Research ArticleNEUROSCIENCE

Neuron-based high-content assay and screen for CNS active mitotherapeutics

See allHide authors and affiliations

Science Advances  08 Jan 2020:
Vol. 6, no. 2, eaaw8702
DOI: 10.1126/sciadv.aaw8702
  • Fig. 1 Assaying mitochondrial morphology in primary neurons.

    (A) Mt mice, work flow, and timeline of the screen. Mitochondria were visualized in primary mouse neurons by the Cre-dependent expression of mitochondria-targeted GFP (Mt-GFP). The image shows the transduced neurons in the field expressing GFP in somatic and neuritic mitochondria (maximum-projection image of confocal z-stacks, three slices, z = 0.7 μm with a 60× objective). (B) Classification of axonal and dendritic mitochondria. Representative images of a neuron (left) expressing Cyto-tdTomato (red) and Mt-GFP (green) and zoomed-in segments of a dendrite and two axons. The frequency distributions of mitochondrial length, located in either dendrites or axons, are plotted in the right panel. Best-fit lognormal distributions (black lines) show that 15% of the mitochondrial lengths can be found in both axons and dendrites (overlap). naxonal = 517, ndendritic = 525. Scale bars, 25 μm (left image) and 5 μm (middle and right images). (C) Image analysis. Maximum z-projection of the green channel (Mt-GFP) containing mitochondria and wide-field image of neurites (Cyto-tdTomato). After image preprocessing (somatic mitochondria removal from both channels, background subtraction and median filtering of mitochondria, and tubeness filtering of neurites), axonal and dendritic mitochondria and neurites were segmented (axonal, 0.5 to 1.4 μm; dendritic, >2.4 μm) by top-hat filtering followed by adaptive thresholding.

  • Fig. 2 Hit parameters for the mitochondrial dynamics screen.

    (A) Representative images showing the effects of FCCP on mitochondrial morphology. FCCP (12.5 μM) for 24 hours induced fragmentation of dendritic mitochondria compared to DMSO treatment. Scale bar, 20 μm. (B to E) Mitochondrial morphology of DMSO- and FCCP-treated neurons. Frequency distribution of the area occupied by individual dendritic (B) and axonal (C) mitochondria. FCCP treatment reduced the average area of dendritic mitochondria (meanDMSO = 1.44 μm2, meanFCCP = 1.06 μm2) and decreased the number of dendritic mitochondria (nDMSO = 28,718, nFCCP = 3713). FCCP increased the number of axonal mitochondria (nDMSO = 33,457, nFCCP = 51,767), while it did not change the average area of axonal mitochondria (meanDMSO = 0.39 μm2, meanFCCP = 0.39 μm2). Average well CA differed significantly between DMSO and FCCP treatment for dendritic (meanDMSO,Dend CA = 1292 ± 213, meanFCCP,Dend CA = 123 ± 44) (B, right) and axonal mitochondria (meanDMSO,Ax CA = 404 ± 79, meanFCCP,Ax CA = 629 ± 83 (C, right). FCCP treatment produced a significant decrease in dendritic length (D), with average medianFCCP = 2.77 ± 0.07 μm compared to medianDMSO = 3.35 ± 0.08 μm (D, right). FCCP treatment also produced a significant increase in the median axonal mitochondrial circularity compared to DMSO treatment (average medianDMSO = 0.83 ± 0.007 and medianFCCP = 0.94 ± 0.004; E, right), also illustrated by their frequency distribution (E, left). Bar graphs represent mean ± SD, nwells = 8 per treatment. Unpaired t tests, ****P < 0.0001, nDMSO,axonal = 33,457 and nFCCP, axonal = 51,767, nDMSO, dendritic = 28,718 and nFCCP, dendritic = 3713, 32 fields, eight wells.

  • Fig. 3 Small-molecule modulators increasing mitochondrial elongation, content, and health.

