View all hidden authors and organizations

Edited by Liqun Luo, Stanford University, Stanford, CA, approved on August 5, 2021 (review received on May 1, 2021)

Neurons have the special ability to grow axons and form synaptic circuits during development, but this is not the case in later life. In adults, the lack of circuit formation may help maintain proficient movement and memory, but it also limits regeneration and repair after injury and illness. Research on developing and damaged neurons has revealed many molecules that contribute to circuit formation and regeneration, but the factors that can induce axon growth and synapse formation in adult brain neurons are still elusive. Here, we searched for such key molecules and found a molecule that alone can induce the formation of a complete circuit. After designing a new circuit in adult mice, we also studied its function and its relevance to memory.

Historically, circuit formation in the central nervous system has been studied during development, after which cell-autonomous and non-autonomous wiring factors are inactivated. In principle, reactivation of the balance of these factors can allow adults to further wiring, but their relative contribution may depend on the circuit and is largely unknown. Here, we studied the sprouting of hippocampal mossy fibers to gain insight into the wiring mechanisms in mature circuits. We found that the only ectopic expression of Id2 in granulosa cells can drive mossy fiber sprouting in healthy adult mice and rats. Mice with the new mossy fiber circuit can solve spatial problems as well as the control group, but seem to rely on local rather than global spatial cues. Our results indicate that the reprogramming connections in mature neurons are through a defined factor and the assembly of new synaptic circuits in the adult brain.

Connectivity is one of the most important characteristics of the identity of neurons in the brain and spinal cord. During development, cell surface receptors and neurotrophic gradients guide the growth of axons and the formation of synaptic connections between neurons (1⇓ –3). After the developmental environment dissipated, inhibitors of neurite outgrowth were upregulated (4, 5), and the axon and dendritic structure of neurons began to build. In the adult brain, the lack of intracellular axon growth, growth support substrates, and chemical attraction not only hinder further wiring, but also hinder the realization of therapeutic rewiring after injury (6). Currently, cell-autonomous mechanisms—including those mediated by transcription factors, such as members of CREB, HDAC, ID, KLF, SMAD, STAT, and SOX family—have been able to partially generalize circuit wiring (7, 8). However, it is believed that additional unit non-autonomous mechanisms are needed to fully generalize the circuit layout (6). However, other studies have shown that under mixed developmental adult conditions, such as after adult neurogenesis (9) and after the transformation of glial/astrocytes into neurons and stem cell transplantation (10), long distances can be achieved. Axon wiring and synapses are integrated in adult circuits. This suggests that certain cell-autonomous mechanisms can overcome obstacles to further circuit formation, but the nature of such permission signals is unclear.

In order to explore the wiring mechanisms in the mature circuits of the adult brain, we studied axonal budding in the hippocampal mossy fiber (MF) system. The naive MF axons formed during normal development are derived from granular cells (GC) in the dentate gyrus and target different cell types in the phylum and CA3 (11). In contrast, during the budding period of MF, which is usually studied in the context of temporal lobe epilepsy (TLE), GCs form new axons and synapses in the dentate, thereby establishing wiring schemes during development (12⇓ –14 ). MF germination can also be induced by different methods, such as rough mechanical (15, 16), electrical (17), and chemical induction (18), or by overactivation of the mTOR pathway in the GC generated after birth (19). However, these operations lead to a variety of other changes in the network (eg, cell death, GC dispersion, abnormal GC dendritic formation, and changes in cell excitability), which to date has hindered understanding of their impact on TLE pathology (20) And the relative contribution of molecular biology. The mechanism of MF germination (21⇓ –23). Regardless of its relationship with TLE, MF budding involves all the key stages of circuit assembly (ie, axon growth, target cell specification, and synapse formation), and therefore also represents a comprehensive model for studying the wiring mechanism of the adult brain.

Our research aims to clarify the transcriptional mechanism behind MF budding and test whether it is similar to the reprogramming of differentiated cells (24) and whether the connectivity of differentiated mature GC can be reprogrammed by cell-autonomous genetic induction. Consistent with the assembly of a completely new circuit, we hypothesized that MF germination is related to extensive transcriptome changes in GC. To fully summarize these changes, we focus on transcription factors and regulators whose activation may initiate the wiring process. Using single-cell RNA sequencing (RNA-seq) screening, we identified Id2, a transcription factor inhibitor, as a master regulator, whose sole activation in GC drives MF sprouting and the formation of functional synapses. Mechanistically speaking, the activation of Id2 leads to the up-regulation of the transcriptome of molecules in the JAK/STAT and interferon signaling pathways, and controls the downstream expression of multiple wiring-related effectors. We further tested whether gene-induced MF sprouting would produce excessive excitement in the hippocampus, as hypothesized by some epileptic generation models, and studied its effects on learning and memory.

First, we used chemical induction (neural excitatory kainic acid [KA], microinjected into the hippocampus) in 2-month-old mice to study the MF germination transcriptome. Using single-cell patch RNA-seq (25, 26), we sampled mature GC (27) at 1 day (representing the acute cellular response to induction) and 14 days after induction (27), when MF will germinate, but Epilepsy does not occur (28, 29), and it develops reliably (Figure 1A). Transcriptomics analysis showed that the expression of several transcriptional regulators was up-regulated (FDR <0.05), including ID (Id2), SMAD (Smad3), SOX (Sox11), STAT (Stat3, Stat1), HDAC (Hdac9), KLF (Klf10) , Klf5) and CREB ​​(Creb1) family (Figure 1B and SI appendix, Figure S1). Among them, Id2, a transcription factor inhibitor with developmental activity, is a candidate of particular concern. Previously, Id2 has been shown to increase its expression in GC after status epilepticus (30), and separate studies have combined this gene with cell and slice culture axon growth (31⇓ –33) and after spinal cord injury (34) Connect. Using immunostaining, we confirmed the up-regulation of Id2 protein levels, which was evident in a small number of GCs 1 day after KA (SI appendix, Figure S1G) but was present in most GCs 3 days after KA (figure 1C and SI appendix) , Fig. S1H). This upregulation continued after 14 days (Figure 1D). Initially, the Id2 signal was present in the nucleus (1 day), but later (3 days and later) it was enriched in the cytoplasm. This is consistent with the model, in which Id2 is either isolated to the cytosol by other factors (35⇓ –37), or it binds and isolates the transcription factor to the cytosol by itself (38). It is worth noting that our analysis also revealed the presence of Id2 protein in the MF axons of CA3, which has nothing to do with its absence (in the control) or presence (after KA) in the GC somatic cells (Figure 1D). Although this may indicate that the role of Id2 in MF axons is not yet recognized, since our focus in this study is on rewiring, we followed up with the up-regulation of Id2 corresponding to sprouting.

