Inducible NMDA receptor knockdown reveals a maintenance phase in dendritic refinement of barrel cortex neurons

Summary

The temporal mechanisms of activity-dependent dendritic patterning during postnatal development remain unclear because appropriate technology is lacking. Here, we demonstrate that the auxin-inducible degron 2 technology enables the rapid knockdown of target proteins at specific time points in the postnatal mouse brain. When N-methyl-D-aspartate-type glutamate receptor (NMDAR) depletion was induced from postnatal day (P)3, barrel cortex layer 4 spiny stellate neurons (barrel cells) failed to form strong asymmetry and a high tree-length variance in the dendritic patterns. Intriguingly, these unique dendritic patterns of barrel cells formed by P6 were rapidly canceled by NMDAR depletion from P6 but not from P12. NMDAR depletion from P6 also extinguished the existing Golgi apparatus’ lateral polarity. These results suggest that the barrel cells’ dendritic refinement involves not only formation but also “maintenance” of specific dendritic patterns during early postnatal development, in which NMDARs play a critical role, likely by regulating the Golgi polarity.

Graphical abstract

Highlights

• •AID2 enabled efficient protein knockdown in the postnatal mouse brain
• •NMDARs were knocked down postnatally in a sparse set of cortical neurons
• •Postnatal NMDAR knockdown rapidly disrupted formed dendritic patterns
• •Postnatal NMDAR knockdown eliminated Golgi lateral polarity

Neuroscience; Molecular Biology; Developmental Biology

Introduction

Precise neural circuits underlying the higher mammalian brain functions are established through activity-dependent refinement during critical periods of postnatal development.^1,2,3,4 Developmental remodeling of dendritic patterns is key for circuit reorganization, whereby individual neurons rearrange and optimize their presynaptic partners.

The whisker-related area in the rodent primary somatosensory cortex (barrel cortex) is a valuable model for dendritic refinement.^5,6,7,8 Barrel cortex layer 4 (L4) has an array of “barrels” that corresponds to the arrangements of whiskers on the face.^9,10,11 Within a barrel, the termini of thalamocortical axons (TCAs) transmitting inputs from the corresponding whisker form a discrete cluster. L4 spiny stellate neurons (barrel cells) in the mouse are mostly located around the barrel edge and exhibit two types of unique features in their dendritic patterns. First, the dendrites are expanded asymmetrically toward single barrels (strong lateral asymmetry of dendritic projection). Second, most dendritic trees are very short and rarely branched, and only one or a few dendritic trees (i.e., “winner” trees) are extremely long and highly branched, resulting in a small percentage of intermediate-length dendritic trees (high tree-length variance). These unique dendritic features of barrel cells, which are the basis of the one-to-one functional relationship between whiskers and barrels,¹² are formed essentially between postnatal day (P3) and P6 in a thalamocortical input-dependent manner.^{13,14,15,16,17,18,19}

Gene knockout (KO) and RNAi approaches have significantly contributed to revealing the mechanisms by identifying molecules that are essential in forming the barrel cells’ unique dendritic patterns. Among these molecules, the N-methyl-D-aspartate-type glutamate receptor (NMDAR), which is a coincidence detector of pre- and postsynaptic activity,^20,21 has received much attention as a key player.^{7,17,22,23,24,25} NMDAR is a tetrameric complex with two essential NR1 subunits and two modulatory NR2(A-D) subunits. In cortex-specific NR1-gene (Grin1) KO mice, barrel cells fail to form asymmetric dendritic patterns.^26,27 Single-cell Grin1-or NR2B (Grin2b) KO also displays rather symmetric dendritic patterns, indicating the cell-autonomous functions of NMDARs in dendritic refinement.^17,23 However, gene-KO impairs gene functions throughout the individual’s life; therefore, phenotypes could be complicated by the accumulation of secondary effects and/or compensations. Additionally, defective phenotypes in earlier developmental stages often conceal important functions carried out by the same gene in later developmental stages.^{17,23,28,29,30} Although some inducible gene manipulation systems, such as CreERT2 and Tet-OFF/ON, have been developed and used for adult brains, the gene product (protein) depletion is too slow (week—month timescales) to be applied to neonatal brains.^31,32,33 During the neonatal stages, barrel-cell dendrites undergo dynamic remodeling and change their patterns drastically, even in a day.^13,23,34 Thus, an innovative methodology that can rapidly (at least within a day) impair the function of the gene product of interest has been long awaited.

This study demonstrated that auxin-inducible degron 2 (AID2), a recently developed protein knockdown (KD) system,³⁵ can rapidly (within hours) and efficiently deplete target proteins in the postnatal mouse brain. We also successfully induced AID2-mediated NR1-protein KD in single barrel cells by combining AID2 with Supernova.^23,36 We demonstrated that barrel cells failed to exhibit the unique dendritic patterns (i.e., strong lateral asymmetry and high tree-length variance) when NMDAR deficiency was induced from P3. Intriguingly, these dendritic patterns formed by P6 were disrupted by NMDAR deficiency induced at P6 but not P12. These results suggest that the dendritic refinement of barrel cells requires the formation and “maintenance” of patterns during early postnatal development, with NMDARs essential for both. Additionally, the lateral polarity of the Golgi apparatus—crucial for dendritic asymmetry^6,12—was also abolished by NR1 KD at P6. These findings, uncovered using a newly emerging inducible protein KD approach, highlight the temporal role of NMDARs in shaping cortical dendrite architecture postnatally.

Results

AID2 induces rapid and efficient protein knockdown in the postnatal mouse brain

AID2 is a recently developed protein KD system.³⁵ When a 7-kD degron derived from the Arabidopsis IAA17 [termed mini-AID (mAID)] is fused to a target protein and OsTIR1(F74G)—a mutant from of Oryza sativa TIR1—is expressed in specific cells, administering the 5-phenyl-indole-3-acetic acid (5-Ph-IAA) ligand depletes the target protein only in OsTIR1(F74G)-expressing cells by activating the endogenous ubiquitin-proteasome pathway.³⁵

To assess the efficiency of AID2-mediated protein KD in the postnatal mouse brain, we used CAG-OsTIR1(F74G)-P2A-mAID.EGFP (mAID.EGFP) reporter mice, which express mAID-fused enhanced green fluorescent protein (mAID.EGFP) and OsTIR1(F74G) (hereafter referred to as TIR1) globally.³⁵ We injected the mice intraperitoneally with 5-Ph-IAA at P6, P15, and P30 and quantified the EGFP intensities in the barrel cortex before (0 h) and 3 h and 6 h after administering 5-Ph-IAA. At all these stages, EGFP signals significantly decreased after 3 and 6 h compared with 0 h (Figure 1). Notably, the decrease was drastic at the earlier postnatal stages, such as P6 and P15. The blood-brain barrier (BBB) maturation in later postnatal development may reduce the efficiency of 5-Ph-IAA penetration into the brain.³⁵ These findings suggest that AID2 is effective in knocking down a target protein during early postnatal development, such as P6 and P15.

Figure 1.

AID2-mediated EGFP knockdown (KD) in the postnatal mouse brain

(A) CAG-OsTIR1(F74G)-P2A-mAID.EGFP (mAID.EGFP) reporter mice, which express mAID-fused EGFP and OsTIR1(F74G) (TIR1) globally, were administered 5-Ph-IAA at postnatal day (P)6, P15, or P30. Brains were dissected 0 h, 3 h, or 6 h after 5-Ph-IAA administration, and coronal slices (400-μm thick) were made. Wild-type (WT) mice were used for the estimation of background autofluorescence. All images in the same age group were obtained under the same conditions. Scale: 1 mm.

(B) EGFP signal intensities in the barrel cortex of mAID.EGFP reporter mice (n = 3–4 mice) subtracted from the background (Tukey’s multiple comparisons test). Mean ± SEM. ^∗∗∗∗p < 0.0001.

Generation of mAID-fused NR1 knock-in mice

To apply AID2 to the inducible disruption of NMDAR function in mice, we generated NR1.mAID mice using the CRISPR-Cas9-mediated knock-in approach. mAID was fused to the cytoplasmic C-terminal of NR1, the essential NMDAR subunit (Figure 2A). Among the seven founders obtained, one (#506) with the desired PCR product sizes and sequences was selected for further studies. Homozygous NR1.mAID (Grin1^mAID/mAID) mice were obtained by intercrossing Grin1^mAID/+ mice in a Mendelian ratio (15 out of 48 pups) and grew normally (Figure S1).

Figure 2.

AID2-mediated NR1 KD during the postnatal development of mice

(A) mAID was fused to the C-terminus of NR1 by the CRISPR/Cas9-mediated knock-in approach. Genomic DNAs of WT and knock-in candidate mice were amplified by PCR with the indicated primers. The #506 (but not #504) mouse showed desired band sizes in electrophoresis. Precise insertion of mAID in the #506 mouse was further confirmed by DNA sequencing. M: size markers.

(B–D) Grin1^mAID/mAID (Control) [n = 3 (P1) and 1 (P2)] and CAG-TIR1; Grin1^mAID/mAID (KD) [n = 4 (P1) and 4 (P2)] pups were administered 5-Ph-IAA. The percentage of Control and KD pups that exhibited the suckling reflex before (0 h) and 3 h after 5-Ph-IAA administration (B). Representative images of Control and KD mice (arrows: stomachs) before (0 h) and 24 h after 5-Ph-IAA administration (C). Changes of body weight (%) in 24 h after 5-Ph-IAA administration (Student’s t test) (D).

