Optogenetic Frequency Scrambling Of Hippocampal Theta Oscillations Dissociates Working Memory Retrieval From Hippocampal Spatiotemporal Codes Part 2

Combining MS optogenetic stimulation and hippocampal calcium imaging

To study the effects of theta manipulations on hippocampal spatial and temporal codes, we combined MS optogenetic stimulation with calcium imaging in CA1. This experimental paradigm raises two potentially important issues: GRIN lens implants involve tissue damage that could alter the physiological state of theta oscillations, and the wavelength spectrum of the excitation LED used to excite GCaMP6f could potentially overlap with that of the opsin located on the terminal fibers of GABAergic MS fibers in the hippocampus.

The hippocampus is a very important structure in the brain. It is mainly responsible for storing and processing memory information. As we all know, memory is an important part of human intellectual activities and an important way for us to communicate and interact with our surrounding environment. Therefore, the function of the hippocampus is crucial to our lives.

Hippocampal space is a way of describing how spatial information is processed in our brains. It refers to the active area of a group of neurons in our brain that processes spatial information. This area is called the parahippocampal area and is closely related to the seahorse. Research shows that the area near the hippocampus processes information related to space and memory and is a key part of our memory process.

Specifically, the hippocampus's processing of spatial information mainly includes two aspects. The first aspect is our sense of direction and ability to navigate. When we walk, the hippocampus records our steps and position, maintaining our orientation in space. If our hippocampus is damaged, it can cause problems such as disorientation or being unable to find our way home.

Another aspect is memory ability. People use the spatial information provided by the hippocampus to help people remember. If we experience something in a certain place, the hippocampus will store this information, and we can learn about these experiences or people through recall.

We are constantly strengthening our memory capabilities through learning, and the hippocampus plays a crucial role in this process. Because of the importance of the hippocampus, we need to pay attention to its health and protect it through some methods. Sudoku, running, learning new skills, etc. can all improve our memory and protect the health of our hippocampus.

Therefore, diet, exercise, and maintaining good sleep can help us protect our hippocampus and improve our memory ability. When we have a clear mind and strong memory, we can better understand the world and create a better life. It can be seen that we need to improve our memory. Cistanche deserticola can significantly improve memory, because Cistanche deserticola can also regulate the balance of neurotransmitters, such as increasing the levels of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrients and energy, thus improving brain vitality and endurance.

Click know supplements to boost memory

To verify whether GRIN lens implants altered the physiology of theta, we first implanted mice with a GRIN lens in the right hippocampus and two electrodes in both left and right hippocampi, and found comparable theta signals in both hemispheres (Fig. 3a). We compared theta oscillations during open field exploration in mice with both a GRIN lens and an attached LFP electrode, to mice with electrodes only, and found no significant differences between both groups (Fig. 3b). We did not find any significant difference between relative theta power in mice implanted with a GRIN lens and LFP electrode (0.14 ± 0.007) versus an electrode only (0.154 ± 0.008; t-test, t54 = 1.060, p = 0.29).

Although previous reports have described that combining optogenetic stimulation of ChrimsonR in cell bodies with imaging of GCaMP in neurons at the vicinity of terminals is possible with minimal crosstalk53, we next monitored any potential opsin activation of MS terminals in the hippocampus by recording CA1-LFP while emitting excitation light with our miniscope through a GRIN lens. After calibrating mini scope light output power (Fig. 3c), we found no effect of the mini-scope blue excitation light on endogenous theta power (1ANOVA, F(4,295)= 0.7729, p = 0.5435; Fig. 3d). Since it has been reported that mini scope excitation light can induce a slight depolarization of terminals transfected with ChrimsonR53, potentially hindering further optogenetic-induced depolarization, we next applied MS optogenetic stimulations while imaging with a ~0.3mW/mm2 mini scope LED power and were able to significantly disrupt or pace theta using scrambled or 8 Hz stimulation respectively (Friedman test χ2 = 6.000, p = 0.0278; Fig. 3e).

Disruption of theta rhythms modulates a small portion of CA1 cells

We then performed phasic (5 s ON, 5 s OFF) optogenetic stimulations of MS while recording CA1 pyramidal cells as mice freely explored an open field (Fig. 4a). We found that a portion of recorded cells were consistently excited in these conditions, while others were inhibited (Fig. 4b; see Methods). The activity of pyramidal cells during stimulations was overall lower compared to baseline, for both scrambled stimulation during running (Pearson correlation, R2 = 0.567, p ≤ 0.0001) and rest (R2 = 0.521, p ≤ 0.0001) periods, as well as for 8 Hz stimulation (R2 = 0.6, p ≤ 0.0001 for rest periods; R2 = 0.632, p ≤ 0.0001 for running periods; n = 1849 cells, N = 5 mice; Fig. 4c). Overall, ~6.42 ± 0.52% of total cells were significantly modulated by scrambled optogenetic stimulations (Fig. 4d). Among those modulated cells, 50.56 ± 6.38% were inhibited while 49.43 ± 6.38% were excited (n = 1849 cells, N = 5 mice; Fig. 4e).

