MC4R Deficiency Causes Dysregulation of Postsynaptic Excitatory Synaptic Transmission as a Crucial Culprit for Obesity

Obesity is a growing global epidemic characterized by abnormal fat accumulation that can usually cause various health problems, such as hyperlipemia, hypertension, hyperglycemia, liver disease, and type 2 diabetes (1–5). Genetic association studies have revealed that the Mc4r gene–encoded melanocortin 4 receptor (MC4R) is involved in the mediation of food intake and energy homeostasis (6–8). Previous studies have shown that MC4Rs were most abundantly expressed in the paraventricular nucleus of the hypothalamus (PVH) in both humans and rodents (9,10). The MC4R expression neurons in the PVH (PVH^MC4R) mainly receive dense projections from the hypothalamic arcuate nucleus (ARC), which contains two functionally distinct populations of neurons, including the pro-opiomelanocortin (ARC^POMC)– and agouti-related peptide (ARC^AgRP)–expressing neurons and commonly regulate feeding behaviors (11,12). ARC^POMC neurons can release α-melanocyte–stimulating hormone (α-MSH), which is deemed an inhibitor of food intake. Meanwhile, ARC^AgRP neurons can release AgRP and neuropeptide Y, which are known to promote food intake (13–17). As a crucial receptor at the ARC→PVH pathway, MC4Rs can interact with α-MSH and AgRP to regulate feeding behaviors, and the loss of MC4R can cause obesity (11,18). Although many studies have shown that MC4Rs participate in modulation of food intake and energy homeostasis, to date, the precise mechanisms causing obesity still remain elusive.

Here, we report that MC4R knockdown (KD) attenuated α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)–mediated postsynaptic responses and caused rapid body weight gain but did not affect the N-methyl-d-aspartate receptor (NMDAR)–mediated postsynaptic responses and γ-aminobutyric acid receptor (GABAR)–mediated postsynaptic responses in PVH^MC4R neurons. We confirmed that the decrease of AMPAR-mediated excitatory postsynaptic current (EPSC) amplitude originated from the decline of GluA1 phosphorylated S845 (pS845), which was believed to be phosphorylated through the protein kinase A (PKA)–mediated signaling pathway. Intriguingly, specific PKA KD in PVH^MC4R neurons not only decreased AMPAR-EPSC amplitude but also led to increased body weight. These results showed that the dysregulation of AMPAR-mediated postsynaptic transmission might be one of the crucial factors for obesity. To confirm this hypothesis, we specifically decreased the AMPAR subunit GluA1 in PVH^MC4R neurons by the RNA interference (RNAi) technique and consistently observed similar phenotypes as in MC4R KD and PKA KD mice. Together, these studies present a view that the dysregulation of AMPAR-mediated excitatory postsynaptic transmission caused by MC4R deficiency is one of the crucial factors for obesity. Meanwhile, our observations elaborate a set of synaptic and molecular mechanisms of how body weight gain is regulated by MC4R in the PVH.

The animal housing conditions and all experimental procedures in this study were in compliance with the guidelines of the institutional animal care and use committee of the Institute of Biophysics, Chinese Academy of Sciences. ∼6–7-week-old male mice on a C57BL/6N background were used for behavioral studies. For all behavioral studies, mice were singly housed and fed a standard rodent diet. After 3–4 weeks of virus injection, the mice were used for electrophysiological studies.

The recombinant adeno-associated viruses (AAVs) were composed of biscistronic expression of shRNA driven by the mouse RNA polymerase III U6 promoter. The following sequences were selected for targeting specific gene mRNA: control shRNA sense strand, 5′-TTCTCCGAACGTGTCACGT-3′; MC4R-specific shRNA sense strand, 5′-GGTCGGAAACCATCGTCAT-3′; PKA catalytic subunit β-specific shRNA sense strand, 5′-GCCCAGATTGTGCTAACAT-3′; and GluA1-specific shRNA sense strand, 5′-GGAAGCTCTCATTAGCATTAT-3′. The vector construction and virus packaging were completed by OBiO Technology (Shanghai) Corp.