    (A to C) Primary screen. Average robust Z-scores (N = 4 replicate plates) (left) and the cumulative distributions (right) for median dendritic mitochondrial length (A), dendritic mitochondrial CA (B), and axonal mitochondrial circularity (C). Populations followed a normal distribution (R2Elongation = 0.986, R2Mt Content = 0.987, R2Health = 0.987; cumulative Gaussian fits, blue lines; right). Hits: robust Z-scores > 2.5 (A and B) or < 2.5 (C). Toxic: median axonal circularity > 3 robust Z-scores and median dendritic length < −3 robust Z-scores. (D to F) Rescreen of primary hits. Relative frequency distributions of average robust Z-scores (N = 10 replicate plates) for median dendritic mitochondrial length (D), dendritic mitochondrial CA (E), and axonal mitochondrial circularity (F). Gaussian fits to frequency histograms of compound (blue, n = 149 compounds) and DMSO-treated (gray, n = 42 wells) neurons yielded population means: MDMSO = −0.05 and MCompounds = 1.07 for median dendritic mitochondrial length (D), MDMSO = −0.015 and MCompounds = 1.36 for dendritic mitochondrial content (E), and MDMSO = 0.046 and MCompounds = −1.01 for median axonal mitochondrial circularity (F). (G) Elongation hits increase dendritic mitochondrial length. Relative frequency distributions of dendritic mitochondrial length (left) and the average of their median values (right, mediansDMSO = 3.39 ± 0.08 μm, mediansHit = 3.62 ± 0.14 μm). (H) Mitochondrial content hits increase dendritic mitochondrial CA. Frequency distribution histograms of individual mitochondrial areas (left) and average CA values (right, meanDMSO = 936 ± 183 μm2, meanHit = 1467 ± 397 μm2). (I) Health hits reduce axonal mitochondrial circularity. Frequency distribution histograms of axonal mitochondrial circularities (left) and averages of their median values (right, mediansDMSO = 0.85 ± 0.0185, mediansHit = 0.82 ± 0.01). (G to I) Bar graphs are means ± SD of 10 well values from the 10 replicate plates (ncompound = 1 well per plate, nDMSO = 42 wells per plate); frequency distribution data were collected from four fields per treatment, total mitochondria nDMSO,Elongation = 1741, nHit,Elongation = 3074, nDMSO,Mt Content = 1365, nHit,Mt Content = 2782, nDMSO,Health = 4262, nHit,Health = 4779. Unpaired t test, **P < 0.01, ***P < 0.001. (J and K) Representative images of compounds that increase elongation and mitochondrial content or improve health. Z-projections of fluorescent images of primary neurons expressing Mt-GFP were collected using a 60× objective. Scale bars, 5 μm (J) and 20 μm (K). (L) Hit enrichment during the screen and hit distribution among mitochondrial parameters. The primary screen of 2400 compounds produced 149 potential hits (6.2%), of which 67 were confirmed in the rescreen (45%). Thirty-five of the confirmed hits were hits for more than one parameter.

  • Fig. 4 Mitochondrial dynamics and function.

