A role for Fis1 in dendritic development

Abstract

Mitochondrial ATP production and calcium handling are critical for metabolic regulation and neurotransmission. Thus, the formation and maintenance of the mitochondrial network is a critical component of neuronal health. Cortical pyramidal neurons contain compartment-specific mitochondrial morphologies that result from distinct axonal and dendritic mitochondrial fission and fusion profiles. We previously revealed that axonal mitochondria are maintained at a small size as a result of high axonal mitochondrial fission factor (Mff) activity. However, loss of Mff activity had little effect on cortical dendritic mitochondria, raising the question of how fission/fusion balance is controlled in the dendrites. Therefore, we sought to investigate the role of another fission factor, fission 1 (Fis1), on mitochondrial morphology, dynamics and function in cortical neurons. We knocked down Fis1 in cortical neurons both in primary culture and in vivo, and unexpectedly found that Fis1 depletion decreased mitochondrial length in the dendrites, without affecting mitochondrial size in the axon. Further, loss of Fis1 activity resulted in both increased mitochondrial motility and dynamics in the dendrites. These results argue Fis1 exhibits dendrite selectivity and plays a more complex role in neuronal mitochondrial dynamics than previously reported. Functionally, Fis1 loss resulted in reduced mitochondrial membrane potential, increased sensitivity to complex III blockade, and decreased mitochondrial calcium uptake during neuronal activity. The altered mitochondrial network culminated in elevated resting calcium levels that increased dendritic branching but reduced spine density. We conclude that Fis1 activity regulates mitochondrial morphological and functional features that influence dendritic tree arborization and connectivity.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-33557-8.

Introduction

Neurons are highly polarized cells that form distinct subcellular compartments termed dendrites, the axon and soma. These domains allow neurons to compartmentalize chemical and electrical signals received from surrounding cells, integrate them, and pass them along in a circuit like process. Neuronal polarization is established by cell-intrinsic and -extrinsic factors, including extracellular gradients, spatially-limited cytoskeletal components, and organelle localization and dynamics^1–4.

Mitochondria are one of the most abundant organelles in neurons, presumably a consequence of the neuron’s high energetic state and its necessity for precise calcium handling. Interestingly, mitochondrial morphology and abundance are strikingly different in the axons and dendrites of cortical pyramidal neurons. In the dendrites, mitochondria have an elongated, tubular morphology with high occupancy of the dendritic space, while in the axon mitochondria are sparse and individual entities positioned at specific locations^5–7. The neurons’ ability to maintain these compartmentalized mitochondrial morphologies appears to be critical for neuronal health as disruption due to environmental factors, genetic mutations or disease results in developmental disorders and potential for neurodegeneration^8–13.

Mitochondrial morphology is established by the combined actions of mitochondrial biogenesis, trafficking, dynamics (fission & fusion), and removal (mitophagy) with mitochondrial dynamics thought to be most prominent^14,15. Mitochondrial fission occurs following the recruitment of the GTPase dynamin-related protein 1 (Drp1) from the cytoplasm to the outer mitochondrial membrane by actin and ‘receptors’ localized at the surface of the mitochondrion^16–19. In eukaryotic cells, four main Drp1 receptors have been identified: fission 1 (Fis1), mitochondrial fission factor (Mff), and mitochondrial elongation factors 1 and 2 (Mief1 & Mief2)^20–25. While Drp1 dependent fission predominates, Drp1-independent fission has also been reported^26–28. Fusion, or the joining together of two mitochondria opposes fission, and is carried out via the outer membrane localized GTPases mitofusion 1 & 2 (Mfn1/2), and the inner membrane localized GTPase optic atrophy 1 (Opa1)^10,29.

Our recent work demonstrated that axonal mitochondria enter the axon and remain small as a consequence of a high rate of fission. This axonal fission relies mainly on Mff-mediated fission as loss of Mff activity results in elongated axonal mitochondria and altered axonal development³⁰. Surprisingly, Mff knockdown had little effect on dendritic mitochondria morphology leaving several open questions about the regulation of mitochondrial dynamics in the dendritic compartment of the neuron.

Fis1 is the second most abundant Drp1 receptor in neurons^31,32, but its role in eukaryotic mitochondrial fission and fusion remains controversial^20,23,33,34. Fis1p was the first fission factor identified in yeast, and along with Mdv4 makes up the only complex to recruit Drp1p^35,36. However, while knockout of Fis1 in cell lines resulted in a small but significant increase in mitochondria size, when compared to Drp1 or Mff knockout, Fis1 appears to play a minor role in eukaryotic mitochondrial morphology and may instead be more involved in mitophagy by end clipping^33,37. Interestingly, recent work shows that human Fis1 can interface with both the fission and fusion machinery and may be important in adjusting the balance between mitochondrial fission and fusion³⁴. While Fis1 is mainly known as a mitochondrial dynamics protein, it has also been proposed to play roles in apoptosis, mitochondrial motility, peroxisomal fission and regulating membrane contact sites³⁸. As most work on Fis1 has employed non-neuronal model systems, it remains unclear whether neuronal Fis1 adopts its canonical role of driving fission, and if it helps establish compartment-specific mitochondria morphologies.

Thus, we explored the role of Fis1 in regulating compartment specific mitochondria morphology in cortical neurons. We report that reduced Fis1 activity surprisingly resulted in smaller dendritic mitochondria without affecting axonal mitochondria size. We show that mitochondrial motility is enhanced in Fis1 knockdown dendrites, resulting in an increased frequency of stochastic mitochondrial dynamics events, and thus lessening the fusion-dominant dendritic profile. Strikingly, the subcellular changes to the dendritic mitochondrial network perturbed cytosolic calcium handling that resulted in aberrant dendritic arborization and spine density.

Results

Loss of Fis1 activity specifically affects the dendritic mitochondrial network

To determine if Fis1 might have a role in compartmentalized neuronal mitochondria morphology, we first visualized mitochondrial Fis1 abundance on dendritic and axonal mitochondria in primary cultured mouse cortical neurons (Supplementary Fig. 1a-c). Interestingly, anti-Fis1 immunofluorescence staining revealed Fis1 abundance is higher on dendritic mitochondria when compared to axonal mitochondria. This result suggested that Fis1 may play a more prominent role in mitochondrial dynamics of the dendrites than axons of cortical neurons.

