Mitochondrial organelles — the energy powerhouses of the cell — must divide and fuse dynamically to function. It emerges that two distinct dynamin enzymes enable mitochondrial division. See Letter p.139
To paraphrase the biologist François Jacob, the dream of every mitochondrion — the cell's energy-producing organelle — is to become two mitochondria. On page 139, Lee et al.1 report that two different dynamin proteins act sequentially in the process of membrane division that divides a mitochondrion into two. One aids the initial mitochondrial-membrane constriction, and the other enables the final stages of lipid constriction necessary for membrane fission. Mitochondria have an essential role in cellular function, and gaining a better understanding of how they divide might shed light on human diseases caused by their dysfunction.
Mitochondria exist as a highly dynamic, interconnected network of organelle structures known as a reticulum, which undergoes repeated rounds of fusion and division2. These remodelling events distribute mitochondria throughout the cell, aid the maintenance of mitochondrial numbers and metabolic capacity, and ensure that mitochondrial genomes are efficiently dispersed throughout the reticulum1.
Cell biologists have long sought to understand how membrane-bound organelles divide, and how they pinch off membranous vesicles to transport proteins and nutrients throughout the cell. Research in this area has often focused on the dynamin family of proteins, GTPase enzymes that assemble in a collar-like structure around the constricting lipid 'necks' of budding membrane-bound vesicles or tubular organelle structures2,3. The hydrolysis of the molecule GTP provides energy that drives changes in dynamin structure; these conformational changes squeeze dynamin-encircled lipid membranes together until membrane division occurs, through a process known as fission3.
Some dynamins2 can constrict vesicles with diameters of 100 nanometres down to the approximately 10-nm diameter required for membrane fission. However, mitochondria have diameters ranging from 500 to 1,000 nm (ref. 2), and two lipid membranes. This mitochondrial scale means that substantially greater remodelling is needed for mitochondrial fission than for the membrane fission of vesicles.
The sites on the mitochondrial-membrane surface at which division will ultimately occur are initially marked by contact with the endoplasmic reticulum (ER), another cellular organelle2. These contacts are thought to initiate mitochondrial-membrane constriction by polymerizing actin-protein filaments, which provides force for remodelling of the mitochondrial tubule2. The mitochondrial dynamin-related protein Drp1 is recruited to these pre-constricted contact sites and assembles in helical and ring-like structures on the membrane surface2. However, whether Drp1 assembly and enzyme activity constrict the mitochondrial tubule to the point of fission was unknown.
Lee et al. noticed that cells lacking dynamin 2 had an excess of fused mitochondria, which provided a hint that dynamin 2 might have a role in mitochondrial division. Using high-resolution light microscopy, the authors monitored dynamin 2 at sites of mitochondrial division. Simultaneously observing dynamin 2 and Drp1 revealed how these proteins act to fully constrict a mitochondrial tubule. Using electron microscopy, Lee and colleagues measured the size of mitochondrial constrictions that accumulated in the absence of dynamin 2. This revealed that Drp1 mediates mitochondrial membrane constriction to a diameter of approximately 100 nm, and dynamin 2 then completes the membrane-tubule constriction to the point of fission (Fig. 1).
This finding was a surprise because dynamin 2 had previously been associated mainly with the fission of vesicles that mediate endocytosis, the process of uptake of extracellular material3. Another surprise was the sequential use of two different dynamins at separate stages of organelle-membrane constriction, which hadn't been observed before.
Drp1 and dynamin 2 share some evolutionarily conserved features with the rest of the dynamin family, particularly in the structural domains that drive assembly of these proteins around a constriction. Each dynamin also binds specific proteins and lipids, which determine where it functions in the cell. The recruitment of Drp1 to the mitochondria occurs through binding to mitochondrial receptor proteins4, including Mff, MiD49 and MiD51. This recruitment is tightly controlled through modifications to Drp1 that alter its conformation and activity, providing a link from cell signalling and metabolism to mitochondrial dynamics2.
A central question raised by Lee and colleagues' work is how dynamin 2 is recruited to mitochondria. This answer might already be close at hand, because mitochondrial isoforms of key dynamin 2 binding proteins are already known. These include the BAR-domain protein endophilin B1 (ref. 5) and the phosphatase enzyme synaptojanin 2A (ref. 6), which targets the lipid phosphoinositide. However, the function of these proteins in mitochondrial division has not been well established. Another protein worth investigating is the dynamin 2 binding partner sorting nexin 9 (SNX9), which regulates formation of vesicles from mitochondria7.
A key area for future work will be determining how the sequential constriction of mitochondria proceeds through the three stages described by Lee and colleagues: from the initial ER-induced constrictions to the Drp1 assembly that constricts the diameter of the mitochondrion, and finally the regulated recruitment of dynamin 2 to the Drp1-constricted site that is required for membrane division. These are separable events, which should allow molecular dissection of each step individually.
For example, microscopy by Lee and colleagues revealed that, in the absence of dynamin 2, Drp1 was recruited to a constricted mitochondrial tubule, giving rise to a 'frustrated' division event in which a narrow mitochondrial tubule covered with Drp1 pulled and contracted, but did not divide. Earlier work8 showed that loss of Drp1 led to ER-mediated partial constrictions, creating mitochondrial structures similar to beads on a string. Studying such intermediate states will help researchers to understand how each of the two dynamins is recruited and assembled to divide mitochondria.
Lee and colleagues' discovery means that mitochondria should be taken into consideration when interpreting observations of dynamin inhibition by genetic or pharmacological means. For example, mutations in dynamin 2 are linked to human disorders such as Charcot–Marie–Tooth disease and centronuclear myopathy9. Although research has focused on errors in endocytosis and vesicle transport to explain these diseases9, alterations in mitochondrial dynamics might contribute to the pathology. Similar neurodegenerative characteristics are also caused by inherited mutations in mitochondrial proteins, including Drp1 (ref. 10). By demonstrating that dynamin 2 has functions in mitochondrial division, Lee and colleagues open up avenues of investigation into how mitochondrial dynamics affect disease.
About this article
For a related paper on mitochondrial dynamics, see page 74.