Mitochondrial fission – changing perspectives for future progress

ABSTRACT

Mitochondrial fission is important for many aspects of cellular homeostasis, including mitochondrial distribution, stress response, mitophagy, mitochondrially derived vesicle production and metabolic regulation. Several decades of research has revealed much about fission, including identification of a key division protein – the dynamin Drp1 (also known as DNM1L) – receptors for Drp1 on the outer mitochondrial membrane (OMM), including Mff, MiD49 and MiD51 (also known as MIEF2 and MIEF1, respectively) and Fis1, and important Drp1 regulators, including post-translational modifications, actin filaments and the phospholipid cardiolipin. In addition, it is now appreciated that other organelles, including the endoplasmic reticulum, lysosomes and Golgi-derived vesicles, can participate in mitochondrial fission. However, a more holistic understanding of the process is lacking. In this Review, we address three questions that highlight knowledge gaps. First, how do we quantify mitochondrial fission? Second, how does the inner mitochondrial membrane (IMM) divide? Third, how many ‘types’ of fission exist? We also introduce a model that integrates multiple regulatory factors in mammalian mitochondrial fission. In this model, three possible pathways (cellular stimulation, metabolic switching or mitochondrial dysfunction) independently initiate Drp1 recruitment at the fission site, followed by a shared second step in which Mff mediates subsequent assembly of a contractile Drp1 ring. We conclude by discussing some perplexing issues in fission regulation, including the effects of Drp1 phosphorylation and the multiple Drp1 isoforms.

Summary: Mitochondrial fission occurs in multiple contexts, mediating a variety of possible responses. This Review discusses how individual aspects of mitochondrial fission might be integrated to give a more holistic understanding.

Introduction

Mitochondria are dynamic organelles, which are capable of both fission and fusion to change their size in a variety of contexts, including during cell growth and division, in response to mitochondrial damage, and to affect changes in metabolism. The mechanisms behind fission and fusion have been the subject of considerable research for several decades, which has revealed many key factors controlling these processes. In this Review, focusing on mitochondrial fission, we argue that it is time for the field to evolve in several ways. First, instead of discussing mitochondrial fission as a single process, we consider the possibility that different ‘types’ of fission exist, which likely differ in initial stimuli and in regulatory mechanisms for key steps. Second, instead of thinking about specific fission factors in isolation, we address how these factors work together in a concerted process. Finally, instead of focusing on only the outer mitochondrial membrane (OMM) as the place where fission occurs, we consider how the intricately structured inner mitochondrial membrane (IMM) undergoes fission in a manner that does not disrupt function. Although we frequently reference the well-studied yeast mitochondrial fission system, which has provided much of the basis for our knowledge, we will mainly address mitochondrial fission in mammals, which possesses key distinctions. We first provide background on the major known players in mitochondrial fission. Then, we address three significant questions – how to quantify mitochondrial fission, how the IMM divides, and how many ‘types’ of fission might exist. Next, we present a model that integrates several known fission regulators. Finally, we discuss outstanding questions in the field.

Molecular and cellular players in mitochondrial fission

Many important players in mitochondrial fission have been identified from a variety of experimental systems. Here, we focus on two systems, mammals and budding yeast.

The dynamin family GTPase Drp1

A central player in mitochondrial fission, the dynamin-related protein 1 Drp1 (Dnm1 in yeast; the human gene is called DNM1L) is highly conserved throughout eukaryotes (Bui and Shaw, 2013) and also mediates peroxisomal division (Koch et al., 2003). Drp1 is a dimeric cytoplasmic protein that is recruited to mitochondria by ‘receptor’ proteins (Fröhlich et al., 2013; Hatch et al., 2016; Rochon et al., 2024). Recruited Drp1 ultimately assembles into a ring around the mitochondrion (Ji et al., 2015). Upon GTP hydrolysis, the ring constricts, leading to mitochondrial fission (Fig. 1A). One report suggests that dynamin-2 provides additional constriction after Drp1-mediated constriction (Lee et al., 2016), but this finding is not universally accepted (Fonseca et al., 2019; Kamerkar et al., 2018; Nagashima et al., 2020).

Fig. 1.

Regulatory factors in mitochondrial fission. (A) Drp1 dimers (red) are recruited to the OMM. OMM-bound Drp1 oligomers assemble into a ring (‘maturation’), which constricts upon GTP hydrolysis, leading to mitochondrial fission. (B) Left, bar diagram of human Drp1 isoform 1 (Rosdah et al., 2020), also called isoform 011 (Strack et al., 2013), including splice inserts 2 and 3 (insert positions are highlighted by arrows 1–3). The G domain, bundle signaling element (BSE), stalk and variable domain (VD), and two phosphorylation sites (S616 and S637) are shown. Right, ribbon diagram of Drp1 dimer, based on PDB: 4BEJ. One subunit is shown using the color scheme shown in the bar diagram; the other subunit is shown in gray. (C) Table displaying phosphorylation site numbers in the nine human Drp1 splice variants. (D) Timelapse imaging of Drp1 puncta (green) dynamics on a mitochondrion (red). Small Drp1 puncta (yellow arrowheads, with size of arrowhead representing size of puncta) assemble into a single punctum, concurrent with membrane constriction, possibly reflecting the maturation phase. Time is shown in seconds. Scale bar: 1 µm. Images reproduced from Ji et al. (2015), where it was published under an CC-BY 4.0 license. (E) The four Drp1 receptors identified to date, on the OMM. (F) Two types of actin affect mitochondria in different ways. Top, Ca²⁺-induced actin (CIA). Ca²⁺ activates the ER-bound formin INF2, which nucleates actin filaments. CIA has two effects: (1) it increases ER–mitochondrial contact, leading to Ca²⁺ transfer and IMM constriction, and (2) it recruits Drp1, leading to OMM constriction. These effects both promote fission. Bottom, acute damage-induced actin (ADA). Mitochondrial damage causes Arp2/3 complex activation around mitochondria. The resulting actin network assembly decreases ER–mitochondrial contact and IMM remodeling, leading to decreased Parkin recruitment. The middle expansion shows the relationship between ER and a mitochondrion in the absence of CIA or ADA. Panels A, B, E and F were created in BioRender by Kamerkar, S., 2025. https://BioRender.com/r59h123. These panels were sublicensed under CC-BY 4.0 terms.

In cell lines, Drp1 knockout or dominant-negative mutants lead to elongated mitochondria and peroxisomes (Smirnova et al., 1998, 2001). However, Drp1 knockout cells remain viable (Ishihara et al., 2009; Otsuga et al., 1998; Wakabayashi et al., 2009), suggesting that Drp1-independent mitochondrial fission mechanisms exist. Drp1-null mice die after embryonic day 12.5 due to brain hypoplasia (Ishihara et al., 2009; Wakabayashi et al., 2009), and multiple missense mutations in Drp1 are associated with a variety of muscular and neuronal disorders in humans (Fahrner et al., 2016; Robertson et al., 2023; Whitley et al., 2018; Verrigni et al., 2019).