Stereotaxic injections were performed as previously described (14). Male mice (6–7 weeks old) were deeply anesthetized with an intraperitoneal injection of tribromoethanol (125–250 mg/kg) (19). The mice received bilateral microinjections of AAVs expressing shRNA against mc4r (National Center for Biotechnology Information [NCBI] Gene ID: 17202), Prkacb (NCBI Gene ID: 18749), or Gria1 (NCBI Gene ID: 14799) in the PVH. Stereotactic coordinates were determined according to the distances from the bregma: anterioposterior, −0.7 mm; mediolateral ±0.18 mm; and dorsoventral, −4.75 mm from the brain surface. ∼ 300 nL of viral vector suspension was injected using a syringe pump (SP101IZ; World Precision Instruments). We also reduced the virus volumes to 40–50 nL for each unilateral injection and then calculated the body weight gain. Stereotaxic injection sites were validated by observing the florescence in the PVH.

Food intake studies were performed as previously described (20). All mice were singly housed and fed a standard rodent diet. The mice and food were weighed every 3 days after virus injection. All measurements were randomized and blind to the experimenter. Control mouse data were applied across experiments for the same experimental conditions.

Brain slices were prepared from control and specific gene KD mice (i.e., MC4R KD, PKA KD, and GluA1 KD mice) after 3–4 weeks of virus injection. Coronary brain slices (300-μm thickness) were cut in ice-cold cutting solution containing 72 mmol/L sucrose, 83 mmol/L NaCl, 26 mmol/L NaHCO₃, 22 mmol/L glucose, 2.5 mmol/L KCl, 1 mmol/L NaH₂PO₄, 5 mmol/L MgCl₂, and 1 mmol/L CaCl₂ (∼325 mOsm) on a vibratome (Leica, Nussloch, Germany). Slices were transferred to artificial cerebrospinal fluid (aCSF) containing 126 mmol/L NaCl, 21.4 mmol/L NaHCO₃, 10 mmol/L glucose, 2.5 mmol/L KCl, 1.2 mmol/L NaH₂PO₄, 1.2 mmol/L MgCl₂, and 2.4 mmol/L CaCl₂ (∼300 mOsm) and incubated at 34°C for 30 min. After recovery, brain slices in aCSF were kept at room temperature (24–26°C) for at least 30 min before electrophysiological recording.

Electrophysiological whole-cell patch-clamp recordings were made from PVH neurons using an EPC10 amplifier (HEKA Elektronik, Lambrecht, Rhineland-Palatinate, Germany). All electrophysiological experiments were performed at room temperature (24–26°C). Miniature EPSCs (mEPSCs) and evoked EPSCs were recorded with the pipette solutions as previously described (21). Miniature inhibitory postsynaptic currents (mIPSCs) and IPSCs were recorded with an internal solution containing 140 mmol/L CsCl, 1 mmol/L BAPTA, 10 mmol/L HEPES, 5 mmol/L MgCl₂, 5 mmol/L Mg-ATP, 0.3 mmol/L Na₃GTP, 10 mmol/L QX314 (∼290 mOsm) in aCSF solution. For IPSCs, 50 μmol/L D-(-)-2-amino-5-phosphonopentanoic acid and 10 μmol/L 6-cyano-7-nitroquinoxaline-2, 3-dione were applied in aCSF. For mIPSCs, an additional 1 μmol/L tetrodotoxin was needed. During mIPSC and IPSC recordings, neurons were clamped at −70 mV.

AMPAR-EPSCs were recorded in voltage-clamp mode at a holding potential of −70 mV. NMDAR-EPSCs were recorded at 40 mV, and the NMDAR-mediated current amplitudes were quantified at 50 ms after the stimulus pulse. Paired-pulse ratio (PPR) was measured at 50-ms intervals.

Western blot was performed as previously described (22). The PVH region was dissected from the slices. The dissected brain tissues from PVH were lysed in ice-cold radioimmunoprecipitation assay buffer. ∼30–50 μg total protein mixture was loaded per lane. Protein samples were separated by 8% SDS-PAGE and blotted onto polyvinylidene fluoride membrane (Millipore). Antibodies were used to probe these proteins (Table 1). The Western blots were scanned by SMART 3.0 and analyzed by ImageJ software.

Electrophysiological data were analyzed offline with Igor 6.2 software (WaveMetrics, Portland, OR). All data are presented as mean ± SEM. Body weights and food intake were analyzed by two-way ANOVA. Unless specifically mentioned, statistically significant differences were evaluated using the two-tailed unpaired Student t test, with P < 0.05 considered significant.

The protocol and statistical code are available from corresponding author S.Z.