To test the role of Fis1 in cortical neurons, we next validated shRNA-mediated knockdown constructs for mouse Fis1. We identified two hairpins (see methods) that provided between 60 and 80% knockdown efficiency (Supplementary Fig. 1d-e). To visualize mitochondrial morphology and test the loss of Fis1 activity in cortical pyramidal neurons in a cell autonomous manner, we used either ex utero (EUE) or in utero (IUE) electroporation at E15.5 to express Fis1 shRNA, cytoplasmic tdTomato and mitochondrial matrix-targeted yellow fluorescent protein (mt-YFP) in cortical layer 2/3 pyramidal neurons^4,30,39,40. At 17DIV (EUE) or P21 (IUE), dendritic mitochondrial length in basal dendrites was surprisingly decreased compared to control neurons both in vitro (3.64 μm ± 0.10 for control, 2.15 μm ± 0.07 for Fis1C, 2.17 μm ± 0.04 for Fis1 386, Fig. 1a, c) and in vivo (4.85 μm ± 0.37 for control, 2.72 μm ± 0.18 for Fis1C, 2.84 μm ± 0.09 for Fis1 386, Fig. 1b, d). This outcome is unlikely to be the result of an off-target effect as the two shRNAs have distinct targets, however to confirm the observed alteration in mitochondrial morphology is a result of Fis1 activity, we designed and subcloned a plasmid to express an shRNA impervious mouse Fis1 tagged with the hemagglutinin (HA) tag (Fis1 imp). After validating that it localized to mitochondria and was resistant to shRNA knockdown (Supplementary Fig. 1e-f), we co-expressed a low level of the plasmid (0.33 µg/µL) in combination with those above and were able to rescue the decrease in dendritic mitochondria length (3.99 μm ± 0.08 for Fis1 rescue, (Fig. 1a-b)). We also designed CRISPR guides to knockout mouse Fis1 and validated that it targets Fis1 and reduces Fis1 expression in vitro (see methods, Supplementary Fig. 2). CRISPR knockout of Fis1 also reduced dendritic mitochondrial length ((3.26 μm ± 0.06 for control, 2.05 μm ± 0.03 for knockout (Fig. 1b, Supplementary Fig. 2e)). While Fis1 loss had a strong impact on dendritic mitochondrial size, loss of Fis1 activity did not alter axonal mitochondria size (1.22 μm ± 0.05 for control, 1.26 μm ± 0.06 for Fis1C, 1.21 μm ± 0.06 for Fis1 386, 1.23 μm ± 0.02 for CRISPR control, 1.23 μm ± 0.03 for CRISPR KO, Supplementary Fig. 3). These unexpected results demonstrate that Fis1 is preferentially active in neuronal dendrites and its role is likely more complex than only promoting fission.

Fig. 1.

Fis1 knockdown results in shorter dendritic mitochondria both in vitro and in vivo. (a) Representative basal dendritic segments from 17DIV cortical neurons showing mitochondrial morphology and dendrite structure via matrix targeted YFP and cytosolic tdTomato respectively for each of the labeled conditions. (b) Representative basal dendritic segments from P21 layer 2/3 cortical neurons showing mitochondrial morphology and dendrite structure via matrix targeted YFP and cytosolic tdTomato respectively for each of the labeled conditions in vivo. (c) Quantification of mitochondrial lengths in vitro for each of the indicated conditions demonstrating that loss of Fis1 results in shorter dendritic mitochondria. (d) Quantification of mitochondrial lengths for each of the indicated conditions demonstrating that loss of Fis1 in vivo recapitulates the mitochondrial phenotype observed in cultured neurons. Control _{in vitro} = 3 independent cultures, 516 mitochondria; Fis1 386 shRNA _{in vitro} = 4 independent cultures, 856 mitochondria; Fis1 C shRNA _{in vitro} = 3 independent cultures, 250 mitochondria; Fis1 386 shRNA + Fis1 HA_imp rescue _{in vitro} = 2 independent cultures, 904 mitochondria; CRISPR control _{in vitro} = 3 independent cultures, 935 mitochondria; CRISPR KO _{in vitro} = 4 independent cultures, 1087 mitochondria; Control _{in vivo} = 3 independent brains, 94 mitochondria; Fis1 386 shRNA _{in vivo} = 3 independent brains, 275 mitochondria; Fis1 C shRNA _{in vivo} = 3 independent brains, 220 mitochondria. p values are indicated in the figure following Kruskal-Wallis tests. Data are shown as individual points on box plots with 25th, 50th and 75th percentiles indicated with whiskers indicating min and max values. Large dots depict the averages of biological replicates. Scale bars, 3 μm in a, 6 μm in b.

Fis1 loss increases both dendritic mitochondria motility and dynamics

To determine how Fis1 loss results in decreased mitochondria size, we employed the use of matrix-targeted, photo-activatable GFP⁴¹ which allowed us to quantify the motility and frequency of mitochondrial fission and fusion in the dendrites of cortical neurons (Fig. 2, Supplementary Video 1). Time-lapse imaging following photo-activation revealed that loss of Fis1 activity results in both increased mitochondrial motility and mitochondrial dynamics in basal dendrites of cortical neurons. As neurons mature, mitochondria in both the axons and dendrites become fixed at specific points along the process leading to a large proportion of stationary mitochondria^42–45. As expected in control dendrite segments at 14DIV, we observed that only ~ 3% (2.5% ± 0.6) of mitochondria were motile over a 15 min imaging session. However, upon Fis1 loss, the percent of motile dendritic mitochondria increased almost four times (11.6% ± 2.1 for Fis1 C, 13.3% ± 1.8 for Fis1 386, Fig. 2a-e). The previously noted oscillating behavior of neuronal mitochondria⁴⁶ was also increased following Fis1 knockdown (Fig. 2d). Concurrently, by quantifying the fission and fusion events of photoactivated mitochondria, we observed a doubling of mitochondrial fission (1.1 ± 0.1 events per 15 min for control, 2.2 ± 0.2 for Fis1 C, 2.4 ± 0.2 for Fis1 386), as well as modestly increased fusion rates (1.6 ± 0.2 events per 15 min for control, 1.9 ± 0.3 for Fis1 C, 2.9 ± 0.3 for Fis1 386) in the dendrites following the loss of Fis1 activity (Fig. 2f-g) leading to a reduction in the fusion dominant phenotype found in control neurons. These results support a novel role of Fis1 in the suppression of dendritic mitochondrial motility that may facilitate its canonical role in dynamics regulation.

Fig. 2.

