Activity-driven formation and stabilization of functional spine synapses

Physical changes in neuronal connections, dictated by the neuronal
network activity, are believed to be essential for learning and
memory. Long-term potentiation (LTP) of synaptic transmission has
emerged as a model to study activity-driven plasticity. The
majority of excitatory contacts between neurons, called synapses,
are found on spines, small dendritic protrusions. LTP is known to
trigger the formation and stabilization of new dendritic spines in
vitro. Similarly, experience-dependent plasticity in vivo is
associated with changes in the number and stability of spines.
However, to date, the contribution of excitatory synaptogenesis to
the enhanced synaptic transmission after LTP remains elusive. Do
new spines form functional synapses with the inputs stimulated
during LTP induction and thereby follow Hebbian co-activation
rules, or do they connect with random partners? Furthermore, at
which time-point are de novo spines functionally integrated into
the network? I developed an optical approach to stably and
exclusively stimulate the axons of a defined channelrhodopsin-2
(ChR2)-transduced subset of CA3 cell in mature hippocampal slice
culture over extended periods of time (up to 24h). I continuously
monitored synaptic activation and synaptic structure of CA1 cells
dendrites using two-photon imaging. To control the dendritic
location where LTP and associated spinogenesis were allowed to take
place, I globally blocked Na+-dependent action potential firing and
directly evoke neurotransmitter release by local light-evoked
depolarization of ChR2-expressing presynaptic boutons (in TTX,
4-AP). I induced optical LTP specifically at this location by
combining optogenetic activation with chemical pairing (in low
[Mg2+]o, high [Ca2+]o, forskolin, and rolipram). Taking advantage
of the NMDA-receptor mediated calcium influx during synaptic
activation I assessed the formation of functional synapses using
the genetically encoded calcium indicator GCaMP6s. I find that
optical LTP led to the generation of new spines, decreased the
stability of preexisting spines and increased the stability of new
spines. Under optical LTP conditions, a fraction of new spines
responded to optical presynaptic stimulation within hours after
formation. However, the occurrence of the first synaptic calcium
response in de novo spines varied considerably, ranging from 8.5
min to 25 h. Most new spines became responsive within 4 h (1.2 ±
0.9 h, mean ± S.D., n = 16 out of 20), whereas the remainder showed
their first response only on the second experimental day (18.2 ±
3.7 h). Importantly, new spines generated under optical LTP were
more likely to build functional synapses with light-activated,
ChR2-expressing axons than spontaneously formed spines (new
responsive spines under optical LTP: 64 ± 4 %; control 1: 0%;
control 2: 13 ± 4 %; control 3: 11 ± 4 %). Furthermore, new spines
that were responsive to optical presynaptic stimulation were less
prone to be eliminated after overnight incubation than new spines
that failed to respond (% overnight spine survival; 81 ± 3 % new
responsive spines; 58 ± 4 % of new unresponsive spines). In
summary, the results from my thesis demonstrate that synapses can
form rapidly in an input-specific manner.