Drp1 has a typical dynamin superfamily architecture comprised of three structural domains – a GTPase domain (G domain), a bundle signaling element (BSE) and a stalk region (Fröhlich et al., 2013) (Fig. 1B). Where other cytoplasmic dynamins also contain a phospholipid-binding pleckstrin homology (PH) domain, Drp1 instead has an unstructured ‘variable domain’ (VD) (Fig. 1B), which lacks a canonical lipid binding domain but is important for membrane interaction and regulation (Fröhlich et al., 2013; Lu et al., 2018; Mears et al., 2011). The stalk domain is essential for Drp1 dimerization. A short linear motif at the C-terminus also might play a role in Drp1 regulation (Pérez-Jover et al., 2024).

Purified Drp1 is dimeric at low concentration in the nucleotide-free state and assembles into larger oligomers with increasing Drp1 concentration (Fröhlich et al., 2013; Hatch et al., 2016; Macdonald et al., 2014). Dimeric Drp1 is auto-inhibited through interaction between the G domain and the BSE (Rochon et al., 2024). In the presence of GTP, purified Drp1 self-assembles into rings that can tubulate isolated mitochondria or model membranes (Basu et al., 2017; Yoon et al., 2001) and can even cause membrane division (Kamerkar et al., 2018). Structural studies show that the presence of the Drp1 receptor mitochondrial dynamics protein of 49 kDa (MiD49; also known as MIEF2) (Kalia et al., 2018) or lipid membrane (Peng et al., 2024 preprint) also enable G domain interaction and Drp1 oligomerization. Drp1 oligomerization is important for efficient GTP hydrolysis, in that the resulting interactions between G domains stimulate their enzymatic activity (Wenger et al., 2013).

In cells, Drp1 appears as puncta of varying sizes overlaying a uniform cytoplasmic background (Ji et al., 2015). These puncta likely represent oligomers, and formation of large Drp1 oligomers from smaller assemblies occurs in live cells (Fig. 1D). Almost all Drp1 puncta appear to be membrane associated, with 63–70% on mitochondria, ∼15% on peroxisomes and ∼15% on endoplasmic reticulum (ER) (Ji et al., 2015, 2017). Interestingly, ER-bound Drp1 might play an under-appreciated role in ER tubulation (Adachi et al., 2020; Ji et al., 2017). On mitochondria, most Drp1 puncta appear to be ‘non-productive’, in that they do not result in mitochondrial fission over a 10 min time course (Ji et al., 2015). These observations led to the hypothesis that there are two stages of Drp1 assembly on the mitochondrion – initial recruitment and oligomerization, followed by maturation of the recruited oligomer into a full ring (Fig. 1A).

In humans, alternative splicing gives rise to nine Drp1 isoforms that can have differential impacts on cellular function, through mechanisms such as association with microtubules (Javed et al., 2024; Rosdah et al., 2020; Strack et al., 2013). How Drp1 isoforms differentially interact with other molecules remains unclear. In addition, the exact number of alternately spliced exons and isoforms of human Drp1 is still unclear, as murine Drp1 has 10 isoforms (Itoh et al., 2018). One brain-specific Drp1 isoform appears to play a role in endocytosis in dendritic spines independently of mitochondrial fission (Itoh et al., 2019).

Drp1 receptors

A fundamental step in mitochondrial fission is recruitment of cytosolic Drp1 to the OMM. Four OMM proteins have been identified as Drp1 receptors – mitochondrial fission protein 1 (Fis1), mitochondrial fission factor (Mff), MiD49 and mitochondrial dynamics protein of 51 kDa (MiD51; also known as MIEF1) (Fig. 1E).

Fis1 is a tail-anchored protein found on mitochondria and peroxisomes. Unlike other Drp1 receptors, Fis1 is broadly expressed in eukaryotes, including in budding yeast, where Fis1 deletion perturbs mitochondrial fission and Dnm1 recruitment (Mozdy et al., 2000). Dnm1 binding by yeast Fis1 requires an adaptor protein, either mitochondrial division protein 1 (Mdv1) or CCR4-associated factor 4 (Caf4) (Griffin et al., 2005; Guo et al., 2012; Tieu and Nunnari, 2000). The function of Fis1 in mammals is more controversial, with initial reports suggesting a role in mitochondrial division but later studies finding little effect of Fis1 depletion in cultured cell lines (Losón et al., 2013; Osellame et al., 2016; Otera et al., 2010). In addition, mammals possess no obvious adaptor proteins that would mediate Fis1–Drp1 interactions. Alternative roles for mammalian Fis1 include targeting defective mitochondrial segments for mitophagy (Kleele et al., 2021; Wong et al., 2018; Yamano et al., 2014) or inhibiting mitochondrial fusion GTPases (Yu et al., 2019). Conversely, peptides that inhibit the Fis1–Drp1 interaction have been found to reduce lipopolysaccharide-induced mitochondrial fragmentation and dysfunction in cultured cardiomyocytes and to mitigate mitochondria damage in traumatic brain injury (Haileselassie et al., 2019; Rios et al., 2023; Sridharan et al., 2024). Although these effects could suggest a role for Fis1 in certain types of fission, more work is required to substantiate this role.

Mff, a tail-anchored protein found only in metazoans, is present on mitochondria, peroxisomes and ER (Gandre-Babbe and Van Der Bliek, 2008; Ji et al., 2017). Loss of Mff in multiple mammalian cell types leads to Drp1 recruitment defects and elongation of mitochondria and peroxisomes (Gandre-Babbe and Van Der Bliek, 2008; Losón et al., 2013; Otera et al., 2010), whereas Mff overexpression causes mitochondrial fragmentation (Otera et al., 2010). Mammalian Mff possesses nine alternately spliced variants (Gandre-Babbe and Van Der Bliek, 2008), all of which contain Drp1-binding N-terminal motifs, a coiled-coil domain (CC) and a transmembrane domain (Liu and Chan, 2015; Otera et al., 2010). The highly conserved CC mediates Mff oligomerization and is necessary for Mff function in mitochondrial fission (Liu et al., 2021). Mff appears as puncta on the OMM, suggesting that it undergoes oligomerization (Liu et al., 2021; Otera et al., 2010). The mechanism by which Mff–Drp1 interactions cause Drp1 activation remains unclear. Mff alone only weakly stimulates Drp1 unless the Drp1 VD is removed (Clinton et al., 2016; Koirala et al., 2013; Liu and Chan, 2015; Otera and Mihara, 2011). In cells, Drp1 does not coimmunoprecipitate with Mff unless chemically cross-linked prior to lysis, suggesting that the interaction is relatively low affinity (Clinton et al., 2016; Gandre-Babbe and Van Der Bliek, 2008; Otera et al., 2010; Strack and Cribbs, 2012). We will discuss possible answers to these puzzling results later in the Review.

