Metaplasticity Flashcards
Problems with the quantal model
Central synapses don’t show the binomial distribution, with quantal peaks ‘smeared out’. Could be because inputs are summed over so many dendrites, and subject to dendritic filtering, that it obscures any binomial distribution that might be there
Non-vesicular release (see another card)
Subquantal events - maybe because Katz was recording too far from the muscle sometimes. Or maybe due to kiss-and-run?
Binomial and Poisson statistics make assumptions (p = at all sites, p is low)
Is n vesicles? Some people said release sites to make it fit the model, but is this a circular argument?
Variation in ‘unit’ amplitude
Lack of consistent correlation between vesicles and release
Vesicles seen at purely electrical synapses
Pre- vs post-synaptic plasticity
If you block all transmission using TTX, you still get mEPSPs, i.e. quantal events. PKC antagonists increase frequency of these, but not amplitude - so must be a presynaptic change. Forskolin increases amplitude, but not frequency - postsynaptic change.
Silent synapses that start expressing post-synaptic receptors will look like a presynaptic change, when in fact it’s just a change in q.
There are plastic changes that affect vesicle size, which will look like postsynaptic change because amplitude increases without frequency.
Non-vesicular release theories
Denker et al - Using styryl dyes at the frog NMJ after PKC antagonist addition showed neurotransmitter release but not destaining, and after zinc showed destaining but not NT release. Suggested that the /reserve/ pool of vesicles acts as buffers for soluble accessory proteins involved in the recycling of vesicles into the readily releasable pool, to stop these proteins from being ‘lost’ into the axon. Calcium influx prevents uptake and causes release of these proteins.
Tauc 1987 - injecting AChE into cytoplasm reduced response, which it shouldn’t if ACh is in vesicles. Later, AMECh was released alongside ACh way quicker than if it had had to be taken up into vesicles. Vesicles in the torpedo electric organ are way too big to be quanta. So he suggested the vesigate theory.
We know that in pathological scenarios there is nonvesicular release (e.g. anoxia causes reversal of transmitter uptake proteins)
Arguments for kiss and run (6)
In the same year as Heuser and Reese showed vesicles at the NMJ using EM, Ceccarelli et al found no evidence of vesicles depletion, only accumulation of new labelled vesicles. They suggested a model similar to Katz’s original, termed ‘kiss and run’. BUT they used different stimulation to H&R, 2Hz vs 10Hz
It requires very strong stimulation to see exo-endocytic intermediates under EM, which may not be physiological.
At hippocampal synapses there are so few vesicles that you never see mid-fusion intermediates, and if they were relying on full recycling this would limit transmission rate.
Hydrophilic styryl dyes should be released preferentially in kiss and run, whereas both hydrophilic and phobic would be released in full exocytosis. One group found preferential release of hydrophilic dyes at hippocampal synapse.
Variation in stoichiometry of NT release
In hippocampal cells, stimulation resulted in a slow partial destaining, suggesting transient fusion.
Cell-attached capacitance recordings showed ‘flickers’ of increased capacitance, between a few msec and 2 seconds, too fast for full exo and endocytosis. This accounted for 20% of vesicles
Arguments against kiss-and-run (8)
The hydrophilic/hydrophobic dye experiment was repeated at physiological temperature, and no difference was found. Also couldn’t be replicated at frog NMJ or goldfish bipolar cell.
Cell capacitance flickers could have been one vesicles exocytosing and another endocytosing
Slow partial destaining seen in hippocampal cells could have been a full release of just a few strongly stained vesicles, rather than partial release of many weakly stained (because it was shown that weakly stained vesicles release all their dye very quickly, and the partial destaining seen was slow).
Even antibody sandwiches can be taken up into vesicles, which would be hard through a fusion pore
Capacitance recordings (measure surface area of terminal) alongside interference reflection microscopy (measuring footprint of terminal) correlate, suggesting all vesicles contribute to footprint.
Very rapid bulk endocytosis could explain why intermediates are rarely seen
Images of recently fused vesicles under cryo-EM have shown pores as big as 20nm in vesicles of diameter 50nm, and widening over time, suggesting full fusion could follow.
