Article Figures & Data
Figures
- Figure 1.
Engram recording stimulation paradigm. A, MAP2 immunostaining of primary hippocampal cultures, illustrating the extensive network connections present. B, Fluo4 calcium dye used to visualize neuronal firing. C, Algorithm detection and indexing of cells. D, Vertical axis gives the index number of the neuron traced, horizontal axis time. Following a test period (to determine functionally active neurons), a control prestimulation period of 60 s is recorded. Then a 10 s stimulation pattern is fired, followed by a poststimulation recording of 60 s. A final test period confirms the continued health of the neurons.
- Figure 2.
Identification of reverberations of the stimulation frequency after the stimulation has ended. A, The trace shows the calcium signal recording from a single neuron in the network. During the prestim period, background noise and minor depolarizations can be seen, and during the 3 Hz stimulation (red bar), the neuron fires reliably. After the stimulation has ended (green arrow), firing continues at a frequency resembling the stimulation frequency. B, The trace shows the same experiment with a 5 Hz stim frequency instead. C, Sample histograms of ISIs show a clear peak at the wavelength of the stim frequency (either 3 or 5 Hz) during the stimulation, as expected. This peak is also evident, along with other frequencies, during the poststimulation period. Quantification of the prevalence of these reverberations will be presented in Figure 3. The baseline of calcium traces normalized for visualization.
- Figure 3.
Frequency of occurrence of reverberations, dependency on synaptic transmission, and neuronal cell types involved. A, The percentage increase in ISIs corresponding to the stimulation frequency in the 60 s poststimulation versus the prestimulation control period. Zero hertz represents a nonstimulated control. A 3, 5, and 8 Hz stimulation results in an increasing presence of stimulation-specific reverberations. Glutamatergic synaptic transmission blockers CNQX, APV, and KYNA reduce the reverberations at 8 Hz, while the glial transmission blockers carbenoxolone and suramin also demonstrate that glia could play a role in the process. B, Data in A show the total amount of reverberations seen in the network, while B indicates the increase in the percentage of the population of individual cells that have reverberations. Again, an increase with frequency is observed. C, Costaining of excitatory (VGLUT) and inhibitory (GABA) neurons, which is then cross-referenced to each cell's activity (Fig. 1C,D). D, Neuronal type of cells expressing reverberations. In the prestim control period, a predominance of excitatory activity is observed, as expected in a normal active network. During the stimulation, all cell types are fired equally, and so the E/I ratio reflects the composition of the network. Poststimulation however, a strong skew to inhibitory cell activity is observed, suggesting inhibitory cells are an important component of the microcircuitry responsible for the reverberations. Full statistical analysis and discussion regarding controls in Extended Data Figure 3A-1, Table 3C-1a,b,D-1.
- Figure 4.
Engrams of multifrequency stimulation patterns; RNN, ISI, and circuitry analysis. Patterns containing two component frequencies are used to briefly stimulate the network. A, Overview of the 10 s stim and the network activity immediately afterward in ∼1,000 neurons. Faint vertical bands can be seen to persist after stimulation has ended. B, Close-up of the stim pattern. C, D, Close-up of two regions in the poststim region, suggesting some cells have low-frequency activity (C), and some have higher-frequency firing patterns (D). E, RNN identification of the stimulation pattern in the entire activity fingerprint. Brightness indicates the degree of similarity to the stim pattern, which was used to train the network. The pattern itself (at 60 s) is the strongest region identified as expected; however, many regions of higher similarity can be seen immediately following the stim in comparison with the prestimulation region. RNN statistics in text. While the RNN analysis produced superior identification of complete engram patterns, histogram analysis allows a more direct analysis of the component frequencies. F, AN example of a single-neuron firing trace during stimulation, illustrating the two component frequencies. G, An example of a neuron in the poststim region repeating a similar pattern. H, Histogram analysis of the ISIs during the stimulation of the entire network. Similar to what we saw with the single frequency, peaks at frequencies corresponding to the high-frequency component (red, 160 ms) and the ISI between the end of one doublet and the start of the next (blue, 300 ms) can be seen, as well as a population at 460 ms (green), the distance from the start of one full pattern to the next. I, Histogram analysis of the poststim region. While the 160 and 300 ms components are detectable, other peaks are frequently observed (green), including the interpattern interval of 460 ms. J, Comparison of the neuron populations in the high-frequency (160 ms) versus low-frequency (460 ms) components. Under nonstimulated, spontaneous network activity, we found a 73.3 ± 2.85% SEM (n = 8) overlap between the neurons, meaning many cells were firing at both frequencies. During stimulation, this number was higher (82.3 ± 2.0%, SEM n = 12), which was expected as all neurons in the network are being driven to fire at both frequencies. In the poststimulation period, there was a substantial decrease in the overlap (43.6 ± 4.3%, n = 15), suggesting different populations of neurons were firing at the two different component frequencies. Full statistical analysis in Extended Data Table 4J-1.
Extended Data
Figure 3
Language genetic influence result without a language mask A. Multiple types of whole-brain regions. After genetic modeling without restricting the language activation map, multiple types of regions were identified. B. The clustering result of the genetic regions without a language activation mask. Download Figure 3, TIF file.
Table 3A- 1a
Functional labels based on Neurosynth for genes contributing to commonality of language clusters Note: For each gene, the functional terms from Neurosynth represent the terms with the most similar meta-analysis whole-brain activation map to the gene’s whole-brain map. The correlation values indicate the correlation of the gene’s whole-brain expression with the term’s meta-analysis result. Download Table 3A, DOC file.
Table 3A - 1b
Functional labels based on Neurosynth for genes contributing to differences of different language clusters Note: For each gene, the functional terms from Neurosynth represent the terms with the most similar meta-analysis whole-brain activation map to the gene’s whole-brain map. The correlation values indicate the correlation of the gene’s whole-brain expression with the term’s meta-analysis result. Download Table 3A - 1b, DOC file.
Table 3D - 1
Download Table 3D - 1, DOC file.
Table 4J - 1
Genetic model results for all the language genetic clusters after excluding specific object domain responsive voxels Note: For each language genetic cluster, the cognitive abilities with existing genetic effects (AE model or DE model, p values < 0.05, compared with the control model E, uncorrected) are marked with yellow. ΔAIC denotes the degree to which the best model is better than the control model. Download Table 4J - 1, DOC file.
In this issue