(E–G) Control and KD mice were administered 5-Ph-IAA at P6 (E and F) or P12 (G), and forebrains were sampled 6 h (E) or 24 h (F and G) later. Western blot analyses were performed using anti-NR1 (filled arrowheads) or anti-actin (open arrowheads) antibodies (n = 3–4 mice, Welch’s t-test).

(H and I) Control and KD mice were administered 5-Ph-IAA at P6, and forebrains were sampled 7 days (H) or 14 days (I) later. Western blot analyses were performed using anti-NR1 or anti-actin antibodies. (P6 + 7 days: 3 Control and 4 KD mice, Mann-Whitney’s U test; P6 + 14 days: 3 Control and 4 KD mice, Student’s t test.).

(J) Changes in NR1 signal 6 h, 24 h, 7 days, and 14 days after 5-Ph-IAA administration in KD mice at P6. Data shown in Figures 2E–2I were used. NR1 signals of control mice were adjusted to 100%. (Tukey’s multiple comparisons test).

(K) Control and KD mice were administered 5-Ph-IAA at postnatal week 1 (PW1: P5–P8), and brains were sampled 6 h later to prepare cortical slices. The amplitude of AMPA- and NMDA-mediated excitatory postsynaptic currents (EPSCs) was plotted (n = 7 and 6 cells from 3 control and 2 KD mice, respectively).

(L) The NMDA/AMPA ratio of the data shown in (h) (Welch’s t-test).

(M) Barrel maps were detected in Grin1^mAID/mAID (Control) (n = 2) but were completely absent in CAG-TIR1;Grin1^mAID/mAID (KD) (n = 3) mice. Both mice were administered 5-Ph-IAA at P3 and sampled at P6. Tangential sections (100-μm thick) of cortices were used for cytochrome oxidase staining. Scale: 1 mm

Mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001, ns: not significantly different.

Previous studies have demonstrated that Grin1 KO (Grin1^−/−) mice cannot drink milk due to suckling reflex defects and die within a day after birth.^37,38 To validate the NR1.mAID (#506) mice, we induced NR1-protein KD during the embryonic stages and compared their phenotypes with the Grin1 KO mouse phenotypes. Global TIR1-expressing (CAG-OsTIR1[F74G]-P2A-mAID.EGFP, referred to as CAG-TIR1) mice were crossed with Grin1^+/− mice to obtain CAG-TIR1;Grin1^+/− mice. Then, these mice were crossed with Grin1^mAID/mAID (or Grin1^mAID/+) mice and the pregnant mothers were administered 5-Ph-IAA twice on embryonic day (E)17.5 and E18.5. Among 26 pups, all five CAG-TIR1;Grin1^mAID/− (NR1-protein KD) pups died within 24 h after birth, and no milk was found in their stomachs. In contrast, all 21 littermate controls, including 3 Grin1^mAID/− pups, survived, and their stomachs were filled with milk. Thus, AID2-mediated NR1-protein KD induced from the embryonic stages reproduced the lethal phenotype of Grin1 KO mice.

AID2 disrupts NMDAR function in the postnatal brain

We administered 5-Ph-IAA to CAG-TIR1;Grin1^mAID/mAID (KD) pups at P1 or P2 to test whether the AID2 system can disrupt the NMDAR function in the postnatal mouse brain. Although these pups initially displayed normal suckling reflexes and milk-filled stomachs, they lost the reflex within 3 h, displayed empty stomachs within 24 h, and failed to gain weight (Figures 2B–2D). These results suggest that AID2 enabled the degradation of a membrane protein (i.e., NR1) that functions at the synapse.

To quantitatively assess the efficiency of AID2-mediated NR1 KD in the postnatal mouse brain, we performed western blot analyses. We administered 5-Ph-IAA to CAG-TIR1;Grin1^mAID/mAID (KD) and Grin1^mAID/mAID (Control) mice at P6 or P12 and quantified NR1-protein levels in the forebrain (Figures 2E–2J). NR1 levels were reduced to 27.2% ± 0.9% after 6 h and to 11.5% ± 0.6% after 24 h of 5-Ph-IAA administration at P6 in KD mice compared with control mice (Figures 2E and 2F). Similarly, 5-Ph-IAA administration at P12 drastically reduced NR1-protein levels (16.5% ± 0.9% after 24 h) in KD mice (Figure 2G). NR1 protein levels returned to 74.9% ± 9.1% and 104.8% ± 11.0% within 7 days and 14 days, respectively, in KD mice administered with 5-Ph-IAA at P6 (Figures 2H–2J). These results indicate that AID2 induced NR1 KD rapidly and efficiently in the postnatal mouse brain and that the effect lasted for more than 24 h, but not 1 week. We also examined the efficiency of NMDAR disruption using electrophysiology. Cortical slices were prepared from CAG-TIR1;Grin1^mAID/mAID (KD) and Grin1^mAID/mAID (Control) pups at P5–P8 6 h after 5-Ph-IAA administration, and whole-cell patch-clamp recordings were performed on barrel cortex L4 neurons. We compared the NMDAR- and AMPAR-mediated excitatory postsynaptic currents; the NMDA-current was almost undetectable (4.4% ± 1.3% of AMPA-current) in the KD mice (Figures 2K–2L).

In the mouse barrel cortex L4, barrel maps emerge during the first postnatal week in an NMDAR-dependent manner.^23,26,39 We previously demonstrated that mice with low levels of NR1-gene expression survived postnatally but failed to form barrel maps.³⁹ We here examined whether postnatal NR1 KD could reproduce a similar phenotype. When CAG-TIR1;Grin1^mAID/mAID (KD) and Grin1^mAID/mAID (Control) mice were administered 5-Ph-IAA at P3, barrel maps were clearly visualized in control mice but completely absent in KD mice at P6 (Figure 2M). These results further confirm that mAID-tagged NR1 constituted functional NMDARs in vivo and that NMDAR functions were impaired by postnatal 5-Ph-IAA administration to NR1.mAID mice with global TIR1 expression.

Taken together, our results demonstrated that AID2 enables rapid, efficient, and reversible NR1 KD in the postnatal mouse brain.

An “immediate” role of NMDARs in forming dendritic patterns revealed by NR1 knockdown induced from P3

Mature barrel cells located at the barrel edges expand the dendrites asymmetrically toward single barrels.^5,9,40 This unique morphological feature of barrel-cell dendrites is formed essentially between P3 and P6.¹³ The inner and outer-dendritic trees of barrel cells are similarly primitive at P3. Then, between P3 and P6, only a few inner dendritic trees are elaborated, while other (both inner and outer) trees remain primitive. This inner tree-specific elaboration produces the barrel cells’ highly asymmetric dendritic projection pattern.¹³ After P6, the overall asymmetry does not change much, although dendrites continue to grow for some time.^13,17,23

First, we analyzed whether single-cell NR1-protein KD induced from P3 displayed similar phenotypes to the single-cell Grin1 KO previously reported.²³ To visualize the dendrite morphology and induce AID2-mediated NR1 KD simultaneously in single cells, we combined the Supernova system^23,36 with the AID2 system. Supernova enables sparse and bright cell labeling and the simultaneous expression of multiple genes in the labeled cells.³⁶ We co-transfected L4 neurons of Grin1^mAID/mAID mice with Supernova-red fluorescent protein (RFP)/TIR1 by in utero electroporation (IUE) at E14.5 (Figure 3A), which induces NR1 KD upon 5-Ph-IAA administration only in RFP-positive neurons (Figure 3B). We administered 5-Ph-IAA three times at P3, P4, and P5 and analyzed barrel cells at P6 (Figure 3C); these neurons were referred to as P6[P3-P6KD]. As controls (P6[Control]), Supernova-RFP/TIR1-labeled barrel cells in Grin1^mAID/mAID mice without 5-Ph-IAA administration were used at P6.

Figure 3.

P6[P3-P6KD] barrel cells fail to form asymmetric dendritic patterns

(A) For the sparse labeling of barrel cells and labeled cell-specific NR1 KD, Supernova-RFP/TIR1 vectors were introduced to Grin1^mAID/mAID mice by in utero electroporation (IUE) on embryonic day (E)14.5. TRE: tetracycline responsive element; LSL: loxP-stop-loxP; ires: internal ribosome entry site; tTA: tetracycline transactivator; WPRE: woodchuck hepatitis virus post-transcriptional regulatory element.

(B) Schematics of single-cell NR1 KD. When barrel cells are sparsely labeled with Supernova-RFP/TIR1, TIR1 is expressed only in RFP-positive cells. As soon as 5-Ph-IAA is administered, mAID-tagged NR1 is rapidly degraded via the ubiquitin-proteasome pathway in RFP-positive cells (red) but not in other cells (gray). TCA: thalamocortical axon.

(C) To induce NR1 KD from P3 onward, 5-Ph-IAA was administered intraperitoneally to Grin1^mAID/mAID mice at P3, P4, and P5. Control mice were not administered 5-Ph-IAA. The brains were collected at P6, and tangential slices (100-μm thick) were prepared. Supernova-RFP/TIR1-labeled barrel cells in Grin1^mAID/mAID mice with and without 5-Ph-IAA administration were referred to as P6[P3-P6KD] and P6[Control] neurons, respectively.

(D) Representative images and traces of P6[Control] and P6[P3-P6KD] barrel cells. Barrels visualized by the TCA clusters were shown in green. Scales: 300 μm (left) and 50 μm (center, right).

(E) A schematic showing the inner (magenta) and outer (cyan) dendritic segments. Dashed green line: barrel border. See STAR Methods for the details.

(F) Comparison of dendritic lengths of P6[Control] and P6[P3-P6KD] barrel cells (n = 7 and 11 cells, from 4 to 6 mice, respectively). Total, inner, and outer-dendritic lengths (p = 0.6120, 0.0380, and 0.0033, respectively; Student’s t test). The dendritic length-orientation bias index (OBI), defined as the ratio of the inner dendrite length to the total length (p = 0.0002, Student’s t test).