    (A) Uncorrelated mitochondrial content and neurite sprouting. Rescreen data of dendritic mitochondrial CA and neurite area of the compounds that significantly increased mitochondrial content (>2 robust Z-scores, dashed line). There was no correlation between increased neurite area and increased dendritic mitochondrial CA (Pearson’s r = 0.13, right). Eleven compounds that increased mitochondrial content (gray) increased neurite sprouting (pink) by >2 robust Z-scores. See table S1 for names of numbered compounds. (B) No correlation between dendritic and axonal mitochondrial content. There was no correlation between axonal and dendritic CA (Pearson’s r = 0.22). (C) Coupling of dendritic mitochondrial content and length. Dendritic mitochondrial length correlated with dendritic mitochondrial content (left, Pearson’s r = 0.76) and dendritic mitochondrial count (right, Pearson’s r = 0.6) after compound treatment. The correlation lines are bounded by a 95% confidence interval (gray shaded area). (D) Representative images showing the coupling of mitochondrial content and elongation. Mt-GFP expressing neurons were treated with a hit compound (12.5 μM) that increased both content and elongation by 12 hours after treatment. White arrows indicate the sites of mitochondrial growth. Scale bar, 20 μm. (E) Axonal mitochondrial length is uncoupled with content and count. There was no correlation between axonal mitochondrial length and axonal CA (left, Pearson’s r = −0.22) or count (right, Pearson’s r = −0.23). (F) Mitochondrial dynamics hits increase mitochondrial function. TMRM signal and ATP production of the rescreened mitochondrial dynamics hits (confirmed hits, n = 67) were measured, and a functional increase was determined as robust Z-scores >2 for TMRM signal or ATP production. Sixty-one increased (91%) one or both functional readouts. (G) Mitochondrial dynamics hit parameters differentiate compounds with increased mitochondrial function. Decreased axonal circularity, increased dendritic length, together with increased dendritic CA identified a population of compounds showing increased mitochondrial function measured by ATP production, TMRM signal, or both (functional increase, blue points, >2 robust Z-scores, average from four replicate plates) among the rescreened primary hits (n = 149). (H) Distribution of the 67 rescreened compounds across content, health, elongation, and function. Venn diagram of compounds that significantly increased ATP production or TMRM signal (n = 61), mitochondrial content (n = 53), elongation (n = 50), and/or health (n = 42). Significance threshold, 2 robust Z-scores.

  • Fig. 5 MnMs protect from oligomeric Aβ(1–42)–, peroxide-, and glutamate-induced mitochondrial damage in primary neurons.

    (A) Effects of the selected MnMs. Parameters were measured 24 hours after treatment with 12.5 μM of compounds. Data are mean robust Z-scores from Fig. 4. (B) Effects of selected MnMs in the presence of disease-related insults. Representative images of mitochondrial fragmentation induced by 10 μM Aβ(1–42), 75 μM peroxide (PO), or 25 μM glutamate (GLUT) in Mt-GFP–expressing neurons, and the mitochondrial protection by 12.5 μM of compounds after 48 hours of cotreatment (top left; scale bar, 20 μm). After cotreatment, axonal mitochondrial circularity (health), axonal mitochondrial average area, dendritic mitochondrial length (elongation), and live-dead cell ratio (survival) were either protected (within 2 Z-scores of vehicle-treated control), improved (>2 Z-scores for axonal mitochondrial area, elongation, and survival and <−2 Z-scores for health), or damaged (<−2 Z-scores for axonal mitochondrial area, elongation and survival and > 2 Z-scores for health) compared to the vehicle-treated control (top right and bottom). Compounds fully protective (all parameters are protected or improved) against an insult are bolded. Color bar represents absolute Z-score values toward either improvement (green) or damage (red) of the parameters. Data are mean Z-scores (n = 12 to 18 wells, four fields per well, two independent experiments). (C) Selection of the top seven MnMs. Summary of the steps for the selection of the top seven compounds.

  • Fig. 6 MnMs enhance ATP production from isolated mitochondria, potentiate basal synaptic activity, and increase respiration of mitochondria in vivo.