Loss of Fis1 activity increases both mitochondria motility and mitochondrial dynamics. (a) Photo-activated mitochondria in a control basal dendrite at t = 0 (red) and at t = 15 min (green). The merged images showing overlap of the two timepoints and a kymograph of the entire 15 min imaging session demonstrating little dendritic mitochondria movement in mature dendrites. (b, c) Photo-activated mitochondria in a Fis1 386 shRNA (b) or Fis1 C shRNA (c) basal dendrite at t = 0 (red) and at t = 15 min (green). The merge images showing overlap of the two timepoints and a kymograph of the entire 15 min imaging session demonstrating increased mitochondria motility following Fis1 knockdown. (d) Pie graphs showing percentages of stationary (dotted area), oscillating (checkered area), and motile (filled area) mitochondria in the dendrites of labeled conditions. (e) Quantification of motile mitochondria percentage for dendrite segments demonstrating increased motility in Fis1 knockdown neurons. (f) Scheme for photo-activation experiments to quantify fission and fusion dynamics in neuronal dendrites. (g) Quantification of dendritic mitochondria fission and fusion rates as events per 15 min for control and knockdown dendrites showing that both fission and fusion are increased upon Fis1 loss in neurons. Control _motility = 37 basal dendrite segments, 663 mitochondria; Fis1 386 shRNA _motility = 15 basal dendrite segments, 279 mitochondria; Fis1 C shRNA _motility = 19 basal dendrite segments, 351 mitochondria; Control _dynamics = 35 basal dendrite segments, 39 fission events, 57 fusion events; Fis1 386 shRNA _dynamics = 31 basal dendrite segments, 75 fission events, 90 fusion events; Fis1 C shRNA _dynamics = 30 basal dendrite segments, 64 fission events, 57 fusion events (All with 3 independent cultures). p values are indicated in the figure following Kruskal-Wallis tests. Data are shown as individual points on box plots with 25th, 50th and 75th percentiles indicated with whiskers indicating min and max values for e, or individual points with mean ± SEM for g. Scale bars, 5 μm.

Fis1 knockdown negatively impacts dendritic mitochondrial membrane potential

To determine whether the altered morphology observed following Fis1 knockdown impacts the mitochondrion’s functional state, we performed EUE to co-electroporate primary cultured neurons with DNA encoding either a control plasmid (pLKO) or the Fis1 386 shRNA and mt-YFP. Fis1 386 shRNA was chosen for the rest of the experiments as Fis1 386 and Fis1 C knockdown effects are conserved but the Fis1 386 has a higher knockdown efficiency (Supplemental Fig. 1). At 14DIV, we assessed mitochondrial membrane potential-dependent uptake of the cationic dye, tetramethylrhodamine methyl ester (TMRM). We observed slightly reduced uptake of TMRM into Fis1 knockdown dendritic mitochondria via signal intensity (680 ± 27 for control, 569 ± 34 for Fis1 386, Fig. 3a-c) supporting a reduction in membrane potential. To confirm this observation, we performed EUE with DNA encoding a matrix targeted fusion protein containing both the pH-sensitive fluorescent protein SypHer and mScarlet. At 14DIV, we measured the ratio of SypHer to mScarlet in dendritic mitochondria and observed a reduction in the ratio of green to red fluorescence in Fis1 knockdown neurons compared to control (1.104 ± 0.01 for control, 1.025 ± 0.01 for Fis1 386, Fig. 3d) representing an acidification of dendritic mitochondrial matrices in Fis1 knockdown. As matrix acidification is potentially associated with impaired proton pumping to the intermembrane space, we tested how dendritic mitochondria responded to complex 3 inhibition by antimycin A. Treatment with 1.25µM antimycin A led to a decrease in TMRM signal in Fis1 knockdown mitochondria as compared to control dendritic mitochondria (-12.7% ± 2.0 for control, -23.0% ± 2.3 for Fis1 386, Fig. 3e-f) demonstrating an increased sensitivity to complex 3 inhibition.

Fig. 3.

Fis1 knockdown mitochondria in dendrites have slightly reduced membrane potential and matrix pH. (a) Representative images of dendritic mitochondria co-electroporated with mt-YFP (top) and control plasmid followed by loading with TMRM (middle). (b) Representative images of dendritic mitochondria co-electroporated with mt-YFP (top) and Fis1 386 shRNA followed by loading with TMRM (middle). (c) Quantification of raw TMRM dendritic mitochondria intensities showing reduced membrane potential in Fis1 knockdown mitochondria. (d) Quantification of normalized mt-SypHer showing that Fis1 knockdown results in a slightly more acidic matrix pH. (e) Graph of relative TMRM intensity following treatment with antimycin A (1.25µM, complex 3 inhibitor) and FCCP (1µM, ionophore to open the inner mitochondrial membrane). (f) Quantification showing dendritic mitochondria upon Fis1 loss are more sensitive to complex three inhibition. Control _TMRMinitial = 3 independent cultures,155 mitochondria; Fis1 386 shRNA _TMRMinitial = 3 independent cultures, 90 mitochondria; Control _mtSypHer = 3 independent cultures,155 mitochondria; Fis1 386 shRNA _mtSypHer = 3 independent cultures,90 mitochondria; Control _TMRManta = 3 independent cultures,113 mitochondria; Fis1 386 shRNA _TMRManta = 3 independent cultures,117 mitochondria. p values are indicated in the figure following Mann-Whitney tests. Data are shown as individual points on box plots with 25th, 50th and 75th percentiles indicated with whiskers indicating min and max values for c, d & f, or mean ± SEM for e. Large dots depict the averages of biological replicates. Scale bars, 5 μm.

Loss of Fis1 activity reduces dendritic mitochondria calcium uptake following evoked neuronal activity

Recent work has shown that dendritic mitochondria play an important role in calcium handling during neuronal activity^45,47–50. As alterations to mitochondrial membrane potential or matrix volume would impact mitochondrial calcium uptake, we tested if Fis1 knockdown had an impact on matrix calcium accumulation during glutamate uncaging to induce synaptically relevant calcium influx⁵¹. Following EUE with plasmids encoding matrix targeted GCaMP6f (mt-GCaMP6f) and matrix targeted mScarlet (mt-mScarlet), we imaged 14–16 DIV neurons following a single, basal dendritic stimulation at 50 microns away from the cell body in imaging media supplemented with 1.3 mM MNI-glutamate. 2–3 s following stimulation, we observed a robust ~ 3 fold increase in dendritic mt-GCaMP6f fluorescence for control neurons, while in Fis1 knockdown neurons the increase in fluorescence occurred on the same time scale but was reduced (2.7 fold ± 0.2 for control, 2.0 fold ± 0.1 for Fis1 386, Fig. 4a-d), arguing that dendritic mitochondria have a reduced ability to take up calcium during neuronal activity following Fis1 loss.

Fig. 4.

