatedly Assembled and Removed at Persistent Sites In Vivo

Katherine L Villa et al. Neuron. .
Free PMC article

Erratum in


Older concepts of a hard-wired adult brain have been overturned in recent years by in vivo imaging studies revealing synaptic remodeling, now thought to mediate rearrangements in microcircuit connectivity. Using three-color labeling and spectrally resolved two-photon microscopy, we monitor in parallel the daily structural dynamics (assembly or removal) of excitatory and inhibitory postsynaptic sites on the same neurons in mouse visual cortex in vivo. We find that dynamic inhibitory synapses often disappear and reappear again in the same location. The starkest contrast between excitatory and inhibitory synapse dynamics is on dually innervated spines, where inhibitory synapses frequently recur while excitatory synapses are stable. Monocular deprivation, a model of sensory input-dependent plasticity, shortens inhibitory synapse lifetimes and lengthens intervals to recurrence, resulting in a new dynamic state with reduced inhibitory synaptic presence. Reversible structural dynamics indicate a fundamentally new role for inhibitory synaptic remodeling--flexible, input-specific modulation of stable excitatory connections.


Figure 1
Figure 1. Triple color labeling of L2/3 pyramidal cells in vivo
(A) Plasmid combination for labeling cell fill (eYFP), inhibitory synapses (Teal-gephyrin), and excitatory synapses (PSD-95-mCherry). (B) Experimental time course. (C) Low-magnification maximum z-projection (MZP) of cell fill pseudo-colored red. (D) MZP of dendritic segment from (C) with labeled inhibitory synapses (cyan) and excitatory synapses (pseudo-colored green). (E and F) Examples of dynamic synapses from boxed regions in (D) on indicated days. Upper, middle, and lower panels show 3 channel merge, Teal-gephyrin alone, and PSD-95-mCherry alone, respectively. Arrows denote dynamic synapses. Inhibitory synapses in (E) appear on days 8 and 9. Excitatory synapse in (F) disappears on day 4 and its spine is removed on day 6. Scale Bars in µm: (C) 10; (D) 5; (E and F) 2.
Figure 2
Figure 2. Triple color imaging resolves three spine types with distinct properties
(A) Proportion of spines without PSD-95, singly innervated spines (SiS) containing only PSD-95, and dually innervated spines (DiS) containing both PSD-95 and gephyrin. (B) Fraction of each subclass which are dynamic or stable. The majority of spine dynamics are due to spines lacking PSD-95. (C and D) Examples of spine dynamics. Left, middle, and right panels show 3 channel merge, Teal-gephyrin alone, and PSD-95-mCherry alone, respectively. Arrows denote dynamic spines, filled when spine is present, empty when spine is absent. (C) Shows the brief appearance and removal of spines without PSD-95 at separate nearby locations. (D) Shows two spines that appear, gain PSD-95, and are stabilized. Scale bars = 2 µm. (E) Stable spines are larger than dynamic spines. Cumulative probability distribution (CPD) comparing between size of dynamic and stable spines, assessed as the ratio of YFP cell fill intensity in the spine head to the average intensity of the dendritic branch (n = 3 cells, 367 stable spines, 54 dynamic spines, p= 4.2×10−13 by KS test). (F) CPD comparing spine size between the 3 spine categories. DiS spines are larger than SiS and spines without PSD-95 (n= 3 cells with 119 DiS, 243 SiS, 59 no PSD-95. p= 6.4×10−13 by KS test).
Figure 3
Figure 3. Inhibitory synapses disappear and appear again in the same location
(A) Fraction of dynamic PSD-95 puncta on SiS and DiS, compared to dynamic gephyrin puncta on shaft or DiS (*p<0.05, **p<0.001 ***p<0.0001 by ANOVA). (B) Comparison of % dynamic structures per day with daily imaging vs 4-day imaging intervals (*p<0.02 **p<0.002 by two tailed student’s T-test) indicates that many events are short-term and go undetected with longer imaging intervals. (C and D) Daily imaging sessions showing examples of recurrent dynamic gephyrin puncta on DiS. Top, second, and third panels show 3 channel merge, PSD-95-mCherry alone, and Teal-gephyrin alone, respectively. Arrows on images denote dynamic synapses, filled when synapse is present, empty when synapse is absent. Scheme below summarizes when inhibitory synapses are present (filled circles) or absent (empty circles). Note that the PSD-95 puncta are stable over all sessions on these spines. Scale bars = 2 µm. (E) Schematic illustrating the classification of one-time dynamic, transient, and recurrent synapses. (F) Percentage of one time dynamic, transient and recurrent synapses (***p<0.0001 by ANOVA). All analyses in this figure were for 1555 spines and 955 inhibitory synapses from 63 dendrites, with statistics based on n= 8 cells All error bars represent SEM. Tukey’s Multiple Comparison Test was used for all ANOVA comparisons.
