Mitochondrial complexity is regulated at ER-mitochondria contact sites via PDZD8-FKBP8 tethering

Abstract

Mitochondria-ER membrane contact sites (MERCS) represent a fundamental ultrastructural feature underlying unique biochemistry and physiology in eukaryotic cells. The ER protein PDZD8 is required for the formation of MERCS in many cell types, however, its tethering partner on the outer mitochondrial membrane (OMM) is currently unknown. Here we identify the OMM protein FKBP8 as the tethering partner of PDZD8 using a combination of unbiased proximity proteomics, CRISPR-Cas9 endogenous protein tagging, Cryo-electron tomography, and correlative light-electron microscopy. Single molecule tracking reveals highly dynamic diffusion properties of PDZD8 along the ER membrane with significant pauses and captures at MERCS. Overexpression of FKBP8 is sufficient to narrow the ER-OMM distance, whereas independent versus combined deletions of these two proteins demonstrate their interdependence for MERCS formation. Furthermore, PDZD8 enhances mitochondrial complexity in a FKBP8-dependent manner. Our results identify a novel ER-mitochondria tethering complex that regulates mitochondrial morphology in mammalian cells.

Fig. 1. PDZD8 shows hotspots in close proximity to mitochondria by high-speed single molecule tracking.

a Diffraction-limited imaging of the ER (cyan) and the mitochondria (red) in the periphery of a representative COS7 cell with the simultaneously measured likelihood of finding a PDZD8 molecule in a 1-min window. The locations of the mitochondria in the probability map are indicated with dotted white lines. Boxes correspond to the mitochondria-associated hotspots (MitoHS, magenta) or non-mitochondria-associated hotspots (OtherHS, green) in b and c. Data are representative of 14 cells from two independent experiments. b Zooms of MitoHS in a showing individual PDZD8 trajectories engaging with the hotspots and the associated PDZD8 probability density. Dotted lines indicate hotspot boundaries as used for subsequent analysis. c Zooms of the OtherHS in a showing individual PDZD8 trajectories engaging with the hotspots and the associated PDZD8 probability density. Dotted lines indicate hotspot boundaries as used for subsequent analysis. The shape and size are consistent with endosomal contact sites, as described in Obara et al.. d Plots of the distance of individual PDZD8 molecules shown in different colors from the center of example hotspots over time. Plots are from MitoHS 2 (top panel) or OtherHS 3 (bottom panel) shown in a. Note the long stretches where single molecules remain engaged in the hotspots. e PDZD8 shows reduced diffusion within both classes of hotspots as compared to freely diffusing in the surrounding ER. n = 89 and 99 hotspots for the MitoHS and OtherHS, respectively. Statistical analysis was performed using two-sided Mann–Whitney test for comparing % reduction in 2D D_eff in MitoHS between OtherHS and one sample Wilcoxon signed-rank test for comparing the hypothetical median (0) and the median of % reduction in 2D D_eff within MitoHS or OtherHS. ns: p > 0.05, ****p < 0.0001. The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values. f Sizes of MitoHS are significantly larger than those of OtherHS in the same cells. n = 89 and 99 hotspots for the MitoHS and OtherHS, respectively. The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values.Statistical analysis was performed using two-sided Mann–Whitney test. **p = 0.0065. g PDZD8 dwell times in individual MitoHS and OtherHS. Inset shows the leaving frequency (k_out) of individual PDZD8 molecules from probability hotspots associated or unassociated with mitochondria. Scale bar: a 5 µm, b, c 100 nm. Source data are provided as a Source Data file.

Fig. 2. Proteomics screening identified a PDZD8–FKBP8 protein complex.