Transcriptomics analysis of chemically induced MF germination. (A) Experimental design. (B) The volcano map shows the acute transcriptomic changes of transcription and translation related molecules in a single GC one day after KA-induced MF germination. Red dots indicate differentially expressed genes (FDR <0.05 and> 2-fold change, or |log2FC|> 1, as shown in the figure). (C) MF germination is already in progress, and Id2 is up-regulated in GCL on day 3 after KA induction. The upper panel shows a control injected with saline; the lower panel shows KA. From left to right: Timm-stained MF axon tracking reconstructed from 80 µm thick sections; Timm-stained brain sections at higher magnification; immunohistochemical staining of ZnT3 and Id2. (D) 14 days after KA, MF germinates and develops, and Id2 remains up-regulated in GCL. The upper panel shows a control injected with saline; the lower panel shows KA. From left to right: Timm-stained hippocampus section showing the entire MF system; immunostaining of Id2 in the MF system (note that Id2 is present in the initial fibers of the hepatic hilum and CA3); immunostaining of ZnT3 and Id2 in the dentate gyrus is higher The magnification is displayed.

To test whether Id2 plays a role in MF germination, we used Cre-dependent AAVDJ/8 virus to selectively clone and ectopic expression of Id2 gene in the GC of the ventral hippocampus of adult Calb1-IRES2-Cre-D mice (note, Although AAV-Id2 was co-injected with AAV-EGFP to confirm the injection site by visualizing the GC, we will refer to this injection as AAV-Id2 hereafter; for the control, only AAV-EGFP with equal volume was used; Figure 2A). One to three months after induction, histological analysis revealed that the newly formed MF axons target the GC and internal molecular layers (GCL and IML, respectively; Figure 2B and C and SI appendix, Figure S2). The quantification of axon length and spots further confirmed these observations. Measurement of MF axon length (AAV-EGFP: 1.3 ± 0.17 cm, n = 6 mice; AAV-Id2, 1 month: 2.5 ± 0.34 cm, n = 4 mice, 2 months: 2.8 ± 0.42 Cm, n = 3 mice, 3 months: 4.7 ± 0.56 cm, n = 5 mice; Figure 2D) and spot size (in GCL, AAV-EGFP: 0.82 ± 0.024 µm2, 737 points/4 mice ; AAV-Id2: 1 month: 1.6 ± 0.038 µm2, 691 puncta/3 mice, 2 months: 1.9 ± 0.14 µm2, 853 puncta/3 mice, 3 months: 2.5 ± 0.25 µm2, 712 puncta/3 Mice; at IML, AAV-EGFP: 0,420 ±60 m puncta/4 mice; AAV-Id2, 1 month: 1.3 ± 0.076 µm2, 309 puncta/3 mice, 2 months: 1.8 ± 0.22 µm2, 534 puncta/3 mice, 3 months: 1.8 ± 0.081 µm2, 358 mice; Figure 2E) reveals the time-dependent increase after delivery of AAV-Id2. Because zinc transporter-3 (ZnT3) is a known marker of naive MFs (39), we tested its expression by immunostaining (Figure 2F) and found an increase in density (in GCL, AAV-EGFP: 1.1 ± 0.16) × 105 puncta/mm3, AAV-Id2: 4.5 ± 0.47 × 105 puncta/mm3; in IML, control: 0.86 ± 0.16 × 105 puncta/mm3; AAV-Id2: 5.0 ± 0.96 × 105 puncta/mm3; AAV-EGFP : From 116 hippocampal mice, AAV-Id2: n = 8 hippocampus from 6 mice; Figure 2G) and size (in GCL, AAV-EGFP: 0.83 ± 0.15 µm2; AAV-Id2: 1.6 ± 0.13 µm2; In IML, AAV-EGFP: 0.83 ± 0.15 µm2; AAV-Id2: 1.2 ± 0.17 µm2; AAV-EGFP: n = 5 hippocampus of 5 mice, AAV-Id2: n = 5 mice 5 hippocampus; Figure 2H) ZnT3 spot inauguration 3 months after AAV-Id2. We also confirmed that AAV-Id2 induces rewiring at the single-cell level through the morphological reconstruction of a single GC, which shows that the recurrent fibers in GCL/IML after AAV-Id2 induction do not have obvious reorganization of the axon and dendritic structures established by cell development (Figure .2I). In order to test whether AAV-Id2 also changes the action potential (AP) emission of GC or its overall synaptic drive to CA3 in addition to the effect on anatomical rewiring, we performed electrophysiological characterization of these characteristics. These indicate that despite the increased input resistance of the cells, AAV-Id2 does not seem to change the AP emission characteristics induced by current pulses or the synaptic drive of CA3 pyramidal cells established by cell development (SI Appendix, Figure S3).

