Pre- and Postsynaptic MEF2C Promotes Experience-Dependent, Input-Specific Development of Cortical Layer 4 to Layer 2/3 Excitatory Synapses and Regulates Activity-Dependent Expression of Synaptic Cell Adhesion Molecules

Early postnatal cell-autonomous and postsynaptic deletion of Mef2c selectively weakens L4 to L2/3 synaptic inputs

We previously reported that early postnatal deletion of Mef2c from postsynaptic L2/3 neurons weakened excitatory synaptic inputs from local columnar circuits but not inputs from the contralateral barrel cortex (S1; Rajkovich et al., 2017). Here, we set out to determine if MEF2C generally promotes local synaptic inputs but not inputs from long-range projections, including those from the ipsilateral cortex. To address this question, we deleted Mef2c in a sparse population (∼3–5%) of early postnatal S1 L2/3 neurons by injecting an AAV expressing a GFP-tagged Cre (AAV9.GFP-Cre) into the left lateral ventricle of postnatal day (P) 1 Mef2c homozygous floxed (Mef2c^fl/fl) mouse pups (Fig. 1A; Kim et al., 2013). We first confirmed that Mef2c deletion in L2/3 neurons weakened excitatory synaptic inputs from local columnar cortical circuits. To do this, we prepared acute coronal slices of the S1 cortex from the AAV-injected hemisphere of Mef2c^fl/fl mice at P18–P25 and performed simultaneous whole-cell patch-clamp recordings from pairs of GFP-Cre(+) (“Mef2c KO”) and neighboring GFP-Cre(−) or “Cre(−)” L2/3 neurons (Fig. 1B). We then mapped the strength of local excitatory synaptic inputs onto recorded cells using laser scanning photostimulation (LSPS) with glutamate uncaging as we have described previously (Rajkovich et al., 2017; Zhang et al., 2021; Fig. 1B). Briefly, slices were bathed in MNI-caged glutamate and stimulated with pseudorandom flashes of a UV laser beam at individual locations within a 16-by-16 grid spanning L1–L5. Laser stimulation focally released glutamate in the slice to evoke action potentials and synaptic transmission at that location. Synaptic transmission was measured through EPSCs recorded from an L2/3 neuron as each location on the map was stimulated. Amplitudes of monosynaptic EPSCs were converted into a color-coded map representing the strength of synaptic inputs from each location, and individual maps were aligned based on the position of the L4 “home,” or within column, barrel to generate an average map for each genotype (Fig. 1C). Responses from direct glutamate activation onto recorded neurons, as described in methods, were excluded from analysis and represented by black pixels. Compared to neighboring Cre(−) neurons, Mef2c KO neurons had a ∼50% decrease in the amplitude of EPSCs evoked from the L4 home barrel and from adjacent barrels in L4, consistent with previous findings (Rajkovich et al., 2017). EPSCs evoked from horizontal L2/3 or from L5 were not different between Cre(−) and Mef2c KO neurons (Fig. 1D). Additionally, Mef2c KO neurons did not differ from neighboring Cre(−) neurons with respect to resting membrane potential, input resistance or intrinsic excitability (Extended Data Tables 1-1, 1-2). In contrast to evoked EPSCs, the frequency of spontaneous miniature EPSCs (mEPSCs) measured in the presence of TTX was increased in Mef2c KO L2/3 neurons [Cre(−), 0.51 ± 0.03 Hz; Mef2c KO, 0.87 ± 0.07 Hz; p < 0.0001; paired t test], as previously shown (Rajkovich et al., 2017); and mEPSC amplitude was unchanged [Cre(−), 15.64 ± 0.42 pA; Mef2c KO, 16.08 ± 0.51 pA; n.s.]. This result indicates that MEF2C in postsynaptic L2/3 neurons selectively promotes evoked excitatory synaptic transmission from L4 neurons.

To measure synaptic inputs from long-range projections, such as the contralateral S1, we performed channelrhodopsin-assisted circuit mapping (CRACM) of labeled callosal axons. Mef2c^fl/fl mice were injected at P1 with AAV expressing mCherry-tagged channelrhodopsin into the right S1 and AAV9.GFP-Cre into the left lateral ventricle to sparsely delete Mef2c in the left S1 (Fig. 1E). At P18–P25, pairs of Cre(−) and Cre(+) Mef2c KO L2/3 pyramidal neurons surrounded by strong, visible ChR2-mCherry–labeled callosal axons were recorded in slices of the left S1 (Fig. 1F). Blue LED flashes centered over recorded somas evoked monosynaptic EPSCs directly from callosal axon terminals in the presence of tetrodotoxin and 4-AP (Petreanu et al., 2009; Zhang et al., 2021). Mef2c KO neurons had a trending increase in the strength of callosal synaptic inputs, as measured by EPSC amplitudes and compared with neighboring Cre(−) neurons (Fig. 1G; p = 0.08; n = 41 cell pairs). To determine if there is a callosal synaptic strength difference localized within a specific subcellular region, we mapped the spatial distribution of callosal synaptic inputs onto postsynaptic L2/3 neurons using blue laser stimulation (sCRACM; Petreanu et al., 2009). Using this method, we observed no difference between Cre(−) and Mef2c KO L2/3 neurons when we plotted the callosal input strength against vertical positions relative to the soma (Fig. 1H).

To determine if MEF2C-regulated ipsilateral long-range synaptic connections, we measured the input strength from the prefrontal/motor cortex to the barrel cortex. AAV9.ChR2-mCherry was injected into the prefrontal cortex ipsilateral to the ventricle injected with AAV9.GFP-Cre (Fig. 1I). At P18–P25, a strong and homogeneous labeling of axons in L1 was observed in the ipsilateral barrel cortex with some labeling in deep layers (Fig. 1J), which is characteristic of long-range projections from the prefrontal/motor cortex within the same hemisphere (Mao et al., 2011). A lack of ChR2-mCherry expression in L2/3 and L4 also indicated that there was no local transfection of ChR2 in the barrel cortex. To recruit more long-range axons, LED stimulation was centered on L1 above the recorded cell pair. Similar to callosal inputs from contralateral S1, we observed normal synaptic strength of ipsilateral long-range cortical inputs onto Mef2c KO L2/3 pyramidal neurons (Fig. 1K). Overall, these results indicate that postsynaptic MEF2C in L2/3 pyramidal neurons functions to selectively promote development and/or strengthening of synaptic inputs from L4.