MiD49 and MiD51 are type 1 membrane proteins that, unlike Mff and Fis1, are found only on mitochondria (Palmer et al., 2011; Zhao et al., 2011). Our own searches using public databases show clear MiD49 and MiD51 homologs only in chordates (i.e. mammals, birds, fish and amphibians) with no apparent homologs in insects, annelids, yeast, plants or protists, although more comprehensive phylogenetic analysis would be useful. Double knockdown or knockout of both MiD49 and MiD51 causes mitochondrial elongation to variable extents, and there are differing reports of redundancy between MiD49 and MiD51 (Losón et al., 2013; Osellame et al., 2016). Similar to Mff, both MiD49 and MiD51 form puncta on mitochondria, suggesting that they also undergo oligomerization (Otera et al., 2016; Palmer et al., 2011, 2013; Zhao et al., 2011). Interestingly, MiD51–Fis1 complexes can mediate inter-lysosomal tethering at lysosome–mitochondrial contact sites (Wong et al., 2022). Both MiD49 and MiD51 contain putative nucleotide-binding pockets similar to nucleotide transferase proteins like cyclic GMP-AMP synthase (cGAS) (Losón et al., 2014, 2015; Richter et al., 2014). Recent work has suggested that long chain fatty acyl-CoAs (LCACA) are ligands that stimulate oligomerization of and activate MiD49 and MiD51 (Liu et al., 2024b). In cells, MiD mutants with disrupted LCACA binding display fewer puncta and increased mitochondrial length, suggesting that MiD-mediated mitochondrial fission is stimulated by increased LCACA (Liu et al., 2024b). Paradoxically, overexpression of either MiD49 or MiD51 leads to mitochondrial elongation, despite increased mitochondrial Drp1 recruitment (Palmer et al., 2011; Zhao et al., 2011). In a later section, we propose a model that might explain this paradox.

Post-translational modifications

Drp1 is extensively post-translationally modified, including by phosphorylation, SUMOylation and S-nitrosylation (Chang and Blackstone, 2010; Xiao et al., 2022). Most of these post-translational modifications (PTMs) occur in the VD. Here, we focus on phosphorylation, the best-characterized Drp1 PTM. Drp1 is phosphorylated at multiple sites. The two best-studied sites, S616 and S637, are in the VD (Fig. 1B). We note that the amino acid numbers of these sites vary depending on the Drp1 splice variant and provide a table of phosphosite numbers for the nine human Drp1 splice variants (Fig. 1C), but will use the most common nomenclature (S616 and S637) here.

Drp1 S616 can be phosphorylated by cyclin dependent kinase 1 (Cdk1–cyclin B) (Taguchi et al., 2007), extracellular signal-regulated kinases 1 and 2 (ERK1/2; also known as MAPK3 and MAPK1, respectively) (Kashatus et al., 2015; Serasinghe et al., 2015; Yu et al., 2011), protein kinase C δ (PKCδ) (Qi et al., 2011), Rho-associated protein kinases (ROCK1 and/or ROCK2) (Brand et al., 2018), Ca^2+/calmodulin-dependent protein kinase II (CaMKII) (Tresse et al., 2021; Xu et al., 2016), Cdk5 (Guo et al., 2018; Strack et al., 2013) and PTEN-induced kinase 1 (PINK1) (Han et al., 2020). Dephosphorylation of S616 might occur through dual-specificity phosphatase 6 (DUSP6) (Ma et al., 2020). In mice, S616 phosphorylation is particularly high in the brain (Ikeda et al., 2024). Interestingly, mice with systemic knock-in of phospho-mimetic (S616D) or phospho-deficient (S616R) do not show systemic defects but do display either reduced (S616D) or increased (S616R) anxiety (Ikeda et al., 2024).

Drp1 S637 is phosphorylated by cyclic AMP (cAMP)-dependent protein kinase (PKA) (Chang and Blackstone, 2007; Cribbs and Strack, 2007), ROCK1 (Wang et al., 2012), Ca²⁺/calmodulin-dependent protein kinase Iα (CaMKIα) (Han et al., 2008), adenosine monophosphate-activated protein kinase (AMPK) (Wikstrom et al., 2013) and protein kinase D (PKD; also known as PRKD1) (Jhun et al., 2018), and is dephosphorylated by protein phosphatase 2A (PP2A) and calcineurin and phosphoglycerate mutase 5 (PGAM5) (Cereghetti et al., 2008; Cribbs and Strack, 2007; Pennanen et al., 2014; Yu et al., 2020).

S616 phosphorylation is widely accepted to increase Drp1-mediated mitochondrial fission, whereas S637 phosphorylation decreases fission. The association of S616 with increased fission is uniformly supported in the literature. In contrast, the effect of S637 phosphorylation is not consistent, with phosphorylation hampering mitochondrial fission in some cases (Cereghetti et al., 2008; Chang and Blackstone, 2007; Cribbs and Strack, 2007; Gomes et al., 2011; Wikstrom et al., 2013) but enhancing fission in others (Han et al., 2008; Jhun et al., 2018; Wang et al., 2012).

PTMs on Mff also contribute to mitochondrial fission regulation. Phosphorylation of Mff by AMPK, PKD and AKT stimulates fission in distinct contexts (Henschke et al., 2024; Pangou et al., 2021; Toyama et al., 2016; Virga et al., 2024). Recent evidence suggests that SUMOylation of Mff is stimulated by AMPK-mediated phosphorylation and induces a rearrangement of complexes between Mff, Drp1 and MiD proteins to enhance mitochondrial fission upon mitochondrial depolarization (Seager et al., 2024).

Actin filaments

Actin has been increasingly associated with mitochondrial dynamics in multiple ways (Fung et al., 2023). However, it is important to appreciate that there is not just one ‘type’ of actin with one specific downstream effect on mitochondria. Here, we focus on two types of actin with contrasting effects on mitochondria – Ca²⁺-induced actin polymerization (CIA), which stimulates mitochondrial fission, and acute damage-induced actin (ADA), which does not stimulate fission.

In CIA, an ER-bound splice variant of the formin INF2 nucleates actin filaments in response to elevated cytoplasmic Ca²⁺ (Fig. 1F) (Ji et al., 2015; Shao et al., 2015; Wales et al., 2016). These ER-associated actin filaments have two distinct effects on mitochondria, both of which stimulate fission. First, CIA promotes ER–mitochondrial contact, allowing increased mitochondrial Ca²⁺ uptake, which stimulates IMM constriction (Chakrabarti et al., 2018). Second, CIA promotes Drp1 recruitment to mitochondria, increasing OMM constriction (Ji et al., 2015; Korobova et al., 2013). Other actin-binding proteins playing roles in CIA-stimulated mitochondrial fission are SPIRE1C, myosin II and fascin (Kage et al., 2022; Korobova et al., 2014; Lin et al., 2019; Manor et al., 2015). One important question is how Drp1, once recruited by actin filaments, transfers to the OMM to drive mitochondrial division. We address this question later in the Review.

