Causal roles of mitochondrial dynamics in longevity and healthy aging

Abstract

Mitochondria are organized in the cell in the form of a dynamic, interconnected network. Mitochondrial dynamics, regulated by mitochondrial fission, fusion, and trafficking, ensure restructuring of this complex reticulum in response to nutrient availability, molecular signals, and cellular stress. Aberrant mitochondrial structures have long been observed in aging and age‐related diseases indicating that mitochondrial dynamics are compromised as cells age. However, the specific mechanisms by which aging affects mitochondrial dynamics and whether these changes are causally or casually associated with cellular and organismal aging is not clear. Here, we review recent studies that show specifically how mitochondrial fission, fusion, and trafficking are altered with age. We discuss factors that change with age to directly or indirectly influence mitochondrial dynamics while examining causal roles for altered mitochondrial dynamics in healthy aging and underlying functional outputs that might affect longevity. Lastly, we propose that altered mitochondrial dynamics might not just be a passive consequence of aging but might constitute an adaptive mechanism to mitigate age‐dependent cellular impairments and might be targeted to increase longevity and promote healthy aging.

Mitochondria form a dynamic, interconnected network that is shaped by fusion, fission and trafficking. This review discusses how aging affects mitochondrial dynamics and how such alterations in turn might help the organism to cope with age‐dependent cellular impairments.

Glossary

ALM neurons

Anterior lateral motor neurons

AMPK

AMP‐activated protein kinase

Bmal1

Brain and muscle arnt‐like protein‐1

CA

Colanic acid

CLEAR

Coordinated lysosomal expression and regulation

CMTD

Charcot–Marie–Tooth disease

CR/DR

Calorie restriction/dietary restriction

CRTC‐1

Cyclic AMP‐responsive element‐binding protein‐regulated transcriptional coactivator

DRP1

Dynamin‐related protein 1

ER

Endoplasmic reticulum

ERMES

ER‐mitochondria encounter structures

ETC

Electron transport chain

FIS1

Mitochondrial fission 1 protein

FUNDC1

FUN14 domain‐containing 1

FZO

Fuzzy onions

HUVEC

Human umbilical vein endothelial cells

IF

Intermittent fasting

IIS

Insulin/IGF‐1 like signaling

IMM

Inner mitochondrial membrane

INF2

Inverted formin 2

IP3R

Inositol 1,4,5‐trisphosphate receptor

MAMs

Mitochondria‐associated membranes

MCU

Mitochondrial calcium uniporter

MDV

Mitochondria‐derived vesicle

MEFs

Mouse embryonic fibroblasts

MFF

Mitochondrial fission factor

MFNs

Mitofusins

MiD49/51

Mitochondrial dynamics protein 49/51

MIRO

Mitochondrial Rho GTPase

mtDNA

Mitochondrial DNA

MTFP1

Mitochondrial fission process protein 1

mTOR

Mechanistic target of rapamycin

NAD

Nicotinamide adenine dinucleotide

NDC

Nondigestible carbohydrates

OMM

Outer Mitochondrial membrane

OPA1

Optic Atrophy 1

PERK

Protein kinase RNA‐like endoplasmic reticulum kinase

PEX

Peroxin proteins

PGC1‐α

Peroxisome proliferator‐activated receptor‐gamma coactivator 1‐α

POMC

Proopiomelanocortin

PTP

Permeability transition pore

PUM2

Pumilio2

ROS

Reactive oxygen species

SCFA

Short‐chain fatty acid

SERCA

Sarco/endoplasmic reticulum Ca²⁺‐ATPase

SIMH

Stress‐induced mitochondrial hyperfusion

SNPH

Syntaphilin

TFAM

Transcription factor A

TFEB

Transcription factor EB

TRAK

Trafficking kinesin protein

ULK1

Unc‐51‐like kinase

UPR

unfolded protein response

UPTmt

Mitochondrial unfolded protein response

Introduction

Mitochondria are key contributors to cellular and organismal health, producing the majority of energy needed to fuel biological processes in eukaryotes. However, while it is undoubtedly true that the estimated 10 million billion mitochondria spread throughout our cells generate the bulk of ATP for cellular processes, the simplistic concept of these organelles as mere “powerhouses” has changed in recent years 1. It is now well‐known that mitochondria form dynamic organellar networks that fuse, divide, and interact with other organelle structures, and direct myriad processes that regulate cellular fitness, longevity, and fate (Fig 1) 2, 3, 4. Conversely, extensive research has shown that mitochondrial dysfunction contributes to several metabolic disorders, cancer, neurodegeneration, and numerous other pathologies 5, 6, 7, 8. Importantly, multiple studies now also suggest that in addition to pathological states, mitochondrial function actively modulates normal aging, and modulating mitochondrial form or function can directly affect organismal longevity 9, 10.