Postsynaptic MEF2C is not required for maintenance of L4 to L2/3 synaptic strength

The development of local L4 to L2/3 synaptic connections in the barrel cortex depends on normal whisker sensory experience during a critical period that occurs in the second postnatal week for rodents (P10–P14; Lendvai et al., 2000; Stern et al., 2001; Shepherd et al., 2003; Wen and Barth, 2011). Based on previous results (Rajkovich et al., 2017), we hypothesized that postsynaptic MEF2C promotes experience-dependent development of L4 to L2/3 synapses. To further test this idea, we examined if postsynaptic MEF2C is required only during the experience-dependent critical period or is also necessary for the maintenance of L4 to L2/3 synapses after P14. To do this, we deleted Mef2c in postsynaptic L2/3 neurons by injecting AAV9.GFP-Cre into S1 barrel cortex of Mef2c^fl/fl mice at P14. Based on our previous study (Rajkovich et al., 2017), we expect Mef2c to be deleted by P20–P21. At P27–P32, we prepared acute slices and measured L4 synaptic input strengths onto neighboring pairs of Cre(−) and postsynaptic Mef2c KO L2/3 neurons with LSPS (Fig. 2A). To isolate the cell-autonomous role of postsynaptic MEF2C, we recorded from L2/3 neurons distanced from the injection site, where only sporadically transfected neurons can be found within the region, using LSPS mapping. With the late deletion of Mef2c, local L4 synaptic inputs onto postsynaptic Mef2c KO L2/3 neurons were of normal strength as compared with Cre(−) neighbors (Fig. 2B,C). This result demonstrates that postsynaptic MEF2C is required for the development, but not maintenance, of L4 to L2/3 synapses during an age that coincides with the experience-dependent critical period.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Postsynaptic deletion of Mef2c in L2/3 neurons after the second postnatal week has no effect on the strength of L4 synaptic inputs. A, Experimental design and timeline for local input mapping. B, Averaged color maps showing spatial distribution of local synaptic input strengths in response to LSPS. Scale bar, 100 µm. C, Mean synaptic input strength from the L4 home barrel (L4H, WT = 12.0 ± 1.1; KO = 10.2 ± 1.2 pA), adjacent barrels (L4A, WT = 4.2 ± 0.5; KO = 3.9 ± 0.5 pA), adjacent L2/3 (WT = 7.8 ± 1.3; KO = 7.4 ± 1.3 pA), and L5A (WT = 7.9 ± 0.8; KO = 8.3 ± 1.1 pA; n = 16 pairs) with values from individual cell pairs (open circles). See Extended Data Table 1-1.

Postnatal hetero- and homozygous deletion of Mef2c in excitatory neurons results in weak synaptic inputs from L4 to L2/3 neurons but does not affect L4 excitability

To study the molecular mechanisms by which MEF2C regulates cortical circuit development, we created a mouse with early postnatal deletion of Mef2c in a large population of cortical excitatory neurons. To do this, we crossed BAC-CaMKIIa-iCre mice (Casanova et al., 2001) to Mef2c^fl/fl mice. CaMKIIa-iCre is expressed beginning approximately P2–P3 and expresses in excitatory neurons in cortical layers 2–4 and 6 (Fig. 3A). Western blots of cortical lysates from CaMKII-Cre-Mef2c KO mice showed loss of full-length MEF2C at P21 (Fig. 3B). To assess the function of local cortical excitatory circuits, we performed LSPS on L2/3 pyramidal neurons in the S1 cortex from CaMKII-Cre-Mef2c^fl/fl (CaMKII-Mef2c KO) or CaMKII-Cre-Mef2c^fl/+ (CaMKII-Mef2c-Het) mice. Controls were Mef2c^fl/fl or Mef2c^fl/+ littermates without Cre [Cre(−); Fig. 3C]. Like postsynaptic, sparse deletion of Mef2c in L2/3 neurons, CaMKII-Mef2c KO mice had weak (∼50% reduction) EPSCs evoked from the L4 home barrel and adjacent barrels (Fig. 3D,E). Surprisingly, CaMKII-Mef2c-Het mice also had weak L4 home barrel inputs (Fig. 3D,E). The strength of synaptic inputs from adjacent L2/3 were unchanged onto CaMKII-Mef2c-Het or KO L2/3 neurons [Cre(−), 3.83 ± 0.71 pA; CaMKII-Mef2c-Het, 3.95 ± 0.77pA; CaMKII-Mef2c KO, 4.1 ± 0.40 pA; n = 20, 20, and 24 cells, respectively; n.s.).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Postnatal homozygous and heterozygous deletion of Mef2c in cortical excitatory neurons weakens L4 to L2/3 synaptic inputs. A, Confocal image (20×) of BAC CaMKII-iCre–driven tdTomato expression in layers (L) 2/3/4/6 of S1 cortex from a P21 mouse. Scale bar, 50 µm. B, Western blot of MEF2C from S1 cortical lysates of CaMKII-Cre-Mef2c littermates at P21. Full-length (FL) MEF2C is observed without Cre. Cre-mediated excision of exon 2 containing the MADS/MEF2 domain results in a truncated C-terminal fragment (arrowhead). C, Graphic representation of experiment paradigm for LSPS (D, E). D, Average color maps of local input strengths onto recorded L2/3 pyramidal neurons in Cre(−), CamKII-Mef2c-Het, and CaMKII-Mef2c KO mice. Scale bar, 100 µm. Bottom, Representative EPSC in response to LSPS in the center of the L4 home barrel from each genotype. Scale bar, 100 pA/25 ms. E, Mean synaptic input strength from the L4 home barrel [L4H, Cre(−) = 17.0 ± 2.0; n = 20; Het = 10.9 ± 1.4; n = 20; KO = 9.8 ± 1.2 pA; n = 25] and L4 adjacent barrels [L4A, Cre(−) = 6.1 ± 1.1; Het = 4.3 ± 0.6; KO = 3.4 ± 0.5 pA]. F, Schematic of recording and LSPS for loose-seal recordings in L4 of CaMKII-Mef2c mice. G, Representative action potentials from a loose-seal recording of L4 neurons in current clamp directly elicited by LSPS at different locations around the center of the recorded soma. H, Expanded trace from inset (dashed line) in G. Scale bar, 5 mV/50 ms. I, Mean (±SEM) of action potential number elicited per neuron by LSPS per genotype [Cre(−) = 6.6 ± 1.6; n = 25; Het = 9.5 ± 2.9; n = 18; KO = 9.8 ± 4.0; n = 19]. See Extended Data Tables 1-1 and 1-2.