In contrast to CIA, ADA causes actin polymerization directly around mitochondria as a rapid response to mitochondrial dysfunction induced by mitochondrial depolarizers, electron transport chain inhibition, hypoxia or ATP synthase inhibition (Fung et al., 2019, 2022; Li et al., 2015). Similar actin networks are associated with chronic damage conditions, such as mitochondrial DNA (mtDNA) depletion and mutations to electron transport chain components (Chakrabarti et al., 2022). The actin nucleating proteins that initiate ADA (the Arp2/3 complex and FMNL formins) are distinct from those used in CIA. ADA does not result in increased mitochondrial fission – in fact, ADA suppresses mitochondrial rearrangements that occur within 20 min of mitochondrial depolarization (Fung et al., 2019, 2022). These rearrangements are not ‘fission’ but are a reorganization of the IMM, as discussed in a later section.

Intriguingly, one effect of ADA is to disrupt ER–mitochondrial contact, which is in direct contrast to the stimulatory effect of CIA on ER–mitochondrial contacts. The result of ADA-induced ER–mitochondrial contact disruption is to decrease mitochondrial recruitment of the E3 ligase Parkin, which in turn reduces mitophagy (Fung et al., 2024 preprint). ADA also acutely stimulates glycolysis (Chakrabarti et al., 2022). One hypothesis for these effects is that the transient ADA response temporarily inhibits large-scale cellular responses like mitophagy, allowing for potential mitochondrial recovery. In summary, there are multiple possible impacts of actin on mitochondrial fission and dynamics, and understanding the context in which they occur is crucial.

Phospholipids

Two phospholipids differentially regulate Drp1 – cardiolipin (CL) stimulates activity (Bustillo- Zabalbeitia et al., 2014; Macdonald et al., 2014) and phosphatidic acid (PA) inhibits activity (Adachi et al., 2016). Although CL is more enriched on the IMM than on the OMM (Horvath and Daum, 2013), local OMM CL content increases dramatically upon stimulation of mitophagy (Chao et al., 2019; Chu et al., 2013). Drp1 binds preferentially to CL-containing membranes of high curvature and readily severs lipid tubes of diameters up to 400 nm (Kamerkar et al., 2018). On the OMM, CL can be converted into PA by mitochondria-localized phospholipase D, with the balance between PA and CL levels influencing mitochondrial fission (Kameoka et al., 2018). The VD of Drp1 is likely the site of phospholipid binding (Adachi et al., 2020).

Organelles involved in mitochondrial fission

Finally, mitochondria communicate extensively with other organelles, and three types of interorganelle contacts have been found to impact mitochondrial fission. Fission events often take place at ER–mitochondria contact sites (ERMCs), and these interactions precede Drp1 recruitment (Friedman et al., 2011). Interestingly, mitochondrial fission is also preceded by mtDNA replication near the fission site (Lewis et al., 2016). ERMCs might provide several factors stimulatory to fission, including INF2, leading to actin assembly for Drp1 recruitment (Chakrabarti et al., 2018). ERMCs also facilitate Ca²⁺ transfer from ER to mitochondria, triggering IMM division prior to OMM division (Chakrabarti et al., 2018; Cho et al., 2017). Mitochondrial contact with lysosomes (Wong et al., 2018) or trans-Golgi network (TGN)-derived vesicles (Nagashima et al., 2020) can also stimulate steps in mitochondrial fission, occurring downstream of Drp1.

How do we quantify fission?

A long-standing problem in the study of mitochondrial fission is the difficulty in quantifying fission events. Even in well-spread mammalian cultured cells, fission quantification is complicated by several issues. Three common techniques are often used for quantifying mitochondria – fixed-cell fluorescence microscopy, electron microscopy (EM) and live-cell fluorescence microscopy (described in Box 1). Here, we outline five issues associated with methods for fission quantification; no current technique is infallible or free of all issues (Fig. 2A–E), but awareness of these potential pitfalls allows minimization of potential artifacts. In our opinion, live-cell imaging currently represents the best approach, but we acknowledge the drawbacks to this and other techniques in Box 1.

Box 1. Common methods of assessing fission – advantages and disadvantages.

In fixed-cell fluorescence microscopy, cells are typically fixed with aldehyde fixative, stained with a fluorescent mitochondrial marker, then imaged. Technical variations to this protocol include the fixative (e.g. paraformaldehyde or glutaraldehyde) and the mitochondrial marker and staining method (e.g. pre-fixation MitoTracker treatment or post-fixation application of antibodies for matrix, IMM or OMM proteins). Quantification methods fall into two categories. First, ‘holistic’ assessment of overall mitochondrial morphology assigns mitochondria into categories such as ‘fragmented’, ‘elongated’, ‘network’, etc. An advantage of this method is that judgements are not based on a limited region of interest. However, this qualitative assessment is subjective, especially when conducted at low resolution. Second, individual mitochondrial features (e.g. length, area and number) can be quantified, generally within a region of interest. This approach provides numerical values for analysis, but these measurements can be confounded by mitochondrial overlap and the possibility of regional mitochondrial heterogeneity (Fig. 2B,D).

In thin-section transmission EM, cells are fixed with glutaraldehyde, processed, embedded and sectioned. Thin sections (<100 nm) are imaged. An advantage of EM is high spatial resolution, allowing resolution of the IMM and OMM. A major disadvantage is uncertainty of whether the section has captured the entire mitochondrial length (Fig. 2C). We consider this a fatal flaw, but one which can be mitigated by electron tomography techniques.

In live-cell microscopy, a mitochondrial marker (e.g. MitoTracker or a transfected fluorescent protein) is introduced into live cells and the cells are imaged over a specified time period at a specified acquisition frequency. Movies are examined for fission events, and high-throughput analysis methods have been published (Theart et al., 2020). Advantages of live imaging are that it allows direct quantification of fission events rather than the inference of fission from mitochondrial size changes, and the ability to identify fission even if subsequent fusion occurs. Disadvantages include low-throughput capacity, as only a limited number of cells can be quantified. The imaging time range also tends to be limited, especially if a high imaging frequency (e.g. an image taken every 5 s) is used to capture transient fission events (‘kiss-and-run’). Live-cell imaging also carries the disadvantage of phototoxicity effects. In fact, imaging using 405 nm excitation can even cause transient mitochondrial depolarization (Abrisch et al., 2020).

Fig. 2.