a Scheme of the immunoprecipitation and LC–MS/MS analysis using the Pdzd8-3× HA KI mice neocortices. The immunoprecipitates from the neocortices of Pdzd8-3× HA mice or the control littermates using an anti-HA antibody were subjected to the LC–MS/MS analysis. b Diagram describing the genomic sequence of Pdzd8-3× HA KI mice. The sequence of a 3× HA tag was knocked-in at the C-terminus of the Pdzd8 coding sequence. c Volcano plot of proteins differentially binding to the PDZD8-3× HA. FKBP8 is labeled in red. Protrudin and VAPA, which have been previously reported to interact with PDZD8, are labeled in blue. The plot represents data from three biological replicates. The p-value was calculated using an unadjusted two-tailed Student’s t-test. d Scheme of labeling proteins in the vicinity of endogenous PDZD8. A Biotin ligase TurboID fused to PDZD8 generates biotin–5′-AMP from biotin and ATP. The biotin–5′-AMP can covalently bind to proteins located within about 20 nm of endogenously expressed PDZD8-TurboID. e Diagram describing the genomic sequence of the PDZD8-TurboID KI HeLa cell. The sequence of TurboID-P2A-Neo^r was knocked-in at the C-terminus of the PDZD8 coding sequence. f Volcano plot of proteins differentially biotinylated with biotin in the PDZD8-TurboID KI HeLa cell. FKBP8 is labeled in red. Protrudin and VAPA, which have been previously reported to interact with PDZD8, are labeled in blue. The volcano plot represents three biological replicates. The p-value was calculated using an unadjusted two-tailed Student’s t-test. g Numbers of proteins highly enriched in the IP–MS (c) and TurboID-MS (f) are shown in a Venn diagram. Twelve proteins are commonly found in the two proteomes. Note that FKBP8 is the only protein annotated with mitochondrial localization. h, i Analysis of the interaction between endogenous FKBP8 and endogenous PDZD8-3× HA from the mouse neocortex (h), or endogenous PDZD8-Venus from NIH3T3 cells. Extracts from neocortex in Pdzd8-3×HA KI mouse (h) or Pdzd8-Venus KI NIH3T3 cells (i) were subjected to immunoprecipitation (IP) with antibodies to HA or GFP respectively. The resulting precipitates as well as the original tissue extracts (Total) were subjected to immunoblot analysis with antibodies to FKBP8, VAPA, MFN2, HA (h), GFP (i), and β-actin (i). Data are representative of three independent experiments. Source data are provided as a Source Data file.

Fig. 3. A direct binding partner FKBP8 is required for PDZD8 to be recruited on mitochondria.

a Sensorgrams of SPR assay. Recombinant human PDZD8 (1, 28–)—FLAG was immobilized on the sensor chip and FKBP8 (1–380)—Histag with indicated concentrations were injected. b SPR responses at equilibrium (R_eq) were plotted against FKBP8 concentration. The plot of R_eq versus FKBP8 concentration was fitted to a monovalent binding model to determine K_D values. c Schematic diagram of the mutants of PDZD8 deleted with various domains. TM transmembrane, SMP synaptotagmin-like mitochondrial-lipid-binding, C2n N-terminal sequence of C2 domain, PDZ PDZ domain, C2c C-terminal sequence of C2 domain, C1 C1 domain, CC coiled-coil region. d Pdzd8^f/f::Cre^ERT2 MEFs expressing a series of deletion mutants of PDZD8-3× FLAG shown in (c) and HA-FKBP8 were treated with 1 μM 4-hydroxy tamoxifen (4-OHT) and cell extracts were immunoprecipitated with anti-HA antibody. Western blotting was performed with anti-HA antibody and anti-FLAG antibody. Data are representative of three independent experiments. e GST-Pulldown assay from the mixture of recombinant GST—Thrombin cleavage site—human PDZD8 (1, 28–506)—HA and recombinant human FKBP8 (1–380)—Histag in vitro. FKBP8—Histag was eluted only from the GST beads incubated with GST- PDZD8 (1, 28–506)—HA. Data are representative of two independent experiments. f HA-Pulldown assay with recombinant human PDZD8 (1, 28–506)—HA and recombinant human FKBP8 (1–380)—Histag in vitro. The FKBP8-Histag was enriched when incubated with hPDZD8 (1, 28–506)—HA, compared to the negative controls (buffer or buffer with BSA). Data are representative of two independent experiments. g Immunofluorescence analysis of Pdzd8-Venus KI NIH3T3 cells knocking out endogenous FKBP8 by confocal microscopy with a Nikon Spatial Array Confocal (NSPARC) detector. The cells were transfected with the control gRNA (upper two rows) or three gRNAs against FKBP8 (bottom two rows), Cas9, and transfection marker mtagBFP2, and stained with antibodies to GFP, and Tomm20 for visualizing endogenous PDZD8-Venus (green) and mitochondrial outer membrane (magenta), respectively. Scale bars: 5 µm (original) 1 µm (magnified). h Quantification of the percentage of endogenous PDZD8-Venus intensity overlapping with mitochondria (Tomm20-positive area). The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values. n = 17, 14 cells for the control and FKBP8 KO cells. Statistical analysis was performed using two-sided Mann–Whitney U test. **p = 0.003. Source data are provided as a Source Data file.

