Mitochondria are double membrane-bound organelles present in most eukaryotic cells. They are traditionally considered to function as the powerhouses of the cell, generating adenosine triphosphate (ATP) through oxidative phosphorylation, used for cellular chemical energy. Mitochondria have their own genome, known as mitochondrial DNA (mtDNA), which in mammalian cells contains 37 genes that are essential for normal mitochondrial function. The mitochondrial genome encodes 2 rRNAs, 22 tRNAs, and 13 protein subunits that are essential for the oxidative phosphorylation process. A reason why these particular genes are found in the mitochondrial genome is that the encoded proteins are highly hydrophobic and thus would have a problem entering the mitochondria (1–3). Besides these proteins, however, most (99%) of mitochondrial proteins in mammalian cells are encoded by nuclear genes, synthesized in the cytoplasm, and imported into mitochondria (4–7). In addition to producing ATP, mitochondria are involved in regulating numerous cellular activities, including cell-cycle progression and cell proliferation, the maintenance and differentiation of cells, mitophagy, and autophagy, the mitochondria-mediated intrinsic cell-death pathway and transduction of calcium signaling. Additionally, mitochondria are the major organelles generating and detoxifying reactive oxygen species (ROS) to adjust cellular redox homeostasis (8).

Mitochondria are highly dynamic and constantly alter their shape through fusion and fission events. The balance of fission and fusion events is termed mitochondrial dynamics, which is controlled by a number of mitochondria-shaping proteins encoded by genes in the nuclear genome. In addition to mitochondrial morphology, mitochondrial dynamics regulates the size, number, distribution, quality control, and transport of mitochondria in cells. Activated fission and/or inhibited fusion can lead to fragmented mitochondria, whereas active fusion and/or inhibited fission conversely can cause mitochondrial elongation (9–11). Over the last decades, emerging evidence indicates that deregulation of mitochondrial dynamics causes mitochondrial dysfunction, impacting on a broad range of cellular functions, and associated with a number of human diseases (11–17).

In this review, we discuss how mitochondrial fusion and fission is regulated by the mitochondria-shaping proteins, and by other co-factors such as the endoplasmic reticulum (ER) and the actin cytoskeleton, as well as how mitochondrial dynamics influences a variety of cellular biological processes such as apoptosis and mitophagy.

An increasing number of studies have suggested that Mff and MIEFs play a significant role in recruitment of the cytosolic Drp1 to mitochondria in mammals, but the action of MIEFs seems to be more complicated than that of Mff in regulating mitochondrial fission and fusion. In general, it is observed that Mff overexpression leads to the extensive recruitment of Drp1 on mitochondria promoting mitochondrial fission, whereas depletion of Mff reduces Drp1 on mitochondria resulting in mitochondrial elongation. In contrast, the mechanisms by which MIEFs can regulate mitochondrial morphology remain puzzling with discordant results. For example, in 293T and HeLa cells, expression of exogenous MIEF1 or MIEF2 induces mitochondrial elongation. Both MIEF1 and MIEF2 can interact with and recruit Drp1 to mitochondria and distribute in a punctate manner on the mitochondrial surface (91, 114). In Cos-7 monkey cells, expression of exogenous MIEF1 or MIEF2 also leads to mitochondrial elongation and accumulation of Drp1 on mitochondria, however, at low levels of the MIEF protein, mitochondrial morphology is relatively normal (92). Further studies show that long-term expression of MIEF1 results in a variety of morphological changes of mitochondria from fragmentation to a network state, suggesting that low levels of exogenous MiD51/MIEF1 can increase fission events, whereas at high expression levels, fission events are inhibited, resulting in mitochondrial fusion (120). On the other hand, knockdown of MIEF1 and MIEF2 by siRNA leads to inconsistent effects on mitochondrial morphology. It was reported that in Cos-7 cells, knockdown of both MIEF genes is required to cause a mitochondrial fusion phenotype (92). In line with this, knockdown of both MIEFs or either of the two genes similarly induces mitochondrial elongation in MEFs (86). However, it was also reported that in HeLa cells, down-regulation of MIEF1 by siRNA resulted in mitochondrial fission (91), while complete knockout of either MIEF alone, or double-knockout of both MIEF1/2 in HeLa cells caused mitochondrial elongation (121). The reasons for these inconsistent results can be that the different cell types used for experiments might have different endogenous levels of the fission/fusion proteins. Furthermore, it has been experimentally shown that introduction of low/intermediate or high levels of MIEF receptor proteins in cells can lead to opposite effects as described in detail below, possibly due to competitive Drp1 binding to the different receptor populations available (122).