(G) A schematic showing the inner (magenta) and outer (cyan) dendritic tips.

(H) Comparison of dendritic tip numbers of P6[Control] and P6[P3-P6KD] cells. Total, inner, and outer tip numbers (p = 0.8240, 0.1087, and 0.0192, respectively; Student’s t test). The tip-OBI, defined as the ratio of the inner-tip number to the total tip number (p = 0.0069, Student’s t test).

Boxplots represent median, 25%–75% range (box) and min-max range (whiskers).^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. ns: not significantly different.

Consistent with previous reports,^13,17 P6[Control] neurons already exhibited clearly asymmetric dendrites, which oriented toward the barrel center (Figure 3D). In P6[P3-P6KD] neurons, this dendritic asymmetry was severely impaired. Quantitatively, the inner-dendrite length was shorter and the outer-dendrite length was longer than in P6[Control] neurons; therefore, the dendritic length-orientation bias index (OBI), defined as the ratio of the inner-dendrite length to the total length (see STAR Methods for details), was much lower than in P6[Control] neurons (Figures 3E and 3F). Conversely, the total dendritic length of P6[Control] and P6[P3-P6KD] neurons was similar. Similar results were obtained after quantifying the number of dendritic tips. The tip number-OBI, defined as the ratio of the inner-tip number to the total tip number, was also much lower in P6[P3-P6KD] neurons than in P6[Control] neurons, although the total tip number was similar (Figures 3G and 3H). These results suggest that NMDARs are indeed required for dendritic asymmetry formation between P3 and P6 when the process is ongoing.¹³

Moreover, our present results, which demonstrated that P6[P3-P6KD] neurons exhibited rather symmetric dendritic patterns similar to previously reported Grin1 KO neurons,²³ confirmed that Supernova-mediated AID2 worked well to diminish NMDAR function in single barrel cells.

Role of NMDARs in “maintenance” of dendritic asymmetry revealed by NR1 knockdown induced from P6

We analyzed what would happen if NMDARs lost their functions after the strong dendritic asymmetry of barrel cells was already formed. We knocked down NR1 from P6 onward and analyzed the dendrite asymmetry at P9. Supernova-RFP/TIR1-labeled barrel cells in Grin1^mAID/mAID mice with and without 5-Ph-IAA administration were referred to as P9[P6-P9KD] and P9[Control] neurons, respectively (Figure 4A). Although P9[Control] neurons displayed clearly asymmetric dendrites, P9[P6-P9KD] neurons did not (Figure 4B). The inner dendrites were shorter, the outer ones longer, and the length-OBI was much lower in P9[P6-P9KD] neurons than in P9[Control] neurons (Figure 4C). Dendritic-tip number analyses also displayed similar results (Figure 4D).

Figure 4.

P9[P6-P9KD] barrel cells fail to “maintain” asymmetric dendrite patterns

(A) Schematic drawing of single-cell NR1 KD from P6.

(B) Representative images and traces of P9[Control] and P9[P6-P9 KD] barrel cells. Scales: 50 μm.

(C) Comparison of dendrite lengths of P9[Control] and P9[P6-P9KD] barrel cells (n = 8 and 4 cells, from n = 4 and 3 mice, respectively). Total (p = 0.2141, Mann-Whitney’s U test), inner (p = 0.0102, Student’s t test), and outer (p = 0.0012, Student’s t test) dendrite lengths. Length-OBI (p = 0.0040, Mann-Whitney’s U test).

(D) Comparison of dendritic tip numbers of P9[Control] and P9[P6-P9KD] cells. Total (p = 0.5698, Student’s t test), inner (p = 0.0240, Student’s t test), and outer (p = 0.0202, Mann-Whitney’s U test) tip numbers. Tip-OBI (p = 0.0003, Student’s t test).

(E) Comparison of dendrite lengths of P6[Control], P9[Control], and 9[P6-P9KD] cells. Total, inner, and outer dendrite lengths (p < 0.05, Steel’s test). The length-OBI (p < 0.05, Steel’s test).

(F) Comparison of tip numbers of P6[Control], P9[Control], and 9[P6-P9KD] cells. Total, inner, and outer tip numbers (p < 0.05, Steel’s test). Tip-OBI (p < 0.05, Steel’s test).

(G) Schematic drawing of single-cell NR1KD from P12.

(H) Representative images and traces. Scales: 50 μm.

(I) Comparison of dendrite lengths of P15[Control] and P15[P12-P15KD] cells (n = 8 and 6 cells, from 5 to 4 mice, respectively). Total, inner, and outer dendritic lengths (p = 0.9561, 0.7224, and 0.2967, respectively; Student’s t test) and length-OBI (p = 0.2696; Student’s t test).

(J) Comparison of tip numbers of P15[Control] and P15[P12-P15KD] cells.Total, inner, and outer tip numbers (p = 0.4314, 0.2643, and 0.8901, respectively, Student’s t test) and tip-OBI (p = 0.8506; Student’s t test).

Boxplots represent median, 25%–75% range (box) and min-max range (whiskers).

^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001. ns: not significantly different.

We then researched what happened between P6 and P9. We compared P6[Control] and P9[Control] neurons and observed that the total- and inner-dendrite lengths were larger in P9[Control] than in P6[Control], whereas the outer-dendrite length and length-OBI were similar between the two groups (Figure 4E). Dendritic-tip number analyses also exhibited similar results (Figure 4F). These results confirmed that the dendritic asymmetry formation of barrel cells was essentially finished by P6. Although the dendrites still grew after P6, the asymmetry did not increase much anymore. Conversely, when we compared P6[Control] and P9[P6-P9KD] neurons, the outer dendrites were longer, and the length-OBI was smaller in P9[P6-P9KD] than in P6[Control] neurons. In contrast, the inner-dendrite lengths were not different between the two groups (Figure 4E). Dendritic-tip analyses also showed similar results (Figure 4F). Thus, in P9[P6-P9KD] neurons, only outer dendrites elaborated between P6 and P9, severely disrupting the dendritic asymmetry. These results were surprising because we expected that both inner and outer dendrites should elaborate in P9[P6-P9KD] neurons if NMDARs are still required for dendritic patterning (see discussion). The findings revealed the critical necessity of the “maintenance” of asymmetric dendritic patterns and the essential role of NMDARs in this maintenance mechanism. Even after the asymmetric dendritic patterns were formed, the dendrites became rather symmetric as soon as the NMDAR function was lost.

Finally, we knocked down NR1 from P12 onward and analyzed the dendritic patterns at P15 (Figures 4G and 4H). We found that the inner- and outer-dendritic lengths and length-OBI did not significantly differ between P15[Control] and P15[P12-P15KD] neurons (Figure 4I). Dendritic-tip analyses exhibited similar results (Figure 4J). These results suggest that P15[P12-P15KD] neurons maintained dendritic asymmetry normally between P12 and P15. It appeared that barrel cells require NMDAR-dependent mechanisms to maintain dendritic asymmetry only transiently during the early postnatal period. However, because target proteins are not completely eliminated by AID2 (e.g., Figures 2E–2G), the possibility that the remaining low levels of NMDARs are sufficient to maintain dendritic asymmetry in later postnatal stages such as P12-P15 cannot be excluded.

Role of neonatal NMDARs in maintaining high dendrite-tree-length variance

Barrel cells at the barrel edge exhibited two unique features of dendritic patterns: the strong asymmetry described above and the high variance of dendritic-tree length (Figure 5A). The latter is also formed essentially between P3 and P6 by detecting the spatially biased inputs from the TCAs.¹³

Figure 5.

Analysis of dendritic-tree length variance of barrel cells A

(A) representative image of a normal barrel cell consisting of six dendritic trees (color-coded). Tree length (%) was defined as the ratio of individual-tree length to the total dendrite length of the cell.

(B–E) Comparison between P6[Control] and P9[Control] barrel cells (n = 35 and 44 trees, from 7 to 8 cells, 4 and 4 mice, respectively). Total tree number (p = 0.4841, Mann-Whitney’s U test) (B). Tree lengths (%) were plotted on histograms, in which each bin represents 5%. Histograms showed the presence of many short (<5%) trees and some extremely long (>50%) trees (C). Cumulative curves of tree lengths (%) (D). Frequencies of intermediate-length (5%–50%) trees. (p = 0.8214, Fisher’s exact test) (E).

(F–I) Comparison between P6[Control] and P9[P6-P9KD] (n = 27 trees, from 4 cells, 3 mice) cells. Total tree number (p = 0.0970, Mann-Whitney’s U test) (F). Histogram (G) and cumulative curve (H) of tree length (%). Frequencies of intermediate-length trees (p = 0.0201, Fisher’s exact test) (I).

(J–M) Comparison between P6[P3-P6KD] (n = 68 trees, 11 cells, 6 mice) and P9[P6-P9KD] cells. Total tree number (p = 0.4899, Student’s t test) (J). Histograms (K) and cumulative curves (L) of tree length (%). Frequencies of intermediate-length trees (p = 0.3792, Fisher’s exact test) (M).

(N–Q) Comparison between P15[Control] and P15[P12-P15KD] cells (n = 31 and 27 trees, 8 and 6 cells, 5 and 4 mice, respectively). Total tree number (p = 0.2370, Student’s t test) (N). Histograms (O) and cumulative curves (P) of tree lengths (%). Frequencies of intermediate-length trees (p = 0.5884, Fisher’s exact test) (Q).

Boxplots represent median, 25%–75% range (box) and min-max range (whiskers).

^∗p < 0.05. ns: not significantly different.