Because CaMKII-iCre is expected to delete Mef2c in L4 neurons, the weak L4 to L2/3 inputs in the CaMKII- Mef2c KO and Het barrel cortex may be due to reduced LSPS glutamate-driven excitability of L4 neurons. To test this possibility, we performed loose-seal cell-attached recordings in L4 neurons from CaMKII-Mef2c KO, Het, and Cre(−) slices and compared action potential (AP) firing in response to LSPS-induced glutamate uncaging (Fig. 3F–I). The number of APs elicited with LSPS averaged between 4 and 6 spikes per cell, and there were no differences between CaMKII-Mef2c KO, Het, and Cre(−) L4 neurons. Consistent with this result, the intrinsic excitability of CaMKII-Mef2c KO and Het L2/3 neurons was unchanged (Extended Data Table 1-2). These results suggest that the reduced L4 to L2/3 LSPS-evoked EPSCs in CaMKII-Mef2c-Het and KO neurons are likely due to the weak synaptic strength of L4 inputs to L2/3.

Presynaptic loss of Mef2c in layer 4 neurons weakens L4 to L2/3 synaptic inputs through decreased probability of neurotransmitter vesicular release

The observation of weak L4 to L2/3 synaptic inputs in the CaMKII-Mef2c-Het mice is in contrast to our previous results that sparse, postsynaptic, and heterozygous deletion of Mef2c in L2/3 neurons had no effect on L4 synaptic input strength (Rajkovich et al., 2017). Thus, we hypothesized that heterozygous deletion of Mef2c in presynaptic L4 neurons results in weak synaptic connections to L2/3. To test this idea, we deleted Mef2c selectively in L4 neurons by crossing Mef2c^fl/fl mice with the Scnn1a-Tg3-Cre (Scnn1a-Cre) driver line, conferring Cre expression in the L4 cortex (Madisen et al., 2010). Scnn1a-Cre is reported to express between P3 and P14 in the cortex (Allen Brain Atlas; https://connectivity.brain-map.org/transgenic/experiment/127170789). To confirm this, we crossed Scnn1a-Cre mice to Ai14-tdTomato Cre-reporter mice (Madisen et al., 2010) or injected P1 pups with AAV9 expressing a Cre-dependent mCherry-tagged channelrhodopsin (ChR2-mCherry). Consistent with reported expression data, there was little to no Cre-reporter expression at P2–P6 in the S1 cortex (data not shown), but robust expression was observed in L4 barrels of S1 at P10 and P22 as visualized by ChR2-mCherry (Fig. 4A) or tdTomato (Fig. 4B), respectively. LSPS was used to map the strength of L4 to L2/3 synaptic inputs in slices from Cre(−), Scnn1a-Mef2c-Het, and Scnn1a-Mef2c KO mice. As in CaMKII-Mef2c-Het and KO mice, weak LSPS-evoked EPSCs from the L4 home barrel were observed in both Scnn1a-Mef2c KO and Scnn1a-Mef2c-Het as compared with Cre(−) mice (Fig. 4C,D). No changes were detected in intrinsic measures (Extended Data Table 1-1). This result suggests that MEF2C functions in presynaptic L4 neurons to promote development and/or strength of synapses onto L2/3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Homozygous or heterozygous deletion of Mef2c in presynaptic layer 4 neurons with Scnn1a-Cre weakens L4 to L2/3 synaptic inputs and decreases presynaptic release probability. A, L4-specific Scnn1a-Cre expression at P10 visualized using AAV9 Cre-dependent ChR2-mCherry. Scale bar, 100 µm. B, L4-specific Scnn1a-Cre expression at P22 visualized using Cre-dependent Ai14-tdTomato reporter mouse. Scale bar, 100 µm. C, Average color maps of local input strengths onto recorded L2/3 pyramidal neurons in Cre(−), Scnn1a-Mef2c-Het, and Scnn1a- Mef2c KO mice. Scale bar, 100 µm, 6 pixels (x-axis) by 5 pixels (y-axis). Bottom, Representative EPSC in response to LSPS in the center of the L4 home barrel from each genotype. Scale bar, 100 pA/25 ms. D, Mean synaptic input strength from the L4 home barrel [L4H, WT Cre(+) = 12.79 ± 1.352, Het = 8.709 ± 1.157, KO = 8.227 ± 0.7579, one-way ANOVA; Dunnett's multiple comparison). E, Top, Example image of ChR2-mCherry–labeled L4 barrels and axons from a Scnn1a-Cre mouse. Scale bar, 250 µm. Bottom, Experimental paradigm for E–G. The blue circle represents blue LED illumination. F, Representative EPSCs (average of 5) evoked with paired pulses of LED light (black arrows) at 200 ms (left) and 400 ms (right) interstimulus intervals (ISI). G, Left, Paired-pulse ratio (PPR) of peak amplitude at 200 and 400 ms ISI [200 ms PPR, WT Cre(+) = 0.58 ± 0.02, Het = 0.61 ± 0.03, KO = 0.71 ± 0.04 (n = 18, 15, 15); 400 ms PPR, WT Cre(+) = 0.64 ± 0.03, Het = 0.69 ± 0.03, KO = 0.79 ± 0.04 (n = 15, 12, 14), two-way ANOVA/mixed-effects model]. Statistics, *p < 0.05, ***p < 0.001. See Extended Data Table 1-1.

