The strength of synaptic connections between neurons needs to be variable, but not too much so. Evidence now indicates that regulation of such synaptic plasticity involves a complex cascade of feedback loops.
Learning is thought to manifest in the brain as physical changes that alter the strength of neuronal contact points called synapses. These contact points allow information to be transmitted from one neuron to another, and understanding the conditions that cause synapses to change strength (a phenomenon known as synaptic plasticity) has been a focus of neuroscience research for many years. Writing in Nature Communications, Tigaret et al.1 challenge the prevailing idea that the local concentration of calcium ions (Ca2+) is the key factor that determines whether a synapse becomes stronger or weaker after repetitive activation. They propose that plasticity involves an intracellular signalling cascade that overrides a safety mechanism. This suggests that the default state of the synapse is not to be plastic.
The main excitatory neurotransmitter in the mammalian brain is the molecule glutamate. Glutamate is released from the presynaptic neuron, and the postsynaptic neuron is excited when the molecule binds to and activates specialized receptor proteins, most of which are ion channels called ionotropic glutamate receptors. When activated, these channels open and positively charged ions enter the cell, depolarizing (reducing the voltage across) the cell membrane. In addition, glutamate receptors that are not ion channels, called metabotropic glutamate receptors, activate various intracellular signalling cascades. Their effect on synaptic transmission is generally slower than that of ionotropic receptors, but they are crucial for healthy brain function2.
In the neuronal structure known as the dendritic spine, which forms a single synaptic contact, the initial depolarization caused by activation of ionotropic glutamate receptors can be amplified by the opening of voltage-gated calcium channels, further depolarizing the spine. A special class of ionotropic glutamate receptors called NMDA receptors have a similar role — they open only when the neuron is already depolarized, forming a positive-feedback loop that increases Ca2+ influx and depolarization3. The activation of NMDA receptors is essential for many forms of long-lasting synaptic plasticity.
However, positive-feedback loops are inherently dangerous for neurons — too much depolarization and Ca2+ can be toxic, eventually triggering cell death. To prevent this from happening, spines have a safety mechanism in the form of calcium-activated, potassium-conducting SK channels4. When intracellular Ca2+ reaches a critical concentration, SK channels open, allowing positively charged potassium ions to exit the cell and so preventing further depolarization (Fig. 1a). This SK mechanism stops the positive-feedback loop, blocking further Ca2+ influx. But a side effect is that the synaptic strength becomes difficult to change5.
Tigaret et al. describe a plasticity-enabling mechanism that inhibits SK channels in individual spines. They found that repeated sequential activation of pre- and postsynaptic neurons in slices of rat brains induced synaptic strengthening, also known as long-term potentiation (LTP). In addition to NMDA-receptor activation, LTP induction required the activity of group 1 metabotropic glutamate receptors (mGlu1). The authors show that activation of mGlu1 triggers a slow-acting mechanism that inhibits SK channels, allowing for sustained depolarization and enhanced Ca2+ entry into the spine (Fig. 1b).
The authors used a strong induction protocol (300 paired activations in 1 minute) to allow the relatively slow metabotropic process to take effect and enable LTP. This might seem unusual — after all, we don't need to be presented with information 300 times before learning a new association. Why was such a strong protocol required?
In an intact brain, specific neuromodulator chemicals such as dopamine and acetylcholine are released when the animal is in an aroused state: for example, when it learns that a certain sound predicts a frightening event. These substances modulate glutamate-activated synapses and have been shown to promote synaptic plasticity by blocking SK channels5,6. The timing window for successful induction of LTP has been shown7 to change radically in the presence of neuromodulators. Thus, it seems that there is not just one rule for how synapses change during learning, but a whole set that are tailored to various occasions such as different mental states. This makes sense from a systems perspective — synaptic potentiation is gated not only by timing, but also by the brain's reward system. From an experimental point of view, the lack of neuromodulatory inputs, which is an inherent limitation of brain-slice experiments, might explain why a strong protocol was required. Functional imaging of single synapses in live, active animals is not yet possible.
The mechanism highlighted by Tigaret and colleagues is not the only way in which synapses can be strengthened. Enzymes called Src tyrosine kinases (which add phosphate groups to proteins) can directly enhance NMDA-receptor function8. This pathway has been shown to cause LTP in the same type of synapse as that analysed in the current study9. The activity of various metabotropic receptors, including mGlu1, can increase glutamate-mediated responses through this pathway10. It will be interesting to investigate whether the mGlu1-triggered blockade of SK channels identified by Tigaret et al. acts together with direct NMDA-receptor phosphorylation to enable LTP, or whether one mechanism is dominant under specific conditions, depending, for instance, on cell type or the age of the animal.
This study also confirms11 that, contrary to general thinking, it is not possible to predict the direction and magnitude of synaptic plasticity by simply analysing levels of Ca2+ in dendritic spines. For example, a pair of presynaptic stimulations triggered a very strong Ca2+ influx into the spine, but no plasticity whatsoever. But before Ca2+ is discarded as the key state variable, we must consider that successful induction of long-term plasticity relies on the interplay of local synaptic Ca2+ signals with Ca2+ signals in the cell body (soma) of the neuron. Indeed, the authors emphasize the importance of postsynaptic electrical activity and the activation of voltage-gated calcium channels for LTP. These processes are not restricted to the active spine; they increase Ca2+ levels throughout the neuron. Thus, it might be possible to predict the future strength of a synapse from simultaneous Ca2+ measurements in the spine and soma. This is certainly not an easy experiment, but sophisticated 3D scanning microscopes could be used to analyse compartmentalized Ca2+ signalling in individual neurons — and perhaps one day in intact animals during learning.Footnote 1
Tigaret, C. M., Olivio, V., Sadowski, J. H. L. P., Ashby, M. C. & Mellor, J. R. Nature Commun. http://dx.doi.org/10.1038/ncomms10289 (2016).
Wang, H. & Zhuo, M. Front. Pharmacol. 3, 189 (2012).
Bloodgood, B. L. & Sabatini, B. L. Curr. Opin. Neurobiol. 17, 345–351 (2007).
Ngo-Anh, T. J., Bloodgood, B. L., Lin, M., Sabatini, B. L., Maylie, J. & Adelman, J. P. Nature Neurosci. 8, 642–649 (2005).
Buchanan, K. A., Petrovic, M. M., Chamberlain, S. E. L., Marrion, N. V. & Mellor, J. R. Neuron 68, 948–963 (2010).
Giessel, A. J. & Sabatini, B. L. Neuron 68, 936–947 (2010).
Brzosko, Z., Schultz, W. & Paulsen, O. eLife 4, e09685 (2015).
Yu, X. M., Askalan, R., Keil, G. J. & Salter, M. W. Science 275, 674–678 (1997).
Lu, Y. M., Roder, J. C., Davidow, J. & Salter, M. W. Science 279, 1363–1367 (1998).
Benquet, P., Gee, C. E. & Gerber, U. J. Neurosci. 22, 9679–9686 (2002).
Nevian, T. & Sakmann, B. J. Neurosci. 26, 11001–11013 (2006).
Related links in Nature Research
About this article
Philosophical Transactions of the Royal Society B: Biological Sciences (2017)
Frontiers in Psychology (2017)
Interface Focus (2017)
Frontiers in Psychology (2016)
Slow-Frequency Pulsed Transcranial Electrical Stimulation for Modulation of Cortical Plasticity Based on Reciprocity Targeting with Precision Electrical Head Modeling
Frontiers in Human Neuroscience (2016)