We analyzed what happened to the dendritic-tree-length variance when NR1 was knocked down postnatally. First, to confirm the high tree-length variance in normal barrel cells, P6[Control] and P9[Control] neurons were analyzed. The total dendritic-tree numbers did not significantly differ between the two groups (Figure 5B). Extremely long dendritic trees, accounting for a large portion (greater than 50%) of the total dendrite length of individual neurons, were often found in both. Short dendritic trees, accounting for less than 5%, were rich in both ages (Figures 5C and 5D). Accordingly, the frequencies of the intermediate-length (5%–50%) dendritic trees were similarly low (nearly half) in both groups (Figure 5E). Thus, the high dendritic-tree-length variance formed by P6 did not significantly change between P6 and P9 during normal barrel-cell development.

We then compared the tree-length variance between P6[Control] and P9[P6-P9KD] neurons (Figures 5F–5I). Intriguingly, in P9[P6-P9KD] neurons, extremely large dendritic trees were undetectable, and the frequency of short dendritic trees was lower than in P6[Control] neurons. Accordingly, the frequency of intermediate-length dendritic trees was higher in P9[P6-P9KD] neurons than in P6[Control] neurons. The high frequency of intermediate-length dendritic trees was a common feature between P6[P3-P6KD] and P9[P6-P9KD] neurons (Figures 5J–5M). Thus, the high dendritic-tree-length variance of barrel cells, which was formed essentially by P6, was quickly canceled by NMDAR depletion induced from P6 onward. On the other hand, NR1 KD from P12 onward did not display detectable differences in the dendritic-tree-length variance (Figures 5N–5Q). These results suggest that the high tree-length variance of barrel cells, once formed, must be maintained during the early postnatal period, and this maintenance mechanism depends on the cell-autonomous NMDAR function.

NMDARs are required to maintain the Golgi lateral polarity

We recently reported that the lateral polarity of the Golgi apparatus localization in the barrel-cell soma, which is oriented toward the barrel center, emerges between P3 and P5¹² (Figure 6A). This Golgi lateral polarization is NMDAR-dependent and required for the formation of asymmetric dendritic patterns in the barrel cells. If the NR1 gene (Grin1) is knocked out in a single barrel cell, the Golgi lateral polarity is not formed, even at P5 (Figure 6A). When Golgi polarity is disturbed by GRASP65 overexpression,⁴¹ barrel cells fail to form asymmetric dendritic patterns.¹²

Figure 6.

NMDARs are required to maintain the Golgi apparatus lateral polarity

(A) Schematics of the Golgi apparatus distribution. In normal barrel cells, the Golgi lateral polarity is formed between P3 and P5. In Grin1 KO barrel cells, the Golgi failed to form the polarity.¹²

(B) To visualize the dendrite morphology and the Golgi apparatus distribution, and to simultaneously express TIR1 only in these labeled cells, Supernova-RFP/TIR1/Golgi.EGFP vectors were introduced to Grin1^mAID/mAID mice by IUE at E14.5.

(C) Representative images of the dendrite morphology and Golgi distribution in P9[Control] and P9[P6-P9KD] barrel cells. High-magnification images are also shown. Dotted lines: boundaries between barrel-inner and outer compartments. Scales: 50 μm (left), 10 μm (right).

(D) The Golgi distribution bias index (Golgi-DBI) was calculated by dividing the Golgi volume in the barrel-inner-half by the total Golgi volume. The Golgi-DBI was decreased between P6 and P9 by NR1 KD (p = 0.0239, Student’s t test).

(E) The percentages of cells having dendritic tree(s) with Golgi deployment were decreased between P6 and P9 by NR1 KD (p < 0.0001, Fisher’s exact test). 0, 1, ≥2: Cells having no, one, and more than one dendritic trees with Golgi deployment, respectively.

(F) The deployed Golgi length, which is the length from the base of the dendritic tree to the distal tip of the elongated Golgi in the trees in each cell, was decreased between P6 and P9 by NR1 KD (p = 0.0087, Mann-Whitney’s U test).

(G) Schematics of the Golgi apparatus distribution. In P9[P6-P9KD] barrel cells, the Golgi apparatus failed to maintain the lateral polarity. (n = 6 Control and 5 KD cells).

Boxplots represent median, 25%–75% range (box) and min-max range (whiskers).

^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗∗p < 0.0001.

Here, we investigated what would happen in the Golgi polarity if the NMDARs lost their functions after the Golgi lateral polarity was already formed. To visualize the Golgi apparatus and cell morphology simultaneously, Golgi-targeting sequence-fused EGFP (Golgi.EGFP)¹² was expressed with RFP and TIR1 in a sparse population of barrel cells in Grin1^mAID/mAID mice using IUE-based Supernova. We knocked down NR1 from P6 onward and evaluated the Golgi polarity at P9 (Figure 6B). In P9[Control] neurons, the Golgi apparatus was polarized toward the barrel center (Figure 6C). We quantified the Golgi polarity as the Golgi distribution bias index (Golgi-DBI), which is the fraction of the Golgi apparatus in the inner-side half of the cell with respect to the barrel.¹² The Golgi-DBI at P9 was quite high (0.88 ± 0.06) (Figure 6D) and similar to what was previously reported in P5 and P7 barrel cells.¹² Conversely, in P9[P6-P9KD] neurons, the Golgi apparatus was found in both the inner- and outer-half of the soma (Figure 6C). The Golgi-DBI was much lower in the KD neurons than in the control neurons (Figure 6D). These results suggest that the Golgi polarity, once formed by P6, disappeared quickly in NR1 KD.

In normal barrel cells in neonates, the Golgi apparatus is also deployed into a few extremely elaborated inner dendritic trees (i.e., “winner” trees).¹² We counted the number of dendritic trees with the Golgi apparatus. In P9[Control] neurons, one or more dendritic trees per cell were deployed by the Golgi apparatus, whereas in P9[P6-P9KD] neurons, nearly half did not show any dendritic trees with the Golgi apparatus (Figure 6E). The deployed Golgi length, which is the length from the base of the dendritic tree to the distal tip of the elongated Golgi in the tree, was significantly shorter in P9[P6-P9KD] neurons than in P9[Control] neurons (Figure 6F). These phenotypes were similar to those of single-cell Grin1 KO barrel cells.¹² Thus, NMDARs are essential not only for the formation but also for the maintenance of the Golgi lateral polarity (Figure 6G). NMDARs likely maintain the strong dendritic asymmetry and high dendritic-tree-length variance of barrel cells by maintaining the Golgi lateral polarity.

Discussion

Specific dendritic patterns are essential for proper circuit function.^5,7,8,12 However, the temporal mechanisms guiding their establishment during postnatal critical periods remain unaddressed due to technological limitations. Using AID2, a recently developed temporally controlled protein KD system,³⁵ we induced NMDAR dysfunction in a sparse population of barrel cells at specific postnatal stages. Our results reveal that dendritic refinement involves the formation and maintenance of specific patterns during early postnatal development. Notably, two distinct features of barrel-cell dendrites—strong lateral asymmetry and high tree-length variance—remain susceptible to the absence of proper inputs beyond the formation phase. This inducible KD approach unveiled a previously unrecognized critical period and a maintenance role for NMDARs in dendritic refinement.

The immediate role of NMDARs during dendrite patterning

Our previous longitudinal in vivo imaging of neonatal barrel cells revealed that the unique features of barrel-cell dendritic patterns are formed essentially between P3 and P6.¹³ The barrel cells’ inner and outer-dendritic trees are equivalently primitive at P3.¹³ Then, between P3 and P6, only a few inner trees are elaborated to be winners, while other trees (both inner and outer) remain primitive.¹³ After P6, the overall dendritic patterns do not change much, although total dendrite lengths continue to increase for a while (See Figures 4, 5, and 7A).

Figure 7.

The “maintenance phase” in dendritic refinement revealed by inducible single-cell NR1 KD

(A) Schematic summary of phenotypes of developmental stage-specific single-cell NR1 KD. In normal barrel cells (wild-type, WT), the inner and outer-dendritic trees are similarly primitive at P3.¹³ Then between P3 and P6, only a fraction of inner tree(s) is elaborated to become “winner(s),” which forms two types of inhomogeneities (i.e., strong asymmetry and high tree-length variance) in barrel-cell dendritic patterns. After P6, total dendritic lengths continue to increase for some time, but the overall dendritic patterns do not change much.^13,17,23 When NMDARs are depleted from P3 by AID2/Supernova-mediated single-cell NR1 protein KD, barrel cells failed to demonstrate dendritic inhomogeneities even at P6, suggesting an “immediate” role of NMDARs in forming dendritic patterns. More importantly, when NMDARs are depleted from P6, but not from P12, dendritic inhomogeneities once formed are rapidly canceled. These results revealed the formation and “maintenance” phases in establishing mature dendritic patterns and critical roles of NMDARs in both mechanisms.

(B) The NMDAR-Golgi model for the maintenance of dendritic inhomogeneities. (Top) In normal barrel cells at P6 onward, NMDARs should be activated by thalamocortical inputs predominantly in inner trees, because the TCA termini are already clustered in the barrel hollow at this age. Presumably, among the inner trees, the “winner” tree(s) should have the strongest NMDAR signals, which somehow maintain the Gogi deployment within the winner tree(s). This highly polarized Golgi localization supports further elaboration of the winner tree(s). By this mechanism, the overall dendritic patterns of the barrel cells could be maintained after P6. (Bottom) When NMDARs are depleted from P6, the Golgi apparatus loses its polarity and leaves the winner tree. It appears that the “depolarized” Golgi is incapable of maintaining the inhomogeneities in the barrel-cell dendrites anymore and the barrel cells are changed to exhibit the symmetric dendritic patterns with low tree-length variance.