To determine if MEF2C in presynaptic L4 neurons strengthened synapses through enhancing presynaptic release probability, we measured short-term synaptic plasticity or paired-pulse ratio (PPR; see Materials and Methods) of evoked EPSCs from L4 neurons. PPR is inversely proportional to presynaptic glutamate release probability. For example, if weak EPSCs are mediated in part by decreases in release probability, we would expect a corresponding increase in PPR. To evoke PPR of EPSCs from L4 neurons, we injected AAV9 expressing a Cre-dependent ChR2-mCherry intraventricularly into the cortex of P1 Scnn1a-Cre mice. ChR2-mCherry expression was robust at P18 and confined to L4 in the cortex (Fig. 4E). At P18–P25, slices were prepared from injected Scnn1a-Mef2c KO, Het, and “WT” [Scnn1a-Cre(+); Mef2c^+/+] mice. EPSCs were evoked onto L2/3 pyramidal neurons with pulses of blue LED stimulation of L4 at two interpulse intervals (200 and 400 ms). At both intervals, EPSCs in all genotypes exhibited paired-pulse depression. However, Scnn1a-Mef2c KO neurons exhibited reduced paired-pulse depression, or increased PPR, compared with WT Cre(+) mice, which is consistent with a lower presynaptic release probability (Fig. 4F,G, Extended Data Table 1-1). An intermediate trend toward increased PPR was observed in Scnn1a-Mef2c-Het neurons, and a two-way ANOVA revealed a main effect of genotype on PPR (F_(1,83) = 7.35; p = 0.008). These results suggest MEF2C in presynaptic L4 neurons strengthens synaptic inputs onto L2/3 neurons by increasing presynaptic release probability.

Germline heterozygous deletion of Mef2c does not affect L4 and callosal synaptic input strength onto L2/3 neurons

Germline haploinsufficiency of MEF2C in humans is associated with MEF2C haploinsufficiency syndrome (MCHS), characterized by intellectual disability, autism, and epilepsy. A mouse model of MCHS was established from a germline deletion of Mef2c (Mef2c-Het) and displays autism-relevant behaviors (Harrington et al., 2020). Because we observe weak L4 to L2/3 synaptic inputs with postnatal, heterozygous deletion of Mef2c, with either CamKII-Cre or Scnn1a-Cre (Figs. 3, 4), we hypothesized that a similar phenotype may occur with germline haploinsufficiency of Mef2c and may be relevant to MCHS. To test this possibility, we measured L4 to L2/3 synaptic strength in P18–P25 WT and Mef2c-Het littermates with LSPS. In contrast to postnatal, cortical deletion of Mef2c, the strength of LSPS-evoked EPSCs from L4 was unchanged in the Mef2c-Het (Fig. 5A,B). This result agrees with previous findings of normal L4 to L2/3 synaptic transmission in Mef2c-Het mice evoked with electrical stimulation. Similar to L4 inputs, callosal input strengths onto L2/3 neurons were normal in Mef2c-Het mice as measured using blue light LED to evoke EPSCs from ChR2-mCherry expressing callosal axons (Fig. 5C–F). Taken together, these results suggest that compensatory mechanisms may function to maintain L4 synaptic strength with germline heterozygous deletion of Mef2c.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Normal L4 and callosal synaptic input strength onto L2/3 neurons from germline Mef2c-Het mice. A, LSPS color maps from germline Mef2c-Het and wildtype (WT) littermates. Scale bar, 100 µm. B, Mean LSPS elicited EPSC amplitude from the L4 home barrel (left; WT = 17 ± 3; Het = 18 ± 3 pA) and L4 adjacent barrels (right; WT = 10 ± 2; Het = 12 ± 2 pA; n = 20 WT and 18 Het cells). For adjacent barrel, average left and right barrel EPSCs were averaged together. C, Experimental design to label and measure callosal synaptic input strengths. D, Example EPSCs evoked with blue LED (2 ms; arrowhead) from callosal axons of a WT and a Mef2c-Het neuron. For each neuron, nine individual responses were recorded (gray traces) with 20–30 s flashing intervals, and an average EPSC response (black trace) was calculated and quantified. E, Raw average EPSC amplitudes from WT and Mef2c-Het neurons (WT = 1,165 ± 124 pA, n = 19; Het = 1,390 ± 132 pA, n = 20). F, Same data as in C, but for each litter, an average value was calculated for all the WT neurons, and values from individual neurons in that litter were normalized to the litter WT average. In this way, variation across different litters due to viral injection was minimized (WT = 1.0 ± 0.075 pA, n = 19; Het = 1.258 ± 0.13 pA, n = 20; Mann–Whitney test, p = 0.089).

Postsynaptic expression of MEF2C-VP16 rescues weak L4 to L2/3 inputs in response to sensory deprivation

