Layer-specific mitochondrial diversity across hippocampal CA2 dendrites

Abstract

CA2 is an understudied subregion of the hippocampus that is critical for social memory. Previous studies identified multiple components of the mitochondrial calcium uniporter (MCU) complex as selectively enriched in CA2. The MCU complex regulates calcium entry into mitochondria, which in turn regulates mitochondrial transport and localization to active synapses. We found that MCU is strikingly enriched in CA2 distal apical dendrites, precisely where CA2 neurons receive entorhinal cortical input carrying social information. Furthermore, MCU-enriched mitochondria in CA2 distal dendrites are larger compared to mitochondria in CA2 proximal apical dendrites and neighboring CA1 apical dendrites, which was confirmed in CA2 with genetically labeled mitochondria and electron microscopy. MCU overexpression in neighboring CA1 led to a preferential localization of MCU in the proximal dendrites of CA1 compared to the distal dendrites, an effect not seen in CA2. Our findings demonstrate that mitochondria are molecularly and structurally diverse across hippocampal cell types and circuits, and suggest that MCU can be differentially localized within dendrites, possibly to meet local energy demands.

Keywords: dendrites; entorhinal cortex; hippocampal CA2; mitochondrial calcium uniporter; mitochondrial localization; mitochondrial morphology.

Figure 1 :. MCU-labeled mitochondria are preferentially enriched in CA2 distal dendrites and larger in size than neighboring mitochondria

A. (i.) 20X epifluorescence tile image of a wild-type hippocampus stained for MCU (yellow) and RGS14 (magenta) with the subregions and dendritic layers labeled. White arrowheads indicate the borders of CA2. (ii.) The MCU channel alone. B. Representative 40X confocal images of MCU staining in SR (i.) and SLM (ii.) of CA2 in a wild-type mouse. All representative 40X images are from a single Z section. Insets in B and C are a zoomed 10 × 10 μm crop of the larger image. C. Representative 40X confocal images of MCU staining in SR (i.) and SLM (ii.) of CA1 in a wild-type mouse. Images were processed identically to the representative images in B for comparison between CA1 and CA2. D. The mean MCU fluorescence of a representative 100 × 100 μm ROI from SR and SLM of CA1 and CA2. Data were normalized to the overall average. SR (closed circles) and SLM (open circles) of the same animal are paired with lines. Two-way RM ANOVA with Sidak’s post hoc; N = 7 animals; 3 hippocampal sections / animal. Overall effect of subregion and layer were significant (subregion: F = 194.1, P = <0.0001; layer: F = 21.8, P = 0.0001; interaction: F = 43.0; P = <0.0001). E. The average area of MCU-labeled mitochondria in the same ROIs from D. Mitochondria were segmented as described in the methods. Data were normalized and plotted the same as in D. Two-way RM ANOVA with Sidak’s post hoc. Overall effect of subregion and layer on area were significant (subregion: F = 102.6, P = <0.0001; layer: F = 23.1, p = <0.0001, interaction: F = 26.2; P = <0.0001). F. The average count of MCU-labeled mitochondria from the dataset in E. Two-way RM ANOVA with Sidak’s post hoc. Overall effect of subregion and layer on count were significant (subregion: F = 42.7, P = 0.0006; layer: F = 17.8, p = 0.0003, interaction: F = 11.0; P = <0.0019). SO = Stratum Oriens; SR = Stratum Radiatum, SLM = Stratum Lacunosum Moleculare. Sidak’s post hoc: * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001. Scale for A = 100 μm; B-C = 20 μm.