It should be stressed, however, that the four Drp1 receptors Mff, MIEF1, MIEF2, and hFis1 are usually expressed simultaneously in mammalian cells. Moreover, each receptor is able to independently interact with and recruit Drp1 to mitochondria. This seems to imply a redundant mode of action for the Drp1 receptors in mammals, but these receptors appear to have an additive effect on Drp1 accumulation on mitochondria. However, whether these receptors work coordinately in the process of Drp1 recruitment is largely unclear. A recent study by Yu et al. reports that MIEFs can interact with both Drp1 and Mff and act as adaptors linking Drp1 and Mff in a trimeric protein complex, thereby facilitating a direct interaction between Mff and Drp1 (122). Ablation of both MIEFs largely reduces the association of endogenous Mff with Drp1, and also greatly impairs the Drp1 recruitment and accumulation on mitochondria mediated by Mff overexpression (122). This indicates that endogenous MIEFs regulate Mff-mediated mitochondrial recruitment of Drp1, although Mff also can serve as an independent receptor for Drp1. In contrast, depletion of endogenous Mff does not affect the association of MIEFs with Drp1 and neither has any evident impact on MIEF overexpression-induced recruitment and accumulation of Drp1 on mitochondria, indicating that Drp1 recruitment by MIEFs occurs independently of Mff (122). However, increased levels of MIEFs can decrease the association between Drp1 and Mff, and conversely, elevated expression of Mff reduces the interaction of MIEFs with Drp1. These results suggest that Mff and MIEFs compete for the binding to Drp1. Based on this, we have proposed a working model to illustrate how MIEFs and Mff coordinately balance mitochondrial dynamics (Figure 3). When higher levels of MIEFs exist in cells, Drp1 is sequestered in Drp1-MIEF-Mff and/or Drp1-MIEF complexes, inhibiting a direct Drp1-Mff binding and resulting in mitochondrial fusion (Figure 3A). When MIEFs are present at moderate levels in cells, Mff can receive sufficient Drp1 via MIEFs acting as a molecular bridge facilitating a direct interaction between Mff and Drp1, thereby maintaining normal and balanced mitochondrial morphology (Figure 3B). However, in the absence of MIEFs, Mff cannot directly capture sufficient Drp1 from the cytosol by itself, thus the balance shifts to mitochondrial fusion, resulting in mitochondrial elongation (Figure 3C).

In addition to the interplay between MIEFs and Mff regulating Drp1 recruitment as mentioned above, accumulating evidence suggests that MIEFs play a role in regulating hFis1-induced mitochondrial fragmentation (91, 114). It is reported that MIEFs exhibit a robust interaction with hFis1, in a manner independent of the interaction between MIEFs and Drp1. Thus, elevated levels of MIEFs partially reverse hFis1-induced mitochondrial fragmentation, whereas increased levels of hFis1 reduce the interaction between Drp1 and MIEF1 or MIEF2, respectively, and also partially reverse MIEF overexpression-induced mitochondrial elongation (91, 114). However, it is still largely unclear how MIEFs and hFis1 can regulate mitochondrial morphology via their mutual interactions.

Overall, the recruitment of cytosolic Drp1 to mitochondria is a key step in Drp1-mediated fission. Due to the simultaneous existence of four Drp1 receptors on mitochondria, the regulation of Drp1 recruitment and subsequent fission induction appears to be subject to a much more complex regulation in mammals than in yeast. Therefore, it will be an important topic for future research to further explore the molecular mechanisms mediating the interplay between different Drp1 receptors, in order to better understand their functions in the regulation of mitochondrial dynamics in mammals.

