Homeostatic regulation through strengthening of neuronal network-correlated synaptic inputs

  1. Samuel J Barnes
  2. Georg B Keller
  3. Tara Keck

  • Curated by eLife

    eLife logo

    eLife assessment

    When sensory inputs, such as vision or sound, are chronically disabled, the loss of input activity is counterbalanced by the upregulation of synaptic activity. In this study, the authors provide evidence that instead of synapses that directly represent the sensory information, synapses that show correlated intrinsic network activity are the ones that undergo the change upon sensory deprivation. This fundamental and important paper will be useful to readers in the fields of experience-dependent plasticity, sensory cortical coding, and homeostatic plasticity. While the key claims of the manuscript are well supported by the data, minor changes are suggested for clarification, including the fact that the present study has addressed homeostatic responses in adult animals rather than in juvenile animals with which homeostatic plasticity has been actively studied to date.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Homeostatic regulation is essential for stable neuronal function. Several synaptic mechanisms of homeostatic plasticity have been described, but the functional properties of synapses involved in homeostasis are unknown. We used longitudinal two-photon functional imaging of dendritic spine calcium signals in visual and retrosplenial cortices of awake adult mice to quantify the sensory deprivation-induced changes in the responses of functionally identified spines. We found that spines whose activity selectively correlated with intrinsic network activity underwent tumor necrosis factor alpha (TNF-α)-dependent homeostatic increases in their response amplitudes, but spines identified as responsive to sensory stimulation did not. We observed an increase in the global sensory-evoked responses following sensory deprivation, despite the fact that the identified sensory inputs did not strengthen. Instead, global sensory-evoked responses correlated with the strength of network-correlated inputs. Our results suggest that homeostatic regulation of global responses is mediated through changes to intrinsic network-correlated inputs rather than changes to identified sensory inputs thought to drive sensory processing.

Article activity feed

  1. Version published to 10.7554/elife.81958 on eLife
  2. eLife

    Author Response:

    Reviewer #2 (Public Review):

    The study is well designed and provides exciting new insights into the plasticity of intracortical connections, (over-)compensating for the partial loss of thalamic inputs. To optically resolve the activity of single synapses in vivo during sensory stimulation is technically very challenging. It would be helpful to know whether the recordings were made in the binocular or monocular region of V1. The results argue against a generalized multiplicative upscaling of all inputs and suggest selective boosting of synapses that are part of sensory-driven subnetworks. However, it is not clear whether homeostatic plasticity occurred at the observed spines themselves or on the level of presynaptic neurons, which could then e.g. fire more bursts, leading to larger postsynaptic Ca transients. The possibility that thalamic inputs from the intact eye in layer 4 could be potentiated should be discussed. It would probably help to explain to the reader the layer-specific connectivity of V1 in the introduction, and why thalamic input synapses themselves were not optically monitored (may require adaptive optics). Technical limitations are a main reason why the conclusions are somewhat vague at this point ("... regulation of global responses"), this could be spelled out better.

    We thank the reviewer for these suggestions. We agree with the reviewer that we cannot determine (due to technical limitations) whether the changes are occurring pre- or post-synaptically or some combination (also related to the reviewer’s point 8). We have added this point to the discussion.

    "Finally, it is important to note that while we made these measurements in layer 5 pyramidal cells, the homeostatic changes mediated by TNF-α could occur outside of layer 5, including changes to upstream inputs or changes to the presynaptic responses, either through changes in presynaptic release (Vitureira et al., 2012) or through a change in activity patterns of the presynaptic cell (e.g., bursts compared to single spikes) (Linden et al., 2009)."

    One important point that was unclear in the earlier version of the manuscript is that the experiments conducted in visual cortex were done in the monocular visual cortex. As explained in comments to reviewer 1, there are not any visually-evoked responses following enucleation in our experiments.

    Reviewer #3 (Public Review):

    Weaknesses are largely restricted to suggested changes to the writing - specifically, there are additional explanations of the data whose discussion may strengthen the long-term impact of the manuscript.