The development of L4 to L2/3 synaptic connections in the barrel cortex depends on normal whisker sensory experience as well as MEF2C (Lendvai et al., 2000; Stern et al., 2001; Shepherd et al., 2003; Wen and Barth, 2011). We previously demonstrated that sensory deprivation by whisker trimming prevents or occludes weakening of L4 to L2/3 synaptic inputs by deletion of Mef2c in L2/3 neurons (Rajkovich et al., 2017). These results suggested that MEF2C and experience function in a common signaling pathway to promote development and strengthening of L4 to L2/3 synapses. Because sensory experience and neuronal activity trigger MEF2C transcriptional activation (Rashid et al., 2014; Assali et al., 2019), we hypothesized that transcriptional activation of MEF2C promotes L4 to L2/3 synaptic input strength and functions downstream of sensory experience. To test this hypothesis, we injected WT mice intraventricularly at P1 with AAV expressing a constitutively transcriptionally active MEF2C (MEF2C-VP16) bicistronically with GFP (Pulipparacharuvil et al., 2008; Fig. 6A). At P17–P21, we performed dual recordings of MEF2C-VP16 [GFP(+)] and neighboring untransfected [GFP(−)] L2/3 pyramidal neurons and LSPS mapping. Surprisingly, we observed no difference in the strength of home barrel L4 synaptic inputs between nontransfected and postsynaptic MEF2-VP16–transfected L2/3 neurons in WT mice (Fig. 6A,B). Thus, overexpression of active MEF2C on a WT background does not strengthen L4 to L2/3 synaptic inputs. This may be due to MEF2C transcriptional activation and/or L4 synaptic input strengths being saturated in sensory-experienced WT mice. If so, then sensory deprivation by whisker trimming may reduce MEF2C transcriptional activity and/or L4 synaptic strength and allow strengthening of L4 inputs in response to MEF2C-VP16. To test this idea, we compared the effects of MEF2C-VP16 on L4 to L2/3 synaptic strength in sensory deprived (unilateral whisker trimming; P9–P17), or contralateral S1, to that in spared, or ipsilateral, S1 of the same WT mice (Fig. 6C). LSPS maps were collected from pairs MEF2C-VP16 [GFP(+)] and neighboring GFP(−) L2/3 pyramidal neurons in slices from deprived or spared hemispheres. In the deprived hemisphere, MEF2C-VP16–expressing L2/3 neurons had increased strength of L4 inputs as compared to their untransfected neighbors, whereas in the spared hemisphere, there was no difference in L4 input strength between MEF2C-VP16–expressing and GFP(−) L2/3 neurons (Fig. 6D–F). A comparison of L4 synaptic input strength across hemispheres revealed that sensory deprivation weakened L4 synaptic inputs onto untransfected neurons, but not in MEF2C-VP16–expressing neurons (Fig. 6D–F). These results indicate that overexpression of transcriptionally active MEF2C can prevent or rescue the weakening of L4 to L2/3 synaptic connections due to sensory deprivation and supports the hypothesis that transcriptional activation of MEF2C promotes L4 synaptic input development downstream of sensory experience.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Postsynaptic expression of MEF2C-VP16 rescues weak L4 to L2/3 synaptic strength in sensory-deprived cortex, but not in normally experienced WT or CaMKII-Mef2c KO mice. A, Averaged color maps of local synaptic input strengths in paired recordings of untransfected and MEF2C-VP16 transfected neurons in WT mice. B, Left, Mean (±SEM) EPSC amplitudes of elicited by LSPS in the L4 home barrel (untransfected = 14 ± 2; MEF2C-VP16 transfected = 16 ± 2 pA; n = 18 pairs). Right, Plot of each individual cell pair (untransfected vs MEF2C-VP16 transfected; open circles). Mean ± SEM (filled red circle). The diagonal line represents equality. C, Experimental design and timeline with sensory deprivation. D, Averaged color maps of local synaptic input strengths in response to LSPS from spared and deprived hemispheres in neighboring WT and MEF2C-VP16-GFP transfected L2/3 neurons. E, Values from individual cell pairs in spared and deprived hemispheres. F, Mean synaptic input strength from the L4 home barrel (L4H; spared, WT = 16 ± 1; MEF2C-VP16 = 16 ± 2 pA, n = 15; deprived, WT = 10 ± 1; MEF2C-VP16 = 13 ± 2 pA; n = 21). G, Averaged color maps of local synaptic input strengths in paired recordings of untransfected and MEF2C-VP16 transfected neurons in CaMKII-Mef2c KO mice. H, Left, Mean (±SEM) EPSC amplitudes of elicited by LSPS in the L4 home barrel (untransfected = 14 ± 2; MEF2C-VP16 transfected = 16 ± 2 pA; n = 18 pairs). Right, Plot of each individual cell pair (untransfected vs MEF2C-VP16 transfected; open circles). Mean ± SEM (filled red circle). The diagonal line represents equality.

To test whether postsynaptic MEF2C-VP16 is sufficient to rescue weak L4 to L2/3 inputs in the CaMKII-Mef2c KO, we sparsely transfected MEF2C-VP16 in the cortex of CamKII-Mef2c KO mice with P1 AAV injection and performed LSPS mapping on pairs of untransfected Mef2c KO [GFP(−)] and transfected MEF2C-VP16 [GFP(+)] L2/3 pyramidal neurons. Surprisingly, L4 synaptic inputs were unaffected by MEF2C-VP16 expression in L2/3 neurons of CaMKII-Mef2c KO mice (Fig.6G,H). Based on the expression pattern of CaMKII-Cre (Fig. 3A), Mef2c is expected to be deleted in L4 and L2/3 neurons. Postsynaptic expression of MEF2C-VP16 may not rescue weak L4 to L2/3 inputs in CaMKII-Mef2c KO neurons because MEF2C is also required in presynaptic L4 neurons. Together these results reinforce the results that MEF2C is necessary in both L4 and L2/3 neurons for development of their synaptic connections.

MEF2C regulates expression of pre- and postsynaptic genes under basal and high activity conditions in cortical slice cultures