    (A) Kinetics of ATP production of isolated mitochondria. Baseline-corrected average curves of the ATP production of isolated forebrain mitochondria of newborn C57BL/6J mice starting 20 min after the addition of vehicle (DMSO), 12 μM FCCP, and the top seven MnMs. Data are means ± SD of four runs from one experiment. Single exponential fits yielded amplitudes of 22,498 ± 358 (DMSO, R2 = 0.9), 17,958 ± 216 (FCCP, R2 = 0.98), 28,383 ± 278 (alverine, R2 = 0.96), 28,882 ± 673 (dyclonine, R2 = 0.91), 49,917 ± 838 (naftopidil, R2 = 0.98), 22,764 ± 315 (orphenadrine, R2 = 0.91), 9853 ± 132 (2′,4′-dihydroxychalcone, R2 = 0.92), 6121 ± 99 (4′-hydroxychalcone, R2 = 0.9), and 10,059 ± 166 (rhamnetin, R2 = 0.9). (B) ATP production of isolated mitochondria of the top seven MnMs. Average ATP production of forebrain mitochondria of newborn mice in the presence of DMSO, 12.5 μM FCCP, or 12.5 μM of the top seven MnMs, normalized to average DMSO treatment. Data are means ± SEM of three independent experiments (four runs per experiment) compared to DMSO by one-way analysis of variance (ANOVA) with Dunnett’s post hoc test, **P < 0.01, ***P < 0.001, ****P < 0.0001. (C and D) Synaptic activity of hippocampal neurons. Example traces (C) and quantified sEPSC amplitude/frequency (D) of cultured hippocampal neurons at DIV8 after 24 hours of incubation with DMSO, 12.5 μM dyclonine, alverine, or naftopidil using whole-cell patch-clamping techniques. Data are means ± SEM (n = 33 to 36 cells) compared to DMSO by one-way ANOVA with Dunnett’s post hoc test, *P < 0.05, **P < 0.01. (E and F) Mitochondrial respiration in primary neurons exposed to dyclonine for 24 hours. Before OCR measurements, DIV13 neurons were treated for 24 hours with DMSO (0.1%, black) or dyclonine (10 μM, red). OCRs were measured at baseline respiration (Base) and after addition of mitochondrial substrates pyruvate (P) and malate (M) or succinate (Succ) along with ADP to the permeabilized (saponin, Sa) neurons (first dashed line, S3). Oligomycin addition (second dashed line) blocked proton re-entry through ATP synthase, slowing ETC function to levels necessary to maintain Δψm (S4o; o = oligomycin induced). Addition of FCCP (third dashed line) short-circuited proton influx to bypass ATP synthase, revealing maximal, uncoupled OCR (S3u; u = uncoupler induced). Addition of rotenone (R)/antimycin A (AA), complex I and complex III inhibitors, respectively, were added to block oxidative phosphorylation and measure nonmitochondrial respiration (fourth dashed line). ns, not significant. (G and H) Respiration of isolated mitochondria from dyclonine-treated mice. OCR of whole brain mitochondria of 9-month-old C57BL/6J mice kept on either dyclonine-supplemented (25 mg/kg) or standard water for 7 months was measured in the presence of complex I (G) or complex II (H) substrates preincubated with mitochondria, yielding results similar to those in (E) and (F). Bar graphs [right panels of (E) to (H)] compare the OCR value of the first measured point in each condition between the water- and dyclonine-treated groups (mean ± SEM, n = 11 to 12 wells). Total protein was used to normalize OCR into pmol O2 min−1 μg−1 total protein. Data in (E) to (H) were analyzed using two-way ANOVA with repeated measures. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Supplementary Materials

  • Supplementary material for this article is available at

    Fig. S1. Optimizing the screen.

    Fig. S2. Mitotoxicity and neurotoxicity with FCCP treatment.

    Fig. S3. False-positive hit rate in the screen.

    Fig. S4. Structural clustering of the MnMs.

    Fig. S5. Dose response of Aβ(1–42) oligomers, peroxide, and glutamate treatment on neuronal mitochondrial features.

    Fig. S6. Scheme of field outlier removal and using neutral wells for robust Z-score calculation.

    Table S1. The effect of MnMs on neurite sprouting.

    Table S2. EC50 values of selected MnMs for processes of mitostasis and neurite sprouting.

    Movie S1. Time-lapse video of elongating and fusing mitochondria.

  • Supplementary Materials

    The PDFset includes:

    • Fig. S1. Optimizing the screen.
    • Fig. S2. Mitotoxicity and neurotoxicity with FCCP treatment.
    • Fig. S3. False-positive hit rate in the screen.
    • Fig. S4. Structural clustering of the MnMs.
    • Fig. S5. Dose response of Aβ(1–42) oligomers, peroxide, and glutamate treatment on neuronal mitochondrial features.
    • Fig. S6. Scheme of field outlier removal and using neutral wells for robust Z-score calculation.
    • Table S1. The effect of MnMs on neurite sprouting.
    • Table S2. EC50 values of selected MnMs for processes of mitostasis and neurite sprouting.
    • Legend for movie S1

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Movie S1 (.mov format). Time-lapse video of elongating and fusing mitochondria.

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article