Challenges with quantification of mitochondrial fission. (A) Uncertainties with fixed-cell imaging of mitochondria (shown in green). Top, mitochondrial shortening could reflect either increased fission or decreased fusion. Bottom, sequential fusion and fission (‘kiss-and-run’) of mitochondria (shown in brown) can result in no apparent change in mitochondrial fission. (B) Overlapping mitochondria. Even when imaged with confocal microscopy (maximal Z-resolution ∼700 nm), multiple mitochondria can overlap. Left, individual mitochondria are shown in various colors. Right, the acquired image can give the impression of more fusion or branching than is actually present. (C) Issues with EM-based quantification. Thin sections (∼60 nm thickness, gray line) might section mitochondria (∼500 nm diameter) several times, resulting in length misinterpretation. Here, the large green mitochondrion appears as four short mitochondria within the EM sectioning plane, whereas the small purple mitochondrion is accurately sectioned. (D) Mitochondrial heterogeneity. Heterogeneous mitochondria might be spatially segregated within a single cell, such that picking a single region of interest only gives part of the story. (E) Uncertainties with matrix markers. If only a matrix marker is used, overestimations of fission can occur if the matrix remodels in the absence of OMM fission. Full-cell views show matrix marker alone, whereas magnifications show matrix and OMM markers. CCCP is a mitochondrial depolarizing agent known to rapidly induce IMM remodeling (‘circularization’) in the absence of OMM fission. Created in BioRender by Kamerkar, S., 2025. https://BioRender.com/e85s383. These panels were sublicensed under CC-BY 4.0 terms. (F) Live-cell microscopy of mitochondria labeled with a matrix marker (red) and OMM marker (green), showing mitochondrial circularization upon CCCP treatment (arrow). Time is shown in seconds. Scale bar: 2.5 µm. Images reproduced from Fung et al. (2019).

First, with fixed-cell techniques (immunofluorescence microscopy or EM), fission events are not observed directly but are inferred by decreased mitochondrial length. However, decreased mitochondrial fusion can also result in shorter mitochondria. In addition, the existence of coupled fusion–fission (also called ‘kiss-and-run’) events means that fixed-cell techniques underestimate the fission rate (Twig et al., 2008) (Fig. 2A). Second, mitochondria often overlap when imaged by light microscopy, making it difficult to distinguish a single large or branched mitochondrion from several overlapping mitochondria. This issue is especially problematic close to the nucleus, where mitochondria often enrich (Fig. 2B). Third, with thin-section EM, it is rarely certain whether the section taken (generally <100 nm thick) cuts through the entire long axis of the mitochondrion (mitochondrial thickness is typically 300–500 nm). In fact, long mitochondria are unlikely to be sectioned from end to end. Thus, ‘length’ measurements are guesses and long mitochondria are underrepresented using this technique (Fig. 2C). The fourth issue arises from mitochondrial heterogeneity. It is clear that distinct mitochondrial populations can exist in the same cell (Benador et al., 2019; Graham et al., 2017; Han et al., 2023; Marlar-Pavey et al., 2025; Ryu et al., 2024; Willingham et al., 2021), meaning that picking a single ‘region of interest’ for quantification might not provide the whole story (Fig. 2D). Finally, mitochondrial matrix markers (such as MitoTracker or most transfected mitochondrial markers) provide a measure of IMM dynamics but not of complete mitochondrial fission. If IMM dynamics occur in the absence of OMM fission, use of a matrix marker alone might lead to an erroneous conclusion that full mitochondrial fission has occurred (Fig. 2E).

Mitochondrial ‘fragmentation’ – proceed with caution

In our opinion, some of the issues described above have resulted in a major misconception in the field: that mitochondrial depolarization causes large-scale mitochondrial fission (also referred to as ‘fragmentation’). Many studies have used mitochondrial depolarization-inducing agents to cause apparent fragmentation (Cereghetti et al., 2008; Fu and Lippincott–Schwartz, 2018; Griparic et al., 2007; Ishihara et al., 2003; Li et al., 2015). However, other studies describe the mitochondria as ‘swollen’, ‘disk-shaped’ or ‘donut-shaped’, and not fragmented (De Vos et al., 2005; Liu and Hajnóczky, 2011; Minamikawa et al., 1999). Studies that have analyzed this process in detail have shown that the major mitochondrial change occurring rapidly after depolarization is not fission but a morphological reorganization, resulting in a donut-shaped IMM within an OMM that is still intact (Fig. 2F) (Fung et al., 2019; Miyazono et al., 2018). This process, which we call ‘circularization’, is Drp1 independent but requires the IMM protease Oma1 (De Vos et al., 2005; Fung et al., 2019, 2022; Miyazono et al., 2018). Although fission does occur upon more prolonged mitochondrial depolarization, the events occurring within the first 30 min are mainly circularization. The techniques often used to assess depolarization effects (generally fixed-cell microscopy at low resolution) can easily mistake circularization for fragmentation.

How does the IMM divide?

In 1949, pioneering work identified mitochondria as central sites for aerobic ATP production (Kennedy and Lehninger, 1949). EM studies in the 1950s revealed mitochondria to be double-membrane organelles, adding a structural dimension to their known function (Palade, 1952; Sjostrand, 1953). In the 1960s, Peter Mitchell proposed the chemiosmotic hypothesis, fundamentally linking the IMM to a major mitochondrial function (Mitchell, 1966). Subsequent research highlighted the sensitivity of IMM morphology to metabolic changes (Hackenbrock, 1968).

Architecturally, the IMM forms invaginations known as ‘cristae’ (Fig. 3A), which are the primary sites of the electron transport chain and ATP synthase. Cristae are dynamic structures that can undergo rapid changes (Kondadi et al., 2020). Furthermore, neighboring cristae can exhibit different membrane potentials (Kondadi et al., 2020), suggesting the presence of structures that can restrict proton flow.

Fig. 3.

IMM fission. (A) Negative-stain EM micrograph of a mitochondrion from a HeLa cell, showing prominent inner membrane folds (cristae) as well as the non-cristae regions of the IMM, the inner boundary membrane (IBM). The cristae junction is the interface of the IBM and the cristae infolding. IMS, intermembrane space. Note the small diameter of the cristae lumen (typically ∼10 nm). Scale bar: 200 nm. Image taken by Radu Stan in the Dartmouth College Electron Microscopy Facility. (B) Schematic illustration of possible consequences of uncoordinated IMM–OMM fission. This could lead to cytosolic leakage of IMS or cristae lumenal components (e.g. cytochrome c) or matrix components (e.g. mtDNA) and subsequent activation of apoptosis or inflammatory pathways, respectively. Please note that these potential consequences are the authors' speculation and that mechanisms of actin IMM fission and coordination between IMM and OMM fission remain unclear. Created in BioRender by Kamerkar, S., 2025. https://BioRender.com/e48m132. This panel was sublicensed under CC-BY 4.0 terms.