Based on our functional results, we hypothesize that neuronal activity stimulates MEF2C-dependent transcription of genes that promote development and function of L4 to L2/3 excitatory synapses. Because MEF2C is necessary for both presynaptic L4 and postsynaptic L2/3 neurons, we hypothesize that MEF2C regulates transcription of genes that promote both pre- and postsynaptic function. To identify activity-dependent genes in the postnatal developing cortex and the regulation of these genes by MEF2C, we established an activity-induced gene expression paradigm in ex vivo cortical slice cultures. S1-containing cortical slices from wildtype mice were prepared at P6–P7 and cultured in vitro for 8 d or equivalent postnatal day (ED) 14–15 (Fig. 7A). We first performed a time course of activity-dependent gene expression to identify a time point for robust gene expression using quantitative (q) RT-PCR for known activity-induced genes or MEF2 target genes (Fos, Egr2, Arc, Nr4a1; Extended Data Table 7-1). At ED14–ED15, cultures were treated with media containing the GABA_A receptor blocker (picrotoxin) and K⁺ channel blocker (4-aminopyridine, 4-AP), for 15 min, 1, or 3 h to increase neuronal activity (Fig. 7B–E). We observed the most robust activity-dependent gene expression at the 3 h time point and used this treatment time in subsequent experiments. To identify activity-induced genes and their regulation by MEF2C, slice cultures were prepared from Cre negative [Cre(−)] and CaMKII-Mef2c KO (or “KO”) mice and treated with either vehicle (“Basal”) or picrotoxin/4-AP (“Activity”) for 3 h. mRNA from cultures were then processed for bulk RNA-seq analysis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Activity-induced gene expression paradigm in WT and CaMKII-Mef2c KO cortical slice cultures. A, Schematic of workflow for the preparation of slice cultures including the growth and treatment timeline and region of cortex isolated for use in RNA extraction. B–E, qRT-PCR of activity-dependent gene expression (Fos, Egr2, Arc, and Nurr77) in wildtype slice culture samples treated with high activity media “Activity” or vehicle “Basal” for 15 min, 1, or 3 h. Values are normalized to “Basal” values for each time point. One-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001). F, Volcano plot of differentially regulated genes (DEGs) in “Cre(−) Activity” normalized to “Cre(−) Basal” from DESeq2 comparison. For this and all volcano plots, upregulated genes, fold change (FC) cutoff of log2 >0.3 and adjusted p-value of <0.05 (green), and downregulated genes, FC cutoff of log2 of <−0.3 and adjusted p-value of <0.05 (blue). G, Same as in F, but with “KO Activity” normalized to “KO Basal.” See Extended Data Tables 7-1, 7-3, and 7-4.

Processed tissue samples expressed, on average, ∼13,000 genes [with counts per million reads (CPM) ≥ 1.0] in Cre(−) samples (Extended Data Table 7-2). Differentially expressed genes (DEGs) were identified using DESeq2 conducted on raw CPMs with a log2 FC cutoff of ±0.3 and an adjusted p-value of <0.05 (Extended Data Table 7-3). We compared four conditions: (1) “Cre(−) Basal” (vehicle treated), (2) “Cre(−) Activity” (picrotoxin/4-AP treated), (3) “KO Basal,” and (4) “KO Activity.” To identify activity-driven genes in the Cre(−) cortex, we normalized the “Cre(−) Activity” condition to those in “Cre(−) Basal” using DESeq2 and identified 714 upregulated and 1,061 downregulated genes in “Cre(−) Activity” (Fig. 7F; Extended Data Tables 7-3, 7-4). Gene ontology (GO) analysis of both up- and downregulated DEGs using ToppFun (J. Chen et al., 2009) revealed enrichment for molecular functions and biological processes related to transcription regulation, DNA, and chromatin binding and the nuclear cellular compartment (Extended Data Table 8-1). Many transcription factors upregulated by activity were canonical activity-dependent immediate-early genes (Fos, Jun, Npas4, Egr1) as well as other known activity-induced genes (Homer1, Bdnf, Arc; Hrvatin et al., 2018; Bach et al., 2023; Pollina et al., 2023). A comparison of gene expression in “KO Activity” to “KO Basal” conditions found 800 genes upregulated and 1,041 genes downregulated by activity in KO (Fig. 7G). GO analysis revealed enrichment for genes related to transcription, DNA, and chromatin binding (Extended Data Table 8-1), many of those similar to what we observed in Cre(−) cultures. Thus, the activity-induced expression of many transcription factors and regulators of transcription does not require MEF2C.

To identify MEF2C-regulated genes in the different activity conditions, we first compared gene expression in the “KO Activity” versus “Cre(−) Activity” conditions. This comparison found 621 upregulated genes and 557 downregulated genes (Fig. 8A; Extended Data Tables 7-3, 7-4), which are both enriched for genes expressed in S1 pyramidal neurons (Fig. 8B; Zeisel et al., 2015). Upregulated genes were also associated with inhibitory neurons suggesting non-cell-autonomous effects of Mef2c deletion in excitatory neurons. GO analysis of upregulated genes revealed the most significant enrichment for biological processes of transsynaptic signaling, synaptic transmission, organization and assembly, and cellular compartments of pre- and postsynapse and dendrites (Extended Data Table 8-1). To refine the GO analysis for synaptic terms, we used an evidence-based, expert-curated database, SynGO (Koopmans et al., 2019). Upregulated genes in “KO Activity” showed enrichment for biological processes of synapse assembly, organization, and transsynaptic signaling in pre- and postsynaptic compartments (Fig. 8C,D). Interestingly, analysis of downregulated genes in the “KO Activity” condition with SynGO found enrichment for genes that function in transsynaptic signaling, synapse adhesion, and organization within the pre- and postsynaptic compartments (Fig. 8E,F; Extended Data Table 8-1). Thus, under high activity conditions, MEF2C primarily regulates expression of genes that function in synaptic processes in pre- and postsynaptic compartments and both suppresses and promotes their expression.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Under high activity conditions, MEF2C promotes and suppresses expression of genes encoding pre- and postsynaptic proteins in cortical slice cultures. A, Volcano plot of differentially expressed genes (DEGs) in high activity CaMKII-Mef2c KO (“KO Activity”) cortex normalized to “Cre(−) Activity” cortex. B, Overlap of up- and downregulated genes in “KO activity” compared with cell-type–specific gene sets obtained from single-cell RNA-seq of murine S1 tissue. In this and bubble plots in other figures, the color denotes significance (−log10 adjusted p-value), and the size denotes the number of overlapping genes in the comparison. C–F, Gene ontology (GO) terms from SynGO analysis for upregulated (C, D) and downregulated (E, F) genes in “KO Activity” normalized to “Cre(−) Activity” conditions for cellular component and biological process. See Extended Data Tables 7-3–7-5.

