Introduction

Neuropeptides are expressed and secreted throughout the mammalian brain where they play key roles in modulating neuronal activities and behaviors (Swaab 1982; Merighi et al., 2011; De Wied 1997). Long-term potentiation (LTP) and long-term depression (LTD) have been considered cellular mechanisms correlated with learning and memory in the central nervous system (CNS) (Grasselli and Hansel 2014). Hippocampus is a unique brain region that is critically involved in learning and memory (Ranganath and Hsieh 2016), and how neuromodulations affect the function of the hippocampal system has attracted lots of interest (Gedankien et al., 2023; Broussard et al., 2016; Li and Gao 2016; Karayol et al., 2021).

During the past decades, many studies have identified the neuronal mechanisms that guide neuropeptides across the hippocampal function and examined whether neuropeptides undergo specific adaptations in response to external innervation. However, most investigations focus on the classical monoamine neuromodulators, including acetylcholine (Ach), dopamine (DP), norepinephrine (NE), serotonin (ST), etc (Gedankien et al., 2023; Broussard et al., 2016; Li and Gao 2016; Karayol et al., 2021). Ach can regulate neuronal excitability and synaptic transmission in the mammalian brain (Takács et al., 2018). Several studies have demonstrated that stress increases Ach release in a brain region-specific manner (Picciotto et al., 2012; Mineur et al., 2013). For example, stress-induced increase in Ach level in the rat hippocampus and cortical area. DP is a critical modulator in neuronal circuitry, which has been shown to be involved in a variety of behavioral phenomena in the hippocampus, such as episodic memory formation (Chowdhury et al., 2012), spatial learning (Kempadoo et al., 2016), and synaptic plasticity (Hamilton et al., 2010). DP innervation from the midbrain is mainly via the dopamine receptor (D1/D5) when it is released into the dorsal hippocampus (Cai and Ford 2018). Additionally, the role of NE in memory retrieval requires signaling through the β1-adrenergic receptor in the hippocampus (Chang et al., 2011). The hippocampus has one of the denser inputs of adrenergic terminals (containing NE) in the CNS, indicating that the adrenergic system plays a role in learning and memory (Goodman et al., 2021). Interestingly, high concentration of serotoninergic fibers in the forebrain is in stratum lacunosum-moleculare (SLM) of hippocampal areas CA1 and CA3, where the axons of layer III neurons in the entorhinal cortex form excitatory synapses with the distal apical dendrites of pyramidal cells (Cai et al., 2013). This temporoammonic pathway is required for some spatial recognition tasks and for long-term consolidation of spatial memory. All these studies implied that neuropeptides play a critical role in the hippocampal system and influence subsequent behaviors.

Cholecystokinin (CCK), one of the most abundant neuropeptides in the CNS (Ma and Giardino 2022), has not received much attention and its function in the hippocampal system has been generally underestimated. Although several reports have shown that CCK enhances the excitatory synaptic transmission in the hippocampus and improves learning performance (Wei et al., 2013; Reisi et al., 2015). However, the exact mechanism by which CCK regulates the hippocampus plasticity and hippocampus-dependent behaviors has not yet been fully elucidated. To address this knowledge gap, we adopted the transgenic mice, optogenetics, GPCR-based sensor, extracellular recording, chemogenetics, RNA interference technique, calcium recording and behavioral task to investigate: 1) the distribution profile of CCK-positive neurons in the CA3 area of dorsal hippocampus (DHP), 2) activation of CA3^CCK neurons secrets the CCK at hippocampal SC-CA1 synapses. 3) the causal relationship between the Ca²⁺ response of excitatory CA3^CCK neurons and hippocampal functions. This study should bring conceptual advances to the foundation of neuropeptide modulation and the studies of hippocampus-dependent learning and memory.

Results

Distribution profile of CCK-positive neurons in the DHP

Although many studies have well documented the role of inhibitory CCK in the hippocampus (Klausberger et al., 2005; Ali 2007; Hefft and Jonas 2005), the function of excitatory CCK in the hippocampus is still unclear. To address this issue, we developed the transgenic CCK Cre::Ai14 (Tdtomato) reporter mice that express the red fluorescent protein tdTomato upon Cre-mediated recombination. This mouse line is suited in examining the distribution profile of CCK positive neurons in the dorsal hippocampus (DHP) (Figure 1A-B). Interestingly, we found that the proportion of CCK positive neurons in area CA1 and CA3 of CCK Cre/Ai14 mice was around 8.50 % and 20.34 %, respectively (Figure 1C-D). We further examined the properties of these CCK neurons in areas CA3, which may send dense CCK⁺ projections to the area CA1. Interestingly, excitatory neuronal marker CamkIIα was intensively colocalized with Tdtomato expressing CCK neurons in the area CA3 (Figure 1E). Moreover. the CCK proteins also show high co-localization in excitatory neurons in CA3 area (Figure 1F).

