Homeostatic Plasticity Scales Dendritic Spine Volumes and Changes The
Total Page:16
File Type:pdf, Size:1020Kb
bioRxiv preprint doi: https://doi.org/10.1101/308965; this version posted April 26, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 Homeostatic plasticity scales dendritic spine volumes and 2 changes the threshold and specificity of Hebbian plasticity 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Authors: 17 Anna F. Hobbiss1, Yazmin Ramiro Cortés1,2, Inbal Israely1,3 18 19 20 1. Champalimaud Centre for the Unknown, 1400-038 Lisbon, Portugal. 21 2. Instituto de Fisiología Celular, Universidad Nacional Autónoma de México, Ciudad Universitaria, 22 Circuito exterior s/n, Ciudad de México 04510, México. 23 3. Department of Pathology and Cell Biology in the Taub Institute for Research on Alzheimer's 24 Disease and the Aging Brain, Department of Neuroscience, College of Physicians & Surgeons, 25 Columbia University, New York, New York, 10032, USA. 26 27 Corresponding author: 28 Inbal Israely. Electronic address: [email protected] 29 1 bioRxiv preprint doi: https://doi.org/10.1101/308965; this version posted April 26, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 30 Abstract 31 Information is encoded within neural networks through synaptic weight changes. Synaptic 32 learning rules involve a combination of rapid Hebbian plasticity with slower homeostatic 33 synaptic plasticity (HSP) that regulates neuronal activity through global synaptic scaling. While 34 Hebbian plasticity has been extensively investigated, much less is known about HSP. Here we 35 investigate the structural and functional consequences of HSP at dendritic spines of mouse 36 hippocampal neurons. We find that prolonged activity blockade induces spine growth, 37 paralleling synaptic strength increases. Following activity blockade, glutamate uncaging- 38 mediated long-term potentiation at single spines leads to size-dependent structural plasticity: 39 smaller spines undergo robust growth, while larger spines remain unchanged. Moreover, we 40 find that neighboring spines in the vicinity of the stimulated spine exhibit volume changes 41 following HSP, indicating that plasticity has spread across a group of synapses. Overall, these 42 findings demonstrate that Hebbian and homeostatic plasticity shape neural connectivity 43 through coordinated structural plasticity of clustered inputs. 44 Introduction 45 Hebbian synaptic plasticity, widely regarded as the leading biological mechanism for 46 information storage, involves activity-dependent changes in synaptic connectivity (Malenka 47 and Bear, 2004). Notably, these changes have a physical component and excitatory 48 postsynaptic current is highly correlated with dendritic spine volume (Matsuzaki et al., 2001). 49 However, left unchecked activity would lead to a positive feedback loop in which changes in 50 synaptic weight are further reinforced by future events. This has been proposed to result in 51 loss of functionality of the system, by either driving synapses towards saturation, or through 52 silencing the population (Miller and MacKay, 1994). How neurons maintain stability in the 53 face of destabilizing cellular and circuit events has been a longstanding question (LeMasson 54 et al., 1993). One method neurons employ to resolve this is Homeostatic Synaptic Plasticity 55 (HSP) (Turrigiano et al., 1998), a feedback mechanism through which a population of synapses 56 can be maintained to function within optimal bounds. During HSP, a decrease in global activity 57 drives a counteracting scalar increase in synaptic strengths to bring the network into an 58 optimal functioning range, and conversely, network activity increases will promote synaptic 59 weakening (Turrigiano, 2011). This scaling is instantiated in part by AMPA receptor trafficking 2 bioRxiv preprint doi: https://doi.org/10.1101/308965; this version posted April 26, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 60 to and from the post-synaptic density, as well as through presynaptic changes which modify 61 neurotransmitter release, leading to concomitant changes in synaptic strength (Murthy et al., 62 2001; O’Brien et al., 1998). HSP has been observed both in vitro and in vivo (Desai et al., 2002), 63 supporting a fundamental role for this mechanism in the proper functioning of a neural 64 system. 