Given the importance of IMM structure to overall mitochondrial function, we speculate that IMM fission is unlikely to be a passive process that merely occurs as a consequence of OMM constriction. Instead, existing evidence suggests the presence of dedicated IMM fission mechanisms. Studies in C. elegans, mouse embryonic fibroblasts and yeast lacking Drp1 show that, although overall mitochondrial fission is defective, mitochondrial matrix markers segregate, suggesting successful IMM fission (Ishihara et al., 2009; Labrousse et al., 1999; Lee and Yoon, 2014; Legesse-Miller et al., 2003; Wakabayashi et al., 2009). The IMM can also constrict or divide independently of the OMM (Breckwoldt et al., 2014; Brustovetsky et al., 2009; Griparic et al., 2004; Jaipargas et al., 2015; MacVicar and Lane, 2014; Skulachev et al., 2004; Stavru et al., 2013; Zhang et al., 2016; Fujioka et al., 2012).

The potential consequences of dysregulated mitochondrial fission are severe. Leakage of mitochondrial content is well known to activate large-scale cellular responses, such as apoptosis in response to release of intermembrane space (IMS)-localized proteins, such as cytochrome c (Czabotar and Garcia-Saez, 2023), or inflammation in response to matrix-localized mtDNA and RNA release (Vringer and Tait, 2023) (Fig. 3B). Also, IMM fission is presumably linked to appropriate distribution of matrix contents, such as mtDNA. Although coordination between mtDNA replication and mitochondrial fission has been shown (Lewis et al., 2016), the specific relationship between mtDNA replication and IMM fission remains unexplored. For these reasons, we feel that mechanisms for controlled IMM fission are likely. Next, we discuss challenges faced in studying IMM dynamics, then describe potential players in IMM dynamics identified to date.

Challenges in studying IMM dynamics

Two major challenges in studying IMM dynamics are its complex organization and small size. Cristae dimensions are typically below the resolution limit of light microscopy, with lumen diameters of ∼10 nm (Siegmund et al., 2018) (Fig. 3A). The IMM is also tethered to the OMM (Anand et al., 2021; Quintana-Cabrera et al., 2018).

A second challenge is that genetic screens, which have been effective in identifying OMM fission factors like Drp1 (Smirnova et al., 1998), are more complex when applied to IMM fission. This complexity arises because disruptions in IMM protein transport (e.g. caused by depletion of Sam50) (Ott et al., 2012) or CL synthesis (Dudek et al., 2013) can cause cristae structural defects that phenotypically mimic fission defects. The fact that the IMM is composed of ∼75% protein (Ardail et al., 1990; Simbeni et al., 1991) means that depletion or deletion of specific IMM proteins might have disproportionate effects and makes in vitro reconstitution of its dynamics particularly challenging. The lack of robust assays to study these processes has thus been a major obstacle in advancing our understanding of IMM dynamics.

Potential mechanisms and players in IMM remodeling and fission

Despite the challenges discussed above, a number of factors have been identified that function in maintaining cristae architecture and could conceivably participate in IMM fission.

Matrix Ca²⁺ appears to be important for initiating IMM fission. Two independent studies have demonstrated that IMM fission occurs before OMM fission, in a Ca²⁺-dependent manner (Chakrabarti et al., 2018; Cho et al., 2017). Constriction of the IMM, which frequently takes place at ERMCs, is Drp1 independent. Instead, these IMM constrictions are initiated by Ca²⁺ influx into the matrix through the mitochondrial Ca²⁺ uniporter (MCU) (Chakrabarti et al., 2018; Cho et al., 2017). The machinery required for IMM constriction and fission downstream of Ca²⁺ influx is unclear. One proposed mechanism involves a Ca²⁺-triggered phase transition of CL, which might facilitate membrane remodeling (Fox et al., 2019; Venkatraman et al., 2023). Alternatively, Ca²⁺ influx might trigger a simultaneous influx of potassium, leading to proteolytic processing of Opa1, a GTPase involved in IMM fusion. This processing, mediated by the mitochondrial metalloprotease YME1L, produces a short form of Opa1 that has been implicated in promoting mitochondrial fission (Anand et al., 2014; MacVicar and Langer, 2016).

Recently, an 8 kDa protein named Mdi1 (also called mitofissin or Atg44) was identified as a key player in yeast IMM fission. Mdi1 contains a putative amphipathic helix that might wedge into the IMM, destabilizing it and lowering the energy barrier required for IMM fission through Drp1-mediated constriction of the OMM (Connor et al., 2023). In addition to its role in IMM fission, Mdi1 is also crucial for mitophagy (Fukuda et al., 2023; Furukawa et al., 2024). Two observations, however, suggest that this mechanism is not universal – IMM constriction still occurs upon Drp1 knockdown in mammalian cells (Chakrabarti et al., 2018; Cho et al., 2017) and no clear mammalian Mdi1 homolog exists. However, another IMM protein that might participate in fission, Mdm33 in yeast and CCDC51 in humans, could potentially play a role in some fission events (Edington et al., 2025).

Many studies have underscored the significance of a large multiprotein complex known as the mitochondrial contact site and cristae organizing system (MICOS) in the structural organization of mitochondrial cristae (Harner et al., 2011; Hoppins et al., 2011; Kozjak-Pavlovic, 2017; Rabl et al., 2009; von der Malsburg et al., 2011). The MICOS complex comprises several proteins that localize periodically at cristae junctions, the regions where cristae meet the inner boundary membrane (Jans et al., 2013). Disruption or loss of MICOS components leads to disorganized and concentric cristae, highlighting its crucial role in maintaining their intricate architecture (Darshi et al., 2011). Mic60, a key MICOS protein, can remodel flat membranes into curved tubular structures, suggesting a specific role in the maintenance of cristae architecture that could potentially contribute to IMM fission (Hessenberger et al., 2017). Recently, the matrix-localized dynamin superfamily protein Mmc1 (also known as MPS2) was reported to interact with MICOS, playing a significant role in cristae architecture (Kumar et al., 2024). Although Mmc1 is present in most fungi, its mammalian homolog has yet to be discovered.

The ATP synthase complex (complex V), which couples proton motive force to ATP synthesis, might also be implicated in IMM fission. Approximately 94% of ATP synthase complexes are localized in cristae, mostly at cristae tips (Gilkerson et al., 2003; Kühlbrandt, 2019). In vitro reconstitution experiments have demonstrated that ATP synthase can induce membrane remodeling, transforming flat membranes into highly curved structures that resemble cristae (Blum et al., 2019; Jiko et al., 2015). Loss of ATP synthase function, whether through complete deletion or disruption of its dimerization, results in abnormal cristae morphology (Habersetzer et al., 2013; Kao et al., 2012; Paumard et al., 2002). Additionally, the electrochemical gradient within cristae might play a role in maintaining cristae structure through effects on CL (Khalifat et al., 2008). Whether these remodeling capabilities contribute to IMM fission remains to be shown.