Mitochondrial morphology is dynamically regulated through the two counteracting apparatuses on mitochondria, i.e., the fission and fusion machineries. It is generally believed that the pro-fission proteins (such as Drp1, Fis1, Mff, and MIEFs) and the pro-fusion proteins (such as Mfn1, Mfn2, and OPA1) separately work in the fission and fusion machineries. However, emerging evidence suggests a crosstalk between the fission and fusion machineries. For example, it was reported that Drp1 colocalizes with Mfn2 and Bax at mitochondrial fission sites during apoptosis (123). Interestingly, another study suggests that Drp1 can interact with Mfn2 on the mitochondrial surface and overexpression of Drp1 facilitates mitochondrial tethering and fusion in Mfn2- or Mfn1-deficient cells, indicating that Drp1 participates in mitochondrial fusion in addition to acting as a pro-fission protein (124). Anand et al. provide additional evidence for a crosstalk between the fission and fusion machineries through studying the actions of the long and short forms of OPA1 in mitochondrial dynamics. The short forms (produced by processing) of OPA1 are found to trigger mitochondrial fragmentation and colocalize with the fission machinery, suggesting that the processing of OPA1 can regulate the balance of mitochondrial fission and fusion (61).

It was recently observed that hFis1 overexpression can induce extensive mitochondrial fragmentation even in the absence of Drp1 or/and Dyn2 (88). This indicates that Drp1/Dyn2 is largely dispensable for hFis1-induced mitochondrial fragmentation, i.e., the existence of Fis1-dependent but Drp1/Dyn2-independent mitochondrial fission pathways in mammals. hFis1 was found to robustly interact with the pro-fusion GTPases Mfn1, Mfn2, and OPA1 at endogenous levels, which supports the idea that hFis1 regulates mitochondrial dynamics via affecting the fusion machinery. In line with this notion, hFis1 reduces the GTPase activity of Mfn1, Mfn2, and OPA1 in vitro. Overexpression of hFis1 reduces mitochondrial fusion, whereas knockdown of hFis1 enhances exchange of the matrix content between mitochondria regardless of whether Drp1 is present or absent in cells. Disruption of the fusion machinery by CRISPR-Cas9 genome editing-based knockout technology phenocopied the hFis1 overexpression-induced mitochondrial fragmentation phenotype. Overall, several lines of evidence suggested that hFis1 can induce mitochondrial fragmentation by inhibiting the activity of the mitochondrial fusion machinery (88). This work emphasizes the importance of hFis1 in bidirectional regulation of the mitochondrial fission and fusion machineries and highlights that hFis1, known as a pro-fission factor, is not limited to promoting mitochondrial fission; it can also actively inhibit mitochondrial fusion (Figure 4).

Collectively, all these data provide evidence for a functional crosstalk between the mitochondrial fission and fusion machineries, and suggest that proteins known as components of the fission machinery may act concurrently in the fusion machinery, and likewise, fusion proteins may also affect the fission machinery. While many details remain largely unknown, the bidirectional regulation of mitochondrial morphology through functional interactions between fission and fusion proteins represents an emerging field of research in mammalian mitochondrial dynamics. Supporting this notion, a recent study reveals that the fission and fusion machineries are present at the same ER-mitochondria membrane contact sites to regulate mitochondrial morphology (125).

The endoplasmic reticulum (ER) communicates with mitochondria through mitochondria-associated ER membranes (MAMs) in yeast and mammals, and these connections have been shown to participate in different physiological functions, such as phospholipid synthesis, Ca²⁺-mediated signal transduction, protein import, mitochondrial distribution and mitophagy (126–128). It is reported that some of the mitochondria-shaping proteins are involved in this ER-MAM intercommunication. The mitochondrial profusion protein Mfn2 is localized on both mitochondria and the ER, and Mfn2 on the ER associates with Mfn1/2 on mitochondria and tethers the ER and mitochondria together to maintain efficient mitochondrial Ca²⁺ uptake, which is essential for ATP production (47).