Because MC4Rs in the PVH play an important role in modulating food intake and energy homeostasis, PVH^MC4R neurons were used to systematically analyze their functional mechanisms. We injected the AAV vector AAV8-CMV-mCherry-U6-MC4R shRNA into the PVH of 6–7-week-old mice to target against MC4R expression; the AAV8-CMV-mCherry-U6-NT shRNA vector encoding nontargeting shRNA served as the negative control (Fig. 1A). The efficiency of AAV-delivered shRNAs was validated by quantitative RT-PCR. The data showed that 73.78% of MC4Rs in the PVH were knocked down (Fig. 1B). In subsequent experiments, we focused first on the effect of MC4R KD on food intake and body weight. MC4R KD mice had increased food intake and accelerated body weight gain starting from 12 days of virus injection (Fig. 1C–E). We also reduced the virus volumes to 40–50 nL for each unilateral injection, and then investigated the body weight gain between the control and MC4R KD mice. Similarly, there was a significant difference between control and MC4R KD mice after 18 days of virus injection (Supplementary Fig. 1A). As MC4R deficiency usually causes hyperlipemia, we also examined serum triglycerides after 18 days of virus injection. Consistent with their excess body weight, the serum triglyceride levels in MC4R KD mice were remarkably increased (Fig. 1F). Considering that prolonged hyperlipemia usually is accompanied by pathological changes of livers, we further examined these features by hematoxylin-eosin staining after 2 months of virus injection. As expected, the typical hepatic fat accumulation was observed in MC4R KD mice (Fig. 1G). Together, these morphological observations suggest that MC4R KD in the PVH can cause marked overweight accompanied by pathological changes of multiple organs, which displayed essentially identical phenotypes as previously reported MC4R-null mice (11).

Considering that disordered excitatory and inhibitory synaptic transmission might lead to acute or chronic alterations in energy homeostasis, we investigated whether MC4R KD also affected synaptic transmission (23). Before studying the synaptic responses, we first examined the existence of AMPAR, NMDAR, and GABA type A receptor (GABA_AR) in the PVH by electrophysiological and pharmacological methods (Supplementary Fig. 2). As expected, these three types of receptors could be observed in the PVH. Next, we performed whole-cell voltage-clamp recordings to investigate the mEPSCs in PVH^MC4R neurons in both control and MC4R KD mice. However, the mEPSC amplitudes and frequencies showed no significant difference. On further analysis of the mEPSC kinetics of rise time and half width, we did not observe significant differences in either (Fig. 2A–E). We then investigated whether MC4R KD affected the evoked excitatory postsynaptic responses. Of interest, we observed a dramatic decrease (52.90%) in AMPAR-EPSC amplitude in MC4R KD mice, but the AMPAR-EPSC kinetics of rise time and half width showed no statistical alteration (Fig. 2F–J). To rule out a potential presynaptic effect owing to MC4R KD in PVH^MC4R neurons, we further analyzed the PPRs at 50 ms intervals, which could reflect the evoked vesicle release probability. MC4R KD had no effect on PPR (Fig. 2K and L) together with mEPSC frequency results (Fig. 2C), suggesting that MC4R KD had no effect on presynapses.

At central synapses, NMDARs also participate in mediating excitatory postsynaptic responses (24). To investigate whether MC4R KD also affected NMDAR-mediated synaptic transmission, we recorded the NMDAR-EPSCs at a holding potential of 40 mV. However, there was no statistical difference between control and MC4R KD mice (Fig. 2M and N). In brief, these observations demonstrated that MC4R KD only affected evoked AMPAR-mediated postsynaptic responses but did not affect the NMDAR-mediated postsynaptic transmission in PVH^MC4R neurons.

The relative contribution of AMPAR and NMDAR to excitatory synaptic currents is believed to play a role in synaptic integration and plasticity (24,25). To further confirm the effect of MC4R KD on excitatory synaptic transmission, we examined elicited NMDAR- and AMPAR-mediated postsynaptic responses in the same PVH^MC4R neurons and analyzed the ratio of AMPAR-EPSCs to NMDAR-EPSCs. We found that MC4R KD resulted in a significant decrease in the AMPAR/NMDAR ratio in agreement with the decrease in AMPAR-mediated postsynaptic responses (Fig. 2H and O). Overall, these results demonstrate that MC4R KD reduces AMPAR-EPSCs but does not change NMDAR-EPSCs in PVH^MC4R neurons. We infer that MC4R might affect the EPSCs by regulating AMPAR numbers and/or activity but do not affect the properties of NMDAR.