Fis1 loss reduces mitochondrial calcium uptake following evoked activity resulting in higher resting cytoplasmic calcium levels. (a) Representative intensity images of somatodendritic mitochondria in a 15DIV neuron co-electroporated with mt-GCaMP6f (intensity gradient) and control plasmid during a time course following glutamate uncaging. (b) Plot of normalized intensity of mt-GCaMP6f in control (grey) or Fis1 386 shRNA (blue) basal dendrites following glutamate uncaging. (c) F_max fold change of mt-GCaMP6f following glutamate uncaging shows decreased calcium influx in Fis1 knockdown mitochondria. (d) Quantification of mt-GCaMP6f extinction after F_max (peak fluorescence reached following stimulation) suggesting that Fis1 loss doesn’t affect mitochondrial calcium extrusion. F_end is the final timepoint t = 30 s (e) Quantification of F_min (lowest fluorescence intensity) Fluo4 showing increased resting cytoplasmic calcium levels in Fis1 knockdown neurons compared to surrounding wildtype neurons. (f) Quantification of the % of F_peak for baseline cytoplasmic GCaMP6f following ionomycin treatment (to reveal maximum GCaMP6f brightness) confirming increased resting calcium levels in Fis1 knockdown neurons. (g) F_max fold change in GCaMP6f signal following glutamate uncaging show no difference in peak calcium levels in Fis1 knockdown neurons. Control _mtGCaMP6f = 39 dendrites; Fis1 386 shRNA _mtGCaMP6f = 33 dendrites; Control _Fluo4 = 16 neurons; Fis1 386 shRNA _Fluo4 = 16 neurons; Control _cytoGCaMP6f = 35 neurons; Fis1 386 shRNA _cytoGCaMP6f = 34 neurons (All with 3 independent cultures). p values are indicated in the figure following Mann-Whitney tests (except e which is a Wilcoxon matched pairs test). Data are shown as individual points on box plots with 25th, 50th and 75th percentiles indicated with whiskers indicating min and max values, except for e which is a paired comparison of values for each pair. Large dots depict the averages of biological replicates. Scale bar, 5 μm.

Decreased mitochondrial uptake of calcium increases resting cytoplasmic calcium

To understand the impact of reduced mitochondrial calcium uptake on neuronal calcium handling, we tested if cytoplasmic calcium levels are altered in Fis1 knockdown neurons. First, we loaded 14DIV cortical cultures with Fluo4-AM, a membrane permeable calcium sensing dye, to visualize resting calcium levels, and observed a significant increase in resting calcium levels in Fis1 knockdown neurons compared to nearby, non-electroporated neurons in the same dish (3053 au ± 247 for control, 3832 au ± 331 for Fis1 386, Fig. 4e). Next, cytoplasmic GCaMP6f was used with the glutamate uncaging protocol to visualize calcium accumulation in the cytoplasm following evoked neuronal activity in cultured 14–16 DIV cortical neurons. Interestingly, the resting fluorescence levels of cytoplasmic GCaMP were higher in Fis1 knockdown neurons following normalization to the maximum fluorescence upon ionomycin addition (22.2% ± 1.4 for control, 28.6% ± 1.9 for Fis1 386, Fig. 4f) confirming the results with Fluo4-AM, even though peak levels of calcium influx were unaffected following stimulation (4.0 fold ± 0.4 for control, 3.8 fold ± 0.3 for Fis1 386, Fig. 4g). These results argue that Fis1 loss does not impact peak cytoplasmic calcium influx following a sufficient stimulus but that these neurons have higher resting calcium levels.

Fis1 knockdown results in decreased ER calcium uptake following evoked activity

As mitochondrial transfer of calcium to endoplasmic reticulum (ER) is known to play an important role in calcium handling and calcium oscillations⁴⁷, we tested whether or not ER calcium dynamics were altered following Fis1 loss. Using the glutamate uncaging protocol on 14-16DIV cortical neurons electroporated with either control plasmid or Fis1 386 shRNA along with an ER localized and optimized GCaMP6 (erGCaMP6-150), we observed a trend towards slightly reduced ER calcium release (-0.11 ± 0.02 for control, -0.08 ± 0.01 for Fis1 386 shRNA), but significantly reduced ER calcium reuptake following stimulation in Fis1 knockdown neurons (0.12 ± 0.04 for control, 0.03 ± 0.02 for Fis1 386 shRNA, Supplementary Fig. 4).

Spontaneous neuronal activity is largely unaffected following Fis1 loss

Since Fis1 knockdown neurons have altered calcium handling and a higher resting calcium level, we imaged 14-16DIV cultured neurons electroporated with cytoplasmic GCaMP6f and either control plasmid or Fis1 386 shRNA to determine if loss of Fis1 altered spontaneous neuronal activity. Following imaging at 2fps for 2 min, we surprisingly did not observe significant differences in the number of calcium transients or transient amplitudes (Supplementary Fig. 5a-d). We did observe an increase in the coefficient of variation between spikes in Fis1 knockdown neurons suggesting an increased variability in neuronal firing (Supplementary Fig. 5e).

Fis1 knockdown results in increased basal dendrite branching

As calcium regulation plays an important role in many cytoskeletal processes, and is known to impact dendritic formation and branching, we next set out to test whether loss of Fis1 activity results in altered dendritic branching^4,52. Thus, we performed Scholl analysis⁵³ on 14DIV cortical neurons expressing cytosolic tdTomato and either control or Fis1 386 shRNA. Interestingly, we found an increase in the number of crossings between 290 microns and 400 microns from the center of the cell body (Fig. 5a-c) demonstrating an increase in dendritic branching of Fis1 knockdown neurons.

Fig. 5.

Dendritic branching and spine density are altered during development following Fis1 knockdown. (a,b) Representative images of 14DIV cultured cortical neurons electroporated with tdTomato and either control (left, a) or Fis1 shRNA (right, b). (c) Quantification of dendritic branching showing that Fis1 knockdown neurons have increased branching. (d) Representative images of P14 layer 2/3 basal dendritic segments electroporated with tdTomato and either control (top) or Fis1 shRNA (bottom) to visualize spine density. (e) Quantification of dendritic spine density at P14 demonstrating that loss of Fis1 results in a reduced spine density in vivo. p values are indicated in the figure following Mixed-effects analysis (c), or Mann-Whitney test (e). Control _branching = 24 neurons; Fis1 386 shRNA _branching = 24 neurons from 3 independent cultures. Control _spines = 51 dendrite segments; Fis1 386 shRNA _spines = 37 dendrite segments from 3 independent brains. Data are shown as a plot of the number of crossings per distance from the cell body (b), or individual points on box plots with 25th, 50th and 75th percentiles indicated with whiskers indicating min and max values (d). Large dots depict the averages of biological replicates. Scale bars, 100 μm for a, 5 μm for c. Note that dendritic tree morphologies are highly variable in culture. No overt polarity phenotypes were observed in vitro.

