Targeted Analysis of Mitochondrial Protein Conformations and Interactions by Endogenous ROS-Triggered Cross-Linker Release

Abstract

The study of in situ conformations and interactions of mitochondrial proteins plays a crucial role in understanding their biological functions. Current chemical cross-linking mass spectrometry (CX-MS) has difficulty in achieving in-depth analysis of mitochondrial proteins for cells without genetic modification. Herein, this work develops the reactive oxygen species (ROS)-responsive cross-linker delivery nanoparticles (R-CDNP) targeting mitochondria. R-CDNP contains mitochondria-targeting module triphenylphosphine, ROS-responsive module thioketal, loading module poly(lactic-co-glycolic acid) (PLGA), and polyethylene glycol (PEG), and cross-linker module disuccinimidyl suberate (DSS). After targeting mitochondria, ROS-triggered cross-linker release improves the cross-linking coverage of mitochondria in situ. In total, this work identifies 2103 cross-linked sites of 572 mitochondrial proteins in HepG2 cells. 1718 intra-links reveal dynamic conformations involving chaperones with ATP-dependent conformation cycles, and 385 inter-links reveal dynamic interactions involving OXPHOS complexes and 27 pairs of possible potential interactions. These results signify that R-CDNP can achieve dynamic conformation and interaction analysis of mitochondrial proteins in living cells, thereby contributing to a better understanding of their biological functions.

Keywords: cross‐linking mass spectrometry; mitochondria; nanoparticles; protein complexes; reactive oxygen species‐responsive.

Figure 1

Schematic illustration of R‐CDNP technique and analysis process for native mitochondrial protein conformation and interaction in intact living cells.

Figure 2

Characterization of R‐CDNP. A) TEM image of R‐CDNP (scale bar = 100 nm). B) CLSM images of HepG2 cells after incubating with Coumarin 6‐labeled R‐CDNP for 3 h. Coumarin 6 (green channel) was utilized to label nanoparticles, and MitoTracker Deep Red (red channel) was employed to co‐stain mitochondria (scale bar = 5 µm). C) Zeta potential changes of R‐CDNP against different H₂O₂ concentrations (20 nM, 100 µM, 1 mm, 10 mm) after 7 h. The general H₂O₂ concentration (20 nM) of normal tissues was considered as the control group. D) Diameter changes of R‐CDNP against different H₂O₂ concentrations (20 nm, 100 µm, 1 mm, 10 mm) after 24 h. The general H₂O₂ concentration (20 nM) of normal tissues was considered as the control group. E) Viability of HepG2 cells for varied concentrations of R‐CDNP upon 7 h treatment (mean ± SD, n = 4 independent experiments). F) Boxplot showing the proteomic change of HepG2 cells. HepG2 cells were treated with 0.6 mg mL⁻¹ R‐CDNP for 7 h, and untreated HepG2 cells were seen as the control group (n = 3 independent experiments, two‐tailed Student's t‐test).

Figure 3

Cross‐linked mitochondrial proteome of HepG2 cells. A) Cα‐Cα distance distribution of identified cross‐links when mapped onto the structures in the PDB database. B) Score distribution of identified PPIs belonging to the STRING database. C) Interaction network between selected groups of mitochondrial proteins. CI = Complex I, CII = Complex II, CIII = Complex III, CIV = Complex IV, CV = Complex V. Nodes were colored based on the submitochondrial localization of proteins, blue for OMM (outer mitochondrial membrane), pink for IMS (intermembrane space), purple for IMM (inner mitochondrial membrane), and green for Matrix. Lines were colored according to interactions found in the STRING database (black) or not (red).

Figure 4

The cross‐linking information among OXPHOS complexes. A) The interaction network. These nodes were indicated using different colors, blue for OXPHOS subunits, green for OXPHOS assembly factors, and red for others. The interactions reported in the STRING database were shown in black, and unreported interactions (unique PPIs) were in red. B) Mapping of identified cross‐links onto CII. Human SDHA structure (PDB: 6VAX shown in palegreen) and SDHB structure (AlphaFold shown in palecyan) were positioned using porcine complex II (PDB: 4YXD shown in gray) as a template. C) Mapping of identified cross‐links onto CI‐CIII2‐CIV. The bovine CIV of CI‐CIII2‐CIV (PDB: 5XTH. CI, CIII, and CIV were in palegreen, palecyan, and gray) was replaced with human CIV (PDB: 5Z62 shown in lightpink). D) Mapping of identified cross‐links onto bovine CV (PDB: 6ZQM shown in palegreen). The numbers in the box from left to right represented the number of identified cross‐links, the number of cross‐links mapped onto structures, and the number of cross‐links satisfying the distance restraint. Blue lines represented the cross‐links satisfying the distance restraint, and red lines represented the cross‐links exceeding the distance restraint.

Figure 5

The cross‐linking information for chaperonin proteins with ATP‐dependent conformation cycles. A) Mitochondrial chaperonin complex. Cross‐links were mapped on the symmetrical football complex (PDB: 4PJ1, left structure) containing HSP60 (shown in palegreen) and HSP10 (shown in palecyan). Cross‐links exceeding distance restraint were mapped on the single ring HSP60 complex (PDB: 7AZP shown in palegreen, right structure). B) Human stress‐70 protein. E.coli HSP70 structure (PDB: 2KHO shown in gray, left structure) was used as a template while positioning SBD (PDB: 3N8E shown in palegreen) and NBD (PDB: 6P2U shown in palecyan) of human stress‐70 protein. Cross‐links were mapped on the constructed protein structure. HDOCK was used to dock SBD and NBD domains by the distance restraints at the domain interface, yielding a domain arrangement (shown in lightpink, right structure). Cross‐links at the domain interface were mapped on HDOCK structure. C) Heat shock 70 kDa protein 1A. Cross‐links were mapped on the inactive “client‐loading” complex (PDB:7KW7, left structure) containing HSP70 (shown in palecyan) and HSP90 (shown in palegreen). HDOCK was used to dock SBD (PDB: 4PO2 shown in bluewhite) and NBD (PDB: 7Q4R shown in lightpink) domains of HSP70 by the distance restraints at the domain interface. Cross‐links at the domain interface were mapped on HDOCK structure. Blue lines represented the cross‐links satisfying the distance restraint, and red lines represented the cross‐links exceeding the distance restraint.