Comparison of gene expression in CaMKII-Mef2c KO and Cre(−) cultures under basal activity conditions revealed 1,452 upregulated and 1,552 downregulated genes in the KO (Fig. 9A; Extended Data Tables 7-3, 7-4). Upregulated genes were enriched for transsynaptic signaling and synaptic transmission in both pre- and postsynaptic compartments and in genes associated with cortical pyramidal neurons and interneurons (Fig. 9C; Extended Data Table 8-1). In contrast to the high activity condition, downregulated genes in KO cultures under basal activity conditions were enriched for processes of cell adhesion, extracellular matrix compartment, and translation at synapses as well as genes associated with astrocytes and microglia (Fig. 9D). In summary, MEF2C suppresses expression of genes that function at synapses and regulate synaptic function under both basal and high activity conditions. However, only under high activity conditions does MEF2C promote the expression of genes involved in synaptic function, assembly, and organization. Up- and downregulated genes in KO cultures overlapped with up- and downregulated genes, respectively, observed in the cortex with embryonic deletion of Mef2c in excitatory neurons (Emx1 Cre; Harrington et al., 2016) or germline Mef2c-Het mice (Harrington et al., 2020; Fig. 9E,F). This overlap was significant under both basal and high activity conditions and suggests that there are similar MEF2C-regulated genes in the cortex in slice cultures and in vivo and may be relevant to MCHS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

MEF2C-regulated genes under basal activity conditions in slice culture and overlap with MEF2C-regulated genes in vivo in mouse and human genes associated with autism and cognitive resilience. A, Volcano plot of all DEGs in “KO Basal” condition normalized to “WT Basal” from DESeq2 comparison. B, Overlap of up- and downregulated genes in “KO Basal” cultures compared with cell-type–specific gene sets obtained from single-cell RNA-seq of S1 tissue. C, D, ToppFun gene ontology terms for “KO Basal” normalized to “WT Basal” of upregulated (C) and downregulated genes (D). E, Overlap of MEF2C up- and downregulated genes in slice cultures (basal activity condition) with published DEGs observed in the cortex with Emx1 Cre-Mef2c KO (Harrington et al., 2016) or germline Mef2c-Het (Harrington et al., 2020). F, Same as in E but comparing overlap of MEF2C up- and downregulated genes in slice cultures in the high activity condition. G, Overlap of MEF2C up- and downregulated genes in slice cultures under basal or high activity conditions with human disease and cognitive resilience associated genes. See Extended Data Tables 7-3–7-5.

While loss of function mutations in MEF2C are associated with autism and intellectual disability (Assali et al., 2019), MEF2C transcriptional activity is positively associated with cognitive function in mice and humans and cognitive resilience with aging (Mitchell et al., 2018; Barker et al., 2021). In support of these associations, upregulated genes in the “KO/Cre(−) Basal” condition significantly overlapped with SFARI ASD genes (p < 0.01; https://gene.sfari.org/), and downregulated genes significantly overlapped with genes associated with cognitive resilience in humans (Barker et al., 2021; Fig. 9G). These results suggest that MEF2C suppresses genes associated with ASD and promotes expression of genes associated with cognitive function in the cortex. Under both basal and high activity conditions, MEF2C-regulated genes overlap with genes associated with human attention-deficit hyperactivity disorder (Zhou et al., 2023) and genes containing broad enhancer-like chromatin domains (BELD; Fig. 9G). BELD genes are highly transcribed during early postnatal brain development and are enriched for genes involved in synapse development and ASD (Zhao et al., 2018). Taken together, these results implicate MEF2C in expression of genes associated with neurodevelopment, cognition, and neuropsychiatric disorders.

Under high activity conditions, MEF2C promotes expression of genes encoding presynaptic proteins that promote presynaptic development and function

Based on our electrophysiology results, we hypothesize that MEF2C transcriptional activation functions downstream of experience and neuronal activity to promote L4 to L2/3 synapse development. Thus, we would predict that there is a set of activity-induced genes that require MEF2C and function to promote excitatory synaptic development and/or function. Genes that are induced by activity in Cre(−), but not in CaMKII-Mef2c KO cultures, would be candidate genes to support our hypothesis. To identify such genes, we compared genes that are induced by activity in Cre(−) cultures [Cre(−) Activity/Basal; 714 genes] with those induced by activity in KO cultures (KO Activity/Basal; 800 genes; Fig. 10A). Of the 714 activity-induced genes in Cre(−) cultures, 245 genes were not induced by activity in KO cultures. Interestingly, GO analysis of these 245 genes with SynGO found enrichment for genes that functioned primarily in presynaptic compartments, including the active zone and synaptic vesicle membrane and in processes of presynaptic assembly (Fig. 10B,C; Extended Data Table 8-1). Of these genes, some encode voltage-sensitive ion channels (Scn1a, Cacnb4, Kcnh4) or regulate synaptic vesicle fusion (Snap25, Synpr, Rims3). Interestingly, many genes in this group encode presynaptic cell adhesion proteins that mediate synapse formation and maturation (Flrt2, Nrxn1, Tenm1, Cbln2, Cbln3) and proteins that regulate cell–cell adhesion and/or axon guidance (EphB2, Pcdh19, Sema3e, Sema4b, Sema7a; Fig. 10D). Therefore, MEF2C is required for the activity-dependent expression of many genes that regulate presynaptic function and development that may contribute to the experience-dependent development of L4 to L2/3 synaptic inputs (Fig. 11E,F). A comparison of the 245 activity- and MEF2C-dependent genes with a MEF2C ChIP-seq dataset (Ma and Telese, 2015) found a significant overlap of genes with MEF2C bound at the promoter (p = 0.0044; Extended Data Table 10-1). This result suggests that many of these genes are direct targets of MEF2C. Of the 469 genes commonly induced by activity in Cre(−) and KO cultures, SynGO analysis only found enrichment in the extrasynaptic compartment and BDNF signaling (Extended Data Table 8-1). Interestingly, these 469 genes also overlapped with MEF2C promoter-bound genes suggesting that many are also direct targets, but MEF2C is dispensable for their induction by activity (Extended Data Table 10-1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