Observing vesicles being endocytosed but without a clathrin coat is not evidence for kiss-and-run, especially as imaging clathrin is v difficult.
LTP and LTD mechanism
High frequency stim opens NMDA for a long time, lets lots of Ca in, activates kinases, causes AMPA insertion (from vesicles) into membrane, and can have long term transcriptional effects.
Low frequency stimulation causes more transient Ca currents, which activate phosphatases (note that some [Linden, 2012] people think it’s about PKC phosphorylating AMPA receptors in the membrane, causing them to unbind GRIP and instead bind PICK1, initiating endocytosis).
These are induction mechanisms - maintenance is unknown
(though some think LTD requires synthesis of Arc, which binds dynamin, to allow further endocytosis. Also there’s a positive feedback loop - PKC disinhibits Raf3, which activates MEK, which activates MAPK, which activates PLA2, which produces AA, which activates PKC.)
Note that LTP at mossy fibre-CA1 synapses appears to be presynaptically mediated!! Not well understood, but known to be NMDAR-independent and perhaps involving L-type Ca channels
Abraham et al 2002 - LTP can last up to a year
Those three examples of metaplastic threshold-shift
5Hz stimulation blocks LTP. Effect (60-90 mins) abolished by AP-5
10Hz stimulation facilitates LTP.
600 x 1Hz stimulation facilitates LTD. Effect abolished by AP-5.
Note my point: there must be two thresholds, since not every stimulation causes long term plasticity
Mechanisms of metaplasticity - retrograde cannabinoid release
Dendrites of glutamatergic synapses release various substances including cannabinoids
Cannabinoids are inhibitory
Cannabinoids bind GABAergic interneurons, cause LTD of this inhibitory synapse onto the glutamatergic neuron
Hence disinhibiting the glutamatergic presynaptic neuron, allowing more LTP.
SO perhaps the priming stimulus is making the postsynaptic dendrites release cannabinoids, and facilitating LTP
BUT my point: this is not strictly speaking metaplasticity. There’s no shifting threshold of LTP, just altering stimulus.
Mechanisms of metaplasticity - astrocytes
Outnumber neurons 10:1, and sit in a reticulum
Kuga et al 2011 - Produce calcium waves that propagate between them in ‘glissandi’, dependent on neural activity
Serrano 2006 - stimulating Schaffer Collateral cells activated GABA-ergic interneurons, which stimulated GABA<b> receptors on astrocytes, caused Ca increase, ATP release, and suppression of SC-CA1 transmission.
Han et al 2012 - grafting human glial progenitors into a neonatal mouse led to increased LTP when it grew up</b>
Hulme et al 2014 - blocking adenosine receptors abolished the blocking effect of stimulation on LTD. Blocking gap junctions with carbenoxolone did the same, and abolished calcium waves seen in the strata radiata after stimulation to strata oriens
Florian et al 2011 - sleep deprivation causes impaired late-phase LTP in rat hippocampus. Effect abolished by adenosine receptors blocker.
Perhaps priming stimulation to the neurons actually affects signaling in the glial network. May work to allow postsynaptic induction to cause presynaptic maintenance, or balance out the network and allow widescale communication of threshold. Specificity is an issue… but maybe metaplasticity /has/ to work on a network-wide level, or else raising threshold at one place wouldn’t matter, because LTP would occur upstream.</b>
Functional examples of metaplasticity
The hippocampus of rats that have ‘learned helplessness’ from uncontrolled footshock cannot undergo LTP, and undergoes LTD more easily.
Cho et al 2009 - monocularly deprived mice have fast LTD in deprived V1, and slow LTP in non-deprived. NR2A subunit KO abolishes the fast LTD component. SO perhaps a reduced ratio of NR2A/NR2B is permissive for LTP
Sidorov et al 2015 - in excitatory V1 synapses of monocularly deprived mice, NMDAR-dependent LTD is reduced in vitro and ocular dominance plasticity is reduced in vivo by mGluR5 antagonism
Florian et al 2011 - sleep deprivation causes impaired late-phase LTP in rat hippocampus. Effect abolished by adenosine receptors blocker.