We next analyzed the effects of optogenetic stimulation on the spatial tuning of hippocampal neurons as mice freely explored the open field. To this end, we computed activity rate maps by using either epoch within or outside of stimulation periods (for the baseline condition, we included epochs that followed the same 5 s ON, 5 s OFF pattern used for actual stimulation; Fig. 4f). Stability was then computed as the correlation between rate maps for baseline and stimulation epochs. Despite the aforementioned changes in overall activity, rate maps displayed no change of spatial stability for either scrambled or 8 Hz stimulations (Kruskal–Wallis H3 = 3.5, p = 0.1773; Fig. 4g).

MS optogenetic stimulation alters behavior but not spatiotemporal codes

To assess the effects of theta disruption on temporal and spatial codes, we monitored CA1 pyramidal neuron activity on the 3-tone linear track (Fig. 5a). Here, we leverage one of the main advantages of calcium imaging, which is the ability to register recorded cells over several days while performing scrambled or 8 Hz stimulations on selected days (Figs. 5b, c, bottom panel). We focused our analysis on pairs of days with the same amount of time (48 h) between testing (Fig. 5c, top panel). For each condition, we assessed the portion of total cells significantly encoding one or several variables and found no effect of either 8 Hz or scrambled stimulation on spatial and temporal encoding (RM-ANOVA; F2 = 0.807, p = 0.453 for the main effect of stimulations; F6 = 1.283, p = 0.285 for the interaction between stimulation and encoded variable, n = 5 mice; Fig. 5d). 

While the portion of cells did not change under stimulation conditions, we assessed the effect of MS stimulation on place-, time-, and distance-modulated cells’ tuning curves stability. To this end, we tracked neurons across days (Fig. 5c; see Methods) and computed the stability of place and time fields as the pairwise correlation between fields across 48 h. Because CA1 is known to display prominent remapping over days, we also used a pair of days with no stimulation to compute a baseline stability score to be used as a reference (Fig. 5e–g). We found no change in stability during MS stimulation for place-modulated cells (1ANOVA, F2 = 1.907, p = 0.1511; n = 205 cell pairs pooled from N = 5 independent mice; Fig. 5h), time-modulated cells (1ANOVA, F2 = 2.201, p = 0.113; n = 227 cell pairs pooled from N = 5 independent mice; Fig. 5i), and distance-modulated cells (1ANOVA, F2 = 0.6962, p = 0.5024; n = 64 cell pairs pooled from N = 5 independent mice; Fig. 5j). We also extended the same analysis to conjunctive neurons (i.e. neurons that can encode more than one variable; Supplementary Fig. 5a–f) and found no effects of optogenetic stimulations on the stability of conjunctive spatial (1ANOVA, F2 = 3.731, p = 0.0661; N = 4 independent mice; Supplementary Fig. 5g), temporal (1ANOVA, F2 = 1.993, p = 0.8228; N = 4 independent mice; Supplementary Fig. 5h), and distance cells (1ANOVA, F2 = 0.469, p = 0.6400; N = 4 independent mice; Supplementary Fig. 5i).

We next assessed the quality of spatiotemporal codes using a naive Bayesian classifier to decode location (Fig. 6a, b), time (Fig. 6cd), and distance traveled (Fig. 6e, f), using random bootstrap samples (n = 50 bootstrap samples, n = 160 cells per sample). Decoded errors were systematically lower than shuffled surrogates, including during 8 Hz or scrambled stimulation for location (2ANOVA, F5 = 2172, p ≤ 0.0001; n = 50 bootstrap samples from one representative mouse; Fig. 6a), time (2ANOVA, F5 = 1292, p ≤ 0.0001; n = 50 bootstrap samples from one representative mouse; Fig. 6c), and distance (2ANOVA, F5 = 1964, p ≤ 0.0001; n = 50 bootstrap samples from one representative mouse; Fig. 6e) indicating that spatiotemporal codes were preserved during stimulation. To estimate the inter-individual significance of spatiotemporal codes, we z-scored decoding error by using both the actual and shuffled results (see Methods) for a given day, for every mouse (Fig. 6b, d, f). MS optogenetic control did not significantly alter the encoding of location (1ANOVA, F2,11 = 2.2332, p = 0.1432; N = 5 mice; effect size η2 = 0.29; Fig. 6b), time (1ANOVA, F2,11 = 0.4561, p = 0.6452; N = 5 mice; effect size η2 = 0.07; Fig. 6d), or distance (1ANOVA, F2,11 = 0.6102, p = 0.5606; N = 5 mice; effect size η2 = 0.09; Fig. 6f).