Since inhibitory GABAergic ARC^AgRP neurons can project to PVH^MC4R neurons and promote food intake (14,26), it poses another question of whether MC4R KD also affects GABA_AR-mediated inhibitory responses. Thus, we detected the mIPSCs and evoked IPSCs in PVH^MC4R neurons. Measurements of mIPSC amplitude and frequency displayed no difference between control and MC4R KD mice (Fig. 2P–R). Furthermore, analysis of the elicited GABA_AR-mediated IPSCs demonstrated no significant difference in amplitude either (Fig. 2S and T). These results suggest that MC4R KD did not affect the GABA_AR-mediated inhibitory synaptic transmission. We believe that MC4R may not affect the activity or numbers of GABA_ARs in PVH^MC4R neurons.

In mammalian brain, the activity and number of postsynaptic receptors are closely associated with synaptic strength (27,28). Our findings revealed that MC4R KD attenuated the AMPAR-mediated postsynaptic responses but did not affect the NMDAR- and GABA_AR-mediated postsynaptic responses. Therefore, we infer that the number and/or activity of AMPAR may be affected in MC4R KD neurons. To address this issue, we first examined total expression levels of the four AMPAR subunits (GluA1, GluA2, GluA3, and GluA4). However, quantitative Western blot results showed no significant difference (Fig. 3A and B). Meanwhile, we also investigated the total expression levels of the main NMDAR subunits (GluN1, GluN2A, and GluN2B) and GABA_AR (Fig. 3C). Likewise, there was no significant difference (Fig. 3D). Thus, we infer that MC4R KD might affect the activity of AMPARs by phosphorylating specific sites. By virtue of reported MC4R-mediated intracellular signaling pathways, quantitative Western blotting was performed to specially detect the phosphorylation levels of crucial sites, including GluA1 S845 and S831 and GluA2 S880 (Fig. 3E). Our data demonstrated that MC4R KD had no impact on GluA1 pS831 and GluA2 pS880 but significantly reduced the level of GluA1 pS845, which was phosphorylated by PKA (Fig. 3F). These findings indicate that MC4R might regulate the AMPAR-mediated synaptic transmission by modulating phosphorylation of GluA1 S845.

Previous studies showed that MC4R regulated hippocampal synaptic plasticity by modulating GluA1 trafficking through GluA1 pS845 in a Gα_s-cAMP/PKA–dependent manner (29). It motivated us to ask whether PKA also participated in feeding behaviors by modulating AMPAR-mediated postsynaptic transmission in PVH^MC4R neurons. To address this question, we applied the RNAi method to target against PKA catalytic subunit-β mRNA by injection of AAV8-CMV-mCherry-U6 PKA shRNA into the PVH. Likewise, Western blot analysis was first performed to quantify the efficiency of shRNA-mediated PKA KD. In addition, protein expression levels, including total GluA1, GluA1 S845, and GluA1 S831, were detected (Fig. 4A). As expected, the PKA and GluA1 S845 expression levels were significantly decreased, but total GluA1 and GluA1 S831 expression were unchanged (Fig. 4B).

Next, we focused on examining elicited AMPAR-EPSCs. While PKA KD did not change the PPR (50-ms stimulus interval), it led to obvious decreases of AMPAR-EPSC amplitude (Fig. 4C–F). Meanwhile, our behavioral results showed that PKA KD in the PVH markedly accelerated body weight gain after 15 days of virus injection (Fig. 4G and H and Supplementary Fig. 1B). In addition, serum triglyceride concentration significantly increased in PKA KD mice (Fig. 4I). All these characterizations in PKA KD mice were similar to MC4R KD mice. Briefly, our data support that PKA KD attenuates AMPAR-mediated postsynaptic responses by regulating GluA1 pS845 and leads to rapid body weight gain. We thus suspect that the dysregulation of AMPAR-mediated postsynaptic transmission might be the crucial cause for obesity in MC4R KD mice.

To investigate whether MC4R affects excitatory synaptic transmission through the PKA-mediated signaling pathway, we used input/output measurements to record the AMPAR-EPSCs in the same brain slices before and after incubation with MC4R agonist α-MSH. Our data demonstrated that the evoked AMPAR-EPSC amplitudes were significantly increased after α-MSH incubation for 3 h (Fig. 5A). However, incubating with MC4R antagonist SHU9119 had no effect on the evoked AMPAR-EPSC amplitudes (Fig. 5B). In addition, we examined the PPR at 50-ms intervals before and after incubation with α-MSH or SHU9119, and there were no significant differences after either treatment (Fig. 5D and E).