Fig. 4. PDZD8 and FKBP8 tether the ER and mitochondria cooperatively.

a Representative electron micrographs of Pdzd8^f/f::Cre^ERT2 MEFs infected with lentivirus carrying shControl or shFKBP8, and treated with or without 0.5 µM 4-OHT. MERCS (yellow arrowheads) were more frequently observed in the Control cells than in Pdzd8 cKO, Fkbp8 KD, and Pdzd8 cKO + Fkbp8 KD cells. Scale bars: 200 nm. b Quantification of the MERCS length normalized by the mitochondrial circumference. The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values. n = 33, 29, 39, 34 cells from two independent experiments for the control, Pdzd8 cKO, Fkbp8 KD, and Pdzd8 cKO + Fkbp8 KD cells, respectively. Statistical analysis was performed using one-way ANOVA and Fisher’s LSD test. ****p < 0.0001, ***p = 0.0003. c The interaction plot corresponding to b. Dots show the mean of each condition. d The results of the two-way ANOVA test. The low (<0.01) variation of the interaction shows that PDZD8 and FKBP8 cooperatively affect the areas of MERCS. Source data are provided as a Source Data file.

Fig. 5. Endogenous PDZD8 and FKBP8 colocalize on mitochondria.

a Immunofluorescence analysis of PDZD8-Halotag KI HeLa cells. The cells were treated with 200 nM of Janelia Fluor 549 for 20 h and then stained with antibodies to FKBP8 and to Tomm20. The boxed regions of the top panels are shown at higher magnification in the corresponding lower panels. Arrowheads indicate PDZD8 colocalized both with FKBP8 and Tomm20. Scale bars, 5 μm (original) or 1 μm (magnified). b The ratios of FKBP8 intensity on or outside (off) the mitochondria were determined for images obtained as described in a. Error bar is mean ± s.e.m. of nine cells from two independent experiments. The average of three cytoplasmic regions cropped from each of the nine cells was used for the analysis. c Distribution of PDZD8 puncta with the indicated distance to the nearest FKBP8 puncta was determined for images obtained as described in a. The distance from centroids of each PDZD8 punctum to the nearest FKBP8 centroids was calculated within mitochondria (on mito) or outside of the mitochondria (off mito) respectively. The scrambled FKBP8 centroids were created by shuffling pixels within mitochondria or outside of the mitochondria in the images showing FKBP8 centroids. Nine cells from two independent experiments were used in the calculation. Two-sided Kolmogorov–Smirnov test was used to test statistical significance. ****P < 0.0001, *P = 0.0254. d The ratios of PDZD8 intensity overlapped with FKBP8 on mitochondria (Mander’s coefficients) were determined for images as described in a. The scrambled FKBP8 images were created by shuffling pixels within mitochondria in the FKBP8 channel. Data are representative of two independent experiments (9 cells). A two-sided paired t-test was used to test statistical significance. *P = 0.0172. e The means of PDZD8 intensity in the FKBP8-present or FKBP8-absent area on mitochondria were determined for images as in a. Data are representative of two independent experiments (nine cells). A two-sided paired t-test was used to test statistical significance. **P = 0.0073. Source data are provided as a Source Data file.

Fig. 6. Overexpression of the mitochondrial FKBP8 recruits endogenous PDZD8 to the mitochondrial proximity.

a Immunofluorescence analysis of PDZD8-Halotag KI HeLa cells overexpressing HA-FKBP8^N403K. Cells transfected with either the control plasmid (upper two rows) or the plasmid encoding HA-FKBP8^N403K (bottom two rows), along with the mitochondrial marker YFP-ActA, were treated with 200 nM of JF549 for 20 h, and subsequently, the fixed cells were observed using a confocal microscope equipped with a Nikon Spatial Array Confocal (NSPARC) detector. Scale bars: 5 µm (upper panels), 1 µm (lower magnified panels). b Quantification of the percentage of endogenous PDZD8-Halotag intensity overlapping with mitochondria (YFP-ActA-positive area). The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values. n = 78, 48 cells for the control and FKBP8^N403K overexpressing cells from two independent experiments. Statistical analysis was performed using two-sided Student’s t-test. ***P = 0.0007. c–e Correlative light and electron microscopy (CLEM) analysis in a PDZD8-HaloTag KI HeLa cell. Cells overexpressing with Venus-FKBP8^N403K or YFP-ActA (for the control) were treated with 200 nM of JF549 for 20 h and then fixed cells were observed by a confocal microscope. After that, ultra-thin sections (50 nm thick) were created and observed in a field emission scanning electron microscope (FE–SEM). Electron micrographs of the serial 8 slices were corresponding to an optical section of fluorescence images. Segmentations and 3-demensional (3D) reconstructions of mitochondria and the ER within 25 nm of mitochondria (MERCS) in electron micrographs were shown in c. 3D reconstruction from electron micrographs (shown as “EM”) were merged with fluorescence images (shown as “LM”) in d. The z projection of mitochondria and MERCS in EM was overlaid with fluorescence images in e. Arrowheads indicate PDZD8 puncta that localize to MERCS. Source data are provided as a Source Data file.