MEF2C is necessary in cortical neurons for activity-dependent expression of genes encoding presynaptic proteins. A, Venn diagram of overlap of DEGs induced by high activity in Cre(−) (green) and CaMKII-Mef2c KO (blue) cultures. B, C, GO terms from SynGO analysis of cellular component (B) and biological processes (C) for the 245 genes that are upregulated with activity in Cre(−), but not in KO cultures. D, Heatmap of normalized CPM values [normalized to average CPM for Cre(−) “Basal”] for all biological samples of candidate synaptic genes. Highlighted genes represent presynaptic genes of interest that regulate neurotransmission (blue) or synapse formation and maturation (green). E, Left, Representative Western blots of FLRT2 (top, arrow) and total protein (bottom) synaptosomes from Cre(−) and KO cultures under basal (B) activity or high activity (A) conditions. Middle, Quantified group data (mean ± SEM) of FLRT2 levels in high activity cultures normalized to basal activity sister cultures for each genotype. Right, Quantified group data of FLRT2 from Western blots of synaptosomes from Cre(−) and CaMKII-Mef2c KO slice cultures under basal and high activity conditions. (n = 6–7 cultures/genotype; two-way ANOVA; treatment–genotype; F_{(1, 12)}= 5.584; p < 0.05; uncorrected Fisher's LSD multiple-comparison test; *p < 0.05). F, G, Representative Western blots (left) and quantified group data (right) of FLRT2 and synapsin 1 (SYN1) from S1 cortical lysates from Cre(−) and CaMKII-MEF2C KO P21 mice. The bottom blot is total protein. H, Left, Representative Western blots of SLIT2 (top) and total protein (bottom) synaptosome fractions from Cre(−) and KO cultures under basal or high activity conditions Middle, Quantified group data of SLIT2 (N-terminal fragment) levels in synaptosomes from high activity cultures normalized to basal activity sister cultures for each genotype (n = 6–7 cultures/genotype; two-way ANOVA; treatment–genotype; F_{(1, 10)} = 7.423; p < 0.05). Uncorrected Fisher's LSD multiple-comparison test; *p < 0.05, **p < 0.01). I, Representative blots of SLIT2 from S1 cortical lysates from Cre(−) and CaMKII-MEF2C KO P21 mice. See Extended Data Tables 7-2, 7-4, 7-5.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

MEF2C promotes expression of genes encoding postsynaptic proteins under high activity conditions. A, Heatmap of normalized CPM values [all samples normalized to average CPM for Cre(−) “Basal”] for all biological replicates of “postsynaptic” genes that have reduced expression in high activity conditions. These genes encode proteins that are localized to the postsynapse and/or function postsynaptically to regulate synapse development or function. Note: Genes in the top and bottom heat maps are plotted on different scales due to differences in normalized expression values. B–D, Representative Western blots (upper) and quantified group data (bottom) of LRRTM1 from S1 cortical lysates (B) from Cre(−) and CaMKII-Mef2c KO P21 mice or total homogenate (C) or synaptosome fractions (D) from Cre(−) and CaMKII-Mef2c KO slice cultures under basal (B) activity or high activity (A) conditions. E, F, Working model of sensory experience- and activity-dependent transcriptional activation of MEF2C in promoting L4 to L2/3 synaptic connectivity. E, Sensory experience- (whisker activity) and activity-dependent transcriptional activation of MEF2C in presynaptic L4 and postsynaptic L2/3 promotes synapse development and connectivity. F, Schematic of presynaptic and transsynaptic proteins (bold text labels), whose genes are upregulated by MEF2C under high activity conditions. See Extended Data Table 7-2.

To determine if changes in expression of activity and MEF2C-dependent genes resulted in altered protein levels, we performed Western blots for selected presynaptic proteins from Cre(−) and KO cultures under basal and high activity conditions (Fig. 10E,H). We selected SLIT2 and FLRT2 for their presynaptic and transsynaptic role in input-specific synapse development (Blockus et al., 2021; Südhof, 2021) and blotted a synaptosome fraction to enrich for synaptic proteins. FLRT2 and SLIT2 protein levels were decreased in KO cultures, as compared with Cre(−), in high activity, but not basal activity conditions (Fig. 10E,H). To determine if FLRT2 and SLIT2 were regulated by MEF2C in vivo, we blotted S1 cortical lysates from P18 CaMKII-Mef2c KO and Cre(−) littermates. We observed decreases in FLRT2, but not SLIT2, in CaMKII-Mef2c KO cortices (Fig. 10F,I). To determine if there were more general decreases in presynaptic proteins, we blotted for synapsin 1 (SYN1) in the culture synaptosomes and cortical lysates and found that SYN1 was unaffected by CaMKII-Mef2c KO in both preparations (Fig. 10G; data not shown). The protein changes are consistent with gene expression results that under conditions of increased neuronal activity, MEF2C promotes expression of select presynaptic proteins that promote synapse development.

MEF2C promotes expression of postsynaptic genes under conditions of high activity

While the analysis in Figure 10 primarily implicated genes that function in the presynaptic compartment, our electrophysiology results also support a role for MEF2C in postsynaptic L2/3 neurons in development of L4 to L2/3 synapses. Since we were able to rescue L4 to L2/3 synaptic strength with postsynaptic expression of MEF2C-VP16, we hypothesized that MEF2C transcriptional activation of genes that functioned in the postsynaptic compartment promotes L4 synaptic input strength. Therefore, we looked more closely at MEF2C-dependent postsynaptic genes. Under high activity conditions, MEF2C promoted expression of genes that were either expressed in the postsynaptic membrane and/or functioned to regulate postsynaptic strength (Fig. 8E,F). Interestingly, some of these genes functioned in synapse adhesion, formation, or maturation such as Lrrtm1, Pcdh8, and Nrg1 (Fig. 11A; Li et al., 2007; Yasuda et al., 2007; Südhof, 2021). To validate changes in protein expression of a candidate postsynaptic adhesion molecule, we performed Western blots for LRRTM1 in S1 cortical homogenates of P21 CaMKII-Mef2c KO and Cre(−) littermates (Fig. 11B), or whole homogenates or synaptosomes from KO and Cre(−) slice cultures under both basal and high activity conditions (Fig. 11C,D). In contrast to changes in Lrrtm1 gene expression, we were unable to detect changes in LRRTM1 protein in the MEF2C KO cortex. Because changes in LRRTM1 may be synapse specific, for example, only at L4 to L2/3 synaptic inputs, we may be unable to resolve these changes with bulk methods, and cell- and input-specific methods may be required.