Figure 1.

Furthermore, to validate the projection of CA1-projecting CA3^CCK neurons in the DHP, we injected the Cre-dependent retrograde axonal transport of AAV (Retro-AAV-EF1α-DIO-EYFP) into the area CA1 (Figure 1G). Unsurprisingly, CA3 neurons were widely labelled with green fluorescent protein (GFP; Figure 1H). Moreover, high degree of colocalization between CamKIIα marker and GFP was observed in the area CA3 (Figure 1I-J; 90.15± 15 %). These anatomy results suggest that excitatory CA3 neurons are densely expressed and may exerts neuronal functions in central nervous system.

Excitatory CA3 neurons can secret the neuropeptide CCK

Next, to investigate the neurotransmission property of the excitatory CCK neurons in CA3 area (CA3^CCK), we utilized the Cre-dependent excitatory AAV (AAV9-DIO-CamKIIα-ChrimsonR-mCherry) to specifically target the excitatory CA3^CCK neurons in CCK-Cre mice (Figure 2A). Four weeks after the AAV injection and expression in vivo, robust expression of ChrimsonR-mCherry was observed in CA3^CCK neurons (Figure 2B-C). Subsequently, we used the 636 nm wavelength light to activate and elicit the field excitatory post synaptic potentials (L-fEPSPs) in ChrimsonR-mCherry expressing brain section (Figure 2D). Intriguingly, we noticed that the L-fEPSPs were almost completely blocked by the AMPA receptor antagonist (CNQX; 20 µM) and NMDA receptor antagonist (APV; 100 µM) in the area CA1, suggesting an excitatory nature of CA1-projecting CA3^CCK pathway (Figure 2E-F).

Figure 2.

To determine whether the neuropeptide CCK can be secreted from the CA3 derived excitatory CA3-CA1 projections, we adopted the GPCR-based CCK-BR sensor to monitor the release of CCK from the CA3-CA1 terminals in CA1 area under the external light stimulation (Figure 2G). To achieve this goal, AAV9-DIO-CamKIα-ChrimsonR-mCherry and AAV9-hSyn-CCK-GFP-2.3 (AAV9-hSyn-GFP as control) were injected into the CA3 and CA1 area to target the CA3-CA1 projection and CA1 pyramidal neurons (CA1 PNs) in CCK-Cre mice (Figure 2H). Subsequently, stimulation fiber and recording fiber were implanted upon the CA3 and CA1 area. Four weeks after the AAVs injection and expression (Figure 2I-J), high-frequency theta burst light stimulation (L-TBS) was delivered to the simulation fiber (Figure 2K), which mimic the neural rhythm in hippocampus in vivo. Interestingly, transient increase in CCK sensor was monitored by L-TBS of ChrimsonR-expressing CA3^CCK neurons, while no significant amplification was found in control (Figure 2L-N; Two samples T-Test, t = 3.79 df = 10, p = 0.004). These results indicates that neuropeptide CCK can be secreted from the excitatory CA3^CCK neurons under the mimic physiological conditions.

CA3^CCK neurons fire actively during hippocampal-dependent tasks

We next wonder about the activity status of CA3^CCK neurons during hippocampal-dependent tasks (Figure 3A). Thus, we injected the AAV9-CamkIIα-DIO-GCaMP6s into the area CA3 of CCK Cre mice to monitor the calcium-response of CA3^CCK neurons during the behavioral task (Figure 3B). In novel object location (NOL) task, mice spent comparable time on object exploration between object 1 (O1) and object 2 (O2) in the training phase, while mice spent significantly more time on interacting with object in the novel place (O2”) compared with the object in the familiar site (Figure 3C; two way mixed ANOVA, Bonferroni adjustment; F_1,10 = 4.58, p = 0.05; Training: O1 11.42 ± 1.56 s v.s. O2 10.62 ± 2.18 s, p = 0.73; Testing: O1 8.62 ± 0.99 s v.s. O2”: 13.95 ± 1.60 s, p = 0.02). Concurrently, we examined the Ca²⁺ responses of excitatory CA3^CCK neurons during object exploration. Interestingly, we observed that object exploration elicited a significant increase in fluorescence intensity in both training phase and testing phase (Figure 3D-E). Moreover, Ca²⁺signal of CA3^CCK neurons elicited by novel locations also show significant difference compared with familiar locations (Figure 3F; two way mixed ANOVA, Bonferrori adjustment; F_1,10 = 1.88, p = 0.20; Training (Δ F/F): O1 0.27 ± 0.07 % v.s. O2 0.31 ± 0.08 %, p = 0.91; Testing (Δ F/F): O1 0.27 ± 0.09 % v.s. O2” 0.41 ± 0.12 %, p = 0.04), which implies that excitatory CA3^CCK neurons are critical for specific spatial memory and learning.