Dendritic spine density is decreased following Fis1 knockdown in vivo

As dendritic branching was increased, we next tested if Fis1 knockdown also affected dendritic spine density. We chose to visualize spines in vivo, as in vitro spine density is notoriously variable based on culture health and density^54–56, by performing IUE at E15.5 with cytoplasmic tdTomato and either control plasmid or Fis1 386 shRNA plasmid. At P14, we perfused the mice, sectioned and stained the brains, then imaged basal dendrites of layer 2/3 cortical pyramidal neurons at high magnification. Upon knockdown of Fis1, we observed a significant reduction in the density of spines on basal dendrites during development in vivo (0.75 spines/µm ± 0.04 for control, 0.57 spines/µm ± 0.03 for Fis1 386 shRNA, Fig. 5d-e). This result coupled with the branching results above demonstrates that loss of Fis1 results in altered dendritic development.

Discussion

In the present study, we reveal that Fis1 activity is critical in the development of the dendritic mitochondrial network. Our results demonstrate that loss of Fis1 surprisingly reduces dendritic mitochondrial length by increasing both the rate of trafficking and dynamics along the dendrites thus tipping the balance away from a fusion dominant phenotype in the dendrites. The resulting dendritic mitochondria have slightly lower membrane potential, increased sensitivity to antimycin A, and a reduced capacity to take up calcium during neuronal activity. Fis1 loss also resulted in increased dendritic branching and reduced spine density presenting a scenario whereby reduced calcium handling by the mitochondrial network alters dendritic development.

Our results argue for a compartment-specific role for Fis1 in the development of the mitochondrial network as the loss of Fis1 activity alters dendritic mitochondrial size, trafficking and dynamics without altering the axonal mitochondrial pool. The question of how Fis1 specifically regulates dendritic mitochondria remains unclear. Our immunocytochemistry data indicates that Fis1 protein is nearly twice as abundant on dendritic mitochondria versus axonal mitochondria likely explaining the preferential impact of its loss on the dendritic pool of mitochondria. Surprisingly though, the loss of Fis1 resulted in decreased, not increased, mitochondrial length. This could potentially arise from a few different possibilities: (1) Fis1 may interact with other ‘Drp1 receptors’ such as Mff or Mief1/2 to alter the recruitment or activity of Drp1 at the outer mitochondrial membrane^23–25,57, (2) Fis1 may preferentially recruit non-active conformations of Drp1 to dendritic mitochondria^14,19,37,58, or (3) human Fis1 was recently shown to interact with the fusion machinery thus the loss of Fis1 may directly impact fusion processes³⁴. Each of these scenarios could support a model where higher Fis1 levels in the dendrites results in increased mitochondrial length, but future work will be necessary to test these possibilities. Further, since Fis1 protein expression is not completely eliminated in our system, it is plausible that the remainder of Fis1 is in an augmented activity state, resulting in a paradoxical increase in Drp1 recruitment. Fis1-mediated recruitment of Drp1 is regulated by protein phosphorylation at Thr34 and Tyr38^59–61. Emerging work posits phosphorylation at either of these sites promotes the oxidation of Cys41 and the formation of Fis1 covalent homodimers that drive mitochondrial fission under oxidative stress⁶². Thus, if Fis1 depletion results in oxidative stress – perhaps secondary to diminished mitochondrial membrane potential and coupling efficiency – there exists a mechanistic explanation for this paradoxical outcome. Future assessment of mitochondrial redox state and bioenergetic capacity under Fis1 depletion will be necessary to test this hypothesis.

Strikingly, Fis1 loss resulted in both increased mitochondrial trafficking and increased mitochondrial dynamics which may suggest a mechanism where Fis1 is a component of the mitochondrial anchoring system. Previous work from multiple groups has shown that mitochondrial trafficking is reduced with dendritic maturation and mitochondria become stably captured at specific points along the dendrites^44,45,50 but the mechanism has remained elusive. Coupled with our results that Fis1 impacts mitochondrial-ER calcium coupling, it is plausible that Fis1 may have a potential role in MERCS formation or dynamics leading to a model where mitochondrial capture and the development of mitochondrial-ER coupling in the dendrites is linked. In fact, recent work in non-neuronal cells have shown that Fis1 is present at MERCS^19,63,64 but its exact role there remains unclear. If Fis1 plays a role in mitochondrial capture, it will be interesting to determine if the mere reduction in mitochondrial motility is sufficient to dampen mitochondrial dynamics in a manner that favors a pro-fusion phenotype.

The link between mitochondrial morphology and function is well documented across multiple cell types with mitochondrial shape impacting membrane potential, ability to utilize different fuels as substrate, calcium dynamics, and potential for turnover^30,65–69. The reduction in membrane potential and acidification of the matrix pH in neurons lacking Fis1 activity suggests a potential reduction in the ability of the mitochondria to efficiently run the electron transport chain to produce ATP. However, the altered matrix pH increases the difficulty to measure ATP levels in the matrix with genetically encoded probes as the more recent and sensitive ATP probes are also affected by pH. Interestingly, recent work from multiple groups argues that dendritic and axonal mitochondria likely have intrinsic differences in substrate usage and ATP production and those techniques may provide the groundwork to isolate the different pools of mitochondria within cells and directly measure their metabolism^70,71. Additionally, reduced mitochondrial membrane potential is a signal for Pink1 dependent mitophagy to remove “damaged” mitochondria^72,73. Mitophagy is closely tied to mitochondrial dynamics as longer, more fused mitochondria are too large to undergo mitophagy and thus are “protected”⁷⁴. Fis1 has been proposed to play both mitochondrial dynamics dependent and independent roles in mitophagy^33,37. One potential hypothesis is that the increased dynamics and transport we observe upon Fis1 loss is a disruption in the signaling between the mitochondrial dynamics and mitophagy machinery. Future work will be required to tease apart the order of operations following Fis1 loss as currently it is not clear whether the functional deficits precede the altered dynamics/transport or if the altered dynamics drive impaired function.

The role of calcium as a cytosolic signal for dendritic branching and postsynaptic development is well established^4,52. Even before evoked activity is present, developing neurons have spontaneous calcium transients that coincide with the timing of dendritic branching⁷⁵. A major pathway connecting dendritic development to calcium signaling is via calmodulin and calcium/calmodulin-dependent protein kinases (CaMKs)⁷⁶. CaMKII is the most well studied with CaMKII alpha and beta shown to alter dendritic outgrowth, filopodial extension, and regulate interaction with actin^77–79. Dendritic spine growth during long-term potentiation (LTP) is also linked to calcium dynamics through the remodeling of spine actin^80–82. Interestingly, a number of groups have recently observed that synaptic activity is directly tied to mitochondrial fission via calcium signaling to directly link neuronal activity to metabolic and calcium handling needs^49,83–85. For instance, higher resting calcium could result in increased activity of Camkk2 which in turn would activate the Ampk-Mff axis to promote mitochondrial fission^84,85. Future investigation will be necessary to determine if Fis1 plays a role, either directly or indirectly with Mff, in this dendritic calcium-dependent fission.