Figure 2:. COX4 labeling is similarly enriched in CA2 and CA1 distal dendrites

A. (i.) 20X epifluorescence tile image of a wild-type hippocampus stained for COX4 (cyan) and RGS14 (magenta) with the subregions and dendritic layers labeled. White arrowheads indicate the borders of CA2. (ii.) The COX4 channel alone. Yellow arrows show examples of COX4 labeling in non-pyramidal cells in the neuropil. B. Representative 40X confocal images of COX4 staining in SR (i.) and SLM (ii.) of CA2 in a wild-type mouse. Insets in F and G are a zoomed 10 × 10 μm crop of the larger image. C. Representative 40X confocal images of COX4 staining in SR (i.) and SLM (ii.) of CA1 in a wild-type mouse. Images in F and G were processed identically to preserve differences between subregions and dendritic layers. D. COX4 mean fluorescence of a 100 × 100 μm ROI from SR and SLM of CA1 and CA2 normalized to the overall cohort average. Lines pair SR and SLM of the same animal. Two-way RM ANOVA with Sidak’s post hoc; N = 7 animals from two cohorts. Overall effect of subregion was not significant for any metric (p > 0.05), thus subregion comparisons are not shown. Overall effect of layer was significant for fluorescence (F = 13.4, p = 0.011). E. Mean area of COX4-labeled mitochondria in the same ROIs in D. Mitochondria were segmented the same as in Figure 1 and data were normalized to the overall cohort average. Two-way RM ANOVA with Sidak’s post hoc. Overall effect of layer on area was significant (F = 15.5, p = 0.008). F. Mean count of COX4-labeled mitochondria from the same dataset as E. Two-way RM ANOVA with Sidak’s post hoc. The overall effect of layer and subregion on mitochondria count were not significant (P > 0.05). G. Representative ProExM images of MCU (yellow) and MT-5 (magenta) in CA2 SLM. Bottom panels show the overlay of both channels (left) and an overlay with GFP+ CA2 distal dendrites (right). H. Representative ProExM images of COX4 staining (cyan) and MT-5 staining (magenta) in CA2 SLM. An ROI was chosen which had a similar amount of MT-5 and GFP staining as in G for comparison. Sidak’s post hoc: * = P < 0.05; ** = P < 0.01; *** = P < 0.001 Scale for A = 100 μm; B-C = 20 μm; G-H = 3 μm (ExM adj.)

Figure 3:. Super resolution imaging and electron microscopy confirm CA2 layer-specific differences in dendritic mitochondrial morphology

A. (I) Representative 20X confocal image of sparse CA2 neuron Mitotag labeling of mitochondria. ROIs (white box) within the 20X image indicate the selection for 60X (II) SO, (III) SR, and (IV) SLM representative images. ROI (white box) within 60X images is representative of 10×10 μm² 4X super- resolution images of Mitotag labeled mitochondria in individual (ii) SO, (iii) SR, and (iv) SLM dendritic branches. B. Representative electron micrographs of (i) SO, (ii), SR, and (iii) SLM dendrites. Examples of mitochondria (arrowheads) within dendrite (arrows) cross sections are indicated. C. Median Mitotag-labeled mitochondrial area in CA2 SO, SR, and SLM dendritic layers. Data are normalized to the average area per mouse with lines connecting SO, SR, and SLM per mouse. Mitochondrial size was significantly increased within the SLM layers compared to both SO and SR in the CA2 (RM One Way ANOVA: F=23.47, P=0.0014, Tukey’s multiple comparison test; N=5 mice; 4–7 hippocampus sections/animal). D. Ratio of the median mitochondrial area between SLM and SR layers in C. Color-filled data points represent the mice selected for representative images (A: blue; E: pink). E. Percent relative frequency plot of individual mitochondrial areas from CA2 SO, SR, and SLM from one representative section from the mouse labeled in pink in D. F. Violin plots of mitochondria area in scanning electron micrographs of CA2 SO, SR and SLM, measured in 100 μm² ROIs. Kruskal-Wallis test with Dunn’s correction. SO n= 573, SR n=475, SLM n= 559 from N = 3 mice. Solid line = median; dashed line = upper and lower quartiles. G. Violin plots of mitochondria Feret’s diameter in the same scanning electron micrographs of CA2 SO, SR and SLM as in F. Kruskal-Wallis test with Dunn’s correction. SO = Stratum Oriens, SR = Stratum Radiatum, SLM = Stratum Lacunosum Moleculare. Post hoc comparisons: * = P <0.05; ** = P <0.01; *** = P <0.001; *** = P <0.0001. Scale = (Al) 100μm, (AIV) 25μm, (Aiv, Bill) 5μm, and (Biii) 1 μm.