Figure 3.

Furthermore, we examined whether the excitatory CA3^CCK neurons are necessary for performing the MWM paradigm (Figure 3G). Intriguingly, in the naive training (trial 1), a robust and persistent Ca²⁺ responses were observed during the initial learning (Figure 3H and Figure 3I). Nevertheless, after the completion of training, CA3^CCK neurons produced a relatively smaller Ca²⁺ signal on the training day 3 with an averaged escape latency of < 10 s to locate the hidden platform (Figure 3H and Figure 3J; Figure 3K: Paired sample t-test, df = 5, t = 2.80; Trial 1: 1.32 ± 0.74 % v.s. Trial 9: 0.75 ± 0.35 %, p = 0.038), this result further supports the conclusion that excitatory CA3^CCK neurons plays a direct role in the animals’ accurate spatial learning and memory formation.

Chemogenetic inhibition of the excitatory CA3^CCK-CA1 pathway impairs behavioral tasks

Next, to further confirm the necessity of excitatory CA3^CCK-CA1 pathway for performance of the hippocampal-dependent behaviors. We adopted the chemogenetic approach, which is a powerful technique for specific disturbance of neuronal activity through virus-mediated DREADD expression in combination with its agonist clozapine N-oxide (CNO) (Gomez et al., 2017), to specifically target the excitatory CA3^CCK neurons. Firstly, we injected the Cre dependent AAV expressing the inhibitory DREADD (hM4DGi) to infect the excitatory CA3^CCK neurons and AAV carrying mCherry as control (Figure 4A) and implanted the drug cannular upon the CA1 area. Four weeks after the AAV injection and expression (Figure 4B), we conducted the MWM task in the two groups of mice to assess the role of CA1-projecting CA3^CCK neurons in hippocampus-dependent spatial learning (Figure 4C). Thus, we delivered the CNO (300 nl; 10 µM) via the cannulas to inhibit CCK⁺ CA3-CA1 projections upon the area CA1 before conducting the MWM. Then, two groups of mice were subjected to the visible platform task to evaluate their swimming capability. Both groups of mice displayed comparable swimming speed and the speed time in finding the visible platform (Figure 4D; two sample t-test, df = 18, t = 0.22, HM4D(Gi): 24.3 ± 1.66 cm/s v.s. mCherry: 23.8 ± 1.52 cm/s, p = 0.83. Figure 4E; two sample t-test, df = 18, t = 1.09, HM4D(Gi): 30.86 ± 2.587 s v.s. mCherry: 34.21 ± 2.12 s, p = 0.29). During the hidden platform task, the mice only expressing mCherry gradually spent less time in locating the platform compared to the mice infected with hM4D(Gi) (Figure 4F; two-way mixed ANOVA, Bonferroni adjustment; F_4,15 = 1.24, p = 0.34; HM4D(Gi): 24.18 ± 4.36 s v.s. mCherry: 15.12 ± 2.13 s on day 5, p = 0.003). Additionally, the control mice showed higher percentage on the target quadrant on the memory retention day, while lower percentage of the occupancy in experimental mice (Figure 4G-H; two-way mixed ANOVA, Bonferroni adjustment; F_3,16 = 1.19, p = 0.34; HM4D(Gi): 27.91 ± 2.75 % v.s. mCherry: 42.31 ± 3.82 % in quadrant 3, p = 0.02). These results demonstrated that excitatory CA3^CCK-CA1 pathway is required for animals to perform adequately in spatial learning.

Figure 4.