The ER also plays an active role in mitochondrial division. ER tubules were observed to wrap around mitochondria, mark the prospective sites of mitochondrial division and reduce the mitochondrial diameter by about 30% before Drp1 recruitment. The ER tubules mainly occur at positions of Drp1 and Mff foci on mitochondria. In fact, ER tubules also mark positions of mitochondrial constriction in the absence of Mff or Drp1 (129). In addition to Drp1 and Mff, two other receptors of Drp1, i.e., MIEF1 and MIEF2, are also observed at mitochondria-ER contact sites, and co-localized with other fission proteins, such as Drp1 and Mff. However, <40% of observed mitochondria-ER contacts at MIEF foci are constriction sites, implicating that MIEFs are not the essential factors to determine ER-mitochondria constriction sites (120). Furthermore, MIEFs require the presence of Drp1 to form foci, whereas Mff can form foci in cells lacking Drp1 (118, 129). Another known Drp1 receptor, Fis1, has been shown in C. elegans to enter into a complex containing Drp1, Mff, and ER proteins at the ER-mitochondrial interface during stimulation of mitophagy (112). Strikingly, it is recently reported that endoplasmic reticulum–associated degradation (ERAD), an ER quality control mechanism, regulates mitochondrial dynamics (130). Additionally, in human cells, a subset of ER-mitochondria contacts spatially couple mtDNA replication with downstream mitochondrial division (131). However, it is still unclear to what extent ER tubules are essential for mitochondrial fission and fusion.

Besides the ER, the actin cytoskeleton is also involved in regulating mitochondrial fission (40, 132). The cycling of actin assembly and disassembly around mitochondrial subpopulations efficiently promotes local Drp1-dependent mitochondrial fission. However, inhibiting Drp1 activity by transfection of Drp1-K38A does not affect actin cycling onto mitochondrial subpopulations as observed through live-cell imaging (133). Furthermore, some actin-related proteins play important roles in mitochondrial fission. An ER-localized actin regulator, INF2 (inverted formin 2) induces actin filaments to drive the initial mitochondrial constriction and then promotes Drp1 recruitment to ER-mitochondria constriction sites (134). An isoform of the actin-nucleating protein Spire1C, localized to the MOM, interacts with INF2, promoting actin assembly at the mitochondrial surface. Disturbing Spire1C or its formin-binding activities can affect mitochondrial constriction and further mitochondrial fission (135). Myosin II plays similar roles in mitochondrial fission, it is enriched at mitochondrial constriction sites, and deletion of Myosin II reduces Drp1 accumulation on mitochondria (136). Interestingly, silencing of the cytoskeletal component septin 2 (Sept2) also decreases the association of Drp1 with mitochondria and increases the average distance between Drp1 clusters (137). In addition, knockdown of the actin-regulating factors cortactin, cofilin, or the Arp2/3 complexes results in elongated mitochondria and disassembles mitochondrial F-actin around mitochondrial subpopulations (133, 138).

The proposed models of mitochondrial fission in mammalian cells are summarized in Figure 2B. Mitochondria are pre-constricted by the ER, together with actin filaments that associate with mitochondria by the action of Spire1C and with the ER by INF2. At the ER constriction site, Drp1 is recruited by its receptors to mitochondria and assembled to form higher-order oligomers around the mitochondrial surface. Then mitochondria are further constricted by Drp1 oligomers, and Dyn2 is recruited to the constricted site instantly to finalize the scission of mitochondria through GTP hydrolysis. In summary, it is a great challenge to understand how the ER, actin cytoskeleton and Drp1 receptors interplay and coordinate their actions to mediate mitochondrial division.