    1. Most importantly, the hypothesis at the heart of this work (subset versus global processes) is framed as orthogonal to the status quo model of homeostatic processes (global). I suspect that adherents to the global argument would quickly point out that the current work is conducted in adult animals, and the majority of the homeostatic plasticity research (which forms the basis of the global model) is conducted in juvenile animals. This is an important distinction because the visual system is enriched in plasticity mechanisms during the ocular dominance critical period. Since Hubel and Wiesel at least, there is extensive evidence to suggest that sensory systems take advantage of critical periods to set themselves up in accordance with the statistics of the world in which they are embedded. The flip side of this is that sensory systems are far less readily influenced by experience once the critical period is closed (Vital-Durand et al., 1978, LeVay et al., 1980; Daw et al., 1992, Antonini et al., 1999, Guire et al., 1999, Lehmann and Lowel, 2008). Through this lens, one might predict that a key feature of the adult cortex is that sensory spines could benefit by being selectively protected from what would otherwise be global homeostatic processes. Either way, the manuscript can be read as if it is framing a show-down between the classical model and a newer, higher-resolution model. I worry that this will be interpreted as misleading without careful presentation/contextualization of the role of development in the introduction and a thorough dissection in the discussion. Currently, the first occurrence of the word, "adult", occurs in the methods, on page 27, line 512. "Juvenile" and "critical period" are not in the manuscript. The age of the animals in this study isn't mentioned until the methods (between P88 and P148 at the time of imaging).
    1. Goel and Lee (2007) seem quite pertinent here: they show that L2/3 neurons give rise to homeostatic regulation of mEPSCs in both juvenile and adult animals, but that the process is no longer multiplicative in nature once the animal is post-critical period. Multiplicity has been the basis of the argument for global change since Turrigiano 1998. Thus, the Goel and Lee finding seems to really bolster the current findings - and also perhaps reconcile the likelihood of a mechanistic difference between CP and adult homeostatic plasticity.

    We fully agree with the reviewer that our results are not in conflict with the developmental synaptic scaling literature. We have changed the text throughout the manuscript to highlight previous studies at different ages and made clear the age of the animals in this work (including in the abstract, introduction, results and discussion). We have also referenced Goel and Lee, 2007, which we agree should be included and thank the reviewer for pointing this out.

  3. eLife

    eLife assessment

    When sensory inputs, such as vision or sound, are chronically disabled, the loss of input activity is counterbalanced by the upregulation of synaptic activity. In this study, the authors provide evidence that instead of synapses that directly represent the sensory information, synapses that show correlated intrinsic network activity are the ones that undergo the change upon sensory deprivation. This fundamental and important paper will be useful to readers in the fields of experience-dependent plasticity, sensory cortical coding, and homeostatic plasticity. While the key claims of the manuscript are well supported by the data, minor changes are suggested for clarification, including the fact that the present study has addressed homeostatic responses in adult animals rather than in juvenile animals with which homeostatic plasticity has been actively studied to date.

  4. eLife

    Reviewer #1 (Public Review):

    In this study, Barnes et al. use chronic two-photon imaging of spine calcium in awake mice to examine the functional response types of synapses that undergo homeostatic spine plasticity elicited by sensory deprivation. Spine plasticity is monitored in apical tuft spines of L5 pyramidal cells in the visual or the retrosplenial cortex, following enucleation/visual deprivation or visual and auditory deprivation, respectively. The authors find that spines that convey sensory stimuli, at least those used for testing, do not change but spines whose activity is correlated to intrinsic network activity undergo compensatory strengthening. The experiments are carefully performed, and the writing is clear and concise. The main findings are important in shedding light on the cellular basis by which a network of neurons compensates for the loss of sensory input activity, specifically suggesting a key role of intrinsic network activity. The study is of significant interest to a broad neuroscience readership. Some of the conclusions are not strongly supported by the data as presented, however, and further considerations involving reanalysis of data and/or presentation are warranted.

  5. eLife

    Reviewer #2 (Public Review):

    Barnes et al. follow individual spines on L5 PC distal tufts in mouse V1 before and after contralateral enucleation. At baseline, some spines show activity driven by visual simulation, others are correlated with network activity (average Ca signal in all other spines). After sensory deprivation (12 h), strongly 'visual' spines had smaller Ca transients while previously weakly 'visual' spines had larger transients, indicating homeostatic boosting. These boosted spines are the ones that were correlated with network activity at baseline. Similar results were obtained in the retrosplenial cortex 48 h after auditory or visual deprivation. As previously described for homeostatic plasticity, a block of TNF-a blocked deprivation-induced boosting of spine responses. Somewhat paradoxically, dendritic sensory-evoked responses did increase after sensory deprivation.

