Long-term storage of information about relevant experiences is essential for successful adaptation of human and animal behavior in a changing environment. A current model of memory formation suggests plastic adaptations in neuronal connections (synapses) caused by relevant experiences. Yet, how such changes in synaptic connectivity lead to the formation of a memory trace remains elusive. How is the processing of external stimuli altered after the formation of a memory? How are we able to continuously store novel memories in a given neuronal circuit without corrupting previously stored memories? In order to understand the mechanisms by which multiple memory traces are coordinated, we are currently applying in vivo imaging techniques to the auditory cortex of mice. The auditory cortex mediates processing of sounds and is involved in the formation of memories of sounds.
Cross section of the auditory cortex of a transgenic mouse used for in vivo spine imaging. A single neuron is highlighted by green fluorescent protein. The dendritic processes are decorated with spines, the morphological correlate of excitatory synapses. Counter staining for synaptic marker proteins ErbB4 (cyan) and Synapsin1 (magenta).
Two-photon laser scanning microscopy in transgenic animals expressing green fluorescent protein in just a small subset of cells permits the same neurons, and even the same individual synapses, to be revisited day after day. This is truly remarkable because we estimate that the brain comprises about 10 trillion (10^13) synapses. We find that neocortical circuits are highly dynamic: remodeling occurs by the formation/elimination of synaptic connections as well as adaptations in the strength of existing connections. We try to understand the rules of these dynamics, which allows us to predict which connections will vanish and which connections are stable elements in the neuronal network. Besides the analysis of synaptic remodeling under basal conditions, we are investigating the impact of auditory learning paradigms on the dynamics of a given set of synapses in the auditory cortex.
• Mongillo G, Rumpel S, Loewenstein Y (2018) Inhibitory connectivity defines the realm of excitatory plasticity. Nat. Neurosci. 21:1463–1470.
• Loewenstein Y, Yanover U, Rumpel S (2015) Predicting the dynamics of network connectivity in the neocortex. J. Neurosci. 35:12535-12544.
• Bathellier B, Tee SP, Hrovat C, Rumpel S (2013) A multiplicative reinforcement learning model capturing learning dynamics and inter-individual variability in mice. Proc. Natl. Acad. Sci. USA 110:19950-19955.
• Moczulska KE, Tinter-Thiede J, Peter M, Ushakova L, Wernle T, Bathellier B, Rumpel S (2013) Dynamics of dendritic spines in the mouse auditory cortex during memory formation and memory recall. Proc. Natl. Acad. Sci. USA 110:18315-18320.
• Loewenstein Y, Kuras A, Rumpel S (2011) Multiplicative dynamics underlies the emergence of the log-normal distribution of spine sizes in the neocortex in vivo. J. Neurosci. 31:9481-9488.
The publicly available dataset containing morphological parameters of >3.000 spines over a 21 day period published in Loewenstein et al., 2015 can be found here.
Left: Confocal image of a cross section of a mouse brain co-transduced with a rAAV coding for nuclear marker H2B::mCherry and genetically encoded calcium indicator GCaMP6m. Countertstain: DAPI; Scale: 1mm. Middle: In vivo two-photon image of layer 2/3 neurons in the mouse auditory cortex. Scale: 100 microns. Right: Registration of fluorescence chances reporting neuronal activity in 19 of the neurons imaged.
In vivo imaging not only permits analysis of synaptic connections, but also monitoring of neuronal activity in hundreds of neurons simultaneously. Action potential-mediated increases in calcium levels can be detected as changes in fluorescence of calcium indicators. We are investigating activity patterns elicited by various sounds in neuronal populations of the auditory cortex in order to learn about the principles how sounds are encoded and recognized in the brain. We observe that activity in layer 2/3 neuronal ensembles is surprisingly strongly constrained into very few response modes. Interestingly, these discrete activity modes can serve as a representational basis to predict generalization behavior in an auditory discrimination task. Our findings point toward a model of neocortical function in which external stimuli are represented in a broad basis set of spontaneous associations into common activity patterns, and classified by sharp transition across the activity patterns. We are investigating the circuit mechanisms that lead to the generation of sounds representations in discrete activity modes, and to what extent auditory learning paradigms cause changes in these neuronal representations of memorized sounds.
• Bathellier B, Ushakova L, Rumpel S (2012) Discrete neocortical dynamics predict behavioral categorization of sounds. Neuron 76:435-449.
Left: In vivo two-photon image stack of an individual pyramidal neuron photolabeled in a transgenic mouse line expressing PA-GFP. Scale: 100 microns. Middle: Confocal image of a coronal brain section of a transgenic mouse expressing PA-GFP that was previously implanted with an optrode (dashed line). The illumination field is directly visualized by the conversion of GFP. Scale: 1mm. Right: Confocal images of the mouse auditory cortex of mice transduced with similar amounts of various rAAV vectors. The differences in the efficacy to drive expression of genes of interest becomes apparent by the expression of the green reporter protein GFP. Counterstain DAPI. Scale: 500 microns.
The analysis of neuronal circuits in the brain is a challenging due to sheer numbers of neurons and their connections as well as their heterogeneity. Up to now it is still unresolved how many cell types exist in the neocortex. We are establishing methods that allow labeling or genetic access to specific neurons in vivo.
To facilitate linking in vivo experiments with a more detailed molecular or physiological analysis in vitro, we have generated and characterized genetically modified mice expressing photoactivatable GFP (PA-GFP) that allow conditional photolabeling of individual neurons in vivo. These mouse models enable the combination of various analytical approaches to characterize living cells, also beyond the neurosciences. Transgenic mouse lines expressing PA-GFP are made available from the Jackson Laboratory Repository with the JAX Stock No. 021069, 021070 and 021071.
Furthermore, we explore adeno-associated virus-derived vectors (rAAV) as tool for gene-delivery in vivo. We recently analyzed in depth the efficacy of various rAAV variants to transduce different cell types in the mouse brain. Despite the fact that the analyzed rAAV variants have the general ability to transduce all major cell types in the brain (neurons, microglia, astrocytes and oligodendrocytes), the expression level of a reporter gene driven from a ubiquitous promoter varies significantly for specific cell type / rAAV combinations. We provide a quantitative dataset to choose a suitable rAAV serotype for various applications to target specific regions and cell types in the mouse brain. In the future we will investigate approaches that will allow rAAV-mediated expression of genes of interest in specific cell types.
• Aschauer DF, Kreuz S, Rumpel S (2013) Analysis of Transduction Efficiency, Tropism and Axonal Transport of AAV Serotypes 1, 2, 5, 6, 8 and 9 in the Mouse Brain. PLoS One. 2013 8:e76310.
• Peter M, Bathellier B, Fontinha B, Pliota P, Haubensak W, Rumpel S (2013) Transgenic mouse models enabling photolabeling of individual neurons in vivo. PLoS ONE 8:e62132.
Jointly, these approaches will pave the way for a series of novel experiments addressing the storage of information in living neuronal networks: a field of research that has been almost exclusively the domain of theoretical neuroscientists thus far.