Finally, FAM92A1 is a recently discovered matrix-localized BAR domain protein, capable of sensing membrane curvature. Deletion of FAM92A1 significantly perturbs cristae morphology. Intriguingly, in vitro assays have demonstrated that FAM92A1 possesses the ability to remodel flat membranes into highly curved tubular structures reminiscent of mitochondrial cristae (Wang et al., 2019).

How many ‘types’ of fission exist?

Mitochondria divide for many reasons. First, mitochondrial fission is important for proper distribution of mitochondria throughout cells. Most mitochondria have multiple genomes, and a typical cultured cell contains several hundred mitochondrial genomes (Bogenhagen and Clayton, 1974; Satoh, 1991). In dividing cells, mitochondrial fission expands mitochondrial mass and ensures that each daughter cell receives an appropriate number of genome-containing mitochondria (Mishra and Chan, 2014) (Fig. 4A). Even in cancer cells engaged in aerobic glycolysis (‘the Warburg effect’), mitochondria are important biosynthetic centers that support cell proliferation (Ahn and Metallo, 2015). In non-dividing cells, mitochondrial fission allows for efficient mitochondrial movement and distribution, which is especially important for polarized cells like neurons (Fig. 4B) in which both mitochondrial length and density can differ significantly between axons and dendrites (Lewis et al., 2018) or even within different regions of the same dendrite (Virga et al., 2024). Mitochondrial fission is an important factor controlling size, distribution and function in these circumstances.

Fig. 4.

Multiple reasons for mitochondrial fission. (A) Fission during cell division. Concerted mitochondrial fission occurs early in mitosis, contributing to proper mitochondrial partitioning between daughter cells. (B) Fission in neurons. (1) Mitochondrial fission in the cell body precedes mitochondrial entry into axons. (2) Fission within axons and dendrites contributes to mitochondrial distribution. (C) Two types of fission that could occur in the same cell. Left, fission of healthy mitochondria (midzone fission) associated with ER contact and INF2 activity. Healthy fission events tend to occur in the middle region of the mitochondrion. Right, fission for mitochondrial disposal by mitophagy (peripheral fission), associated with lysosomal contact and a drop in mitochondrial membrane potential, occurs at the ends of mitochondria. (D) Fission and production of mitochondrially derived vesicles (MDVs) and compartments. Fission is required to pinch off MDVs or MDCs from the mitochondrion. Created in BioRender by Kamerkar, S., 2025. https://BioRender.com/k19k990. This figure was sublicensed under CC-BY 4.0 terms.

Second, mitochondrial fission plays a crucial role in mitophagy (Pickles et al., 2018). Mitochondria are susceptible to damage from oxidative stress as well as accumulation of mutations to the mitochondrial genome. Fission at the ends of mitochondria enables damaged material to be shed into smaller mitochondria destined for mitophagy, which suggests a distinct fission signature for mitochondrial degradation (Kleele et al., 2021; Twig et al., 2008) (Fig. 4C). We should note that although mitochondrial fission is important for mitophagy, fission is inhibited during bulk autophagy (Gomes et al., 2011).

Third, mitochondrial fission is associated with changes in cell metabolism in a variety of contexts (Liesa and Shirihai, 2013; Mishra and Chan, 2016). One observation is that elongated mitochondria are more metabolically active, suggesting that mitochondrial fission can be associated with lower mitochondrial ATP generation. Conversely, mitochondrial fission can also be associated with higher mitochondrial oxidative activity in cells like brown adipocytes and pancreatic β-cells as well as in certain cancers, where excess fatty acids can trigger mitochondrial fission, which is necessary for increased mitochondrial fatty acid oxidation (Ngo et al., 2023). A possible mechanism of fatty acid-induced mitochondrial fission might be through activation of MiD49 and MiD51 by fatty acyl-CoA (Liu et al., 2024b). Finally, hyperglycemia triggers mitochondrial fragmentation in several cell types (Paltauf-Doburzynska et al., 2004; Shenouda et al., 2011; Wang et al., 2012; Yu et al., 2006).

Fourth, the fission machinery is involved in production of small mitochondrial fragments called mitochondrially derived vesicles (MDVs) (König and McBride, 2024), as well as release of multilamellar OMM-containing mitochondrially derived compartments (MDCs) (Wilson et al., 2024a,b). One clear function of both structures is transport of mitochondrial components to lysosomes or vacuoles for clearance and degradation (Sugiura et al., 2014; Wilson et al., 2024a). These structures might thus serve as a more subtle mechanism than mitophagy to maintain mitochondrial homeostasis, but MDVs and MDCs also function in signaling (König and McBride, 2024). MDVs additionally contribute mitochondrially derived components to developing peroxisomes (Sugiura et al., 2017). MDVs and MDCs are released from mitochondria owing to the actions of mitochondrial fission proteins, including Drp1, Mff, MiD49 and MiD51 (Hughes et al., 2016; König et al., 2021) (Fig. 4D). Intriguing recent studies show that portions of the IMM can be extruded from (McArthur et al., 2018) and even pinched off from the OMM (Prashar et al., 2024), but these structures are not Drp1-dependent and therefore might be distinct from MDVs and MDCs.

Finally, Drp1 and other components of mitochondrial fission are implicated in release of cytochrome c and other pro-apoptotic proteins during apoptosis. Although actual mitochondrial division does not necessarily occur during apoptosis, suppression of Drp1 delays cytochrome c release, caspase activation and cell death (Suen et al., 2008).

Assigning specific mitochondrial fission factors to specific fission events

It is reasonable to expect that the diverse purposes for mitochondrial fission would be accompanied by distinct regulatory mechanisms given the plethora of regulatory factors described above, many of which are unlikely to regulate 100% of fission events. In addition, the effects of some regulatory mechanisms vary from cell to cell. For example, Drp1 S637 phosphorylation inhibits fission in some cases but activates fission in others (see above). Finally, the functional and morphological heterogeneity of mitochondria both between cells and within a single cell (Benador et al., 2019; Graham et al., 2017; Han et al., 2023; Monzel et al., 2023; Willingham et al., 2021) might suggest differential regulatory mechanisms for fission.

A current challenge is to determine the regulatory factors that are relevant in specific contexts. A recent study has revealed two distinct fission processes, designated ‘midzone’ and ‘peripheral’ fission (Fig. 4C) (Kleele et al., 2021). Midzone fission occurs in ‘healthy’ mitochondria and is associated with cell growth, whereas peripheral fission separates segments of unhealthy mitochondria destined for mitophagy. Midzone fission is dependent on INF2-polymerized actin and is associated with ER contacts, whereas peripheral fission shows a dependence on Fis1. We suspect that midzone and peripheral fission might be only the tip of the iceberg and that other categories of mitochondrial fission might exist that are involved in controlling MDV and MDC assembly, mitochondrial distribution and metabolic switching.