In the NOL task (Figure 4I), both groups of mice showed equal motivation to explore objects (Figure 4J; two sample t-test, df = 18, t = -0.31, hM4D(Gi): 81.18 ± 9.14 s v.s. mCherry: 85.35 ± 9.67 s, p = 0.76). Compared with the control mice, the experimental group was unable to distinguish the novel and familiar object location in the test phase, and presented low discrimination ratio (Figure 4K; two-way mixed ANOVA, Bonferroni adjustment; F_1,18 = 4.58, p = 0.04; hM4D(Gi): Familiar 40.83 ± 5.27 s v.s. Novel: 43.46 ± 7.40 s, p = 0.66; mCherry 34.38 ± 3.58 s v.s. Novel: 55.35 ± 6.11 s, p = 0.003. Figure 4L; two sample t-test, df = 18, t = 0.77, hM4D(Gi): -0.003 ± 0.09 v.s. mCherry: 0.22 ± 0.06, p = 0.03), indicating suppression of excitatory CA3^CCK-CA1 pathways impaired the spatial memory. Taken together, we can conclude that excitatory CA3^CCK neuron is an essential and functional component in the hippocampal system.

Chemogenetic inhibition of the excitatory CA3^CCK-CA1 pathway attenuates LTP formation

Since the LTP in the hippocampus is commonly considered a key cellular mechanism that underlies spatial learning and memory (Ge et al., 2010). To confirm whether excitatory CA3^CCK-CA1 pathway are directly involved in LTP formation, we further conducted the electrophysiological recording in vitro. Four weeks after the AAV injection and expression (Figure 5A-B), hippocampal slices were subjected to electrophysiological recording to test the effect of CNO on the neuroplasticity of excitatory CA3^CCK-CA1 pathway (Figure 5C).

Figure 5.

After the baseline recording, the slope of fEPSPs decreased gradually with the CNO treatment while no significant changes were observed in controls (Figure 5D-E; two-way mixed ANOVA, Bonferroni adjustment; F_1,13 = 78.32, p < 0.001; hM4D(Gi): Pre_ex 100.42 ± 0.18 % v.s. Post_ex 78.07 ± 1. 52 %, p < 0.001; mCherry: Pre_ctrl 100.15 ± 0.48 % v.s. Post_ctrl 98.90 ± 2.90 %, p = 0.60, Post_ex v.s. Post_ctrl, p < 0.001), indicating that neuronal activities of the excitatory CA1-projecting CA3^CCK neurons were remarkably suppressed. Then, we further employed the E-TBS protocol for the LTP induction under this condition (Figure 5F). Intriguingly, the extent of LTP formation in the slices expressing hM4D(Gi) was significantly smaller than those in the control group (Figure 5G-H; two way mixed ANOVA, Bonferrori adjustment; F_1,13 = 9.46, p = 0.01; HM4D(Gi): Pre_ex 99.65 ± 0.31 % v.s. Post_ex 111.52 ± 2.40 %, p < 0.001, mCherry: Pre_ctrl 100.41 ± 0.29 % v.s. Post_ctrl 125.17 ± 3.43 %, p = 0.001, Post_ex v.s. Post_ctrl, p = 0.005), hinting that CA3^CCK neurons are involved in the maintenance of hippocampal neuroplasticity.

RNA interference of excitatory CA3^CCK expression impairs hippocampal functions

Next, to verify whether neuropeptide CCK in excitatory CA3^CCK neurons are also involved in spatial learning and memory, RNA interference (RNAi) strategy was utilized to downregulate CCK expression in CA3^CCK neurons (Figure 6A). We thus injected AAV9-CamkIIα-DIO-mCherry-shRNA (CCK) into area CA3 to knockdown target gene and AAV9-CamkIIα-DIO-mCherry-shRNA (Scramble) as control. The experimental group was Cre on anti-CCK to knock down the expression of CCK in the area CA3 of CCK Cre mice, while the control group was Cre on anti-scramble. We confirmed the expression of shRNAs in the hippocampus at 4 weeks after AAV injection (Figure 6B). Moreover, we further verified that anti-CCK shRNAs faithfully downregulated CCK mRNA levels in vivo by using the qPCR technique (Figures 6C; two sample t-test, df = 10, t = 4.95, anti-scramble 100 ± 7.03 % v.s. anti-CCK 58.73 ± 7.23 %, P = 0.001). Subsequently, we conducted the MWM task in the two groups of mice to assess the role of CA1-projecting CA3^CCK neurons in the hippocampus-dependent spatial learning (Figure 6D).

Figure 6.