Id2 delivered by AAV induces axon growth and target-specific rewiring in mature hippocampal GC. (A) Experimental design. The Calb1-IRES-Cre-D transgenic line allows genetic access to dentate GC through Cre expression. Confocal images showed that Id2 overexpression was confirmed by immunostaining. (B) Timm staining shows the dentate gyrus 3 months after Cre-dependent AAV-EGFP (control) and AAV-Id2 injection. After AAV-Id2, the dark ring of precipitation around the GCL represents the newly formed MF (red arrow). (C) Timm staining shows GCL/IML 1, 2 and 3 months MF after AAV-EGFP and AAV-Id2 injection. (D) Quantification of total axon length after AAV-EGFP and AAV-Id2 injection (one-way analysis of variance, AAV-EGFP and AAV-Id2, 1 month, *P = 0.043; 2 months, *P = 0.023; 3 months, ****P <0.0001). (E) Quantification of the size of Timm positive spots in GCL/IML at 1, 2, and 3 months after AAV-EGFP and AAV-Id2 injections (data points represent an individual; two-way analysis of variance, GCL: AAV-EGFP and AAV- Id2, 1 month, ***P = 0.0003; 2 months, ****P <0.0001; 3 months, ****P <0.0001. IML: AAV-EGFP and AAV-Id2, 1 month , P = 0.053; 2 months, ****P <0.0001; 3 months, ****P <0.0001). (F) 3 months after injection of AAV-EGFP and AAV-Id2, the EGFP marker of GC (green) in GCL/IML and ZnT3 staining of MF synapse (red) (note that AAV-Id2 and AAV-EGFP were co-injected to show GC (G) Quantification of ZnT3 positive spot density in GCL/IML 3 months after AAV-EGFP and AAV-Id2 injection (two-way analysis of variance, GCL: AAV-EGFP and AAV-Id2, ****P <0.0001; IML: AAV-EGFP and AAV-Id2, ****P <0.0001). (H) Quantification of the size of ZnT3 positive spots in GCL/IML 3 months after AAV-EGFP and AAV-Id2 (two-way analysis of variance, GCL: AAV-EGFP and AAV-Id2, **P = 0.0022; IML: AAV-EGFP And AAV-Id2, P = 0.089). (I) A single GC was reconstructed 3 months after the injection of AAV-EGFP and AAV-Id2. After AAV-Id2, newly formed axons extending to GCL and IML are shown in red. The original MF projection of CA3 is marked with a blue arrow. For clarity, CA3 is omitted.

Because the ventral and dorsal hippocampus support different brain functions and show different molecular patterns (40⇓ –42), and the left-right asymmetry of hippocampal function has also been identified (43), in the next set of experiments , We tested that AAV-Id2 induced MF rewiring at hippocampal locations other than the ventral side. For this reason, we expressed AAV-Id2 in the dorsal and ventral GCs of the left and right hippocampus (Figure 3A), which led to the development of MF rewiring in the entire mouse hippocampus regardless of the anatomical location (Figure 3A). 3B and SI appendix, Figure S2A).

Id2 delivered by AAV induced rewiring of MF throughout the hippocampus of mice and rats. (A) The experimental design shows that Id2 is overexpressed in the dorsal and ventral hippocampus of mice. (B) Image showing Timm stained sections collected from different levels of dorsal hippocampus (bregma, –2.0 mm and –3.2 mm) after injection of AAV-EGFP (control) and AAV-Id2. The high-magnification image at the bottom shows the GCL and IML germination of AAV-Id2 mice. (C) The experimental design shows that Id2 is overexpressed in the ventral hippocampus of rats. (D) Example image of rat Timm staining after Id2 overexpression. A coronal section of the rat ventral hippocampus (bregma, –6.2 mm) was collected from the area where AAV infection was confirmed by EGFP expression. The hippocampus not infected with AAV was used as a control. (E) Quantification of Timm staining intensity. Relative to the measured intensity of the hilar signal in the same part (two-way analysis of variance, GCL: control and AAV-Id2, **P = 0.0017; IML: control and AAV-Id2, ****P <0.0001). (F) ZnT3 staining of MF synapses in GCL/IML 3 months after AAV-Id2 injection.

Finally, in addition to mice, since MF germination has been widely characterized in rats (12⇓ ⇓ ⇓ ⇓ ⇓ –18, 21, 22), we tested whether Id2 has the ability to induce MF germination in different species. To this end, we injected Cre-independent AAV-Id2 into the hippocampus of wild-type rats. Similar to observations in mice, this operation revealed powerful MF rewiring (Figure 3 CF). Two to three months after the injection, the relative intensity of Timm staining in the GCL and IML of the dentate gyrus injected with AAV-Id2 was significantly higher than that of the control (control, GCL: 18.4 ± 4.3%, IML: 30.3 ± 4 %, 6 are from hippocampus of three rats; AAV-Id2, GCL: 40.4 ± 2.8%, IML: 69.9 ± 5.5%, from seven hippocampus of five rats; percentage represents the signal relative to the hilum in the same section strength). These observations indicate that AAV-Id2 can uniformly activate GC wiring programs in different hippocampal segments and species.

To test whether the MF sprouting induced by the AAV-Id2 gene is involved in the formation of functional synapses, we labeled MF boutons by ZnT3 immunostaining and characterized them using electron microscopy. This indicates that ZnT3 boutons are formed on GC dendrites and spines, each containing one or more release sites and an abundant supply of synaptic vesicles (Figure 4A and B). In order to detect physiological transmission, we used AAV-Id2 to induce MF rewiring, while also expressing channel rhodopsin (AAV-ChR) in GC in vivo, and prepared brain slices for electrophysiology 3 months after induction. Because the large conductance ChR current will mask the relatively small synaptic current, we used Rbp4-Cre transgenic mice to restrict ChR expression and reconnected to a subset of GCs, of which only about 30% of GCs express Cre, and therefore rely on Cre's Id2 and ChR (Figure 4C). In separate experiments on the dorsal and ventral hippocampus, we performed patch clamp recordings on ChR non-expressing cells. Regardless of the anatomical location, compared with the control, the activation of ChR by blue light after AAV-Id2 injection caused larger and more frequent excitatory synaptic events in the slices (in the dorsal hippocampus, control: -9.9 ± 1.9 pA, AAV -Id2: -40 ± 11pA; control, n = 8 cells from 6 mice; AAV-Id2, n = 17 cells out of 55 cells, from 7 mice; Figure 4D; in the ventral hippocampus, Control: -11 ± 2.8 pA, AAV-Id2: -36 ± 9.7 pA; control, n = 10 out of 36 cells from six mice; AAV-Id2, out of 39 cells from seven mice N = 6; Figure 4E). In summary, the AAV delivery of a single transcriptional regulator Id2 leads to the formation of new MF circuits, including the formation of functional synapses.