Before conducting NR1[P3-P6KD] experiments, we considered two possibilities: (1) if NMDARs are needed only for the commitment of dendritic refinement before P3, KD barrel cells should form normal dendritic patterns between P3 and P6; and (2) if NMDARs have immediate roles in dendritic refinement, KD barrel cells should “mildly” elaborate all (both inner and outer) trees between P3 and P6 and form rather symmetric and tree length-homogeneous dendritic patterns, such as Grin1 KO barrel cells.²³ Our results validate hypothesis 2 and suggest that NMDARs indeed contribute to dendritic patterning when it is ongoing.

During early neonatal stages, such as P5, “patchwork-type” spontaneous activity originating from the periphery coactivates TCA termini and L4 neurons within the same barrels.^42,43,44,45 This spontaneous activity likely activates NMDARs as co-incidence detectors^20,21 primarily in inner dendrites, which are synchronized with presynaptic TCAs. NMDAR activation in inner dendrites may drive spatially polarized dendrite distribution by inducing Golgi apparatus polarization in the barrel cells.¹² Additionally, sensory feedback from self-generated whisker movements (twitches)^45,46,47,48 and passive whisker sensation^48,49 during early postnatal development may also contribute to polarized dendritic patterning. Further studies are needed to explore the detailed mechanisms of NMDAR-dependent dendritic patterning.

“Maintenance” phase in dendritic refinement

Before conducting NR1[P6-P9KD] experiments, we considered two possibilities: (1) if NMDARs are not needed anymore after P6, KD barrel cells should continue to elaborate primarily the winner inner trees and (2) if NMDARs are still required for dendritic refinement after P6, KD barrel cells should stop distinguishing the inner and outer dendrites at P6 and elaborate the inner and outer dendrites equally. Intriguingly, our results indicated that both hypotheses were incorrect. Instead, only outer dendrites were elaborated between P6 and P9 in the KD barrel cells, making them rather symmetric. Thus, barrel-cell dendritic patterns that were formed by P6 were not gradually degraded but actively canceled when NR1 was knocked down. These results revealed the critical necessity of the “maintenance” of the polarized dendritic patterns once formed and the essential role of NMDARs in the maintenance mechanism.

How do NMDARs maintain the highly polarized dendritic patterns? We recently reported that barrel cells displayed an apical-to-lateral Golgi apparatus polarity shift during the dendritic patterning period (between P3 and P5).¹² This Golgi polarity shift is key in the NMDAR-dependent formation of the barrel cells’ dendritic asymmetry. When the NMDAR function was impaired in barrel cells by single-cell Grin1 KO, the Golgi apparatus failed to form a lateral polarity.¹² Furthermore, when the Golgi polarity was disturbed by overexpressing GRASP65,⁴¹ dendritic asymmetry was not formed.¹² Here, we demonstrated that the Golgi lateral polarity, once formed by P6, was disrupted quickly (within three days) by the induction of NMDAR dysfunction at P6 (Figure 6). These results indicated that even after P6, the Golgi lateral polarity must be maintained and that NMDARs are essential to maintain it.

Between P5 and P9, spontaneous activity in barrel cortex L4 gradually shifts from patchwork to broadly synchronized (across multiple barrels) patterns.⁴² However, by P6, TCA termini transmitting input from a single whisker are already clustered within the corresponding barrel, and barrel-cell dendrites are localized within that barrel. As a result, TCA inputs likely activate NMDARs exclusively in the inner dendrites, regardless of the spatial organization of the spontaneous and sensory activity transmitted via TCAs. It is reasonable to assume that among the inner dendritic trees, the winner trees have the strongest NMDAR signals. This strong NMDAR signaling may anchor the Golgi apparatus to the winner tree(s). The NMDAR-mediated maintenance of Golgi deployment could further elaborate only on the winner tree(s), resulting in the maintenance of a strong orientation bias of the dendritic projection and high length variance among the dendritic trees (Figure 7B).

Interestingly, the NR1 KD induction at P6 quickly disrupted the barrel cells’ polarized dendritic patterns and made them rather symmetric. How does this happen? Our previous in vivo time-lapse imaging revealed that barrel-cell dendrites are highly dynamic during the neonatal stages.^13,23 Furthermore, recent higher spatiotemporal-resolution imaging of barrel cells at P4, which detected precise changes in dendritic morphologies, revealed that the barrel inner and outer dendrites (trees and branches) emerged/elongated and disappeared/retracted at similarly high frequencies.³⁴ These results suggested that barrel-cell dendrites are highly dynamic; however, positive and negative motilities are balanced at the border of the presumable asymmetric “dendrite territory” at that age in the short term. In the long term, wild-type neurons may expand the territory without changing the overall asymmetric pattern. On the other hand, NMDAR-deficient barrel cells may have a symmetric dendritic territory and, in the long term, expand the territory without changing the overall symmetric pattern. It is likely that the polarity of the Golgi—whether polarized or unpolarized—determines whether the dendritic territories are asymmetric or symmetric, respectively. In this scenario, when NMDAR function is impaired at P6, the territory could suddenly shift from an asymmetric to a symmetric pattern, because the Golgi apparatus loose polarity rapidly. Then, the balance between positive and negative dendritic motility could be disrupted until the dendritic tips reach the border of the new dendritic territory, which could be determined by the unpolarized Golgi apparatus.

NR1[P12-P15] KD failed to disrupt the normal patterns of the barrel-cell dendrites (Figures 4G–4J and 5N–5Q), suggesting that the maintenance necessity could be transient only during the early postnatal stages. This was consistent with our recent finding indicating that barrel cells do not display lateral Golgi polarity at P15.¹² It is likely that at this age, asymmetric and highly tree-length variable dendrite morphologies are already stable and do not need to be maintained by NMDAR-dependent mechanisms anymore. Additionally, remaining low levels of NMDARs might be sufficient to maintain the polarized dendritic patterns in later postnatal stages such as P12–P15. Between P6 and P12, many factors intrinsic or extrinsic to barrel cells change their characteristics in the barrel cortex L4, which may affect differences in the stability of the dendritic patterns. For example, barrel-cell dendrites increase spine densities drastically between P9 and P11.⁴² The AMPAR/NMDAR ratio significantly increases, and the NMDAR subunit composition changes in barrel cells between P3-P6 and P9-P11.⁵⁰ Spatiotemporal patterns of spontaneous activity in the barrel cortex L4 exhibit a patchwork-type pattern at P1–P5 but widely synchronized and sparse patterns around P9 and P11 onward, respectively.^42,45,51 Barrel nets, which are L2/3 neuron axons in L4 septal regions, are invisible at P6 and become visible at P10.⁵² Subplate neurites are clustered in the barrel hollow at P6 but are reorganized to be located at the barrel septa by P10.⁵³ The number of microglia significantly increases in the L4 barrel hollow between P5 and P9.⁵⁴ These developmental changes in the features of the barrel cells and barrel cortex L4 may affect the stability and/or NMDAR-dependency of the barrel-cell dendritic patterns.

Gene KO and RNAi approaches have identified dozens of molecules involved in forming asymmetric dendritic patterns in barrel cells, including metabotropic glutamate receptor 5,⁵⁵ FGF receptors,⁵⁶ protein kinase A,⁵⁷ tropomyosin receptor kinase A,⁵⁸ adenylyl cyclase 1,⁵⁹ Ras GTPase-activating proteins,⁶⁰ RhoA,²⁴ BTBD3.⁶¹ Furthermore, the IUE-mediated expression of the dominant-negative forms of Rac1 or Tiam1 in barrel cells impairs the dendritic asymmetry and Golgi lateral polarity.¹² Some of these and/or similar molecules could be involved in maintaining the polarized dendritic patterns.

AID2 system: A newly emerging approach for understanding postnatal brain development in a high-temporal-resolution

Highly organized neuronal circuits in the neocortex underlie higher brain functions in mammals. Neocortical circuits are coarse when animals are born but are reorganized postnatally to establish mature connectivity in an activity-dependent manner.^2,3,4 Gene-KO approaches have contributed significantly to identifying molecules that are essential for postnatal circuit refinement, including dendritic refinement.^10,39,62,63 Region- and/or cell-type-specific gene KO using the Cre/loxP recommendation has identified “where” these molecules work.^26,59,64,65 Single-cell gene KO and RNAi further revealed cell-autonomous functions.^17,23,61 However, in these previous approaches, gene functions were impaired from embryonic stages; therefore, the phenotypes obtained at a specific time point could be a complex accumulation of abnormalities and compensations during the earlier development. Moreover, phenotypes in earlier developmental stages may conceal crucial mechanisms in later development. Thus, the temporal mechanisms of developmental circuit refinement remain unexplored.^30,66,67

Some methods for the temporal regulation of gene functions, including CreERT2 and Tet-OFF, have been used in adult mouse brains.^68,69 However, with these methods, the depletion of the gene product (protein) is very slow. For example, in a study that used Tet-OFF-based inducible BDNF-gene KO, the authors could use mice for the experiments three months after inducing the KO.³¹ In another study that used a different Tet-OFF-based system for inducible Grin1 KO in mice, NR1 protein levels decreased to 27.5% of those in control mice 5 days after inducing the KO.³² The CreERT2-based inducible gene-KO system is similarly slow and often inefficient.^33,70

Here, we propose that AID2, a recently developed protein KD system,³⁵ is an ideal tool to study the temporal mechanisms of developmental circuit refinement. Because this system targets the gene product (protein) directly, the induction of gene-function impairment is much faster than that of previous methods targeting the DNA or RNA. AID2 does not require a high ligand concentration and thus works in many organ types in living mice, including the heart, lung, and kidney. Although AID2 works less efficiently in the adult mouse brain, possibly because 5-Ph-IAA is blocked by the BBB,³⁵ we here demonstrated that it can efficiently induce protein degradation in the postnatal mouse brain, where BBB could be immature^71,72,73 (Figures 1 and 2).