Fig. 7. Overexpression of the mitochondrial FKBP8 narrows the MAM–OMM distance.

a Overexpression of FKBP8^N403K increases the intensity and the abundance of the PDZD8-Venus puncta. At cryogenic temperatures, autofluorescence of the NIH3T3 cells in the green channel is strong, hence the presence of puncta in the control cells. The Venus fluorophore also emits light in the red channel. The images represent two stacks (field of view: 638.9 µm²) for the control and four stacks (field of view: 638.9 µm²) for FKBP8^N403K OE, respectively. b–e For the cryo-ET analysis, mScarlet-FKBP8^N403K overexpression (OE) was used to increase the number of associations between mitochondria and mitochondria-associated membrane (MAM) captured in cryo-FIB milled lamellae. An SEM image of a target cell before Cryo-FIB milling is shown (b). Cryo-fluorescence imaging of lamellae confirmed the presence of mScarlet-FKBP8^N403K in the target cells. Note that the apparent mismatch between the cryo-fluorescence image and cryo-TEM is due to factors including autofluorescence, differences in resolution, registration errors, distortions in the imaging plane, and ice-crystal contaminations (c). Using medium-mag high-resolution TEM montages of the lamellae, mitochondria with MAM were targeted for high-resolution tilt series acquisition (d). Eighty tomograms containing MAM were obtained for the OE condition (of which 20 were fully segmented and labeled), and 10 tomograms containing MAM were obtained for the control (of which 6 were fully segmented and labeled). Two representative tomograms from the OE condition corresponding to the arrows in panel (d) are shown (e). The image of b represents more than 35 cells. The image of c represents 13 lamellae. FLM: fluorescent light microscopy. f A surface morphometrics analysis was used to calculate the MAM-outer mitochondrial membrane (OMM) distance. The distances are shown as a heatmap. Mammalian OMM and MAM show a great deal of heterogeneity in their membrane ultra-structure. g Aggregate analysis of the area-weighted MAM–OMM distance histogram shows a shift to smaller distances in the overexpression condition compared to the control. Source data are provided as a Source Data file.

Fig. 8. Volume EM analysis revealed that PDZD8 promotes mitochondrial complexity by inhibiting FKBP8.

a Schematic of mitochondrial morphology analysis by the volume EM. Serial electron micrographs were obtained by imaging the serial sections with an FE–SEM. After cropping the volume, the mitochondria in the volume were semi-automatically extracted with an AI-assisted pipeline. b Formula of calculating MCI (mitochondrial complexity index). MCI is calculated as SA³/(16π²V²), where SA is the surface area and V is the volume of each mitochondrion (see  details in the Methods section). c Representative 3D reconstruction of mitochondria extracted from serial EM images acquired by array tomography in Pdzd8^f/f::Cre^ERT2 MEFs infected with lentivirus carrying shControl or shFKBP8, and treated with or without 0.5 µM 4-OHT. Scale bars: 1 µm. d Quantification of MCI (mitochondrial complexity index). The data are presented as individual points on box plots, with the center indicating the median, and the 25th and 75th percentiles represented by the box. Whiskers extend to the minimum and maximum values. n = 63, 90, 54, and 52 mitochondria from 5, 5, 5, and 4 cells for the control, Pdzd8 cKO, Fkbp8 KD, and Pdzd8 cKO + Fkbp8 KD cells, respectively. Statistical analysis was performed using one-way ANOVA and Fisher’s LSD test. ****p < 0.0001, ***p = 0.0006, *p = 0.033. Source data are provided as a Source Data file.