Subcellular synaptic connectivity of layer 2 pyramidal neurons in the medial prefrontal cortex

Abstract

Pyramidal neurons in the prefrontal cortex (PFC) are important for the control of cognitive and emotional behavior. The medial PFC (mPFC) receives diverse long-range excitatory inputs from the midline thalamus, contralateral mPFC, basolateral amygdala, and ventral hippocampus. While axons from these different regions have distinct distributions in the mPFC, their functional connections at the cellular and subcellular levels remain unknown. Here, we use optogenetics to show that layer 2 pyramidal neurons in acute slices of the mouse mPFC receive excitatory inputs from each of these regions. Using a combination of optogenetics and two-photon microscopy, we then determine the subcellular properties of these inputs. We find that different types of inputs make selective contacts at the levels of both dendrites and spines. Using two-photon uncaging, we show that this subcellular targeting strongly influences synaptic efficacy in these neurons. Together, our results show that functional connectivity is finely tuned, with important implications for signal processing in the mPFC.

Figure 1.

Anatomical overlap of axons, dendrites, and spines. A, Left, Two-photon image stack of an L2 pyramidal neuron. Dashed lines designate pia and L1/L2 border. Scale bar, 50 μm. Right, Magnified views of spines in L1 (top, blue) and L2/3 (bottom, green) from the boxed regions on left. Scale bars, 1 μm. B, Confocal images showing mCherry-labeled axons arriving from MD, cmPFC, BLA, and VH. To the right of each image is the average normalized fluorescent intensity profile as a function of distance from the pial surface (n = 6). Scale bar, 50 μm.

Figure 2.

L2 pyramidal neurons receive diverse excitatory inputs. A, Average light-evoked EPSCs for the different inputs, recorded in voltage-clamp at +40 mV in control conditions (solid lines), after wash-in of 10 μm CPP (long-dash lines) and subsequent addition of 10 μm NBQX (short-dash lines). Arrows indicate the time of light pulses. B, Quantification of the data in A, expressed as a fraction of control EPSC for each condition. C, AMPA/NMDA ratios from the data in A.

Figure 3.

Mapping functional connections at the subcellular level. A, Two-photon image stack of an L2 pyramidal neuron. Scale bar, 30 μm. B, Top, Magnified view of the orange boxed region in A with no ChR2 stimulation. Middle, The same segment overlaid with ΔG/R signal (green) following ChR2 stimulation. Bottom, Positions of interleaved line scans through active spines (red and green) and a nearby neighbor (blue). Scale bar, 1 μm. C, Left, Individual line scans through spines (top) and dendrites (bottom) indicated in B, where morphology is color coded and Ca signals are white. Right, Quantified Ca signals, showing successes (color) and failures (black) of synaptic transmission at the three spines. Scale bar: top, middle = 0.25 ΔG/G_sat, 75 ms; bottom = 0.05 ΔG/G_sat, 75 ms. Arrows indicate stimulus times. D, Reconstruction of the recorded neuron, with sampled dendrites in black and active spines indicated by colored dots. Scale bar, 30 μm. Stim, Stimulation.

Figure 4.

Unbiased detection of connections onto spines. A, Left, Two-photon image of dendrite (d) and spine (s), showing two-photon uncaging spot (red dot) and line-scan path (red dashed line). Scale bar, 1 μm. Top right, Example line scan showing morphology (red) and ΔG/R signal (green) in the spine and dendrite. White arrow indicates the uncaging time. Bottom right, Mean (thick lines) and individual (thin lines) ΔG/G_sat transients in the spine (red) and dendrite (black). Arrow indicates the uncaging time. B, Amplitudes (left) and SNR (right) of uncaging-evoked synaptic Ca signals as a function of radial distance from the soma. Solid and dashed lines represent linear fits and 5–95% confidence intervals, respectively. Orange dashed line indicates the detection threshold. Correlation coefficients (r values) are labeled in the upper right-hand corners. C, Left, Summary of light-evoked synaptic Ca signal amplitudes for each input from mapping experiments. Right, Summary of probability of release (Pr) for each input from these light-evoked synaptic Ca signals.

Figure 5.

Different inputs target select dendritic locations. A, Reconstructed L2 pyramidal neurons contacted by MD, cmPFC, BLA, and VH inputs, showing sampled dendrites (black) and active spines (colored dots). B, Probability distributions of synapse position as a function of vertical (y) displacement from the soma for each input. C, Cumulative probability distributions of synapse position as a function of y displacement from the soma for each input. Orange line shows the position of the L1/L2 border. D, Summary of y displacement for each input. Asterisks indicate significant differences. Color of asterisks and lines indicates which inputs are compared. E, F, Similar to C and D but in the horizontal (x) dimension.

Figure 6.

Anatomy does not predict functional connectivity. A, Axon density maps (left) are multiplied by spine density maps (middle) and normalized to generate maps of predicted synapse location (right). Boxed regions are one-dimensional profiles of y and x displacements. B, Probability distributions of predicted (lines) and measured (bars) synapse location as a function of vertical (y) displacement from the soma for each input. C, Measured cumulative probability distributions plotted against predicted cumulative probability distributions for each input in the y dimension. Dashed gray line indicates unity. Orange dots indicate L1/L2 border (closed) and soma (open). D, Correlation coefficients between measured and predicted distributions for each input in the y dimension. E, F, Similar to C and D but in the horizontal (x) dimension.

Figure 7.

Different inputs target biased populations of spines. A, Head volumes of all reconstructed spines as a function of vertical (y; left) and horizontal (x; right) displacement from the soma. Solid black lines represent linear fits to the data, and dashed black lines indicate the 5 and 95% confidence intervals. Correlation coefficients (r values) are labeled in the upper right-hand corners. B, Two-photon image stacks of representative dendrites and spines contacted by each input, where green indicates evoked Ca signals (ΔG/R). Scale bars, 1 μm. C, Cumulative probability distributions for each input, showing the head volumes of detected spines (colored lines) and the entire nonactive population (gray dashed line). D, Summary of head volumes of spines contacted by different inputs. Gray line shows median of entire nonactive population. Asterisks denote significant differences between the head volumes contacted by a given input and the entire nonactive population.

Figure 8.

Spine location and morphology impact synaptic efficacy. A, Top, Two-photon image stack of a dendritic segment, showing heterogeneous spine morphologies, where colored dots indicate two-photon uncaging targets. Scale bar, 1 μm. Bottom, uEPSPs evoked with a brief pulse of 725 nm light. Black lines are the average of 10 trials, and gray regions indicate the SEM. Colored arrows indicate uncaging times at indicated spines. Scale bar: 0.2 mV, 25 ms. B, Top, Two-photon image stack of an L2 pyramidal neuron, showing two regions (green, blue) targeted for two-photon uncaging. Scale bar, 50 μm. Bottom, Average uEPSPs generated at the two regions. Colored arrows indicate uncaging times at different locations. Scale bar: 0.4 mV, 25 ms. C, D, uEPSP amplitude as a function of spine head volume (C) and radial distance from the soma (D). Solid and dashed lines represent linear fits and 5–95% confidence intervals, respectively. Correlation coefficients (r values) are labeled in the upper right-hand corners. E, Predicted unitary EPSP amplitude for each input, calculated by combining results from the mapping and uncaging experiments. The x-axis indicates the variables included in the prediction calculation (head volume, location, or both). Orange dotted and faint lines represent the median and SE of the average response expected if no selective targeting occurs on spines or dendrites. Asterisks denote significant differences between the full and reduced predictions.