Id2 provided by AAV induces functional synapse formation. (A) Electron microscope image showing ZnT3 positive button (red) on GC dendrites and spines (green) 3 months after AAV-Id2. Illustrations a and b are shown enlarged, while c and d show other examples that do not exist in panel A. (B) 3D electron tomography reconstruction number 2 of the ZnT3-positive MF Burton located in the GCL/IML boundary 3 months after AAV. Red: axon axis, green: mitochondria, blue: synapse formed by this bouton terminal. (C) Experimental design and injection schedule to test the physiological transmission after MF rewiring. In acute brain slices, patch clamp recording was performed by ChR-negative GCs, while ChR-positive GCs were activated with blue light 3 months after AAV-ChR (control) and AAV (30% of the total number of GCs in these experiments)- Id2 (mixed with AAV-ChR) injection. (D) Cumulative probability plot showing the fraction of recorded cells in the dorsal hippocampus GC and the light-induced EPSC amplitude (recorded in 10 µM Gabazine; Kolmogorov-Smirnov test, P = 0.57). The inset shows the light-induced EPSC amplitude (Mann-Whitney U test, *P = 0.048). (E) Cumulative probability plot showing the relationship between the score of recorded cells in the ventral hippocampal GC and the amplitude of the light-induced EPSC (recorded in 10 µM Gabazine; Kolmogorov-Smirnov test, *P = 0.041). The inset shows the light-induced EPSC amplitude (Mann-Whitney U test, **P = 0.0075).

Next, we tried to better understand the molecular mechanism behind AAV-Id2-induced MF rewiring. Although Id2 has been widely characterized for its role in transcriptional regulation (31, 38, 44), a previous study showed that Id2 may directly promote axon growth and growth cone formation, and has nothing to do with transcriptional regulation (33) . This function depends on Akt-mediated phosphorylation of Serine 14 in Id2 (33). To test this possibility first, we expressed the phosphorylated and ablated mutants Id2 (S14A) and AAV-Id2 (S14A) in the ventral hippocampus. We assume that if Akt/Id2 signaling is involved, MF rewiring will not develop using this mutation. However, AAV-Id2 (S14A) still induces MF rewiring (SI Appendix, Figure S4), indicating that Id2 is unlikely to directly form growth cones.

Second, we studied the transcriptome changes induced by Id2. By directly binding transcription factors, Id2 inhibits their DNA binding, thereby inhibiting their transcriptional activity (31, 44) (Figure 5A). Therefore, the increase in Id2 expression may lead to the up-regulation and down-regulation of gene expression, and its identity depends on the transcription factor suppressed by Id2. In order to study the transcriptomics results, we sequenced a single GC one month after AAV-Id2 induction (Figure 5B), at which time axon growth was already obvious (Figure 2C). Consistent with the role in transcriptional regulation, this reveals a wide range of transcriptome changes, which are mainly caused by JAK-STAT (Stat1, Stat3, and Irf9) (45) and interferon signaling pathways (for example, Irf1, Irf7, Irf9, Isg15, Usp18). ) Members dominate) (46) and multiple other molecules related to axon wiring (for example, Tle1, Nefm, Slit1, Adcy1; Figure 5C and SI appendix, Figure S5; see discussion).

Single-cell transcriptomics revealed a comprehensive rewiring program induced by Id2 provided by AAV. (A) The drawing depicts the transcription function of Id2. (B) Experimental strategy. (C) The volcano plot shows the difference in gene expression between AAV-EGFP (n = 59 cells) and AAV-Id2 (n = 71 cells)-a single GC provided. The horizontal and vertical dashed lines show FDR = 0.05 and 2 times the change (|log2FC|>1), respectively. The gene names highlighted in red belong to the JAK-STAT and interferon pathways. (D) Enrichr transcription factor target enrichment analysis based on 285 up-regulated (red) and 848 down-regulated genes (blue) that are differentially expressed (P <0.05) between AAV-EGFP or AAV-Id2 datasets. The recognized transcription factors (E: coding, C: ChEA) and their expression rate in GC are shown at the bottom. The size of the circle represents the number of target genes present in the input data. (E) Gene regulatory network activated by AAV-Id2. Nodes represent molecules from C and D; edges represent interactions. GAF and ISGF3 refer to the proteome assembly of Stat1 homodimer and Stat1, Stat2 and Irf9, respectively (45).

We next asked what the key mediator of the rewiring process induced by AAV-Id2 might be, that is, the molecule that connects Id2 to the above-mentioned genes. Since Id2 inhibits the activity of transcription factors without changing their expression, the transcriptome changes found may represent downstream effects and do not indicate which transcription factors Id2 directly acts on. Therefore, we conducted a transcription factor-target enrichment analysis with the purpose of identifying transcription factors that are directly or indirectly inhibited or indirectly de-suppressed by Id2. Using Enrichr (47, 48), we found 26 such factors whose known regulatory networks match the observed gene expression patterns and are expressed in at least 30% of GCs (Figure 5D). Together with this analysis, our results outline the comprehensive transcriptome model behind MF rewiring, where Id2 exerts control over members of the JAK-STAT, Wnt, cAMP, and Slit/Robo signaling pathways related to wiring (Figure 5E and SI Appendix , Figure S5-S7).