The observation that Fis1 results in opposite phenotypes of increased dendritic branching but reduced spine density is surprising. The idea that an increased resting calcium level would be detrimental to spine density is not novel and in fact multiple groups have shown in beta-amyloid induced models of Alzheimer’s disease that there is a link between the increase in cytoplasmic calcium and spine loss^86–89. On the other hand, most reports of increased cytoplasmic calcium show decreased dendritic branching via reduced filopodial growth or growth cone collapse^52,90,91. One potential explanation is that cultured neurons, which were used to check branching due to technical limitation, respond differently than neurons in vivo as a result of extrinsic factors that are not recapitulated in vitro. Alternatively, this increased branching maybe an attempt of Fis1 deficient neurons to perform homeostatic rescaling to promote maintenance of normal firing rates^92–94. Future experiments will need to look at whether the reduced spine density coincides with altered spine maturation or spine-head size, and whether the branching phenotype is conserved in vivo. It will also be interesting to determine whether Fis1 deficient neurons have calcium-dependent alterations in postsynaptic receptor trafficking or membrane docking.

Taken together, our data demonstrate that Fis1 plays a compartment-specific, yet unconventional, role in the development of the dendritic mitochondrial network.

Methods

All experiments were conducted in accordance with the ARRIVE guidelines.

Animals

Animals in the study were handled and experiments conducted according to Institutional Animal Care and Use Committee (IACUC) approved protocols at the Oklahoma Medical Research Foundation (OMRF, 20–23, 23 − 18). Time-pregnant females of CD-1 IGS strain (Strain Code: 022) were purchased at Charles River Laboratories and used for in utero electroporation experiments and primary neuronal cultures. When exsanguination was used the animals were rendered fully unconscious before the exsanguination process was initiated. This study was approved by the Institutional Animal and Use Committee (IACUC) at the Oklahoma Medical Research Foundation (OMRF, protocols 20–23 and 23 − 18).

Plasmids

pCAG:mtYFP-P2A-tdTomato, pCAG:tdTomato and pCAG:mt-YFP were previously published in⁹⁵. pCAG:2xmtpaGFP p2a 2xmtmScarlet was created by cloning a gene block encoding (from IDT) 2xmtpaGFP p2a 2xmtmScarlet into pCAG via restriction digest. pCAG:mt-SypHer was created by PCR of mt-SypHer from Addgene plasmid 48251 (a gift from Nicolas Demaurex) and cloning it 3’ to the CAG promoter. pCAG GCaMP6f was created by PCR of GCaMP6f from Addgene plasmid 40755 (a gift from Douglas Kim) and cloning it 3’ to the CAG promoter. pCAG mt-GCaMP6f was created by excising the YFP from pCAG mt-YFP and inserting GCaMP6f in its place. pCAG erGCaMP6-150 was created by PCR of GCaMP6-150 from Addgene plasmid 91777 (a gift from Douglas Kim) and cloning it 3’ to the CAG promoter while adding a KDEL retention motif to the C-terminus via the primers. pCAG HA-mouse Fis1 and pCAG HA-mouse Fis1 impervious were created by ordering gene blocks encoding a HA tag N-terminal to mouse Fis1 variant 1 or a mouse Fis1 variant 1 with the following sequences mutated (site 1 for Fis1 386 shRNA: cctggttcgaagcaaatacaa to tgccttgtgaggtctaagtacaa and site 2 for Fis1 C shRNA: ccaaggagctggaacgcctgattgataag to caaaggaacttgagcggctcatagataaa) and cloning them into the pCAG vector via restriction digest. pLKO.1 cloning vector was purchased from Addgene (plasmid 10878, a gift from David Root) and used as a control plasmid. Fis1 386 shRNA plasmid was ordered from Sigma (pLKO.1 TRCN0000124386, target sequence: CCTGGTTCGAAGCAAATACAA). Fis1 C shRNA plasmid was ordered from Origene (pGFP-C-shLenti clone C, target sequence: CCAAGGAGCTGGAACGCCTGATTGATAAG). Fis1 CRISPR guides were generated using CHOPCHOP online software targeting exon 3, 4 or 5. Oligos of each selected guide were ordered through IDT and cloned into pOrange (Addgene plasmid 131471) between Bbs1. Exon 3 guide: CGTGGGCAACTACCGGCTCAAGG, Exon 4 guide: AAGGCTCTAAAGTATGTGCGAGG, Exon 5 guide: CTGGTAGGCATGGCCATCGTTGG. Targeting and cutting efficiency of Fis1 guides 3 and 4 were tested using pCAG-EGxxFP (Addgene plasmid 50716) with the genomic DNA for the 3rd thru 4th exon of mouse Fis1 cloned between Nhe1 and EcoR1 (xx region). If a guide targets and leads to efficient cutting of the xx region, this will allow for efficient recombination of the EGFP without the xx region and thus the ability to fold into a fluorescent protein.

Cell lines

Human embryonic kidney cells (HEK293T/17) were purchased from ATCC (CRL-11268). 1x105 cells were suspended in media (DMEM, Gibco) with penicillin/streptomycin (0.5x; Gibco) and FBS (Sigma) were seeded in 6 well tissue culture dishes (Falcon). Transfection with plasmid DNA (1 mg/mL) using jetPRIME® reagent (Polyplus) according to manufacturer protocol was performed 24 hours after seeding. For the EGxxFP assay, cells were live imaged 48 hours after transfection. For western blotting 72 hours following transfection, cells were carefully washed with 1xPBS (Gibco) then collected into RIPA buffer with protease inhibitor cocktail.

Western Blotting

Aliquots of the collected samples were separated by SDS-PAGE and then transferred to a polyvinylidene difluoride (PVDF) membrane (Amersham). After transfer, the membrane was washed 3X in Tris Buffer Saline (10 mM Tris-HCl pH 7.4, 150 mM NaCl) with 0.1% of Tween 20 (T-TBS), blocked for 1 h at room temperature in Odyssey Blocking Buffer (TBS, LI-COR), followed by 4 °C overnight incubation with the appropriate primary antibody in the above buffer. The following day, the membrane was washed 3X in T-TBS, incubated at room temperature for 1 h with IRDye secondary antibodies (LI-COR) at 1:10,000 dilution in Odyssey Blocking Buffer (TBS), followed by 3X T-TBS washes. Visualization was performed by quantitative fluorescence using an Odyssey CLx imager (LI-COR). Signal intensity was quantified using Image Studio software (LI-COR). Primary antibody used for Western-blotting was Rabbit anti-HA (CST3724) or Rabbit anti-GAPDH (PA1-16777). Total protein was assessed using the Revert 700 Total protein stain (LI-COR 926-11010). Blots were cut based on the ladder before addition of primary antibody to allow for blotting for Gapdh and HA on the same loaded lanes.