Furthermore, our results demonstrated that NR1 proteins returned to their original levels within 2 weeks after 5-Ph-IAA administration (Figures 2H–2J). This recovery was not as rapid as the decrease in protein levels after ligand application. Nevertheless, this reversible feature of AID2 will allow us to address the effects of abnormal circuit reorganization on adult brain functions in future research.

To the best of our knowledge, this study provides the first evidence that AID2 works for a membrane protein that functions at synapses in living mice. If an inducer that efficiently penetrates the BBB and induces better degradation in the adult brain is developed in the future, we expect this methodology to be even more powerful for neurobiology.

NMDAR functions in suckling behavior and barrel formation

Apart from the roles of NMDARs in dendrite refinement, postnatal NR1 KD highlighted additional insights into NMDAR functions. (1) Gene-KO mice of Grin1 or Grin2b exhibit impaired suckling responses and die within 24 h without milk in their stomach.^37,38,74 In the current study, when the NR1 protein was knocked down globally in neonatal mice, they stopped drinking milk (Figures 2B–2D). This result excluded the possibility that the suckling defect phenotype of KO mice is ascribed to an abnormality in forming neuronal circuits for the suckling response. Instead, NMDAR function at synapses in postnatal mice is critical. (2) Global gene KD and cortex-specific gene KO of Grin1 result in impaired barrel map formation,^26,39 whereas the pharmacological blockade of NMDARs in the newborn rat barrel cortex fails to impair barrel map formation.⁷⁵ An interpretation of this discrepancy is that NMDARs are required not postnatally but only during the embryonic stages for the “commitment” of barrel formation. Our results (Figure 2M) excluded this possibility and provided direct evidence for the postnatal role of NMDARs in barrel map formation.

Limitations of the study

This study has several limitations. First, the AID2 system cannot fully eliminate the target protein, so NMDAR-dependent mechanisms requiring minimal NR1 protein may have been overlooked. Second, we did not find the downstream mechanisms by which NMDARs form and maintain Golgi polarization and its deployment in the winner dendrite(s). Finally, the precise role of Golgi polarity in determining dendritic patterns remains unclear. These issues represent important areas for future research.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Takuji Iwasato (tiwasato@nig.ac.jp).

Materials availability

The knock-in mouse line and plasmids generated in this study will be made available upon request.

Data and code availability

• •Data: Accession numbers are listed in the key resources table. Original western blot images and microscopy data reported in this article will be shared by the lead contact upon request.
• •Code: DOIs are listed in the key resources table.
• •Additional information: Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We thank Tatsumi Hirata, Shingo Nakazawa, Naoyuki Matsumoto and Hiroki Uno for the critical reading of the article; Iwasato lab members for valuable discussion and comments; Noriko Yamatani (Division for Development of Genetic-Engineered Mouse Resource, NIG) for providing technical support in CRISPR-Cas9-mediated genome editing; Masahiro Takagi, Minako Kanbayashi, Satoko Kouyama, Michiyo Sato, Misaki Ebisawa and Mika Sugimoto for technical assistance; A.N. was supported by the SOKENDAI Special Researcher Program, JSPS Fellowship (DC2), the Teijin-Kumura scholarship and Sugiyama Hokokai; This work was supported by JST SPRING (JPMJSP2104) and JSPS KAKENHI (JP23KJ1001) to AN; JSPS KAKENHI (JP21H05702, JP23H04242 and JP24K02127) to NN; JSPS KAKENHI (JP21H04719 and JP23H04925) to MTK; JSPS KAKENHI (JP24H00586, JP24H02310, JP21K18245, JP20H03346), and the Cooperative Study Program (24NIPS103) of the National Institute for Physiological Sciences to T.I.

Author contributions

A.N. and T.I. conceived the project and designed the experiments with input from N.N. and M.T.K.; A.N. conducted most experiments and data analyses with the assistance of N.N., T.S., L.W., and T.I.; M.Y. and Y.Y. performed electrophysiological experiments and data analyses; A.N., T.S., R.A., Y.S., and T.I. designed and generated the NR1.mAID mouse; R.A. and Y.S. provided mAID.EGFP mice; M.T.K. provided AID2-related plasmids, reagents, and information; A.N. and T.I. wrote and revised the article with input from all authors.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-Actin	Millipore	Cat# MAB1501; RRID: AB_2223041
Mouse monoclonal anti-NMDAR1	Invitrogen	Cat# 32-0500; RRID: AB_2533060
Anti-mouse IgG, HRP-linked antibody	Cell Signaling Technology	Cat# 7076; RRID: AB_330924
Rabbit polyclonal anti-VGluT2	Synaptic Systems	Cat# 135403; RRID: AB_887883
Alexa Fluor 647 Goat anti-rabbit IgG	Invitrogen	Cat# A21244; RRID: AB_2535812
Chemicals, peptides, and recombinant proteins
5-Ph-IAA	BioAcademia	Cat# 30-003
TrueCut Cas9 protein v2	Invitrogen	Cat# A36496
NEBuilder HiFi Assembly	New England Biolabs	Cat# E2621
Guide-it Long ssDNA Production System	TaKaRa Bio	Cat# 632666
cytochrome C	Sigma	Cat# C7752
3–3′-diaminobenzidine tetrahydrochloride	Nacalai Tesque	Cat# 11009-54
PRO-PREP Protein Extraction Solution	iNtRON Biotechnology	Cat# 17081
EzWestLumi Plus	ATTO	Cat# WSE-7120S
DAPI	Roche	Cat# 10236276001
Experimental models: Organisms/strains
Mouse: CAG-OsTIR1(F74G)-P2A-mAID.EGFP	Yesbolatova et al.35	N/A
Mouse: NR1.mAID	This study	N/A
Mouse: NR1-null	Li et al.37	N/A
Mouse: TCA-GFP	Mizuno et al.23	N/A
Oligonucleotides
See method details for a complete list of sequences
Recombinant DNA
pTRE-Cre (pK031)	Mizuno et al.23	RRID: Addgene_ 69136
pCAG-LSL-RFP-ires-tTA-WPRE (pK029)	Mizuno et al.23	RRID: Addgene_ 69138
pCAG-LSL-Golgi-EGFP-WPRE (pK359)	Nakagawa and Iwasato12	N/A
pCAG-LSL-OsTIR1(F74G)-WPRE (pAN025)	This study	N/A
Software and algorithms
CRISPOR	Concordet and Haeussler76	RRID: SCR_015935
CLUSTALW	Thompson et al.77	RRID: SCR_017277
LAS-AF	Leica Microsystems	RRID: SCR_013673
Prism (v10.2.0)	GraphPad	RRID: SCR_002798
Imaris	Bitplane	RRID: SCR_007370
MEPHAS	Zhou et al.78	DOI:10.1186/s12859-020-3494-x
Fiji (ImageJ)	Schindelin et al.79	RRID: SCR_002285
Excel	Microsoft	RRID: SCR_016137

Experimental model and study participant details

Mice

All experiments were performed according to the guidelines for animal experimentation of the National Institute of Genetics (NIG) and the National Institute for Physiological Sciences (NIPS) and were approved by the animal experimentation committee of both institutes [study approval numbers: R3-10, R4-16, R5-9, R6-6 (NIG) and 23A052, 24A067 (NIPS)]. All mice were housed in a room temperature-controlled at 23°C ± 1°C, with 50% ± 4% humidity and maintained on a 12-h light/dark cycle with food and water available ad libitum. The day at which the vaginal plug was observed was considered as embryonic day (E)0 and E19 was defined as the day of the birth (postnatal day [P]0). Under continuous mating conditions, if pups were born by noon, that day was defined as P0. Mice of both sexes were used in the experiments. Genotypes were determined by PCR with the following primer sets: PRM045/PRM047, PRM050/PRM 051 and KS53/KS54 for NR1.mAID mice, CAG-OsTIR1(F74G)-P2A-mAID.EGFP (referred to as CAG-TIR1-mAID.EGFP or mAID.EGFP) reporter mice⁸⁰ and NR1-null mice,³⁷ respectively.

Method details

Analysis of reporter mice

mAID.EGFP reporter mice were administered 5-Ph-IAA (5 mg/kg) at P6, P15 or P30. Before (0 h) or 3 h or 6 h after 5-Ph-IAA administration, mice were anesthetized with an intraperitoneal injection of sodium pentobarbital (50 mg/kg) in saline and perfused with 0.9% NaCl and 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB) (pH 7.4). Wild-type (WT) mice at the same ages were used to determine background fluorescence. After the brains were rinsed with 0.1 M PB, coronal sections (400-μm thick) were made with a vibratome (Microslicer ZERO1, DSK, Kyoto, Japan). Images were taken with a fluorescence microscope (Leica M205 FCA) equipped with a digital camera (Leica DFC 7000-T). Fluorescence intensities (in the barrel cortices) were measured using the Fiji/ImageJ software (National Institutes of Health). The averaged fluorescence intensity of WT mice (n = 4, 4, and 3 mice for P6, P15 and P30, respectively) was used as the background for the same age. The EGFP signal of a reporter mouse was calculated by subtracting the background from the fluorescence intensities of the reporter mice.