The ability of genes to induce MF germination allows us to examine the network effects of this observed rewiring event. MF sprouting has been observed in human TLE and is a sign of experimental TLE, but the question of whether it is the cause or effect of seizures has been controversial (20, 21). Therefore, by recording with multi-channel silicon probes in the hippocampus of freely moving mice, we observed whether pathological brain dynamics features (49) have been formed 3 months after AAV-Id2-induced MF rewiring ( Figure 6A and B). In the local field potential range of 1 to 400 Hz, our recordings did not record pathological oscillations or epileptiform activities (Figure 6C). Specifically, θ, β, slow and fast gamma, and ripple and fast ripple range oscillations remain unchanged. In addition, we analyzed CA1 sharp wave ripple (SWR) and dent spike (DS) events, because their natural frequencies and occurrence rates increase in TLE (50, 51), respectively. None of these pathologies exist in our data. Incidence rate (AAV-EGFP: 0.31 ± 0.033 Hz, n = 6 mice; AAV-Id2: 0.27 ± 0.037 Hz, n = 6 mice) and natural frequency (AAV-EGFP: 152 ± 7.8 Hz, n = 6 mice; AAV-Id2: 152 ± 2.7 Hz, n = 6 mice) SWR was not different between AAV-EGFP and AAV-Id2 injected mice (Figure 6D and SI appendix, Figure S8). In contrast, the incidence of DS1 (DS1) was reduced (AAV-EGFP: 0.59 ± 0.12 Hz, n = 6 mice; AAV-Id2: 0.32 ± 0.19 Hz, n = 6 mice; P = 0.06 , Mann-Whitney U test), and selective loss of type 2 DS (DS2) after delivery of AAV-Id2 (AAV-EGFP: 0.22 ± 0.089 Hz, n = 6 mice; AAV-Id2: 0.021 ± 0.020 Hz, n = 6 mice; P = 0.015, Mann-Whitney U test, Figure 6E and SI appendix, Figure S8). In summary, complete oscillation and SWR indicate that the network dynamics of AAV-Id2 mice are different from that of TLE, while the reduction in DS indicates that active routing is still effectively changed in the dentate network.

AAV-Id2 induces hippocampal dynamics after MF rewiring. (A) Experimental design. After the injection of AAV-EGFP and AAV-Id2, each mouse was implanted with a linear silicon probe in the dorsal hippocampus. Three months after the delivery of the AAV, the recording was performed on a freely moving mouse. (B) Histological image of the silicon probe passing through the hippocampus (Pyr: pyramidal layer, Rd: radiatum, LM: lacunar molecule, Mol: molecular layer, Hil: hilus). (C) Delta, theta, beta, slow gamma, midgamma, fast gamma (all during exercise), and ripple and fast ripple (during rest) range frequency (AAV-EGFP, n = 7 mice). Field potential (LFP) power; AAV-Id2, n = 6 mice). Using a two-way ANOVA test, the pairwise comparisons between the AAV-EGFP and AAV-Id2 groups did not show statistically significant (P <0.05) differences. (D) Left: Current area distribution related to SWR in the ripple peak trigger CSD diagram; the average LFP waveform (black trace) is superimposed and displayed. Right panel: quantification of ripple occurrence (Mann-Whitney U test, P = 0.31) and the internal frequency of ripples in CA1 Pyr and Rd (Mann-Whitney U test, P = 0.45; AAV-EGFP, n = 6 mice; AAV-Id2, n = 6 mice). (E) Left: CSD configuration files of DS1 and DS2. Right panel: quantification of DS1 (Mann-Whitney U test, P = 0.065) and DS2 (Mann-Whitney U test, P = 0.015; AAV-EGFP, n = 6 mice; AAV-Id2, n = 6) .

The DS1 and DS2 events are thought to be triggered by the bursts of layer II stellate cell populations in the lateral entorhinal cortex (LEC) and medial entorhinal cortex (MEC), respectively (52). Because LEC and MEC are proposed to support local ("egocentric reference frame") and global ("different-centric reference frame") landmark navigation (53⇓ –55), and the dentate gyrus is heavily involved For some forms of navigation object correlation and spatial learning and memory (11, 41, 42), we hypothesize that the network effects related to the budding of MF will be manifested in the behavioral process. To test this hypothesis, we used eight different previously validated tests to perform phenotypic analysis of mice 3 months after bilateral, dorsal, and ventral AAV-EGFP or AAV-Id2 injections (about each For details of this test, see Methods).

Before testing, the animals were subjected to light cycle inversion. In order to check the light cycle adaptation, motor activity and free movement behavior, we monitored each mouse in the cage for 13 days and in the new open environment for 20 minutes. In the rearing cage, the two groups of mice were equally well-adapted to the inverted daylight cycle, and there was no difference in their overall activity levels (SI Appendix, Figure S9A). In addition, in open ground, there is no significant difference between the total travel distance (AAV-EGFP: 66 ± 2.7 m, n = 12; AAV-Id2: 75 ± 4.4 m, n = 12) and the time spent in the central area (AAV-EGFP : 29 ± 2.2%, n = 12; AAV-Id2: 25 ± 2.5%, n = 12) between the two groups (SI Appendix, Figure S9B). Taken together, these indicate that AAV-Id2 mice did not exhibit hyperactivity or anxiety-like behaviors.

Next, we used the new object recognition and T-maze test to test the long-term and short-term memory performance of the hippocampus. Both tests take advantage of the rodent’s strong preference for novel things, whether it is object-related or environment-related. . The new object recognition test showed that although both groups showed a preference for new objects, the performance of AAV-Id2 mice was lower than that of the control group (discrimination index [DI]; AAV-EGFP: 71 ± 3.2%, n = 8 ; AAV-Id2: 50 ± 8.7%, n = 7; mice with DI> 25% during training were excluded from the analysis) (Figure 7A and SI appendix, Figure S9C). In the T maze, the probability of alternating occurrence of the two groups is higher than (AAV-EGFP: 71 ± 5.6%, n = 12; AAV-Id2: 61 ± 7.5%, n = 12) (Figure 7B), which is also expected The result of untreated rodents. However, AAV-Id2 mice did not show the normally occurring increase in selection delay throughout the experiment, and their entry delay into the arm was still significantly shorter than that of AAV-EGFP mice (in experiment 6: AAV-EGFP: 35 ± 8.3 s, n = 11; AAV-Id2: 10 ± 2.2 s, n = 11) (Figure 7B).