In the behavioral paradigm used to analyze temporal modulation of neuronal activity, mice are water-scheduled and trained to collect rewards. Based on this assumption, we quantified the number of returns to an empty reward site as errors, and computed the percentage of correct trials to total trials as a proxy for performance (Fig. 6g). In these conditions, we found a significant effect of optogenetic stimulation on performance across days (Friedman test χ2 = 6.000, p = 0.0278) and in particular a significant difference in performance between during baseline (78.70 ± 2.45%) and scrambled stimulation (44.44 ± 5.56%; multiple comparisons, p = 0.0429; N = 3 mice; Fig. 6h). Only mice with a minimum of 12 runs were included in this analysis. Because of the limitations of this task (low cognitive load and a low number of mice tested), we next set out to assess the effect of MS optogenetic stimulation in standardized memory tasks.

Disruption of theta signals impairs spatial recognition and working memory retrieval

To test the role of theta signals in spatial memory, we used a dedicated group of mice injected with ChrimsonR and implanted with a fiber optic in the MS (Fig. 7a, left panel). Mice were subjected to a novel place object recognition (NPOR) task (Fig. 7a, right panel). To assess the role of theta oscillations in memory encoding and retrieval, we Fig. 7 | MS optogenetic stimulations disrupt maintenance and retrieval, but not encoding of episodic and working memory. a Mice were implanted with fiber optics in the MS after transfecting ChrimsonR (top). They were then subjected to the novel object place recognition task (bottom, see Methods). b In this task, both scrambled (red) and 8 Hz (blue) stimulations during retrieval, as well as 8 Hz (green) but not scrambled (yellow) stimulations during encoding disrupted memory performance (2ANOVA, F4,31 = 3.283 for the main effect of treatment, p = 0.0097; effect size for the main effect of group, η2 p = 0.306; N = 12 mice). c 

To further examine the effect of MS stimulation on memory encoding, maintenance, and retrieval, mice were trained in a delayed non-match to sample (DNMTS) task in an automated T-maze. d Mice transfected with ChrimsonR (red) or YFP (black) were trained (in the absence of stimulation) to choose the correct, non-matching arm until the performance exceeded a criterion of 0.8 portions of correct choices per day, for at least two consecutive days (green band; RM-ANOVA, F8 = 8.738, p ≤ 0.0001 for the main effect of training days; pairwise Tukey multiple comparison tests between groups, p = 0.420; N = 17 mice). e–g Performance during stimulation at different phases of the task for mice injected with ChrimonR (top) or YFP controls (bottom). Red shading indicates the stimulated regions of the maze. e Average daily performance when performing MS stimulations during encoding only (RMANOVA, F11 = 2.197, p = 0.547; N = 17 mice). f Average daily performance when performing MS stimulations during the 10 s delay period only (RM-ANOVA, F11 = 3.483, p = 0.0495; effect size for the main effect of treatment, η2 p = 0.109; N = 17 mice). g Average daily performance when performing MS stimulations during retrieval only (RM-ANOVA, F11 = 3.265, p = 0.050; N = 17 mice). 

All bar plots and line plots represent the mean ± SEM of at least three independent experiments. Article https://doi.org/10.1038/s41467-023-35825-5 Nature Communications | (2023) 14:410 10 performed optogenetic stimulations specifically during the sample and test phases and computed the recognition index (RI; see Methods). When stimulating during retrieval, memory performance was significantly reduced for both scrambled (0.472 ± 0.048 RI, n = 6 mice) and 8 Hz stimulations (0.435 ± 0.057 RI, n = 6 mice) groups, compared to YFP controls that displayed a significant increase in object exploration during testing (0.61 ± 0.018 RI; 2ANOVA, F4,31 = 3.283 for the main effect of treatment, p = 0.0097; effect size for the main effect of group, η2 p = 0.306; N = 12 mice; Fig. 7b). On the other hand, scrambled stimulations during encoding did not impair memory at test time (0.60 ± 0.034 RI, p = 0.0157, n = 6 mice) but 8 Hz stimulation during encoding lowered memory performance to chance levels (0.39 ± 0.074 RI, p = 0.8740, n = 6 mice).