The two groups of mice injected with anti-CCK and anti-scramble showed comparable swimming speed (Figure 6E; two sample t-test, df = 18, t = -0.40, anti-CCK: 23.3 ± 2.75 cm/s v.s. anti-scramble: 24.1 ± 1.41cm/ s, p = 0.69) and swimming time in locating the visible platform above the water surface of the swimming pool (Figure 6F; two sample t-test, df = 18, t = -0.51, anti-CCK: 33.4 ± 2.21 s v.s. anti-scramble: 34.9 ± 1.93 s, p = 0.62). Then, mice infected with the anti-CCK showed deficits in spatial learning during the 5 days training (Figure 6G; two way mixed ANOVA, Bonferrori adjustment; F_4,15 = 0.30, p = 0.88; anti-CCK 25.48 ± 2.12 s v.s. anti-scramble 18.05 ± 3.10 s on day 5, p = 0.007), and also exhibited deficiency in memory retention (Figure 6H-I; two way mixed ANOVA, Bonferrori adjustment; F_3,16 = 1.20, p = 0.34; anti-CCK 28.38 ± 1.78 % v.s. anti-scramble 35.76 ± 3.68 % in quadrant 3, p = 0.03). These results suggest that knockdown of CCK expression in area CA3 attenuates spatial learning ability.

In another spatial learning protocol, the novel object location (NOL) task (Figure 6J), two groups of mice exhibited equal motivation to explore objects (Figure 6K; two sample t-test, df = 18, t = 0.28, anti-CCK: 83.82 ± 5.85 s v.s. anti-scramble: 81.22 ± 7.25 s, p = 0.78). Compared with the control mice (anti-scramble), the mice infected with anti-CCK failed to distinguish between the novel and familiar object location in the test phase, and displayed low discrimination ratio (Figure 6L; two-way mixed ANOVA, Bonferrori adjustment; F_1,18 = 2.97, p = 0.10; anti-CCK: Familiar 42.63 ± 5.21 s v.s. Novel: 44.87 ± 6.33 s, p = 0.67; anti-scramble: Familiar 34.41 ± 5.30 s v.s. Novel: 49.22 ± 8.87 s, p = 0.01. Figure 6M; two sample t-test, df = 18, t = -2.0, anti-CCK: 0.02 ± 0.06 v.s. anti-scramble: 0.19 ± 0.05, p = 0.04), indicating that downregulation of CCK expression in area CA3 was sufficient to impair spatial memory. Taken together, these results support our hypothesis that excitatory CCK acts as a functional neuromodulator in the hippocampal system.

Additionally, for the electrophysiology recording, E-TBS adequately induced LTP at CA3-CA1 synapses in control slices containing the anti-scramble shRNAs, but the same protocol elicited remarkably smaller LTP on slices with anti-CCK shRNA targeting the sequences of CCK gene (Figures 6N-O; two way mixed ANOVA, Bonferrori adjustment; F_1,14 = 10.06, p = 0.007; Pre_ctrl: 100.93 ± 0.57 % v.s. Post_ctrl: 127.78 ± 2.47 %, p < 0.001; Pre_ex: 100.44 ± 0.34 % v.s. Post_ex: 115.76 ± 2.96 %, p < 0.001; Post_ctrl v.s. Post_ex, p = 0.008). Moreover, we noticed that the amplitude of the electrically evoked fEPSPs in anti-CCK group is similar to those of the control group (Figures 6N, insert), suggesting normal neurotransmission in hippocampal slices from both groups. Therefore, the impairment of LTP is likely due to the reduced expression of CCK in CA1-projecting CA3^CCK terminals which impaired the neuromodulation.

Discussion

In the present study, we reported the distribution profile of CCK positive neurons in the dorsal hippocampus, a key brain region associated with spatial memory formation and transformation (Martin and Clark 2007). Although the role of inhibitory CCK neurons is well studied in the hippocampus, the function of excitatory CCK neurons is still unclear. Optogenetic activation of the CCK inhibitory populations produces a prominent IPSC and controls the input and output gain of CA1 pyramidal neurons, and this kind of modulation is mediated by the presynaptic CB1 receptors (Hartzell et al., 2018). A recent study uncovered that systemic activation of CCK-GABA neurons minimally affects emotion but significantly enhances cognition and memory (Whissell et al., 2019). In our study, we combined the Ai14::CCK Cre mice and immunochemistry method to show the distribution of CCK positive neurons in the hippocampus. Moreover, we adopted the pharmacological approach to diminish the tendency for light-evoked fEPSPs of excitatory CA1-projecting CA3^CCKneurons. Additionally, using the patch clamp technique to record the EPSP of these neurons also can directly validate this conclusion. However, we are unable to conduct whole cell recording due to the limitation of available techniques. To our knowledge, this is the first study that demonstrated the function of excitatory CCK neurons in the hippocampus.