AAV-Id2 induces learning and memory after MF rewiring. (The statistical test is a two-way analysis of variance, unless otherwise specified.) (A) New object recognition. From left to right: experimental design, DI (AAV-EGFP and AAV-Id2, training P=0.87, test*P=0.021). (B) T-shaped maze. From left to right: experimental design, change (Mann-Whitney U test, P = 0.33) and selection delay (AAV-EGFP and AAV-Id2, trial 1, P = 0.87, trial 2, P = 0.66, trial 3, P = 0.31, test 4, **P = 0.0048, test 5, *P = 0.011, test 6, ****P <0.0001). (C) Morris Water Maze. From left to right: experimental design, escape latency [FDay (4, 84) = 13, P <0.0001; FTreatment (1, 21) = 0.056, P = 0.82; FTreatment × Day (4, 84) = 1.4, P = 0.23], quadrant time (adjacent and target, first detection test, AAV-EGFP: *P = 0.017, AAV-Id2: *P = 0.014; second detection test, AAV-EGFP: P = 0.82, AAV -Id2: **P = 0.0026), the length of the swimming path parallel to the wall (acquisition and reversal, AAV-EGFP: **P = 0.0025, AAV-Id2: P = 0.45) and the number wall method (acquisition and reversal, AAV-EGFP: **P = 0.0079, AAV-Id2: P = 0.57). (D) The Barnes Labyrinth. From left to right: Experimental design, main path length [FDay (4, 88) = 6.6, P = 0.0001; FTreatment (1, 22) = 0.4, P = 0.52; FTreatment × Day (4, 88) = 0.3, P = 0.85], the main error (Mann-Whitney U test, **P = 0.0038), the poke ratio in the probe test after acquisition (AAV-EGFP and AAV-Id2, angle = 0°: ****P <0.0001 , Angle = 18°: P = 0.52, angle = 36°: P = 0.54, angle = 54°: P = 0.98, angle> 72°: P = 0.72), average acquisition and reversal during AAV-EGFP and AAV-Id2 Strategy used: direct, P = 0.31, serial, **P = 0.0047, mixed P = 0.059). (E) Eight-arm radial maze. From left to right: Experimental design, memory error of decoy consumed every day [FDay (2, 44) = 9.0, P = 0.0005; FTreatment (1, 22) = 2.8, P = 0.11; FTreatment × Day (2, 44) = 0.53, P = 0.59], the memory error of each decoy consumed (AAV-EGFP and AAV-Id2: decoy 1 to 4, P = 0.88, decoy 5 to 6, P = 0.86, decoy 7 to 8, **P = 0.0029), preferred angle [FAngle (2, 44) = 16, P <0.0001; FTreatment (1, 22) = 0.056, P = 0.81; FTreatment × Angle (2, 44) = 2.6, P = 0.086], and Selection at the preferred angle (AAV-EGFP, Days 1 to 2 and Days 3 to 8, **P = 0.0071, Days 1 to 2 and Days 9 to 10), *P = 0.020, Days 3 to 8 8 days compared to days 9 to 10, P = 0.68; AAV-Id2, days 1 to 2 compared to days 3 to 8, **P = 0.0049, days 1 to 2 compared to days 9 to 10,* ** *P <0.0001, 3 to 8 days vs. 9 to 10 days, *P = 0.026). (F) Situational and implied fear conditioning. From left to right: experimental design, freezing during the context retention (AAV-EGFP and AAV-Id2: baseline, P = 0.60, context, ****P <0.0001), suggesting freezing during the retention (AAV-EGFP and AAV-Id2 : Pretone, P = 0.58, tone, P = 0.032, q = 0.064; does not meet FDR standards) and freezing during extinction (AAV-EGFP and AAV-Id2: baseline, P = 0.24, first tone, P = 0.38 , The last tone, P = 0.73).

To test memory performance related to spatial information, we used Morris Water, Barnes, and the eight-arm radial maze test, in which goal-oriented navigation is enhanced by disgusting, natural, and positive stimuli, respectively. In the Morris water maze, it was shown that the spatial learning was successful, and the escape latency (Figure 7C) and swimming path length (SI appendix, Figure S9D) were significantly reduced in both groups during the acquisition process. In addition, both groups showed a strong preference for the original target quadrant in the first detection experiment of reverse learning, which is to test space retention (day 4; the time in the quadrant as a percentage of the total time, AAV-EGFP : Target quadrant 36 ± 4.8%, adjacent quadrant 21 ± 2.2%; n = 12; AAV-Id2: target 39 ± 3.0%, adjacent quadrant 23 ± 1.6%, n = 12) (Figure 7C). However, during the second detection experiment of reverse learning (also on day 4), which tested the reverse learning itself, AAV-Id2 instead of AAV-EGFP, the mice still showed resistance to the original target quadrant (AAV- EGFP: target quadrant 25 ± 4.5%, adjacent quadrant 24 ± 2.4%, n = 12; AAV-Id2: target 39 ± 3.1%, adjacent quadrant 21 ± 1.2%, n = 12) (Figure 7C), indicating space insist. In addition, AAV-Id2 mice reverted to a wall-facing non-spatial swimming strategy (Figure 7C). Consistent with the observations in the Morris water maze, both groups learned the Barnes maze task (Figure 7D and SI appendix, Figure S9E). However, AAV-Id2 mice made more mistakes when looking for escape rooms (main mistake, in all trials; AAV-EGFP: 7.6 ± 0.52, n = 12; AAV-Id2: 10 ± 0.69, n = 12) And it does not show the preference for the original target when the escape room is removed in the probe test (the poke ratio of the original target, angle = 0°, AAV-EGFP: 3.6 ± 0.75, n = 12; AAV-Id2: 1.3 ± 0.28, n = 12). This may be because AAV-Id2 mice are better than AAV-EGFP mice (sequence strategy, AAV-EGFP: 16 ± 4.1% of all attempts, n = 12; AAV-Id2: 33 ± 5.7%, n = 12). Similarly, in the eight-arm radial maze, the two groups have the same learning effect in reducing memory errors, which is to enter one arm every few days (test 9th to 10th, AAV-EGFP: 0.47 ± 0.095, n = 12; AAV-Id2: 0.47 ± 0.13, n = 12) (Figure 7E). However, memory errors during the collection of the last two baits in AAV-Id2 mice were higher (average of all test days, AAV-EGFP: 1.7 ± 0.15, n = 12; AAV-Id2: 2.2 ± 0.2, n = 12; Figure 7E), indicating that the memory load of these mice increased. As the preferred strategy, both groups tend to enter the adjacent arm after visiting one (angle = 45°) (Figure 7E). However, AVV-Id2 mice, instead of AAV-EGFP, strongly increased their choice at this preferred angle within a few days (AAV-EGFP: Days 1 to 2 = 38 ± 2.0%, Days 3 to 8 = 52 ± 3.7%, 9th day) to 10 = 50 ± 4.3%, n = 12; AAV-Id2: 1st to 2nd day = 32 ± 3.5%, 3rd to 8th day = 47 ± 4.1%, 9th to 10 days = 59 ± 6.7%, n = 12) (Figure 7E).