Figure 4
Figure 4. Presence or absence of Teal-gephyrin puncta on dually innervated spines reflects presence or absence of functional GABAergic synapses
(A) Representative image of a dendritic segment from an L2/3 pyramidal neuron in organotypic slice culture expressing tdTomato (red) and Teal-gephyrin (green). Arrowhead marks a gephyrin positive spine. Crosses indicate two-photon GABA uncaging spots. Scale bar= 1µm. (B) Representative uIPSC traces evoked from a spine positive for gephyrin (1, green) and its neighboring gephyrin-negative spine (2, red). Arrowheads indicate onset of GABA uncaging. (C) Summary graph of uIPSC amplitudes from gephyrin positive spines (green bar; 42 spines, n = 18 cells) and spines lacking gephyrin puncta (red bar; 44 spines, n = 18 cells. **p < 0.01 by two-tailed student’s t-test). (D) Segregation of uIPSC amplitudes between spines with gephyrin puncta (green circles) and lacking puncta (red circles). (E) Gephyrin negative spines (red) clearly segregate from gephyrin positive spines (green). Positive correlation between uIPSCs and Teal-gephyrin signal intensity (r = 0.53 by Pearson’s correlation; p < 0.01) and no correlation between uIPSC and background gephyrin intensity of gephyrin negative spines (r=−0.09 by Pearson’s correlation; p=0.56; n = 42 spines, 18 cells). Error bars are S.E.M.
Figure 5
Figure 5. Kinetics of inhibitory synapse dynamics are altered during experience dependent plasticity
(A) Experimental protocol, MD was performed immediately following the first imaging session. (B) Spine dynamics are unaffected by MD as compared to NE (p>.05). (C) MD increases the % of dynamic inhibitory shaft, but not inhibitory spine synapses. (D) Dynamic inhibitory shaft synapses are more recurrent after MD. (E) Dynamic inhibitory synapses on DiS are more recurrent after MD. (F) MD causes an increase in the number of dynamic events per location for both inhibitory shaft and inhibitory synapses on DiS. (G) MD decreases the number of consecutive days present for both dynamic inhibitory shaft and inhibitory spine synapses. (H) MD increases the number of days between synapse reappearance for inhibitory synapses on DiS. (B–H) Error bars = SEM. *p<0.05, **p<0.01, ***p<0.0005. All p-values from 2 tailed student’s T-test based on n=7 cells NE (304 inh on DiS, 534 inh shaft) and n=7 cells MD (556 inh on DiS, 717 inh on shaft). (I) Survival fraction of observed synaptic events (points), fit with the exponential and constant term, SF = fe–t/τ + s (lines), allowing visualization of how MD affects the survival of each population. (J) Schematic illustrating how inhibitory synaptic dynamics are altered by MD.
Figure 6
Figure 6. Different logic for excitatory vs. inhibitory synaptic changes
(A–C) Schematics illustrating the most prevalent categories of dynamic events for spines without PSD-95, spines with PSD-95, and inhibitory synapses on DiS and on the shaft. (A) Dynamics of spines without PSD-95 are rapid and sample different locations, to test different potential partners. (B) Spines that lose PSD-95 are destabilized, while those that gain a PSD-95 are stabilized and persist. In both cases they represent a local rewiring of excitatory circuits. (C) Inhibitory synapses on the shaft or on DiS are removed and reassembled at stable locations, providing a mechanism for reversible inhibitory modulation of excitatory circuits. (D and E) Breakdown of spines carrying an excitatory synapse (D) and of all inhibitory synapses (E) broken down by dynamic category illustrate that recurrence is a feature of inhibitory synapses. (D) Of 71 dynamic spines with PSD-95 from the NE dataset, 64 (87%±6%) were dynamic once, 4 (5%±3%) were transiently present, and 3 (7.9%±5.6%) returned to the same location after the spine was eliminated. (E) Of 169 total inhibitory synapses from the NE dataset,110 (44%±6%) were one time dynamic, 76 (32%±5%) were transiently present, and 72 (24.2%±7.1%) were recurrent.

Similar articles

See all similar articles

Cited by 45 articles

See all "Cited by" articles

Publication types

MeSH terms

LinkOut - more resources