To examine the effect of optogenetic control of the MS on specific phases of working memory function (encoding, maintenance, and retrieval), mice were transfected with ChrimsonR and implanted with fiber optics in the MS and trained in a delayed non-match to sample (DNMTS) task. In the sample phase, mice were forced to run to a randomly designated arm to collect a reward. After a delay (10 s), they could either run in the opposite arm (correct choice) to receive another reward or run in the same, unrewarded arm (incorrect arm; Fig. 7c). The advantage of this task is to allow for repetitive testing, specific isolation of task phases (training, delay, testing) and within-subject controls. Mice were trained in this task with no stimulation until a criterion performance of 0.8 (a portion of correct trials) was reached for at least two days. Both ChrimsonR and YFP control mice displayed significant improvement over time (RM-ANOVA, F8 = 8.738, p ≤ 0.0001 for the main effect of training days). Importantly, we found no learning rate differences between the two groups (pairwise Tukey; p = 0.420; Fig. 7d). Once mice had learned the rule associated with the DNMTS task, we assessed performance while delivering either scrambled (0.792 ± 0.045) or 8 Hz (0.850 ± 0.036) optogenetic stimulation in the encoding phase (forced choice) only, and did not observe any difference in performance compared to baseline (0.825 ± 0.030, RMANOVA, F11 = 2.197, p = 0.547, N = 17; Fig. 7e). When stimulated only during the delay period, only scrambled stimulations significantly decreased memory performance (0.733 ± 0.057) compared to baseline (0.825 ± 0.030, RM-ANOVA, F11 = 3.483, p = 0.0495; effect size for the main effect of treatment, η2 p = 0.109; Fig. 7f). In contrast, when stimulated during retrieval, mice stimulated with 8 Hz displayed significantly reduced memory performance (0.675 ± 0.049) compared to baseline (0.825 ± 0.030, RM-ANOVA, F11 = 3.265, p = 0.050; Fig. 7g). In contrast, YFP control mice were not affected by stimulations during encoding (RM-ANOVA, F2 = 0.1314, p = 0.8781), the delay period (RMANOVA, F2 = 0.2020, p = 0.8197), or retrieval (RM-ANOVA, F2 = 0.0454, p = 0.9557; effect size for the main effect of treatment, η2 p = 0.196) of working memory.

Optogenetic control of MS neurons does not alter locomotion

Importantly, it was previously reported that pacing theta oscillations could decrease locomotor speed and its variability10, which could explain at least in part the effects of optogenetic stimulations on working and episodic memory. To thoroughly assess the specificity of MS optogenetic stimulation on memory, we performed additional experiments to rule out the direct effects of the optogenetic stimulations on locomotor velocity. A subset of mice injected with ChrimsonR and implanted with fiber optics in the MS as well as LFP electrodes in CA1 were allowed to explore an open field freely while being subjected to 5 s ON, 5 s OFF optogenetic stimulations (Supplementary Fig. 6a). 8 Hz stimulation led to consistent pacing of hippocampal oscillations to that frequency (Supplementary Fig. 6b). Those stimulations were not associated with any apparent change in locomotor behavior (Supplementary Fig. 6c), including mean speed (unpaired, two-tailed ttest, t116 = 0.4140, p = 0.6796; Supplementary Fig. 6d) and speed coefficient of variation (CV; unpaired, two-tailed t-test, t116 = 0.8296, p = 0.4095, n = 59 stimulation epochs; Supplementary Fig. 6e). Similarly, scrambled stimulation consistently led to abolished theta oscillations (Supplementary Fig. 6f) but no apparent changes in locomotor behavior (Supplementary Fig. 6g), including mean speed (unpaired, two-tailed t-test, t118 = 0.2268, p = 0.8210; n = 60 stimulation epochs; Supplementary Fig. 6h) and speed CV (unpaired, two-tailed t-test, t118 = 1.838, p = 0.686; n = 60 stimulation epochs; Supplementary Fig. 6i).

While natural theta frequency and locomotor speed are correlated59, the exact direction of causation between these variables remains poorly understood. To answer this question, we performed optogenetic stimulation using a 5s ON, 5s OFF paradigm, and a random frequency was selected for each stimulation epoch (Supplementary Fig. 6j). These stimulation frequencies covered the entirety of the theta band spectrum (Supplementary Fig. 6k). As expected, we found that natural theta oscillations frequencies are directly correlated to running speed for one example mouse (R2 = 0.1114, p ≤ 0.0001; Supplementary Fig. 6l). Importantly, when theta oscillation frequency results from MS optogenetic control, the correlation drops below chance level (R2 = 0.0006779, p = 0.6244; Supplementary Fig. 6m) suggesting that locomotion dictates theta frequency, but not the opposite. We systematically replicated these results across mice and observed a drop in the correlation between theta frequency and locomotor speed under optogenetic stimulation control (paired t-test, t3 = 3.922, p = 0.0295; N = 4 mice; Supplementary Fig. 6n). Altogether, these results support that our optogenetic stimulation does not alter locomotor behavior, which is highly relevant for the next behavioral assays.

For more information:1950477648nn@gmail.com