A model of coordinated function of multiple mitochondrial fission regulators

At some point, the multiple regulatory factors involved in fission (multiple Drp1 receptors, actin, CL, post-translational modifications and organelles) must be linked with the multiple fission stimuli (cell division and/or growth, cell polarization, mitophagy and metabolic changes) in a cohesive fashion. It is unlikely that all regulatory mechanisms are acting in all fission events and equally unlikely that single regulatory mechanisms are acting in isolation. We propose a new model, based on the following findings, as a start towards an integrated view of mammalian fission regulation. First, Mff appears universally important in mammalian mitochondrial fission because knockout and knockdown studies show nearly equivalent effects of Mff and Drp1 on mitochondrial length (Gandre-Babbe and Van Der Bliek, 2008; Ji et al., 2017; Losón et al., 2013; Otera et al., 2010). Second, Mff can synergize with three other factors (actin filaments, MiD49 and MiD51, and CL) to activate Drp1 (Ji et al., 2015; Liu et al., 2021,b; Macdonald et al., 2016). By contrast, actin, MiDs and CL do not synergize with each other. Third, Mff appears to prefer Drp1 oligomers to dimers (Liu and Chan, 2015), although this is not universally accepted (Clinton et al., 2016). Fourth, treatments that induce mitophagy can cause increased CL concentration on the OMM (Chao et al., 2019; Chu et al., 2013).

Based on these findings, we propose a two-step Drp1 recruitment process, in which there are three distinct mechanisms for initial Drp1 recruitment. All three initial recruitment mechanisms pass Drp1 oligomers to a common factor, Mff, which mediates the second step – maturation of the Drp1 oligomer into a ring capable of OMM contraction (Fig. 5).

Fig. 5.

Model of three regulatory routes for mammalian mitochondrial fission. Three conditions trigger distinct mechanisms for initial Drp1 recruitment to and oligomerization on the OMM. Subsequently, Drp1 matures into a contractile ring, mediated by Mff. In healthy fission (‘midzone fission’), increased cytoplasmic Ca²⁺ triggers ER-associated actin polymerization through INF2. These actin filaments recruit Drp1 and initiate the oligomerization process. In metabolic fission, increased cytoplasmic fatty acids lead to increased fatty acyl-CoA, stimulating oligomerization of MiD49 and MiD51, which in turn recruit Drp1 to the fission site. In therapeutic fission (‘peripheral fission’), mitochondrial damage or dysfunction leads to exposure of CL on the OMM. Increased OMM CL leads to Drp1 recruitment at the fission site. Created in BioRender by Kamerkar, S., 2025. https://BioRender.com/a47u992. This figure was sublicensed under CC-BY 4.0 terms.

The three initial recruitment possibilities occur in three distinct contexts, with two of the contexts described previously (Kleele et al., 2021). The first context is ‘healthy fission’ (called midzone fission in Kleele et al., 2021), linked to cell growth and division. Initial activation of healthy fission occurs through increased intracellular Ca²⁺, which stimulates ER-associated actin polymerization through INF2, driving ER–mitochondrial contact. This contact stimulates both IMM dynamics (through increased matrix Ca²⁺) and OMM dynamics (through Drp1 recruitment). The second context is metabolic fission, linked to an increased need for mitochondrial fatty acid oxidation. Initial activation of metabolic fission occurs through increased fatty acyl-CoA, which activates MiD49 and MiD51 to initiate Drp1 recruitment. The third context is ‘therapeutic’ fission, linked to mitophagy (called peripheral fission in Kleele et al., 2021). An important stimulus for this fission might be CL exposure on the OMM.

We realize that this model might be overly simplistic. There are likely to be other categories of fission and there might be mechanisms in which some of these regulatory factors work outside of this mechanistic structure. However, as with any model, we hope that it provides the basis for further experimentation and, as such, drives the field forward.

Conclusions and perspectives

Although the identified regulatory mechanisms for mitochondrial fission are many and varied, it is possible to create integrated models for how these mechanisms work together in specific contexts. Going forward, full models will eventually require several additional facets – accounting for other regulatory mechanisms not included in our model, accounting for IMM fission as well as OMM fission, and including a more detailed understanding of downstream processes following Drp1-mediated OMM constriction.

The model shown in Fig. 5 neglects several known regulatory mechanisms that certainly play roles in fission. In particular, Drp1 phosphorylation at S616 and S637 (Fig. 1B) is clearly important but the mechanistic basis for this regulation is still unclear. Here, we raise five issues concerning Drp1 phosphorylation. First, although S616 phosphorylation has been uniformly associated with increased fission, this phosphorylation does not stimulate the GTPase activity of Drp1. In fact, S616 phosphorylation decreases or eliminates the ability of Mff, MiD49, actin or CL to stimulate Drp1 (Liu et al., 2024a). One possibility is that S616-mediated Drp1 stimulation requires additional factors that have not yet been identified. Second, S637 phosphorylation is associated with decreased fission in some cases (Cereghetti et al., 2008; Chang and Blackstone, 2007; Cribbs and Strack, 2007; Gomes et al., 2011; Wikstrom et al., 2013) but increased fission in others (Han et al., 2008; Jhun et al., 2018; Wang et al., 2012). More detailed studies are required to define the roles of these phosphorylation events in specific contexts. Third, the same kinase can phosphorylate different sites depending on cell type. For example, ROCK proteins phosphorylate S616 in cardiomyocytes (Brand et al., 2018) but S637 in podocytes (Wang et al., 2012). Curiously, both phosphorylation events enhance fission. Fourth, the crosstalk between S616 and S637 phosphorylation is unclear. One study shows that S637 phosphorylation promotes S616 phosphorylation (Valera-Alberni et al., 2021), whereas other studies report a lack of correlation or an inverse correlation between S616 and S637 phosphorylation (Cha et al., 2021; Guo et al., 2018; Lee and Kim, 2018; Ma et al., 2020; Xu et al., 2016). Fifth, the coordination between phosphorylation and other identified Drp1 PTMs or with Mff PTMs is unclear.

Another interesting question is how the multiple splice variants of both Drp1 and Mff play context-specific roles in mitochondrial fission. Recent evidence suggests that a Drp1 splice variant associated with poor cancer outcomes causes decreased apoptosis-induced mitochondrial fission as well as decreased sensitivity to chemotherapeutics (Javed et al., 2024). Finally, although Drp1-mediated constriction alone can induce membrane fission in vitro (Kamerkar et al., 2018), it is likely that events subsequent to this constriction are required for mitochondrial fission in cells, as evidenced by the role for Golgi-derived vesicles late in the fission process (Nagashima et al., 2020).

Overall, integrating individual regulatory factors for mitochondrial fission into context-specific fission processes will enable a more holistic understanding of fission regulation for specific purposes, rather than a ‘one size fits all’ mechanistic understanding.