Material below summarizes the article Adenosine Shifts Plasticity Regimes between Associative and Homeostatic by Modulating Heterosynaptic Changes, published on February 8, 2017, in JNeurosci and authored by Nicholas M. Bannon, Marina Chistiakova, Jen-Yung Chen, Maxim Bazhenov, and Maxim Volgushev.
Plasticity is a universal property of synapses, vital for fundamental operations of the nervous system. The Hebbian rule for associative plasticity postulates that inputs that consistently take part in firing of a postsynaptic neuron — that is, inputs that produce postsynaptic potentials that closely precede spikes and thus contribute to their generation — should be potentiated. By extension, inputs whose activity does not help drive firing should be depressed.
These Hebbian-type learning rules explain a multitude of plastic phenomena in the nervous system, ranging from refinement of connectivity during development (“neurons that fire together wire together”) to extraction of causal relations between events in the environment in Pavlovian conditioning and other types of associative learning.
However, Hebbian-type learning rules introduce positive feedback on synaptic strength (or “weight”) changes. Potentiated synapses have a higher probability to induce spikes and thus to be further potentiated, while depressed synapses are less likely to induce spikes and thus tend to be further depressed. This positive feedback causes runaway synaptic dynamics leading to the saturation of synaptic strengths at extreme values.
In real neurons, however, synaptic weights are not saturated but distributed continuously over an operational range. This implies the existence of additional mechanism(s) preventing runaway dynamics. The need for such mechanisms has been appreciated since early theoretical studies.
A simple and robust method commonly used in modeling studies to constrain runaway dynamics of synaptic weights and activity is normalization: Following induction of plasticity at activated synapses (homosynaptic plasticity), all synaptic weights on the cell are readjusted so that their sum remains constant. Normalization postulates heterosynaptic plasticity: changes at synapses that were not active during the induction. The biological mechanisms that prevent runaway dynamics in neurons have been elusive.
In prior work on layer 2/3 pyramidal neurons in slices from rat neocortex, we described a novel form of plasticity — weight-dependent heterosynaptic plasticity — that can serve this homeostatic function (see recent reviews). Heterosynaptic plasticity occurs alongside associative plasticity induced by spike-timing-dependent plasticity (STDP) protocols— repetitive pairing of presynaptic activation with bursts of postsynaptic spikes, or can be induced by a purely postsynaptic protocol of intracellular tetanization: bursts of postsynaptic spikes without presynaptic stimulation.
The direction of heterosynaptic changes is correlated with the initial paired-pulse ratio (PPR), a measure inversely related to the release probability at the presynaptic terminal. Weak inputs (high PPR) get stronger and strong inputs (low PPR) tend to get weaker. Model simulations show that this experimentally observed form of weight-dependent heterosynaptic plasticity robustly prevents runaway dynamics of synaptic weights for a broad range of STDP rules and activity patterns. Moreover, it equips model neuronal networks with essential computational features: enhancement of synaptic competition, facilitation of development of specific connectivity, and the ability for relearning.
As a metabolite of ATP, adenosine might be the most widespread neuromodulator in the brain. Adenosine is involved in regulation of the sleep-wake cycle, as one factor mediating increases in sleep pressure. Adenosine levels rise with activity and throughout waking and decrease during subsequent sleep. At the same time, adenosine modulates release probability (and PPR) in excitatory and inhibitory connections to layer 2/3 pyramids in rat visual cortex. Because PPR is one of the factors determining the outcome of heterosynaptic plasticity, adenosine may provide a link between changes of brain state and modulation of plasticity.
We induced homosynaptic and heterosynaptic plasticity in layer 2/3 pyramids on the background of differing degrees of adenosine receptor activation. For plasticity induction, we used an STDP protocol or intracellular tetanization. In STDP experiments, the overall outcome of homosynaptic plasticity at paired inputs was potentiation, although depression and no change also occurred. At unpaired inputs in STDP experiments, or following intracellular tetanization, the net outcome was no change, though individual synapses could express heterosynaptic potentiation or depression.
In control conditions (no manipulations to adenosine receptor activation), both homosynaptic and heterosynaptic plasticity were weight-dependent, as the magnitude of synaptic changes correlated with the initial PPR. Inputs with a low PPR (and thus a higher release probability) tended to undergo depression, while inputs with a high PPR (lower release probability) tended to undergo potentiation.
Adenosine strengthened the weight-dependence of plasticity and blockade of adenosine A1 receptors (A1Rs) weakened it. These effects were especially strong for heterosynaptic plasticity. Under adenosine, the correlation between initial PPR and heterosynaptic changes was higher and thus the initial state of the synapse was more predictive of plastic outcome (R2 = 0.36 vs. 0.16 in control). Surprisingly, blockade of A1Rs completely abolished weight dependence of heterosynaptic plasticity. Although cases of potentiation and depression still occurred, they were unrelated to the initial state of the synapse.
In model neurons, these experimentally-observed changes in heterosynaptic plasticity fundamentally shifted the plasticity regime. Heterosynaptic plasticity lacking weight-dependence, as in experiments with A1R blockade, cannot constrain runaway dynamics imposed on synaptic weight changes by Hebbian-type rules. In this regime associative plasticity dominates and learning can lead to profound segregation of synaptic weights, driving them toward extreme values. With increasing adenosine tone, as in control experiments and with added adenosine, heterosynaptic changes become progressively more weight-dependent and their homeostatic effect on synaptic weights become stronger. In this regime, the homeostatic effect of heterosynaptic plasticity dominates, synaptic changes are tightly constrained, and activity drives synaptic weights away from extreme values.
Adenosine modulation of plasticity regimes strongly supports the hypothesis proposed by Tononi and Cirelli about state-dependent synaptic changes. In the awake state (low adenosine tone), associative changes dominate, unbalancing synaptic weights. With increasing sleep pressure and subsequent sleep (high adenosine tone), homeostatic action of weight-dependent plasticity dominates, restoring synaptic balances.
We propose that adenosine modulation may fine-tune the mode of operation of neurons and translate changes of brain state into shifts in regimes of synaptic plasticity and learning.
Adenosine Shifts Plasticity Regimes between Associative and Homeostatic by Modulating Heterosynaptic Changes. Nicholas M. Bannon, Marina Chistiakova, Jen-Yung Chen, Maxim Bazhenov, Maxim Volgushev. JNeurosci Feb 2017 DOI: https://doi.org/10.1523/JNEUROSCI.2984-16.2016