Ex utero electroporation

A mix of endotoxin-free plasmid preparation (1–2 mg/mL) and 0.5% Fast Green (Sigma) mixture was injected using FemtoJet 4i (Eppendorf) into the lateral ventricles of isolated heads of E15.5 mouse embryos. Embryonic neural progenitor cells were electroporated using an electroporator (ECM 830, BTX) and gold paddles with four pulses of 20 V for 50 ms with 500 ms interval and an electrode gap of 1.0 mm. Dissociated primary neuron culture was performed after ex utero electroporation.

Primary neuronal culture

Following ex utero electroporation, embryonic mouse cortices (E15.5) were dissected in Hank’s Buffered Salt Solution (HBSS) supplemented with Hepes (2.5 mM), CaCl₂ (1 mM, Sigma), MgSO₄ (1 mM, Sigma), NaHCO₃ (4mM, Sigma) and D-glucose (30 mM, Sigma), hereafter referred to as cHBSS, and incubated in cHBSS containing papain (Worthington; 14 U/mL) and DNase I (100 µg/mL) for 15 min at 37 °C with a gentle flick between incubation. Samples were washed with cHBSS three times, and dissociated by pipetting on the fourth wash. Cells were counted using Countess™ (Invitrogen) and cell suspension was plated on poly-D-lysine (1 mg/mL, Sigma)-coated glass bottom dishes (MatTek) in Neurobasal media (Gibco) containing FBS (2.5%) (Sigma), B27 (1 X) (Gibco), and Glutamax (1 X) (Gibco). After 7 days, media was changed with supplemented Neurobasal media without FBS.

In utero electroporation

A mix of endotoxin-free plasmid preparation (1-2 mg/mL) and 0.5% Fast Green (Sigma) was injected into one lateral hemisphere of E15.5 embryos using FemtoJet 4i (Eppendorf). Embryonic neural progenitor cells were labelled using the electroporator (ECM 830, BTX) with gold paddles at E15.5. Electroporation was performed by placing the anode (positively charged electrode) on the side of DNA injection and the cathode on the other side of the head. Five pulses of 38 V for 50 ms with 500 ms interval and an electrode gap of 1.0mm were used for electroporation. Embryos were randomly assigned to control or knockdown without regard to sex.

Intracardial perfusion

Animals were anesthetized using 5% isoflurane mixed with air and exsanguinated 14 or 21 days after birth (P14 or P21) by terminal intracardial perfusion. Fixative was kept on ice during the entire procedure. Direct perfusion was performed with 30 mL of fixative: 2% PFA/0.075% GA in PBS. Animals were then dissected to isolate brains, that were later subjected to 20 h post fixation in the same fixative that was used for perfusion.

Immunohistochemistry

Cultured neurons were fixed for 10 min on ice in 2% PFA (Alfa Aesar) with 0.075% GA (Electron Microscopy Science, EMS), and then washed with PBS (Sigma). Cells were permeabilized with 0.2% Triton X-100 in PBS and incubation in 0.1% BSA and 2.5% goat serum in PBS was followed to block nonspecific signals. Primary and secondary antibodies were diluted in the blocking buffer described above and incubated at 4 °C overnight. Coverslips were mounted on slides with Fluoromount G (SouthernBiotech). Primary antibodies used in these experiments were mouse anti-HA (Biolegend, 1:500), chicken anti-GFP (Aves Lab 1:1000), rabbit anti-dsRed (Abcam 1:1000), rabbit anti-Fis1 (Proteintech 1:100) and all secondary antibodies were Alexa-conjugated (Invitrogen) and used at 1:1000 dilution. For visualizing endogenous Fis1, after fixation the cultures were washed with PBS with 22 mg/mL of glycine (Sigma) for 10 min. Cells were then placed in blocking buffer for 1 h with 0.2% Saponin (Sigma) in 1xPBS with 5% NGS and 22 mg/mL glycine. Primary (1:100 O/N at 4 C) and secondary (1:500 2 h at RT) antibodies were incubated in 1xPBS with 1% BSA (Sigma), and 0.2% Saponin.

Immunocytochemistry

Following fixation and washing, brains were embedded in 3% low melt agarose (RPI, A20070) in 1x PBS. Brains in agarose cubes were sectioned using a vibratome (Leica VT1200) at 120 μm. Sections were then incubated with primary antibodies (chicken anti-GFP Aves Lab 1:1000, rabbit anti-dsRed Abcam 1:1000) that were diluted in the Blocking buffer (1%BSA, 0.2%TritonX-100, 5%NGS in PBS) at 4 °C for 48 h. Subsequently sections were washed 6 times for 10 min in PBS and incubated with secondary antibodies (Alexa conjugated goat anti-chicken488 and goat anti-rabbit568 1:1000) at 4 °C for 48 h. The excess of secondary antibodies was removed by six, 10 min washes in 1x PBS. Sections were then mounted on slides and coverslipped with Aqua PolyMount (PolyMount Sicences, Inc.) and kept at 4 °C.

Live imaging of cultured neurons

Cultured neurons, or basal dendrite segments, were imaged on a Nikon Ti2 widefield system equipped with a Hammamatsu ORCA-Fusion CMOS camera, a custom penta-band cube for 378/474/554/635/735 (± 25 nm) excitation and associated emission (432/515/595/681/809 ± 25 nm) with an Aura III light engine, and 60 × (1.4NA) oil objective, and live imaging chamber from OXO (UNO-T-H-CO2) with objective warmer. In addition, a 405 nm laser (LUN-F, 50mW) with XY galvo control (Opti-microscan) is connected to perform targeted ROI based stimulation. The whole system is controlled by Nikon Elements. For live imaging, culture medium was removed and replaced with warmed cHBSS. For each experimental paradigm both controls and experimental conditions were imaged with the same light powers (5 to 10%), camera exposures (50 to 200ms) and all other microscope settings.

For mitochondrial dynamics photo-activation experiments, 8 micron ROIs were selected along multiple basal dendrites of each cell imaged. A before stimulation image was taken, then each box was individually stimulated with 405 nm light at 2% laser power for 100µs, immediately followed by timelapse imaging every 30 s for 15 min with 488 and 561 nm light.