Additionally, to fully validate the involvement of excitatory CA1-projecting CA3^CCK neurons in hippocampal processing involving learning and neuroplasticity, we used the chemogenetic technique to examine the function of CA3^CCK neurons. Interestingly, we found that specific inhibition of these neurons significantly impaired the CA1-CA3 LTP and attenuated spatial learning. Many studies have also demonstrated that neurotransmitters facilitate the hippocampal plasticity and further modulate the behavior performance (Birthelmer et al., 2003; Ohta et al., 2003; Lopes et al., 2002), because deficiency of neurotransmitters in the CNS is known to be able to decrease the extent of synaptic transmission in the CA3-CA1 pathway. For instance, Mlinar reported that endogenous 5-hydroxytryptamine (5-HT) potentiates the LTP in the hippocampus and its positive effects on cognitive performance (Mlinar et al., 2015). Also, another study used a viral-mediated approach to delete brain-derived neurotrophic factor (BDNF) specifically in CA3-CA1 pathway, and further demonstrated that presynaptic and postsynaptic BNDF are essential for LTP induction and maintenance, as well as contextual memory impairments (Mohajerani et al., 2007). In this study, we used RNA interference technique (RNAi) to fully characterize the function of CA3^CCK neurons. Intriguingly, the knock down of CCK expression significantly impaired LTP formation at excitatory CA1-projecting CA3^CCK synapses and further affected the hippocampus dependent-spatial learning and memory. Therefore, we established the essential role of CCK as an active neuromodulator that regulates hippocampal plasticity and its dependent behaviors (Figure 7).

Figure 7.

Establishing a causal link between neural plasticity and spatial learning is crucial for understanding the status of CA3^CCK neurons in the hippocampus. Our investigation verified that CA3^CCK neurons modulate hippocampal long-term plasticity in a homosynaptic manner, and their activity underlies spatial learning and memory formation. We demonstrated that the CA3^CCK projections to the CA1 region are required and sufficient to induce the LTP at CA3-CA1 synapses. In the behavioral setup, we also observed that the calcium response of CA1-projecting CA3^CCK neurons increased during the exploration. The Ca²⁺ activities were also raised throughout the MWM task. Nevertheless, perturbation of the activities of CA3^CCK neurons impaired spatial learning both in the NOL task and MWM task. It was likely that the CA3 input to the CA1 conveys spatial information. Thus, we can conclude that CA3^CCK neurons may specifically regulate the hippocampal state and plasticity during behavioral performance.

Finally, further experimental works could be undertaken to investigate the function of CA3^CCK neurons. For instance, the change of CCK concentration during the light high frequency stimulation of CA1-projecting CA3^CCK neurons by using the GPCR-based CCK sensor. Several high-sensitive GPCR-based sensors for detecting the neurotransmitters have been well developed, including the dopamine, norepinephrine, serotonin and acetylcholine. Unfortunately, evidence for CCK release in the hippocampus is not present in this study. Additionally, real time monitoring of CCK release is also critical when animals are subjected to the spatial learning. However, solving these questions still needs high-sensitive CCK sensor in the future.

Methods

Animals

Adult Cck-IRES-Cre (CCK Cre: RRID:IMSR_JAX:012706) mice and Ai14 mice (RRID:IMSR_JAX:007914) were used in this study. In behavioral experiments, all male mice were housed in a 12-h light/dark cycle and were provided food and water ad libitum. All experimental procedures were approved by the Animal Subjects Ethics Sub-Committee of the City University of Hong Kong.

Viruses

Adeno-associated virus (AAVs) were purchased from the BrainVTA, Wuhan, China: AAV9-CaMKIIa-DIO-GCaMP6S-WPRE-hGH-pA (PT-0090); AAV9-CaMKIIa-DIO-(mCherry-bGH pA-U6)-shRNA(CCK)-WPRE-hGH pA (PT-9086); AAV9-CaMKIIa-DIO-(mCherry-bGH pA-U6)-shRNA(Scramble)-WPRE-hGH pA (PT-3088); AAV9-CaMKIIa-DIO-hM4D(Gi)-mCherry-WPRE-hGH polyA (PT-1143); AAV9-Syn-DIO-mCherry-WPRE-hGH polyA (PT-0115); AAV9-hSyn-CCK sensor 2.3-GFP; AAV9-hSyn-GFP; Addgene, Cambridge, MA, USA: Retrograde AAV-EF1a-DIO-EYFP (27056). Taitool BioScience, Shanghai, China: AAV9-mCaMKIIa-DIO-ChrimsonR-mCherry-ER2-WPRE-pA (S0728-9).