    The study is well designed and provides exciting new insights into the plasticity of intracortical connections, (over-)compensating for the partial loss of thalamic inputs. To optically resolve the activity of single synapses in vivo during sensory stimulation is technically very challenging. It would be helpful to know whether the recordings were made in the binocular or monocular region of V1. The results argue against a generalized multiplicative upscaling of all inputs and suggest selective boosting of synapses that are part of sensory-driven subnetworks. However, it is not clear whether homeostatic plasticity occurred at the observed spines themselves or on the level of presynaptic neurons, which could then e.g. fire more bursts, leading to larger postsynaptic Ca transients. The possibility that thalamic inputs from the intact eye in layer 4 could be potentiated should be discussed. It would probably help to explain to the reader the layer-specific connectivity of V1 in the introduction, and why thalamic input synapses themselves were not optically monitored (may require adaptive optics). Technical limitations are a main reason why the conclusions are somewhat vague at this point ("... regulation of global responses"), this could be spelled out better.

  6. eLife

    Reviewer #3 (Public Review):

    In this work, the authors address the question of whether sensory deprivation drives homeostatic responses in all dendritic spines (the standard model/status quo) or is restricted to a functional subset of spines. The key claims of the manuscript are well supported by the data, the writing is clear, and the conclusions are both thoughtful and restrained. The contrast/comparison of the current results to prior work, specifically the difference between homeostatic responses in adult versus critical period animals, should be presented early and often.

    Strengths:
    This manuscript builds on prior work from the authors that seek to understand compensatory plasticity in cortical circuits in the intact animal. Here, the authors present clear evidence that, instead of a global homeostatic response, circuit rebalancing may be the result of a selective strengthening of intra-network connections. Crucially, this rebalancing via network tuning does not involve homeostatic adjustment of sensory-related spines. More specifically, by tracking the same spines over 3 d, the authors reveal a functional separation between those spines that faithfully respond to sensory input and those spines that are network-correlated. The amplitude of calcium transients in network-correlated spines is increased following enucleation, which the authors suggest forms the basis of the global (network-wide) sensory-evoked responses. This is quite interesting as it is somewhat counterintuitive; absent these data, it would be reasonable to assume that increased network responses are reflective of homeostatic processes in the sensory-related spines and synapses. To reach these conclusions, the authors employ GCaMP6s-based calcium imaging of L5 pyramidal neurons in visual and retrosplenial cortices prior to and during sensory deprivation (enucleation or ear-plugging).
    This manuscript is well written. It is clear and not overstated. The work is presented in a linear and approachable style that should be accessible to readers outside of the field. These findings are a meaningful advance for the field and raise foundational questions about the neurobiology of the cortex. Specifically, homeostatic regulation of neuronal activity may be constrained to a subset of processes, or alternatively, adult sensory processes are somehow shielded from the impact of homeostatic change.

    Weaknesses:
    Weaknesses are largely restricted to suggested changes to the writing - specifically, there are additional explanations of the data whose discussion may strengthen the long-term impact of the manuscript.
    1. Most importantly, the hypothesis at the heart of this work (subset versus global processes) is framed as orthogonal to the status quo model of homeostatic processes (global). I suspect that adherents to the global argument would quickly point out that the current work is conducted in adult animals, and the majority of the homeostatic plasticity research (which forms the basis of the global model) is conducted in juvenile animals. This is an important distinction because the visual system is enriched in plasticity mechanisms during the ocular dominance critical period. Since Hubel and Wiesel at least, there is extensive evidence to suggest that sensory systems take advantage of critical periods to set themselves up in accordance with the statistics of the world in which they are embedded. The flip side of this is that sensory systems are far less readily influenced by experience once the critical period is closed (Vital-Durand et al., 1978, LeVay et al., 1980; Daw et al., 1992, Antonini et al., 1999, Guire et al., 1999, Lehmann and Lowel, 2008). Through this lens, one might predict that a key feature of the adult cortex is that sensory spines could benefit by being selectively protected from what would otherwise be global homeostatic processes. Either way, the manuscript can be read as if it is framing a show-down between the classical model and a newer, higher-resolution model. I worry that this will be interpreted as misleading without careful presentation/contextualization of the role of development in the introduction and a thorough dissection in the discussion. Currently, the first occurrence of the word, "adult", occurs in the methods, on page 27, line 512. "Juvenile" and "critical period" are not in the manuscript. The age of the animals in this study isn't mentioned until the methods (between P88 and P148 at the time of imaging).
    2. Goel and Lee (2007) seem quite pertinent here: they show that L2/3 neurons give rise to homeostatic regulation of mEPSCs in both juvenile and adult animals, but that the process is no longer multiplicative in nature once the animal is post-critical period. Multiplicity has been the basis of the argument for global change since Turrigiano 1998. Thus, the Goel and Lee finding seems to really bolster the current findings - and also perhaps reconcile the likelihood of a mechanistic difference between CP and adult homeostatic plasticity.

  7. Version published to 10.1101/2022.08.02.502580v1 on bioRxiv