For TMRM experiments, neuron cultures were preloaded with 20nM TMRM for 20 min then placed in cHBSS with 5nM TMRM. Cells were allowed to equilibrate for 20 additional minutes before imaging. Antimycin A and FCCP (Sigma) were used at 1.25µM and added at the timepoints indicated while timelapse imaging basal dendrite segments every 30 s with 561 nm light.

For mt-SypHer experiments, mitochondria in basal dendrite segments were imaged live with 488 and 561 nm light.

For Fluo4-AM experiments, neuron culture were loaded with 5µM Fluo4-AM (Life Technologies Corp, F14217) in imaging media for 20 min then imaged live with 488 nm light.

For glutamate uncaging experiments, imaging media (cHBSS) was supplemented with 1.3mM MNI-glutamate (MNI-caged-L-glutamate, Tocris 1490) and 0.001mM TTX (to block spontaneous activity, Tocris, 1078/1). After an appropriate basal dendrite with a similar number of mitochondria was identified, an ROI was masked at 50 microns from the cell body followed by timelapse imaging at 2fps for 30 s with 488 nm light. After 10 s, the 405 laser (2% laser power for 100µs) was pulsed to stimulate the selected dendrite. Mitochondria could be visualized via mt-mScarlet to confirm similar numbers of mitochondria were present. ER was not visualized, thus an assumption was made that ER is similar in these dendrite segments.

For spontaneous activity, neuronal cultures were switched to cHBSS and neuronal cell bodies imaged at 2fps for 30 s with 488 nm light to visualize GCaMP activity.

Imaging of fixed brain sections and cultured neurons

Fixed samples were imaged on a Zeiss LSM 880 confocal microscope controlled by Zeiss Black software. Imaging was performed with two lasers that excite at 488 nm and 561 nm and associated GFP and RFP excitation/emission mirrors, together with Zeiss objectives 40 × (1.2NA) with 2x digital zoom (mitochondria), or 100x oil (1.25NA) with 3x digital zoom (spines). Z-stacks were acquired at Nyquist based on wavelength and optical settings. For dendrite branching experiments, tdTomato expressing neurons were imaged with a 20x objective and 561 nm light. Z-stacks were acquired at 2 micron steps through the entire dendrite tree by manually selecting the lowest and highest z plane with detectable fluorescence. For each experimental paradigm both controls and experimental conditions were imaged with the same light powers (5 to 10%), scan speed (6) and all other microscope settings.

Axons and dendrites are identified by their morphological features. These include dendrites being thicker, shorter and containing spines. Axons are thinner, longer and their mitochondria are much smaller and sparser even following Fis1 manipulations.

Quantification and statistical analysis

For data in Fig. 1 and supplementary Fig. 3, individual mitochondria were measured via the mt-YFP signal from isolated, basal dendrite segments or axonal segments (~ 100 microns from the cell body) with NIS Elements AR (Nikon) using the length measurement tool on the raw images.

For data in Fig. 2, photo-activated dendritic mitochondria were manually tracked in NIS Elements AR to calculate motility and fission/fusion dynamics. Motility was assessed as movement greater than 5 microns, oscillating mitochondria are those that moved but less than 5 microns, while stationary mitochondria remained in place throughout the timelapse. Fission was counted if a single mitochondrion split into two or more individual mitochondria, while fusion was visualized and counted by the mixing of matrix content resulting in an altered green to red ratio.

For data in Fig. 3a-c, regions of interest were drawn around individual mitochondria in basal dendritic segments. Average ROI intensity was collected in NIS Elements AR (Nikon). For 3d, regions of interest were drawn around individual mitochondria labeled with mt-SypHer-mScarlet in basal dendrite segments. Average ROI intensity was collected in NIS Elements AR for both green and red channels and the ratio of green to red plotted. For 3e, regions of interest were drawn around basal dendrite segments containing multiple mitochondria and TMRM intensity normalized to the starting value was plotted over time following the addition of the indicated drugs. For 3f, the fold change is plotted (delta F/F initial) at the time point directly before FCCP addition (15 min) to quantify the impact of Antimycin A addition on membrane potential. FCCP was used as a positive control driving dramatic mitochondrial membrane potential loss.

For data in Figs. 4 and supplementary Fig. 4, regions of interest for analysis of GCaMP intensity were drawn over stimulated basal dendrite segments at 100 microns from the cell body. Average ROI intensity normalized to the starting value was collected over the time via the Time Measurement plugin in NIS Elements AR. Calculations used to visualize/normalize the data (such as ∆F/F₀) are presented on the y-axis of each graph in the figures.

For data in Fig. 5, branching was measuring by using the Scholl analysis plugin in FIJI on optically isolated 14DIV cortical neurons. Crossings were plotted and analyzed in GraphPad Prism. For spine data, 2 basal dendrites per P14 neuron were imaged at high magnification. In NIS Nikon Elements AR, individual spines were manually counted and the length of the dendritic segment was calculated. Spine density was then quantified as spines per micron of dendrite.

For data in supplemental Fig. 5, average GCaMP6f intensity was measured via regions of interest drawn over the cell body of control or Fis1 knockdown neurons. Following extraction of intensity values via the Time Measurement plugin in NIS Elements AR, the maximum and minimum values of each cell were set to 1 and 0 respectively to normalize for differences in GCaMP6f expression level. These values were plotted vs. time to visualize spikes. The amplitudes are displayed as a cumulative frequency plot and area under the curve is determined for the independent cultures. Spikes were manually counted for each neuron with a threshold of 0.2 set for a spike. Coefficient of variability was calculated (standard deviation/mean) between spikes for each neuron to determine the dispersion in spike frequency.

Statistical analysis was done in GraphPad’s Prism 9. Statistical tests, p-values, and (n) numbers are presented in the figure legends. Gaussian distribution was tested using D’Agostino & Pearson’s omnibus normality test. All analyses were performed on raw imaging data without any adjustments. Images in figures have been adjusted for brightness and contrast (identical for control and experimental conditions in groups compared), and have been processed with Nikon’s proprietary denoise.ai for visualization purposes only.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: TL. Formal analysis: KS, TP, PK, PS, AM, TL. Funding acquisition: TL. Investigation: KS, TP, PK, PS, JW, KC. Methodology: KS, TP, PK, TL. Visualization: KS, TP, PK, TL. Writing – original draft: KS, TL. Writing – review & editing: KS, TP, PK, TL.

Funding

This research was supported by grants from NIGMS (R35GM137921) and the Presbyterian Health Foundation to TL.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.