To further elucidate the correlation between MC4R and PKA, we examined the synaptic response in PKA KD mice before and after incubation with α-MSH. As expected, α-MSH had no effect on the AMPAR-EPSC amplitudes in PKA KD mice (Fig. 5C). Similarly, the PPR at 50-ms intervals displayed no difference before and after α-MSH incubation in PKA KD mice (Fig. 5F). We assumed that α-MSH ineffectiveness in PKA KD mice might originate from the decrease of GluA1 phosphorylation due to impairment of the PKA-mediated signal pathway. To further support our hypothesis, Western blots were performed to investigate the levels of GluA1 pS845 before and after incubation with α-MSH or SHU9119. Consistent with our electrophysiological results, we found that levels of GluA1 pS845 were significantly enhanced in control mice after α-MSH incubation rather than after SHU9119 treatment (Fig. 5G and H). In addition, incubation with α-MSH had no effect on levels of GluA1 pS845 in PKA KD mice (Fig. 5I). Together, our results support that the MC4R could modulate excitatory synaptic transmission through the PKA-GluA1 pathway in the PVH.

To further confirm the above hypothesis, we performed stereotactic injection of AAV8-CMV-mCherry-U6 GluA1 shRNA to target against GluA1 mRNA in the PVH with the intention to artificially attenuate AMPAR-mediated postsynaptic responses. Quantitative Western blots showed that the expression level of GluA1 significantly decreased (Fig. 6A and B). Although GluA2, GluA3, and GluA4 subunits demonstrated a slightly compensatory increase in MC4R KD mice, there were no significant differences (Fig. 6A and B). Further electrophysiological and behavioral results showed that GluA1 KD in the PVH also led to an obvious decrease of AMPAR-EPSC amplitude, overweight, and increased serum triglyceride concentration, which were consistent with our hypothesis (Fig. 6C–I and Supplementary Fig. 1C). Overall, our study suggests that the dysregulation of postsynaptic transmission was caused by MC4R-mediated alteration of GluA1 pS845, which could be considered a crucial pathogeny of obesity (Fig. 7).

As a crucial receptor in the ARC→PVH neural circuit, MC4R plays an important role in maintaining energy homeostasis (7,14,30). In this study, we demonstrated that MC4R affected neurotransmission by regulating AMPAR phosphorylation through the PKA-mediated signaling pathway. MC4R KD in PVH^MC4R neurons caused AMPAR-mediated neurotransmission dysfunction and led to aberrant food intake and body weight gain. Our study systemically revealed the functional mechanism of MC4R and provides a synaptic and molecular explanation of how body weight is regulated by MC4R in the PVH. It is noteworthy that the MC4R-mediated intracellular PKA signaling pathway could be a promising therapeutic target for obesity and associated diseases.

Previous studies reported that MC4R could modulate synaptic plasticity in the hippocampus by regulating GluA1 pS845 through the PKA-dependent pathway (29,31). Glucagon-like peptide 1 (GLP-1) receptors in the PVH can also affect food intake by the same PKA-dependent signaling pathway. Interestingly, MC4R KD mice in our study also suffered from obesity because of deceased levels of GluA1 pS845 through the PKA-mediated signaling pathway (20). Thus, we speculate that the MC4R-PKA-GluA1 S845 pathway may play a crucial role in modulating food intake by regulating excitatory neurotransmission in the ARC→PVH neural circuit. However, it is known that PVH^MC4R neurons receive dense projections from ARC^POMC and ARC^AgRP neurons (14,32). ARC^AgRP neurons not only release AgRP/neuropeptide Y, but also release the inhibitory neurotransmitter GABA (17,33,34). Although our findings showed no significant changes on GABA_AR-mediated inhibitory synaptic transmission in the PVH, we still could not exclude the potential effects of GABA_ARs in modulating feeding behaviors. So, what is the primary role of GABA_AR-mediated neurotransmission in the ARC→PVH neural circuit? It is possible that GABA released from ARC^AgRP neurons could directly inhibit the activity of ARC^POMC neurons as previously reported (35,36). On the other hand, potential GABA_AR-related signaling pathways might exist in the PVH that can bypass MC4R and modulate the synaptic transmission. In addition, the PVH contains heterogeneous cell populations, including PVH^Glp1r, PVH^MC4R, PVH^Oxt, and PVH^CRH, which can release different neuropeptides and affect feeding behavior or body weight by modulating downstream brain regions (37). Thus, the related underlying functional mechanisms should be complex, multiform, and confluent.