Finally, we use prompt and contextual fear condition tests to analyze associative learning. Here, although both groups showed freezing response during the context retention test, the response of AAV-Id2 mice was smaller (AAV-EGFP: 12 ± 2.1%, n = 12; AAV-Id2: 3.1 ± 1.7%, n = 12) (Figure 7F). In contrast, although the response of AAV-Id2 mice to pitch preservation still seems to be low, there was no significant difference in the freezing response during pitch preservation and extinction tests between the two groups (pitch preservation: AAV-EGFP: 21± 3.1%, n = 12; AAV-Id2: 13 ± 3.6%, n = 12; Extinction test: first tone: AAV-EGFP: 23 ± 4.2%, n = 12; AAV-Id2: 19 ± 4.6%, n = 12, the final tone: AAV-EGFP: 11 ± 1.4%, n = 12; AAV-Id2: 9.1 ± 2.4%, n = 12) (Figure 7F).

In this study, we systematically analyzed MF sprouting in order to understand the transcriptomic mechanism that can promote axon growth and circuit formation in the adult brain. The motivation for our research design is to hypothesize that the activation of certain cell-autonomous mechanisms may be sufficient to drive rewiring in the adult brain, where further axon growth is usually inhibited. Our results suggest three main conclusions that not only affect the circuit assembly in the adult brain, but also affect the pathophysiology and information processing of TLE in the dentate gyrus.

In the developing nervous system, Id2 enhances cell proliferation and inhibits the activities of basic helix-loop-helix (31) and other transcription factors (44). In terms of mechanism, the developmental regulation and degradation of Id2 is related to the up-regulation of axon growth inhibitors (31), which indicates that the activation of Id2 can counteract growth inhibition and enhance axon growth, but this prediction is only carried out in the culture system and after the spinal cord Test injury (31⇓ ⇓ –34). Compared with previous studies, we used unbiased single-cell RNA-seq screening to determine Id2 and asked about the consequences of Id2 activation in mature, uninjured neurons in healthy adult mice and rats. We found that the sole activation of Id2 in mature GC can drive MF sprouting and the formation of functional synapses.

In our preliminary analysis, we found that during the chemically induced MF germination, both Id2 mRNA and protein are rich in GC (Figure 1). This suggests that Id2 may play a role in MF germination, and we tested it through the cell-autonomous activation of Id2 in GC. We found that AAV-Id2 has the extraordinary ability to induce MF rewiring throughout the hippocampus, and its circuit structure is similar to the previous description of MF germination (for example, References 12⇓ –14, 21) but no GC death or layer dispersion Signs (Figure 2-4).

As the master regulator, Id2 activates a comprehensive transcriptome program for rewiring. Because Id2 is an inhibitor of transcription factors, MF rewiring seems to be caused by the inhibition of the continuously active transcription program. Consistent with this, we found that its function is related to the activation and silencing of circuit-level recombination molecules (Figure 5). Up-regulation of JAK-Stat3 is related to the promotion of axon growth after visual and spinal system injury (56⇓ ⇓ –59). Although previous work linked Id2 and Stat3 to axon growth, our results established a connection between the two, so that the activation of Id2 promotes the downstream expression of Stat3. Other up-regulated genes include Phf11d, which is a regulator of the transcription factor Bex1 (60) and promotes axon regeneration (61).

Down-regulation molecules are also consistent with the increase in axon organization and circuit formation ability, including 1) Tle1, the core inhibitor of Wnt signaling in axon mode (62, 63); 2) neurofilaments Nefl and Nefl, which determine the structure and caliber of mature axons Nefm (64); 3) Slit1, a regulator of developmental axon guidance and mode (65), its downregulation may allow MF to enter the dentate gyrus; 4) Adcy1, whose deletion leads to developmental axon retraction arrest (66), Axon branching in the sensory area is vigorous (67, 68) and recovery after spinal cord injury (69); 5) Fmr1, a synaptic protein local translation regulator (70); and 6) Rreb1, waller axon degeneration after injury Conditioner (71).

Delineating the sequential activation/inactivation and stepwise action of related molecules is important to understand the MF rewiring process in detail. Other aspects of MF rewiring may still be controlled independently of Id2, but can still be controlled autonomously. For example, AAV-Id2 induced MF budding developed for several months (Figure 2), which is similar to other observations (21), but slower than the realization of non-cell-specific chemical induction (Figure 1). Whether the factors that control the speed of circuit formation are cell-autonomous and individually identifiable, and how to reprogram synaptic targeting to other cell types instead of GC remain open questions. Regarding other pathways previously specifically related to axon growth, regeneration or MF budding, our data from a single GC did not reveal Id2-induced TGF-β/BMP-Smad (axon growth) transcriptome changes (72 ⇓ –74), BDNF (MF sprouting) (21, 23), PTEN-mTOR (MF sprouting in the GC produced after birth) (19, 22) and p38/JNK (axon regeneration) (75) pathways (SI appendix, Figure S5) . Some of them may be located upstream of Id2 (for example, TGF-β/Smad2) (72), are controlled by translation, function independently of Id2, or do not participate in the MF germination of mature GC at all.