AAV injection

Following deep anesthesia induced by an intraperitoneal injection of pentobarbital sodium (50 mg/kg; Ceva Sante Animale Co., France), each adult mouse was secured in a stereotaxic frame. A midline skin incision exposed the skull, and the stereotaxic apparatus was adjusted to level the skull using the bregma and lambda as landmarks. Using a nanoliter injector (Micro4 system) fitted with a quartz glass micropipette, we delivered the AAV suspension. A small craniotomy was drilled at the predetermined coordinates for the unilateral CA3 region (AP: -1.70 mm, ML: 2.35 mm, DV: -1.85 mm). The AAV was infused at a constant rate of 30 nL/min (volume and titer are specified in the respective figure legends). To prevent backflow, the micropipette remained in place for 5 minutes post-injection before careful removal. The incision was then sutured and treated with erythromycin ointment to inhibit infection. Mice recovered on a heating pad and were subsequently returned to their housing facility.

Brian slice recording

To prepare for brain slice recording, mice were deeply anesthetized with 1.4% gaseous isoflurane (Wellona Pharma) and decapitated. The brain was rapidly extracted and submerged in ice-cold, oxygenated (95% O₂ / 5% CO₂) artificial cerebrospinal fluid (aCSF) containing (in mM): 124 NaCl, 3 KCl, 1.25 KH₂PO₄, 1.25 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose (pH ∼7.4). This same aCSF solution was used for all subsequent steps, including dissection, slicing, incubation, and recording. Coronal sections (300 µm thick) containing the target region were prepared using a vibrating microtome (Leica VT1000S) and were subsequently incubated in aCSF at 32°C for recovery.

For electrophysiological recording, we used a system (Alpha MED Sciences) integrated with a light generator (Inper) to record both electrically- and optically-evoked field excitatory postsynaptic potentials (fEPSPs). Individual hippocampal slices were placed on a multi-electrode array (MEA) probe. One electrode was designated as the stimulating electrode for electrical stimulation. For optical stimulation, a 2-ms pulse of red light (635 nm) was delivered via an optical fiber (200 µm diameter) targeted to the CA1 region. After identifying stable fEPSP responses, input/output (I/O) curves were generated by plotting the fEPSP slope against varying stimulus currents or light intensities. All fEPSP slopes were normalized and analyzed using the MED Mobius software.

Real-time PCR

Tissues of the target brain region (area CA3) were dissected freshly from animals after AAV expression for 4 weeks (AAV carrying anti-CCK or anti-Scramble shRNA). The mRNAs were extracted from the bilateral area CA3 using TRIZOL reagent and reversed transcribed into cDNA with Maxima Reverse Transcriptase (Thermo). CCK gene was then amplified specifically with QuantStudio™ 3 Real-Time PCR System (ABI). The sequences of the primers: CCK-forward, 5’ - AGC GCG ATA CAT CCA GCA G - 3’; CCK-reverse, 5’ - ACG ATG GGT ATT CGT AGT CCT C - 3’. The relative expression level of the CCK gene was analyzed by using the comparative threshold cycle (C_t) technique (2 -ΔΔC_t method). Three biological replicates were performed and a reference gene (GAPDH) was used to normalize the data of gene expression in this study.

Morris water maze task

Morris water maze (MWM) task follows the Vorhee’s protocol (Vorhees and Williams 2006). Briefly, all mice were subjected to the visible platform to access their swimming ability in 122 cm radius swimming pool. Then, during the 5 days training, mice were allowed to locate the hidden platform within 60 seconds which was submerged into the water. The light intensity was set bright enough for the video recording. On day 6, the platform was removed from the target site (quadrant three) of the swimming pool, and all mice were trained to assess the spatial reference memory in the memory retention task. Lastly, the swimming-tracking paths of each mice in different quadrants were analyzed by using the MATLAB software.