In this study, we focused on the function of MC4R in the ARC→PVH neural circuit. However, the PVH also constructs the connection with other brain regions, such as the dorsomedial nucleus of the hypothalamus, hindbrain nucleus tractus solitarius neurons, and catecholaminergic neurons in the brainstem (20,38). All these nuclei were proven to participate in regulating food intake and energy homeostasis. For example, nucleus tractus solitarius GLP-1–producing neurons send robust projections to the PVH region. The depletion of corresponding GLP-1 receptors in the PVH can also lead to a decreased level of GluA1 pS845 through the PKA-dependent signaling pathway, which promotes food intake and causes obesity (20). In addition, the depletion of vesicular GLUT-2 in MC4R-bearing Sim1 neurons in the PVH region could cause obesity, which has been attributed to the decrease of glutamate release (39). All these results strongly suggest that the attenuation of excitatory synaptic transmission is a crucial factor for obesity. Here, we propose that the dysregulation of AMPAR-mediated excitatory synaptic transmission in the PVH is a crucial pathogeny of obesity. Nevertheless, it is puzzling that not all excitatory synaptic transmission dysregulation can cause obesity. We speculate that multiple neural circuits may jointly maintain neuronal homeostasis and finally achieve metabolic balance or that the existence of undiscovered intracellular signaling pathways may participate in modulating downstream or upstream synaptic strength. Hence, the effect from a single factor is weakened.

It is noteworthy that two neural melanocortin receptors MC3R and MC4R, coexpressed in the PVH region. MC3R was expressed presynaptically in AgRP projections to MC4R neurons in the PVH (14), which could regulate GABA release from AgRP neurons onto MC4R target sites, exerting boundary control on the activity of MC4R neurons (40,41). In our study, we knocked down the MC4R in the PVH, which might have impaired the boundary control of MC3R.

In brain, α-MSH and AgRP are deemed, respectively, as the natural agonist and antagonist of the MC4R, which commonly maintain the energy homeostasis. However, some studies reported that constant agonist treatment could cause MC4R tachyphylaxis by upregulation of AgRP mRNA levels and desensitization (42,43). Thus, we speculate that the activation of MC4Rs in the PVH might be flexibly adjustable according to physiological and nutritional demands. In addition, it is known that α-MSH augments cAMP levels and that the antagonist (SHU9119) displays opposite effects (44,45). Previous studies reported that the increase of the intracellular cAMP levels induced by NDP-α-MSH disappeared after coapplication with SHU9119. However, the sole application of SHU9119 only showed a tendency of cAMP decrease (45,46). Combined with our electrophysiological results, we speculate that SHU9119 might have a long-term effect on modulating synaptic transmission through regulating intracellular cAMP levels.

While we propose that AMPAR-mediated excitatory postsynaptic transmission dysregulation caused by MC4R KD is one of crucial pathogeny of obesity, further studies are still required to delineate whether other presynaptic or postsynaptic aberrations can also cause a similar phenotype as MC4R KD. In addition, we used the RNAi method to knock down PKA and GluA1, which caused an increase in body weight, but off-target effects of the shRNAs in our study could not be excluded. Meanwhile, compensation responses about other AMPAR subunits or isomers during the KD process also should be considered. Furthermore, the PVH contains multiple heterogeneous cell populations, which may modulate one another. Thus, further functional studies for concrete regulatory mechanisms are of great interest.

Acknowledgments. The authors thank Dr. Xinhua Qiao, Erzhong Wu, and Ruijin Zhang of the National Laboratory of Biomacromolecules, Key Laboratory of RNA Biology, Institute of Biophysics, Chinese Academy of Sciences, for technical assistance. The authors also thank Yanrulin Li at State Key Laboratory of Brain and Cognitive Sciences, Institute of Biophysics, Chinese Academy of Sciences, and Kexuan Duan of the Neurobiology and Behavior Department, Cornell University, for critical reading of the manuscript.

Funding. This work was supported by the Ministry of Science and Technology of the People’s Republic of China (2021ZD0203800), CAS Key Laboratory of Brain Connectome and Manipulation (2019DP173024), Instrument Developing Project of the Chinese Academy of Sciences (YJKYYQ20180028), National Natural Science Foundation of China (32070987 and 31871033), and Strategic Priority Research Program of Chinese Academy of Sciences (XDB37030303).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Xi.W., X.C., and F.L. contributed to the investigation, methodology, and writing of the original draft of the manuscript. Ya.L., Yu.L., J.D., H.H., and Xu.W. contributed to the methodology. J.S. and Y.Y. contributed to the supervision and review and editing of the manuscript. S.Z. contributed to the conceptualization, supervision, investigation, and review and editing of the manuscript. S.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.