Our ability to genetically induce MF rewiring in GC allows us to examine the consequent network effects in the context of two related but independent hypotheses. According to one hypothesis, MF budding may produce an over-excited network state. MF sprouting has been observed in human TLE and is one of the hallmarks of chemically induced (non-specific and widely affecting different cell types) circuit changes in experimental TLE (21). Whether MF sprouting is the cause or effect of seizures has been controversial (20). Therefore, a special question is whether pathological brain dynamics occur after gene-induced MF rewiring, such as in epileptic brains. However, our in vivo electrophysiological records did not record pathological oscillations, SWR, or epileptiform activity 3 months after AAV-Id2 delivery (Figure 6). In contrast, it has been reported that after chemically induced epilepsy, theta and mid-gamma oscillations are reduced (49) and the internal SWR frequency is increased (50). Therefore, our results reveal a different network dynamics from epilepsy and show that MF rewiring does not produce seizures (20).

According to another assumption, the dentate gyrus is assumed to process information about content (for example, objects) and local spatial landmarks, passed from LEC, and about global spatial landmarks, passed from MEC (53⇓ –55). Our in vivo recordings in AAV-Id2 mice showed a decrease in the occurrence of DS1 and the loss of DS2 events, which are thought to be triggered by population outbreaks in LEC and MEC, respectively (52). Although the circuit mechanism behind the DS event is still elusive, our data shows that MF sprouting has different interferences to the generation of DS1 and DS2 events. Therefore, information about content and local landmarks from LEC and information about the global spatial context of the experience from MEC may be affected differently by the budding of MF.

By using a variety of different memory analyses to evaluate object and space-related information processing, we discovered the cyclical pattern of AAV-Id2 animal behavior with MF sprouting. Regarding objects, we found that AAV-Id2 mice showed a preference for new objects, although the preference was lower than that of the control group (Figure 7). It is worth noting that the decrease in preference for new objects is associated with a decrease in the incidence of LEC-related DS1 events in AAV-Id2 mice. These represent further disagreements with TLE, as new subject preferences have not changed (76, 77) and the incidence of DS in epileptic mice has increased (51) (note that DS1 and DS2 were not separately analyzed by subsequent studies). Nevertheless, these are not contradictory to our findings, indicating that the DS1 event may be related to the identification of new objects. In terms of spatial information, AAV-Id2 mice showed no defects in the performance of their main tasks, but their way of solving spatial problems seemed to be different from that of the control group. Specifically, Morris water, Barnes, and eight-arm radial maze tests showed that AAV-Id2 mice learned and performed well without space retention defects (Figure 7). Please note that these findings represent another divergence from TLE, where the same test shows learning and memory impairment (49, 78⇓ –80). However, AAV-Id2 mice showed a higher level of spatial perseverance and chose a non-spatial, wall-facing swimming strategy during the reversal learning of the Morris water maze, and performed it at the expense of the Barnes maze and mixed experiment More continuous trials. The eight-arm maze collectively indicates that AAV-Id2 mice tend to use strategies based on local landmarks and possibly self-reference navigation, rather than relying on global spatial cues. Since DS2 events related to MEC (Global Spatial Information) are particularly lost in AAV-Id2 mice, our findings indicate that DS2 events themselves are not necessary to solve the main spatial task, but may help to handle guided navigation to enhance performance. One explanation for these findings should also be considered as the potential confusion of pattern separation caused by repeated redistribution of neural activity between GCs after MF germination. Pattern separation is a major function performed by the dentate gyrus (11, 81) and is thought to be controlled by young but immature adult GCs (82). Since in these experiments the Calb1-IRES2-Cre-D system was used to deliver Id2 to the GC and the beginning of Calb1 expression marked the transition of the adult-born GC to a more mature state (83), the possibility of changing the pattern separation is unlikely to be young Of adult-born GCs have made a significant contribution to our observations. In general, our results are consistent with the assumption that MEC and LEC support navigation based on global and local cues, respectively (54, 55). Finally, we also found that unlike the control group, AAV-Id2 mice did not increase their selection delay in subsequent trials in the T maze. This phenotype may be the result of a self-centered navigation strategy that uses left and right sequences with high precision, but it may also be interpreted as an inability to get used to (although our other data does not indicate this) or faster decision-making, Therefore, the relationship between its MF germination is currently not clear.

Repair is a key goal, usually in the context of pathology or injury to study the factors that promote the rewiring of circuits in the adult nervous system. Our findings provide evidence for cell-autonomous activation of axon growth and circuit formation in healthy mature neurons in the absence of development or injury signals. A more detailed understanding of how the molecules involved contribute to this process will help advance circuit engineering methods that can induce axon growth and control the formation of specific circuits in target cells in the adult brain.

All mouse protocols and feeding practices were approved by the Canton Veterinary Office in Zurich. Full description of 1) animals, 2) plasmids and viruses, 3) stereotactic injection, 4) single-cell RNA-seq and bioinformatics, 5) histology and neuroanatomy, 6) electron microscope, 7) in vitro electrophysiology, 8) In vivo electrophysiology, 9) behavior and 10) statistical analysis, please refer to the SI appendix.

RNA-seq data has been deposited in the National Biotechnology Information Center Gene Expression Synthesis (GSE161619) (27).

This research was funded by the Swiss National Science Foundation (310030_188506 to CF), Dr. Eric Slack-Gyr-Stiftung Award (to CF), Novartis Medical Foundation funding (to CF), 20017-1.2.1- NKP-2017- 00002 grant (for CV), EFOP-3.6.2-16-2017-00008 grant (for CV), 20765-3/2018/FEKUTSTRAT grant (for CV), ERC Consolidator grant (nanoAXON #772452 for CV)), and Forschungskredit Scholarship at the University of Zurich (for WL). We thank Dr. Jean-Charles Paterna and Melanie Rauch (viral vector facility, Zurich University/ETH Zurich) discussed and virus production, and the Zurich Center for Functional Genomics provided RNA-seq support.

Author contributions: research on WL and CF design; research on WL, ME, AD, LQ, NAC-O., SS, CS, AA, ES, IA, JW, TL and CV; WL, AD, DL, JS, DPW, CV, and CF analyzed the data; CF developed the concept and supervised the research; WL and CF wrote this paper.

The author declares no competing interests.

This article is directly contributed by PNAS.

This article contains online support information at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2108239118/-/DCSupplemental.

This open access article is distributed under the Creative Commons Attribution-Non-Commercial-No Derivative License 4.0 (CC BY-NC-ND).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.