Novel object location task

The novel object location (NOL) task was divided into three parts: 10 mins inhabitation for each mouse in the apparatus on day 1, and 5 mins training setup followed by a 5 mins testing setup on day 2. All mice were allowed to interact with two different objects during the training setup, after which animals were transferred into home cage. After 1 hour, mice were subjected to the testing setup in which they were put into the apparatus with the previous two objects, while one object was replaced with a new site. The following formula was carried out to analyze the discrimination index (Pérez-García et al., 2016):

Note: NOL: novel object location; FOL: familiar object location. A positive/negative value indicates more/less time on exploring the object in a novel location, while zero suggests equal time spent with both objects in familiar and novel locations.

Chemogenetic manipulation

To enable chemogenetic manipulation, CCK-Cre mice were bilaterally implanted with guide cannulas targeting the CA1 hippocampal region (AP: -1.75 mm, ML: ±1.25 mm, DV: -1.20 mm) following viral injection. The surgical protocol was identical to that described previously. Four weeks post-injection, mice underwent hippocampal-dependent memory testing (MWM and NOL tasks) and electrophysiological assessment. Prior to behavioral tasks, the experimental group (expressing hM4D(Gi)) and the control group (expressing mCherry) received a bilateral microinfusion of Clozapine N-oxide (CNO; 10 μM in 0.1% DMSO) into the CA1 region. Each site received 300 nL at a rate of 50 nL/min via a programmable injector. For in vitro slice recordings, 50 μM CNO was added to the perfusing artificial cerebrospinal fluid (aCSF) during fEPSP recording sessions.

Fiber photometry recording

A fiber-photometry system (Doric Lenses) was employed to record neural activity. The system utilized sinusoidally modulated LEDs (473 nm at 220 Hz and 405 nm at 330 Hz) to excite the GCaMP sensor and capture an isosbestic autofluorescence control signal, respectively. These light signals were conveyed through an implanted optical fiber, with the output intensity calibrated to 10 μW at the fiber tip. Emitted fluorescence was collected via the same fiber and was subsequently focused onto two independent photoreceivers (Newport Corporation, 2151). The entire process-LED control and autonomous demodulation of the 473 nm and 405 nm fluorescence signals-was managed by an RZ5P acquisition system (Tucker-Davis Technologies; TDT) with a built-in real-time processor.

For behavioral correlation, object exploration during the Novel Object Location (NOL) task was defined as the mouse approaching, sniffing, or touching an object. Furthermore, a simplified version of the Morris Water Maze (MWM) task was adapted for compatibility with photometric recordings (Qin et al., 2018). All acquired photometry data were processed using the pMAT open-source software suite (Bruno et al., 2021). The 473 nm signal, representing GCaMP activity, was normalized by fitting the 405 nm isosbestic signal to generate a motion-artifact-corrected ΔF/F trace according to the standard formula: ΔF/F = (473 nm signal - fitted 405 nm signal) / fitted 405 nm signal.

Anatomy and histology

All animals were anesthetized with sodium pentobarbital and were perfused with phosphate-buffered saline and fixed with paraformaldehyde solution. Then, mouse’s brain was detached and submerged into 4 % PFA solution for fixations at 4 °C in the refrigerator. The brains were sectioned into 40 mm-thick slices via the vibratome machine. The prepared brain sections were counter-stained with DAPI (1:10000) and mounted onto grass slides with 70% glycerol in PBS to observe the AAV expression and fiber track. Finally, the coverslips were used to cover the brain slices and sealed with adhesive glue.

For the histological procedure, brain slices were washed with 0.01 M PBS and blocked with blocking solution (15 % goat serum mixed with 0.4 % Triton X-100) at room temperature for 2-3 h. Then, the primary antibody was prepared and incubated with brain slices in 24-well cell culture plates overnight at 4 °C. Then, brain slices were washed by 0.01 M PBS and coupled to a secondary antibody at room temperature for 2-3 h. Next, brain slices were washed with PBS and were stained with DAPI. Finally, fluorescence image of brain slices was captured by using a Nikon Eclipse fluorescence microscope and a Nikon A1HD25 confocal microscope.

Statistical analysis

All statistical analyses (including two sample t-test, and two-way mixed ANOVA) were done in SPSS (IBM, USA). Statistical significance was set at p < 0.05.

Acknowledgements

We thank Prof. Jufang He for providing resources. This work was supported by funding from the following: Hong Kong Research Grants Council, General Research Fund: CityUHK 11101521, CityUHK 11103922, CityUHK 11104923, CityUHK 11104524.

Additional information

Author contributions

Conceptualization: H.F; Methodology: H.F; Investigation: H.F and A.B; Writing: H.F. and S.T.B.