In addition to the roles of the ER and actin cytoskeleton in regulating mitochondrial dynamics, mitochondrial membrane phospholipids also communicate with the fusion/fission machinery. Phospholipids are the fundamental building blocks of mitochondrial membrane bilayers. Emerging evidence indicates that at least two types of phospholipids, phosphatidic acid (PA) and cardiolipin (CL), play important roles in mitochondrial dynamics. PA is transported mainly from the ER to the mitochondrial inner membrane, where PA can be converted to CL via multiple enzymatic reactions (139). Some of the CL synthesized in the inner membrane can also be transported to the outer membrane. It has been reported that the pro-fission GTPase Drp1 on the outer membrane directly interacts with CL, and the B insert (also called the variable domain) of Drp1 is responsible for its interaction with CL (140, 141). This interaction between Drp1 and CL stimulates both oligomerization and GTPase activity of Drp1, enhancing mitochondrial division (82, 142, 143). In the outer membrane, CL can be converted back to PA by the MOM-localized phospholipase D (MitoPLD) (144). Drp1 also interacts with PA in the outer membrane, which restrains Drp1 activity, leading to mitochondrial fusion (142). Along these lines, MitoPLD interacts with Drp1, Mfn1, and OPA1, inhibiting Drp1-mediated fission and promoting mitofusin-dependent fusion (142, 144). There is evidence showing that CL located in the inner membrane is directly involved in OPA1-mediated fusion (145), while PA that is generated through the cleavage of CL by MitoPLD in the outer membrane participates in Mfn-mediated fusion (144). Moreover, overexpression of MitoPLD induces mitochondrial fusion, while its knockdown inhibits fusion resulting in mitochondrial fragmentation (144). In line with this, overexpression of lipin 1b [a phosphatase that dephosphorylates PA to form diacylglycerol (DAG)] results in mitochondrial fragmentation (142, 146). Similarly, overexpression of PA-PLA₁ (PA-preferring phospholipase A1), which hydrolyzes PA to produce lysoPA, triggers mitochondrial fragmentation, whereas knockdown of PA-PLA₁ induces elongation (147). CerS6-derived sphingolipids interact with Mff and promote mitochondrial fragmentation (148).

Overall, it has become increasingly clear that the interactions between the fusion/fission proteins and the phospholipids in mitochondrial membranes add another layer of complexity to the regulation of mitochondrial fusion-fission dynamics and contributes significantly to our understanding of this process (149–151).

Accumulating data suggest that apart from the canonical Drp1-dependent mitochondrial fission, there may be additional Drp1-independent mechanisms that contribute to mitochondrial division in mammalian cells, as partly described above (Figure 5). Several previous studies show that Fis1 is involved in such a Drp1-independent mitochondrial fission pathway. Onoue et al. reports that Fis1 interacts with TBC1D15 (a GTPase activating protein for the small GTPases Rab7 and Rab11) and recruits the cytosolic TBC1D15 to mitochondria. Fis1, together with TBC1D15, regulates mitochondrial morphology independently of Drp1 (152). This finding is further supported by a recent study showing that Fis1 mediates crosstalk between mitochondria and lysosomes via recruiting TBC1D15 to the mitochondrial surface, and in turn TBC1D15 interacts with lysosomal GTP-bound Rab7 at mitochondria-lysosome contact sites to drive hydrolysis of Rab7-GTP, thereby regulating both mitochondrial and lysosomal dynamics independently of Drp1 (87, 153). Although these findings that the Fis1-TBC1D15-Rab7 interaction involves mitochondria-lysosome contact sites can largely promote our understanding of how Fis1 functions in regulating mitochondrial dynamics, it still seems difficult to fully understand the mechanisms by which Fis1 overexpression induces extensive mitochondrial fragmentation.

In a more recent study mentioned above, hFis1 overexpression was reported to induce extensive mitochondrial fragmentation regardless of the presence or absence of Drp1 (88), further suggesting that Drp1 is dispensable for hFis1-mediated mitochondrial fragmentation. Further analysis showed that hFis1 binds to and inhibits the pro-fusion GTPases Mfn1/2 and OPA1, thereby reducing the activity of the fusion machinery in Drp1-deficient cells, shifting mitochondrial morphology to a fission phenotype (88). Given the importance of the Fis1-TBC1D15-Rab7 involvement in a Drp1-independent mitochondrial fission pathway, ablation of TBC1D15 would be expected to rescue the hFis1 overexpression-induced fragmentation phenotype, but this still remains to be explored. Thus, the underlying mechanisms by which hFis1 can induce mitochondrial fragmentation in the absence of endogenous Drp1 remains to be further explored.