65 It is well established that HSP modulates synaptic function; however, it is currently unknown 66 how this process also correlates with spine structural changes. Moreover, how HSP affects 67 the induction and longevity of subsequent forms of Hebbian plasticity at individual inputs 68 remains to be determined. In this study, we examine how the induction of HSP affects the 69 structure and function of hippocampal pyramidal neuron synapses, and what are the 70 consequences of these changes for future Hebbian plasticity. We find that the induction of 71 HSP through activity blockade leads to an overall increase in the size of spines, corresponding 72 to a structural scaling that matches the functional scaling of synapses. Through precise 2- 73 photon mediated glutamate uncaging, we further investigate how these homeostatic 74 functional and structural plasticity changes impact the ability of individual inputs to undergo 75 subsequent plasticity. We demonstrate that after HSP, spines express increased longevity of 76 LTP, an increased growth rate after stimulation, and a reduced plasticity threshold. We find 77 that HSP enhances the magnitude of synaptic potentiation through the preferential 78 modulation of small spines and by promoting structural plasticity at clustered inputs following 79 Hebbian activity at single synapses. Together, these changes provide a mechanism by which 80 homeostatic plasticity can modulate synaptic efficacy and enhance future learning without 81 compromising previously stored information. 82 83 Results 84 Structural correlates of homeostatic plasticity 85 Hebbian plasticity at an individual input is linearly correlated with volume changes in the 86 corresponding spine (Govindarajan et al., 2011; Harvey and Svoboda, 2007; Matsuzaki et al., 87 2004). Since HSP is known to change the functional properties of the spine, we reasoned that 88 homeostatic modifications of efficacy would be accompanied by structural plasticity of inputs. 89 We induced HSP using prolonged activity blockade through Tetrodotoxin (TTX) inhibition of 3 bioRxiv preprint doi: https://doi.org/10.1101/308965; this version posted April 26, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 90 sodium channels, for either 0 h, 24 h, 48 h or 72 h (Figure 1a), as this has been shown to 91 induce scaling of synaptic strengths in a variety of systems (Karmarkar and Buonomano, 2006; 92 Turrigiano et al., 1998). We chose to conduct our experiments in mouse hippocampal 93 organotypic slice cultures, which maintain a physiologically relevant tissue architecture and 94 are amenable to genetic manipulation. To verify that functional synaptic scaling occurred, we 95 recorded spontaneous miniature excitatory post-synaptic currents (mEPSCs) from both TTX 96 treated and Control CA1 hippocampal pyramidal neurons at 48 h after the beginning of the 97 HSP induction period (Figure 1b-d). All recordings were performed in the presence of acute 98 TTX to block action potentials. As expected, we found a significant increase in mEPSC 99 amplitude following 48 h of activity blockade (Figure 1b,c; Control = 18.6 ± 0.25 pA, TTX = 100 20.72 ± 0.28 pA, p = 1.55e-8 Mann Whitney). The distribution of the TTX mEPSCs scaled 101 linearly to overlay with the control distribution (Figure 1d), in accordance with the synaptic 102 scaling theory. We next investigated whether homeostatic plasticity leads to structural 103 modification of synapses by live, 2-photon imaging of GFP-labelled CA1 dendrites (Figure 1e). 104 The volumes of all visible spines within the image were measured using SpineS, a custom in- 105 house developed Matlab toolbox (Erdil et al., 2012). We found that spines from chronically 106 TTX treated cells were significantly bigger than those from control cells beginning at 48 h 107 (Figure 1e,f; Control = 0.136 ± 0.0062 µm³, TTX = 0.211 ± 0.0123 µm³). Surprisingly, as opposed 108 to the linear functional scaling of mEPSCs seen in response to HSP, we found that structural 109 scaling of spine volumes at 48 h are best fit to controls with a second order equation (Figure 110 1g). This superlinear scaling resulted from a preponderance of large spines after TTX 111 treatment, for which correspondingly large mEPSCs were not observed. 112 113 After 72 h of activity blockade, synapses maintained the enhancement of mEPSC size (Figure 114 1h; Mean ± SEM: control 72 h = 20.35 ± 0.27 pA, TTX 72 h= 22.18 ± 0.239 pA, p = 1.88e-15, 115 Mann-Whitney) and spine volume increases (Figure 1i; Mean ± SEM: control 72 h= 0.16 ± 116 0.0073 µm³, TTX = 0.19 ± 0.0093 µm³, p = 0.0008 Mann-Whitney). However, at this time, 117 scaling was instantiated in a linear fashion at both the level of mEPSC amplitude and spine 118 volumes (Figure 1j,k). This suggests that the expression of functional and structural scaling 119 evolves dynamically over time in response to prolonged activity blockade. 120 121 4 bioRxiv preprint doi: https://doi.org/10.1101/308965; this version posted April 26, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 122 Reversibility of homeostatic plasticity mediated structural changes 123 Synaptic strength modifications that occur in response to activity blockade are reversible 124 upon re-exposure of a circuit to activity (Desai et al., 2002; Wallace and Bear, 2004). We 125 tested whether the structural changes that resulted following activity blockade were also 126 reversible when activity is restored.