In kittens raised in darkness, ocular dominance columns do not develop, and LTP is much easier to evoke in cortical slices. This change may be due to subunit expression shift in NMDA receptors. NR1 is obligatory, but NR2B (longer open time) shifts to NR2A in normal kittens. NR2A may function to ‘fix’ plasticity at a certain level. NR2B is more plasticity-amenable.
Proctolin and RPCH in STG
Proctolin only - no effect
RPCH only - depolarisations but no spiking
Both at once - strong burst of spiking
RPCH, washout, proctolin - depolarisations but no spiking
Nusbaum and Marder, 1988 - RPCH reversibly strengthens the reciprocal inhibitory LP-PD synapse, which alters CPG output.
Dickinson et al 1990 - RPCH also strengthens synapses from IVN in cardiac sac onto nerves in the gastric mill circuit, coupling cardiac sac and gastric mill, and causing rhythmogenesis. CF Sharp et al 1996 - dynamic clamp, where there was no regular rhythm until the two spontaneously firing cells were linked by an artificial inhibitory synapse.
MM and SCP in STG
Brezina et al - Two motorneurons onto the tongue in the Aplysia feeding circuit. Small cardiogenic peptide increases amplitude and narrow width of action potential. Contraction is harder but briefer. Myomodulator increases amplitude at first, but decreases it a lot at higher concentrations
Both increase cAMP to open Ca channels [increase amplitude and rate of relaxation], and activate K channels as a shunt [decrease amplitude]. SCP is better at the former, MM at the latter.
The combination of the two increases the space of effect, and allows more flexibility, e.g. increasing contraction amplitude without changing rate of relaxation.
How do networks remain stable in the face of such varied modulation?
1) Some modulators have voltage-dependent effects, e.g. Proctolin causes an inward current similar to NMDA-dependent, which is blocked by external Ca and has a reversal potential of 0mV. It can increase frequency and amplitude of firing, but will not depolarise baseline - unlike nicotine.
2) Modulators acting on the same neuron or target channel will occlude each other, so there’s a ceiling
3) Modulation of different network components will match up and balance out effects, e.g. Brezina suggested that the muscle, NMJ and CPG are all modulated as one.
4) Modulators can have opposite effects so are self-limiting - e.g. dopamine alters both inward and outward currents, to keep neurons within their firing range
5) Thirumalai et al 2006 - Post-synaptic neuron saturates - RPCH can strengthen the LP-PD synapse several-fold, but not necessarily cause a change in pyloric rhythm
6) Modulators can have state-dependent effects, or depend upon co-modulators. Yet many networks with different underlying parameters will reliably respond to the same neuromodulator.
Substance P in lamprey locomotor network
Parker and Grillner 1998 - Substance P increased frequency of spiking in excitatory interneurons, via a calcium-independent, presynaptic mechanism. Could have been increase in amplitude that allowed detection of spikes previously indistinguishable from noise (because these were v small interneurons, hard to record from), but unlikely as amplitude change was independent from frequency change, as it could be blocked by cadmium without affecting frequency. Amplitude was increased via a post-synaptic, calcium-dependent, NMDAR-dependent mechanism. The presynaptic effect was short term, the postsynaptic effect lasts over 24hrs.
Substance P, 5-HT and dopamine in lamprey locomotor network
Parker and Bevan 2007 - Substance P decreased variability in depolarisations in excitatory locomotor-related interneurons, and in spiking in motorneurons during network activity. Did not increase variability when depolarising pulses were applied to motorneurons, so definitely affecting the interneurons not the motoneurons.
Increased variability in depolarisations in small intrinsic inhibitory interneurons, perhaps to maintain network flexibility.
Applying 5-HT blocked pre- and post-synaptic effects of SP
Applying DA did nothing
Applying 5-HT and DA blocked only postsynaptic effect, so DA is a gating agent.