The ER and the actin cytoskeleton have been reported to be involved in mitochondrial dynamics in cooperation with Drp1 (129, 132, 134), but whether the ER and actin cytoskeleton play a role in Drp1-independent fission is unknown. Emerging evidence, however, suggests that F-actin depolymerization significantly prevents hFis1 overexpression-induced mitochondrial fragmentation in a Drp1-independent manner, implying that the actin cytoskeleton plays a Drp1-independent role in regulating mitochondrial fission (88). The ER is generally considered to be the first step of Drp1-mediated mitochondrial fission, promoting mitochondrial constriction before Drp1 recruitment (129), but whether the ER induces mitochondrial scission directly, independently of Drp1 needs to be further explored.

Dyn2 has been reported to work at the final step of Drp1-dependent mitochondrial scission (94), but another study suggests that Drp1-mediated mitochondrial division still occurs in Dyn2-deficient cells (95), indicating Dyn2 is not essential for Drp1-dependent fission. However, a recent report presents that Dyn2 knockdown partially reversed the mitochondrial fission phenotype that was induced when both fission and fusion machineries had simultaneously been destroyed by double knockout of Drp1 and OPA1 through CRISPR/Cas9 technology (88), indicating that Dyn2 may act via a separate mechanism to promote mitochondrial fission when Drp1-mediated fission is blocked.

Collectively, it is becoming increasingly clear that there may be multiple Drp1-independent mechanisms regulating mitochondrial division in mammalian cells. Although substantial progress has been achieved through studies of Drp1-mediated mitochondrial fission, little is known about the regulation of mitochondrial division in mammalian cells lacking Drp1. Undoubtedly, elucidating the underlying mechanisms that drive Drp1-independent mitochondrial division is of fundamental importance to fully understand mammalian mitochondrial dynamics.

Accumulating evidence indicates that the carefully orchestrated balance of mitochondrial fusion and fission is critical for maintaining a healthy population of mitochondria. Fission allows the exclusion of damaged mitochondria via mitophagy and mitochondrial biogenesis, whereas fusion allows mitochondria to exchange their substance including proteins and mtDNA among individual mitochondria (154). Abnormal mitochondrial dynamics is directly associated with mitochondrial dysfunction, which impacts on a wide range of cellular processes, such as mitochondrial transport and biogenesis, cell-cycle regulation, cell proliferation and differentiation, energy metabolism, ROS production, Ca²⁺ signaling, mtDNA maintenance, mitochondrial quality control via autophagy (mitophagy), and ultimately programmed cell death (155–160). Moreover, some mitochondrial fission and fusion proteins are known to be critical for embryonic development in mice (45, 161–163), further emphasizing that functions of mitochondrial dynamics go beyond the appearance of mitochondria and have important physiological consequences. In this review, we will focus on the potential roles of mitochondrial dynamics in apoptosis and mitophagy.

In the past 20 years, considerable progress has been made in terms of identifying core components and regulatory factors of the mitochondrial fission/fusion machineries in mammals, and huge efforts have been made to elucidate the potential molecular mechanisms that control and regulate mitochondrial fission and fusion events. One outcome of these efforts is a shift from the view that the mitochondrial fission and fusion machineries are evolutionarily conserved from yeast to human, to an understanding that these processes are much more complicated in mammals than in single-cell yeast. So far, at least four Drp1 receptors (Fis1, Mff, MIEF1, and MIEF2) in mammalian cells are responsible for recruitment of Drp1 to mitochondria with an additive effect, although they are not functionally redundant. Increasing evidence indicates that these receptors also coordinately work in Drp1-mediated mitochondrial fission. Thus, further research is warranted to elucidate the interplay between these Drp1 receptors in health and disease. Although the pro-fission and pro-fusion proteins are thought to regulate mitochondrial fission and fusion events separately, emerging data suggest that there is also extensive crosstalk between the fission and fusion machineries. It is also becoming increasingly clear that there are additional Drp1-independent mechanisms that drive mitochondrial division in mammalian cells. Finally, research into the mechanisms of apoptosis and mitophagy has excitingly placed the mitochondrial fission and fusion machineries at the center of these physiological processes. Elucidation of the underlying molecular mechanisms involved in these processes will be critical for better understanding of mammalian mitochondrial dynamics.

RY and JZ conceptualized and wrote the review with contributions from MN and UL. RY and JZ made the figures. All authors discussed and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.