Figure 4:. MCU preferentially localizes to the proximal dendrites of CA1 after MCU overexpression.

A. (i) Representative 20X epifluorescence image of GFP (left) and MCU (right) staining in CA1 of an AAV-GFP treated mouse. Bottom: Representative 40X confocal images of MCU staining in SR (ii) and SLM (iii) of CA1 in a AAV-GFP treated mouse. Inset is a zoomed 10 × 10 μm crop. A LUT was applied to highlight the dynamic range of intensity. B. (i) Representative 20X epifluorescence image of GFP (left) and MCU (right) staining in CA1 of an AAV-MCU treated mouse. Bottom: Representative 40X confocal images of MCU staining in SR (ii) and SLM (iii) of CA1 in an AAV-MCU treated mouse. A and B were processed the same to maintain differences between AAV-MCU and AAV-GFP mice. C. Mean MCU fluorescence of a 100 × 100 μm ROI from SR and SLM of AAV-GFP and AAV-MCU mice normalized to GFP control average. N = 5 AAV-GFP and 6 AAV-MCU mice. Two-way RM ANOVA with Sidak’s post hoc comparing dendritic layers. Overall effect of AAV treatment and layer were significant (Treatment: F = 113.4, P = <0.0001; Layer: F = 4.81, P = 0.021). See supplemental Table 1 for post hoc comparisons. D. Mean MCU-labeled mitochondrial area in SR and SLM of the AAV-MCU and AAV-GFP mice normalized to GFP control average. Two-way RM ANOVA with Sidak’s post hoc. Overall effect of treatment and layer were significant (Treatment: F = 124.4, P <0.0001; Layer: F = 3.84, P = 0.041). E. (i) Representative 20X epifluorescence image of GFP (left) and COX4 (right) staining in CA1 of an AAV-GFP treated mouse. Bottom: Representative 63X epifluorescence images of COX4 staining in SR (ii) and SLM (iii) of CA1 in a AAV-GFP treated mouse. Inset is a zoomed 10 × 10 μm crop. F. (i) Representative 20X epifluorescence image of GFP (left) and COX4 (right) staining in CA1 of an AAV-MCU treated mouse. Bottom: Representative 63X epifluorescence images of COX4 staining in SR (ii) and SLM (iii) of CA1 in an AAV-MCU treated mouse. E and F were processed the same to maintain differences between AAV-MCU and AAV-GFP mice. G. Mean COX4 fluorescence of a 100 × 100 μm ROI from SR and SLM of the same AAV-GFP and AAV-MCU mice in C-D normalized to GFP control average. N = 5 AAV-GFP and 6 AAV-MCU mice. Two-way RM ANOVA with Sidak’s post hoc comparing dendritic layers. Overall effect of AAV-treatment was not significant (F = 0 02, P = 0.89). Overall effect of layer was significant (F = 10.4, P = 0 001). H. Mean COX4-labeled mitochondrial area in SR and SLM of the AAV-MCU and AAV-GFP mice normalized to GFP control average. Two-way RM ANOVA with Sidak’s post hoc. Overall effect of AAV-treatment was not significant (F = 0.28, P = 0.61). Overall effect of layer was significant (F = 28.9, P = <0.0001). SR = Stratum Radiatum; SLM = Stratum Lacunosum Moleculare. Sidak’s post hoc: * = P < 0.05; ** = P < 0.01; *** = P < 0.001. Scale for A-B(i.) = 50 μm; Scale for A-B(ii.-iii.) = 20 μm; Scale for G = 3 μm, inset = 1 μm (ExM adj.)