Synaptic and circuit mechanisms prevent detrimentally precise correlation in the developing mammalian visual system
Curation statements for this article:-
Curated by eLife
eLife assessment
The authors use computational modeling of the mammalian visual system to address an important and understudied problem: how precise temporal properties of synaptic transmission might impact the kinds of neuronal correlations that instruct development. The present description of the simulations provides mixed evidence for the authors' conclusions. That slow NMDA currents help to minimize rapid timescale correlations is compelling, but other aspects of the simulations, such as neuronal heterogeneity may also contribute.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Neuroscience (eLife)
Abstract
The developing visual thalamus and cortex extract positional information encoded in the correlated activity of retinal ganglion cells by synaptic plasticity, allowing for the refinement of connectivity. Here, we use a biophysical model of the visual thalamus during the initial visual circuit refinement period to explore the role of synaptic and circuit properties in the regulation of such neural correlations. We find that the NMDA receptor dominance, combined with weak recurrent excitation and inhibition characteristic of this age, prevents the emergence of spike-correlations between thalamocortical neurons on the millisecond timescale. Such precise correlations, which would emerge due to the broad, unrefined connections from the retina to the thalamus, reduce the spatial information contained by thalamic spikes, and therefore we term them ‘parasitic’ correlations. Our results suggest that developing synapses and circuits evolved mechanisms to compensate for such detrimental parasitic correlations arising from the unrefined and immature circuit.
Article activity feed
-
-
eLife assessment
The authors use computational modeling of the mammalian visual system to address an important and understudied problem: how precise temporal properties of synaptic transmission might impact the kinds of neuronal correlations that instruct development. The present description of the simulations provides mixed evidence for the authors' conclusions. That slow NMDA currents help to minimize rapid timescale correlations is compelling, but other aspects of the simulations, such as neuronal heterogeneity may also contribute.
-
Reviewer #1 (Public Review):
The authors address an important and understudied problem: how precise temporal properties of synaptic transmission might impact the kinds of neuronal correlations that instruct development. The methods used to characterize and simulate retino-thalamo-cortical development are carefully carried out and yield convincing results. Based on these simulations, the authors argue that features such as slow NMDA receptor-mediated currents are able to prevent aberrant development which might otherwise result from rapid timescale correlations that lack meaningful information about visual topography.
-
Reviewer #2 (Public Review):
The paper starts with the premise that given the broad immature connectivity between the retina and thalamus during development, locally homogeneous synaptic currents should generate precise spike correlations (on a millisecond timescale) which are not seen in development and could be bad for developmental refinements and "network diversity". Rather, the correlations during development are over much longer timescales. The authors propose that two main factors, the dominance of NMDA (over AMPA) currents and the absence of recurrent connections prevents these precise correlations and preserves diversity.
The paper consists of three parts: (I) develop a biophysical model for a thalamic neuron, (II) use the model to determine which factors govern precise correlations, and then (III) simulate a cortical network …
Reviewer #2 (Public Review):
The paper starts with the premise that given the broad immature connectivity between the retina and thalamus during development, locally homogeneous synaptic currents should generate precise spike correlations (on a millisecond timescale) which are not seen in development and could be bad for developmental refinements and "network diversity". Rather, the correlations during development are over much longer timescales. The authors propose that two main factors, the dominance of NMDA (over AMPA) currents and the absence of recurrent connections prevents these precise correlations and preserves diversity.
The paper consists of three parts: (I) develop a biophysical model for a thalamic neuron, (II) use the model to determine which factors govern precise correlations, and then (III) simulate a cortical network and demonstrate loss of network diversity when precise correlations are used. While all parts are interesting, there are several claims in each (and the links between them) that are not fully justified.
What is commending about the paper is that it is one of few theoretical/modeling papers that focuses on neural circuit development and it manages to link experimental results to principles of circuit function. The authors apply quite a few modeling approaches ranging from single-neuron models (including building a database of thalamic neurons based on experimental data) and network models. Some of the claims regarding the timescales of correlations (long in development) are unjustified because the authors use a fixed-timescales kernel to compute these correlations and mainly investigate their amplitude or level, not their timescales. It is also not clear how important is the heterogeneity among thalamic neurons. Are the effects on the correlations the result of NMDA currents or the neuronal diversity from their database? Are precise correlations generated because of the diverse/heterogeneous neurons, or because of the levels of convergence? What happens to homogeneous neurons? The authors also propose that precise correlations impair network diversity but never show this impairment directly beyond a diversity of excitatory-to-excitatory connection strengths. If the authors were to clarify these issues then the paper could be a valuable contribution to the field of developmental systems and computational neuroscience.
-
Reviewer #3 (Public Review):
This paper addresses an apparent contradiction related to the interaction between spontaneous neural activity and neural plasticity during circuit development. It is well established that spontaneous activity contains instructive information for developmental downstream circuit refinement, for instance for the formation of retinotopic projections in the visual system. These developmental cues are contained in activity correlations, and thus can be picked up by Hebbian mechanisms. However, previous work has shown that informative correlations in spontaneous activity can be found on slow time scales (100s of milliseconds). Here it is shown that in this case correlations on fast time scales arise simultaneously from the integration of unrefined inputs and that these likely interfere with developmental …
Reviewer #3 (Public Review):
This paper addresses an apparent contradiction related to the interaction between spontaneous neural activity and neural plasticity during circuit development. It is well established that spontaneous activity contains instructive information for developmental downstream circuit refinement, for instance for the formation of retinotopic projections in the visual system. These developmental cues are contained in activity correlations, and thus can be picked up by Hebbian mechanisms. However, previous work has shown that informative correlations in spontaneous activity can be found on slow time scales (100s of milliseconds). Here it is shown that in this case correlations on fast time scales arise simultaneously from the integration of unrefined inputs and that these likely interfere with developmental plasticity. The paper shows that such fast, "parasitic" correlations can appear during retinal wave activity, and provides evidence that NMDA receptors can suppress them to avoid their influence during developmental plasticity.
This work is based on detailed biophysical models of thalamic relay neurons, which were fit to data from in-vitro whole-cell recordings. A genetic algorithm was used to fit these models, which provides a family of suitable models instead of a point estimate of good parameters. This approach is a real strength as it shows that the effects generalise well to many plausible models, and do not depend on specific parameter choices. The model neurons are then placed in simulated networks (without or with anatomically informed recurrent connections) and driven by retinal wave activity recorded from the mouse. Together the simulation and analysis show under which conditions fast, undesirable correlations do and do not appear. Specifically, the key model ingredient this work identifies for the suppression of fast correlations is the presence of NMDA receptors on the recipient synapses of the relay neurons.
An open question is how the parasitic correlations are actually suppressed by NMDA receptors. Is it correct that a stronger NMDAR contribution to the transmitted activity simply low-pass filters the incoming spike trains, and that fast correlations are smoothed out as a result? So are developing circuits tuned to slower activity? This could also explain why AMPA receptors subunits with slower kinetics are expressed during development in many circuits.
Taken together, I think this paper presents a very interesting set of results. The issue with parasitic correlations is quite obvious in retrospect, and clearly, a problem developing circuits will generally face. Additionally, the presence of NMDA receptors is often linked to plasticity and is seen as less important in shaping postsynaptic integration. Although no developmental plasticity has been modelled, would it be possible to predict the possible effects of experimental manipulations?
-