Generation of mAID-fused NR1 knock-in mice

mAID-fused NR1 knock-in (NR1.mAID) mice were generated using the CRISPR-Cas9 system. mAID was fused to the C-terminus of NR1, which is located at the cytoplasmic side. Guide RNAs (gRNAs) were designed using the CRISPOR program (http://crispor.tefor.net/), and screened based on high MIT and CFD scores (>90), low off-target mismatches, and their proximity to the NR1 stop codon. Based on these criteria, the target sequence (CAGAGCCCCGGAGCACGACG) was selected and gRNA.NR1-mAID KI_F was synthesized by Integrated DNA Technologies (IDT).

The targeting vector for NR1.mAID knock-in (pAN009) was constructed as follows: A DNA fragment that spans from 0.85 kb upstream to 0.69 kb downstream of the NR1 stop codon was amplified from the C57BL/6J mouse genomic DNA by PCR using PRM031/PRM032 primers and Tks Gflex DNA polymerase (TaKaRa Bio) and cloned into the TOPO vector (Invitrogen). Resulting two plasmid clones (pAN016 and pAN017) were confirmed to be identical by sequencing using M13F/M13R primers. The 5′ arm, mAID and 3′ arm were isolated from pAN016, pMK287.mAID-Hygro,⁸¹ and pAN017 by PCR with PRM019/PRM004, PRM005/PRM029 and PRM030/PRM022 primer sets, respectively, and the KOD One DNA polymerase (Toyobo). Obtained three PCR products were inserted into HindⅢ/EcoRI sites of pUC18 vector using NEBuilder HiFi Assembly (NEW ENGLAND BioLabs), and the pAN009 was obtained. This vector was confirmed by sequencing using M13F/M13R/PRM033/PRM034 primers. Single-strand DNA (ssDNA) of pAN009 targeting vector was made using the Guide-it Long ssDNA Production System (632666, TaKaRa Bio) with phosphorylated PRM035/PRM036 primer sets. Alt-R CRISPR-Cas9 crRNA and Alt-R CRISPR-Cas9 tracrRNA were ordered from IDT.

B6C3F1 (C57BL/6N×C3H/HeN) female mice were super-ovulated and mated with B6C3F1 males, and fertilized embryos were collected from oviducts. To generate NR1.mAID mice, mixtures of the ssDNA, annealed crRNA and tracrRNA, and TrueCut Cas9 protein v2 (Invitrogen) were dissolved in injection buffer [10 mM Tris-HCl and 0.1 mM EDTA (pH 7.4)], and microinjected into pronuclei of fertilized eggs of B6C3F2 using manipulation system (NARISHIGE) equipped with FemtoJet 4i microinjector (Eppendorf). The injected zygotes were cultured in KSOM (Millipore) at 37°C under 5% CO₂ until they reached the two-cell stage after 1 day. Thereafter, two-cell stage embryos were transferred into the uterus of pseudo-pregnant MCH females at 0.5 days postcoitum and a total of 56 pups were obtained. Genomic DNAs of these pups and WT mice (Control) were amplified with PCR using four primer pairs: PRM031/PRM046, PRM044/PRM032, PRM047/PRM045 and PRM031/PRM032 (Figure 2A). PCR products were loaded on 2% agarose in TAE with a DNA ladder (DM2300, SMOBiO) to identify candidates of NR1.mAID founder mice, in which mAID is fused to C-terminus of endogenous NR1. Selected pups were further screened by direct sequencing. PCR products (300–500 ng) for PRM044/PRM032 and PRM031/PRM046 were treated with Tks Gflex DNA Polymerase, and then reacted with the BigDye Terminator v3.1 Cycle sequencing kit (Applied Biosystems) with PRM044 and PRM046 primers, respectively, according to the manufacturer’s instructions. The products were subsequently analyzed with the Applied Biosystems 3500 Genetic Analyzer (Applied Biosystems). Peaks were read manually or using the multiple sequence alignment software CLUSTALW (https://www.genome.jp/tools-bin/clustalw). We identified seven founders (#0506, #0513, #0766, #0770, #0772, #0777 and #0781), in which homologous recombination occurred at the correct locus and mAID was fused to the C-terminus of NR1. The #0777 pup exhibited a congenital eye abnormality. The #0506 line (NR1.mAID knock-in mouse) was used for all experiments after 2–3 times backcrosses with C57BL/6J mice.

Drug administration

5-Ph-IAA dissolved in phosphate-buffered saline (PBS) was administered intraperitoneally to the mice at a dosage of 5 mg/kg body weight³⁵ using a microsyringe (NanoFil syringe, World Precision Instruments, USA) fitted with a 33G needle.

Suckling reflex

Mouths of pups were gently touched with a pipette tip containing a small amount of milk and it was checked whether rhythmic movements of the lower jaws were induced or not. To confirm the impairment of sucking reflex by postnatal NR1 KD, body weights and the amounts of milk in the stomach of KD and control mice were also examined before and after the 5-Ph-IAA administration.

Cytochrome oxidase staining

Mice were anesthetized with pentobarbital, and perfused with 0.9% NaCl and 4% PFA in 0.1 M PB. Brains were collected and postfixed in 4% PFA in 0.1 M PB at 4°C overnight. Right hemispheres were flattened and cryoprotected in 30% sucrose in 0.1 M PB at 4°C for 24–48 h. Tangential sections (100-μm thick) were obtained with a ROM-380 freezing microtome (YAMATO, REM-710). Cytochrome oxidase (CO) staining was performed as previously described³⁹ Briefly, sections were incubated with CO stain solution [0.05% cytochrome c (Sigma)/0.08% 3–3′-diaminobenzidine tetrahydrochloride (Nacalai Tesque)/30% sucrose in 0.1 M PB] for 4 h at 37°C. Following visual detection of stain, sections were washed three times with 0.1 M PB and mounted on the slide glasses (FF-001, Matsunami Glass Ind., Ltd.) using EUKITT (Kindler).

Western blotting

Pups were perfused with 0.9% NaCl and brains were collected on ice. The right forebrain was homogenized in a protein extraction solution (PRO-PREP, iNtRON Biotechnology, Inc.) using a homogenizer (T25 ULTRA-TURRAX, IKA) for 5 s and clarified by centrifugation. Proteins (20 μg/lane) were separated by SDS-polyacrylamide gel electrophoresis with MULTIGEL II Mini 10 (Cosmo Bio, Tokyo, Japan) at 100 V for 90 min employing a minislab-type electrophoresis apparatus (DPE-1020, Cosmo Bio) and electroblotted to PVDF membranes (Hybond-P PVDF, Amersham Biosciences, Sweden) at 15 V for 30 min using a semi-dry transfer cell (Trans-Blot SD Cell, BioRad, USA). The membranes were stained with anti-actin monoclonal antibody (1:5,000, mouse, MAB1501, Millipore) or anti-NMDAR1 monoclonal antibody (1:500, mouse, 32–0500, Invitrogen), and anti-mouse IgG, HRP-linked antibody (actin; 1:3,000, NMDAR1; 1:500, 7076, Cell Signaling Technology). The bound antibodies were detected by chemiluminescence using EzWestLumi Plus (WSE-7120S, ATTO) with the ChemiDoc Touch Imaging System (BioRad). Intensities of bands were quantified using the Fiji/ImageJ software and background signals (the average of signals in two regions without samples in the same image) were subtracted. The NR1 protein levels were normalized with respect to the actin protein levels.

Electrophysiology

Coronal slices of barrel cortex (400-μm thick) were prepared from P5-P7 mice under deep anesthesia with isoflurane, recovered in a normal artificial cerebrospinal fluid (ACSF) containing (in mM): 126 NaCl, 3 KCl, 1.3 MgSO₄, 2.4 CaCl₂, 1.2 NaH₂PO₄, 26 NaHCO₃, and 10 glucose at 33°C for 1h, and maintained at room temperature (24°C-26°C). For whole-cell recording, patch pipettes (4–6 MΩ) were filled with a solution containing (mM) 130 Cs-gluconate, 8 CsCl, 1 MgCl₂, 0.6 EGTA, 10 HEPES, 3 MgATP, 0.5 Na₂GTP, and 10 Na-phosphocreatine (pH 7.3 with CsOH). Excitatory postsynaptic currents (EPSCs) were recorded from visualized layer 4 (L4) neurons under infrared-differential interference contrast (IR-DIC) optics. EPSCs were evoked by electrical stimulation using a pair of bipolar tungsten electrodes placed at the border between layer 6 and white matter. During EPSC recordings, 10 μM SR95531, a GABA_A receptor antagonist, was added to the ACSF to block inhibitory postsynaptic currents. AMPAR-mediated EPSCs were recorded at −70 mV, and NMDAR-mediated EPSCs were recorded at +40 mV in the presence of 10 μM 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX), a non-NMDA receptor antagonist. We selected cells with a high seal resistance (>1 GΩ) and a series resistance <30 MΩ.

In utero electroporation

In utero electroporation (IUE) was performed as described previously.²³ IUE was conducted on E14 morning to target L4 excitatory neurons. Pregnant mice were anesthetized via intraperitoneal injection of a combination of three anesthetics [medetomidine (0.3 mg/kg, Zenoaq, Domitor), midazolam (4 mg/kg, Maruishi Pharmaceutical, Dormicum), and butorphanol (5 mg/kg, Meiji Seika Pharma, Vetorphale) in saline]. A DNA solution, diluted in ultrapure water and containing 5% trypan blue (T8154, Sigma) was injected into the lateral ventricle of the embryonic brain using a pulled glass capillary (2-000-050, Drummond). The head of the embryo was then clasped with tweezer-electrodes (CUY650P5, Nepa Gene, Chiba, Japan) and a series of three to five electric pulses (35 V for 50 ms, with 950 ms intervals) were delivered using an electroporator (Nepa Gene). Following the procedure, atipamezole (0.3 mg/kg, Zenoaq, Antisedan) in saline was administered intraperitoneally to reverse the effects of anesthesia.

