Visualizing multiple inter-organelle contact sites using the organelle-targeted split-GFP system

Abstract

Functional integrity of eukaryotic organelles relies on direct physical contacts between distinct organelles. However, the entity of organelle-tethering factors is not well understood due to lack of means to analyze inter-organelle interactions in living cells. Here we evaluate the split-GFP system for visualizing organelle contact sites in vivo and show its advantages and disadvantages. We observed punctate GFP signals from the split-GFP fragments targeted to any pairs of organelles among the ER, mitochondria, peroxisomes, vacuole and lipid droplets in yeast cells, which suggests that these organelles form contact sites with multiple organelles simultaneously although it is difficult to rule out the possibilities that these organelle contacts sites are artificially formed by the irreversible associations of the split-GFP probes. Importantly, split-GFP signals in the overlapped regions of the ER and mitochondria were mainly co-localized with ERMES, an authentic ER-mitochondria tethering structure, suggesting that split-GFP assembly depends on the preexisting inter-organelle contact sites. We also confirmed that the split-GFP system can be applied to detection of the ER-mitochondria contact sites in HeLa cells. We thus propose that the split-GFP system is a potential tool to observe and analyze inter-organelle contact sites in living yeast and mammalian cells.

Figure 1

Schematic diagram of the split-GFP-fusion proteins used in this study. FL represents the full-length protein. Tom20_1–33, Tom70_1–70, ERj1_1–200, Sec63_1–240 and Pex3_1–60 mean the N-terminal segments of the indicated proteins.

Figure 2

Split-GFP fragments on the ER and MOM label their contact sites. (A) Diagram of the split-GFP system for detecting inter-organelle interactions. (B) Yeast cells expressing Mmm1-GFP and mitochondria-targeted RFP (Su9-RFP) were imaged by confocal fluorescent microscopy. Maximum projection images reconstituted from the z-stacks were shown. Scale bar represents 5 µm. (C) Yeast cells expressing Tom71-GFP1-10 and Ifa38-mCherry-GFP11 were imaged by confocal fluorescent microscopy. Maximum projection images reconstituted from the z-stacks were shown. Scale bar represents 5 µm. (D,E) Yeast cells expressing split-GFP proteins, Ifa38-mCherry-GFP11 and Tom71-GFP1-10 (D) or Ifa38-GFP1-10 and Tom71-GFP11 as well as ER-targeted mCherry (E) were imaged by confocal fluorescent microscopy. Single focal plane images were shown. Dotted lines indicate plasma membranes. Scale bars represent 5 µm.

Figure 3

Co-localizations of assembled split-GFP fragments on the ER and mitochondria with ERMES. (A,B) Yeast cells expressing Mdm34-RFP from the multi-copy plasmid (pMY3) (A) or Mdm12-mScarlet from the chromosome (B) were transformed with plasmids coding for the indicated split-GFP proteins and imaged by confocal fluorescent microscopy. Maximum projection images reconstituted from z-stacks were shown. Dotted lines indicate plasma membranes. Scale bar represents 5 µm. Box and whisker plots (minimum and maximum) show the maximum length of the indicated fluorescent signals. *P = 0.0111, n = 52 (Mdm34-RFP) and n = 122 (split-GFP). **P = 0.0231, n = 232 (Mdm12-mScarlet) and n = 119 (split-GFP) (unpaired two-tailed t-test). (C) mmm1∆ and mdm12∆ cells harboring a CEN-URA3-plasmid expressing Mmm1 (pYC1) or Mdm12 (pYC4) were transformed with plasmids expressing the indicated split-GFP fusion proteins and Vps13-D716H, Mcp1, ChiMERA or empty vectors. Serial 10-fold dilutions of the resulting transformant cells were spotted onto selective synthetic media plates with or without 5′-FOA (5′-fluoroorotic acid). (D) HeLa cells transiently expressing Tom20N-FLAG-GFP1–10 and ERj1N-V5-GFP11 were stained with MitoTracker. Scale bar represents 10 µm. (E) HeLa cells used in (D) were subjected to immunofluorescence with anti-V5 antibodies. Scale bars represent 10 µm.

Figure 4

Loss of ERMES does not lead to a decrease in the number of split-GFP signals between the ER and mitochondria. (A) Wild-type and mmm1∆ cells expressing Ifa38-GFP1-10, Tom71-GFP11 and Su9-RFP were imaged by confocal fluorescent microscopy. Maximum projections reconstituted from z-stack GFP images were shown (middle panel) or overlaid with DIC images (lower panel). Scale bar represents 5 µm. (B) The maximum projection images surrounded by a rectangular frame in (A) were expanded and shown with Su9-RFP. Dotted lines indicate plasma membranes. Scale bar represents 2 µm. (C) The number of GFP dot signals was counted. Error bars represent standard deviation. n = 87 and 115 for wild-type and mmm1∆ cells, respectively. (D) Yeast cells expressing ER-targeted GFP were stained with MitoTracker Red CMXRos and imaged by confocal fluorescent microscopy. Single focal plane images were shown. Scale bars represent 2 µm.

Figure 5

Loss of Mmm1 does not affect the number of split-GFP signals generated at the vacuole-mitochondria contact sites. (A) Wild-type and mmm1∆ cells expressing Vph1-GFP1-10, Tom71-GFP11, and Su9-RFP were imaged by confocal fluorescent microscopy. Maximum projections reconstituted from z-stack GFP images were shown (middle panel) or overlaid with DIC images (lower panel). Scale bar represents 5 µm. (B) The maximum projection images surrounded by a rectangular frame in (A) were expanded and shown as merged images with Su9-RFP. Dotted lines indicate plasma membranes. Scale bars represent 2 µm. (C) The number of GFP dot signals was counted. Error bars represent standard deviation. n = 87 and 115 for wild-type and mmm1∆ cells, respectively.

Figure 6

A number of inter-organelle contact sites visualized with the split-GFP system. (A) Diagram of inter-organelle contact sites we tested. The alphabet X in squares corresponds to Fig 6X showing the corresponding inter-organelle contact sites. (B–J) Wild-type cells expressing the indicated split-GFP proteins were imaged by confocal fluorescent microscopy. Su9-RFP, mCherry-PTS1, BipN-mCherry-HDEL and Erg6-mCherry were used to label mitochondria, peroxisomes, the ER and LDs, respectively (organelle marker proteins were summarized in Materials and Methods). The vacuole and LDs were stained with fluorescent dyes, FM4-64 and BODIPY 558/568 C₁₂, respectively. Maximum projection was performed to show split-GFP signals on mitochondria, peroxisomes and LDs while a single focal plane was shown for the ER and vacuole. Dotted lines indicate plasma membranes. Scale bars represent 2 µm.

Figure 7

The peripheral ER and vacuole contact sites visualized by the split-GFP probe. Wild-type cells expressing the Ifa38-GFP1-10, Dpp1-V5-GFP11 and BipN-mCherry-HDEL were imaged by confocal fluorescent microscopy. Different single focal plane images are shown. Scale bars represent 5 µm.