For Figures 3, 4, and 5, a mixture of pK031.TRE-Cre (5 ng/μL),²³ pK029.CAG-loxP-STOP-loxP (LSL)-RFP-ires-tTA-WPRE (1 μg/μL)²³ and pAN025.CAG-LSL-OsTIR1(F74G)-WPRE (1 μg/μL) (constructed in this paper as described below), was used. For Figure 6, a mixture of pK031 (5 ng/μL), pK029 (1 μg/μL), pAN025 (1 μg/μL) and pK359.CAG-LSL-Golgi.EGFP-WPRE (1 μg/μL)¹² was used.

Plasmid construction

To construct pAN025, the WPRE sequence was extracted from pK038.CAG-LSL-EGFP-ires-tTA-WPRE³⁶ by NotI digestion and subsequently inserted into the NotI site of pAN018. To construct pAN018, the OsTIR1(F74G) sequence was excised from pAN011 via SalI/NotI restriction digestion and ligated into the SalI/NotI sites of pK038. To construct pAN011, the OsTIR1(F74G) sequence from pK349.UC57v-CcvOsTIR1(F74G) was PCR amplified using the primers PRM038 and PRM039, and then inserted into the SalI/NotI sites of pK068.CAG-FRT-stop-FRT-EGFP-ires-tTA-WPRE.³⁶ All coding sequences in each construct were verified by DNA sequencing.

Analyses of the dendrite patterns

Mice were anesthetized with pentobarbital or isoflurane and perfused with 0.9% NaCl and 4% PFA in 0.1 M PB. Brains were collected and postfixed in 4% PFA in 0.1 M PB at 4°C overnight. Right hemispheres were flattened and cryoprotected in 30% sucrose in 0.1 M PB at 4°C for 24–48 h. Slices (100-μm thick) were obtained with a ROM-380 freezing microtome. The TCA clusters were visualized using TCA-GFP transgenic mice²³ or VGluT2-immunohistostaining. For VGluT2 immunohistochemistry, slices were incubated in a blocking buffer [PBS (pH 7.4) containing 3% goat serum and 0.2% Triton X-100] for 1 h at room temperature. Slices were then incubated with anti-VGluT2 antibody (1:1,000, 135403, SYSY) diluted in the blocking buffer for two overnight at 4°C followed by incubation with Alexa Fluor 647 goat anti-rabbit IgG antibody (1:1,000, A21244, Invitrogen) at room temperature for 2 h. To visualize the barrel walls, slices were incubated with 4′,6-Diamidino-2-phenylindole (DAPI) (2 μg/mL, 10236276001, Roche) at room temperature for 15 min. Slices were mounted on slide glasses (PLATINUM PRO, PRO-04, Matsunami Glass Ind., Japan) with an anti-fade mounting medium.⁸²

Three-dimensional (3D) fluorescence images were acquired using a confocal microscope (TCS SP5, Leica). Images of the barrel map and neurons were captured using a 10×/0.85 objective lens with a step size of 10.0-μm. For fine structure imaging, a 40×/0.85 objective lens was utilized with a step size of 1.0-μm. All images were acquired at a resolution of 1024 × 1024 pixels. The serial z stack confocal images were traced and analyzed using the IMARIS Filament Tracer (Bitplane) and Fiji/ImageJ.

For the quantitative analysis, we selected RFP⁺ neurons that met the following criteria, based on previous methods²³ with slight modifications: (1) Neurons that composed barrel walls (barrel-edge L4 neurons) in main barrels (rows A to E, arcs 1 to 5) and had adjacent barrels; (2) neurons that were brightly labeled with RFP, not overlapped with other RFP⁺ cells, and not located near surfaces of the sections. These are important for clear visualization of dendritic morphologies; (3) neurons that had no apical dendrites (barrel cells).

The “barrel edges” were delineated along the threshold boundaries of the TCA clusters determined by the Otsu’s threshold algorithm for each image. To determine barrel inner and outer dendrite lengths of a cell, a line (“barrel border”; Green dotted line in Figure 3E) that crossed the center of the soma and was parallel to the barrel edge was delineated. For analyses of Figures 3F, 4C, 4E, 4G, and 4K, the sum of lengths of dendritic segments located inside and outside the barrel border were defined as the inner and outer lengths of dendrites of the cell, respectively. If a dendrite segment crossed the barrel border, it was categorized based on the side where a larger proportion of the segment resided. The number of dendritic tips was also quantified using the same barrel border (Green dotted line in Figure 3G). If the distance from a tip to the branchpoint was less than 5 μm, it was not counted as a tip. The dendritic length-orientation bias index (OBI) was defined as the ratio of the dendrite segment length inside the barrel to the total dendrite length.²³ Similarly, the tip number-OBI was defined as the ratio of the number of dendritic tips inside the barrel to the total number of tips. For Figure 5, “tree length (%)” was defined as the percentage of the length of individual dendritic trees to the total dendrite length of the cell. Individual dendritic trees were classified into three categories [short trees (<5%), intermediate-length trees (5%–50%), and extremely long trees (>50%)] according to values of “tree length (%).”

The Golgi apparatus labeling and quantification

The Golgi apparatus lateral polarity was quantified as previously reported.¹² First, a circle of a diameter of 100 μm was drawn, centered on the soma. Second, a straight line (Line 1) was drawn through the intersections of the circle and the barrel edge. Finally, a second straight line, parallel to Line 1 and passing through the center of the soma, was drawn and used as the boundary between barrel-inner and outer compartment. The volume of the Golgi apparatus (Golgi.EGFP signal) located in the barrel inner or outer compartment was quantified using the “3D Objects Counter” in Fiji/ImageJ in 3D reconstructed images. The Golgi-distribution bias index (Golgi-DBI) was calculated by dividing the Golgi volume in the soma and dendrites belonging to the barrel inner compartment by the total Golgi volume. The lengths of Golgi deployment were measured from the base of the dendritic tree to the distal tip of the Golgi apparatus that enters the dendrite. If a cell contained Golgi apparatus in several dendrites, the total length was considered as the cell’s deployed Golgi length.

Oligo DNAs

PRM004: 5′-CACCGGATCCGCTCTCCCTATGACGGGAAC-3′

PRM005: 5′-TAGGGAGAGCGGATCCGGTGCAGGCGCC-3′

PRM019:

5′-TAAAACGACGGCCAGTGCCAAGCTTTGTGGTTTGCACCAACTTG-3′

PRM022:

5′-CAGCTATGACCATGATTACGAATTCGTCAGGTTGCAGAGAGCAGGC-3′

PRM029:

5′-CTCCCCCGACCCCGTCGACTTATTTATACATCCTCAAATCGATTTTCC-3′

PRM030: 5′-TATAAATAAGTCGACGGGGTCGGGGGAGGAGCACC-3′

PRM031: 5′-CCAGAAAACGGTTCCCTGTC-3′

PRM032: 5′-GGACACTGGGAGAGCTAGTC-3′

PRM033: 5′-GTCCAGCTGCTGAGATCC-3′

PRM034: 5′-GACCGAGGGATCTGAGAGG-3′

PRM035: 5′-CGTTGTAAAACGACGGCCAGTG-3′

PRM036: 5′-TCACACAGGAAACAGCTATGACC-3′

PRM038: 5′-CATGTCGACGATATCGCCACCATGACATACTTTCC-3′

PRM039: 5′-GCCGCGGCCGCTCACAGAATCTTCACAAAGTTGG-3′

PRM044: 5′-GTCATAGGGAGAGCGGATCC-3′

PRM045: 5′-CCGTGATATCAGTGGGATGG-3′

PRM046: 5′-GTGCTCCGTCCATTGATACC-3′

PRM047: 5′-CGCGACGCGCTATTGAGAGG-3′

PRM050: 5′-CCTGGTCGAGCTGGACGGCGAC-3′

PRM051: 5′-TCACGAACTCCAGCAGGACCATG-3′

KS53: 5′-TCAAGCAGACATGATCGTGG-3′

KS54: 5′-CTTCCATTTGTCACGTCCTG-3′

M13F: 5′-GTAAAACGACGGCCAGT-3′

M13R: 5′-CAGGAAACAGCTATGAC-3′

sgRNA

sgRNA.NR1-mAID KI_F:

5′-CAGAGCCCCGGAGCACGACGGUUUUAGAGCUAU-3′

Quantification and statistical analysis

Statistical analyses were performed using the Prism 10 software (GraphPad) or MEPHAS. For comparisons between two groups, Student’s t test was applied when the data were normally distributed and the variances were not significantly different. Welch’s t-test was used when the variances were different between the two groups. The Shapiro-Wilk test was used for the assessment of normality. The F-test was used to compare the variances. For parametric multiple comparisons, Tukey’s multiple comparison test was used. For non-parametric comparison of two groups, Mann-Whitney’s U test was used. For non-parametric multiple comparisons between the control group and other groups, Steel’s test was applied. Fisher’s exact test was used to determine if the proportions of categories in two group variables were significantly different from each other.

The asterisks in the figures indicate statistical significance as follows: ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001. P-values of <0.05 were considered statistically significant. ns indicates not significantly different. Error bars in the bar graphs and line graphs present standard error of the mean (SEM). In boxplots, the upper and lower limits of the box represent the 75th and 25th percentiles, respectively; horizontal lines represent the median, and upper and lower whiskers represent the maximum and minimum within 1.5 interquartile range. Pre-determination of sample sizes using statistical methods was not performed. Exact sample sizes for all the results were described in the figure legends.

Published: July 29, 2025