Figure 1. Mitochondria perform a variety of functions that are essential for cellular health.

As the metabolic hub of the cell, mitochondria have many roles beyond ATP generation such as the production of intermediate metabolites, inter‐organelle signaling, and Ca²⁺ and ROS homeostasis. All of the functions listed here are interconnected, dependent upon both the integrity of individual mitochondria and the integrity of the entire mitochondrial network. Additionally, these processes are dysregulated with age. Age‐related changes in mitochondrial fission and fusion may contribute to the impairment of these mitochondrial functions.

Links between mitochondria and the aging process have long been recognized. The earliest and perhaps most prominent theories proposed that decreased mitochondrial function, increased mitochondrial reactive oxygen species (ROS) production, and increased mitochondrial DNA (mtDNA) mutations are causally linked to aging 11, 12, 13, 14. Recently, the detrimental role of mitochondrial‐derived ROS in aging has been challenged, with mitochondrial hormesis—low levels of ROS that are pro‐longevity—gaining traction as a link between mitochondrial function and healthy aging 15, 16, 17, 18. Coinciding with new roles for cellular mitochondria has been a growing appreciation that, like other organelles, mitochondria do not exist in isolation in the cell and instead show dynamic remodeling and interactions both with each other and with alternate cellular compartments 19. Interestingly, studies investigating mitochondrial network dynamics have found that the state of mitochondrial morphology, whether fused or fragmented, has a direct consequence on all of the above‐mentioned mitochondrial functions. Mitochondrial dynamics themselves undergo alterations with age 20, 21, 22, 23, 24, 25. Studies in nematodes 26, flies 27, and mice 28 have shown that the mitochondrial network of aged animals is often more heterogeneous, fragmented, and comprised of large, swollen mitochondria that cannot be eliminated by mitophagy. However, much less is known about the origin of these changes and whether they actively drive aging by modulating mitochondrial health or are a passive consequence of aging.

In this review, we will discuss how mitochondrial dynamics are altered with age. We will examine whether or not a particular mitochondrial morphology renders pro‐ or anti‐aging effects, including findings from both invertebrate and vertebrate systems. Importantly, we will review emerging research on distinct factors, which directly and indirectly influence mitochondrial dynamics to affect longevity and discuss their impact on the field of aging. Beyond defining one network state as being strictly pro‐aging or pro‐longevity, we describe the contextual roles for fusion and fission in healthy aging and the underlying functional outputs behind these effects. To better understand these associations, we first describe the components of the mitochondrial fission and fusion machinery, their role in mitochondrial morphology maintenance, and their dysregulation with age.

Fission and fusion: molecular modulators and their age‐related dysregulation

Mitochondrial fission

It was previously hypothesized that mitochondria undergo growth and division in the cell primarily to ensure that new daughter cells inherit a parental mitochondrial complement 26. However, beyond mtDNA replication and transmission to daughter mitochondria, it is now appreciated that mitochondrial fission is essential for removal of defective mitochondria via mitophagy, mitochondrial transport, and programmed cell death via stress‐induced hyperfission 29, 30, 31, 32.

Mitochondrial fission is coordinated by two classes of regulatory components: cytosolic proteins that translocate to the mitochondria to initiate fission and mitochondrial membrane proteins that anchor the cytosolic factors. The central protein which orchestrates mitochondrial fission is the cytosolic dynamin‐related protein 1 (DRP1, Table 1). DRP1 is primarily localized in the cytosol but assembles on mitochondrial networks in response to phosphorylation at Ser⁶¹⁶ and is known to mediate membrane constriction and scission 33. Recent studies have found that other more canonical dynamins such as dynamin‐2 assist DRP1 in terminal steps of mitochondrial scission as well 34. Imaging studies and protein–protein interaction analyses have found several mitochondrial outer membrane‐bound proteins that recruit DRP1 to initiate scission. These proteins include mitochondrial fission factor (MFF), fission protein‐1 (FIS1), mitochondrial dynamics protein‐49 (MiD49), and mitochondrial dynamics protein‐51 (MiD51) (Fig 2A) 33, 35. In addition to the outer mitochondrial membrane (OMM) fission machinery, proteins such as MTP18 that face the inter‐membrane space (IMS) by tethering to either the OMM or the inner mitochondrial membrane (IMM) assist in IMM fission to complete mitochondrial separation 36.

Table 1.

Mammalian proteins and their eukaryotic homologs involved in mitochondrial dynamics. Functional homologs with no apparent sequence homology are marked with *. — indicates no currently identified homolog

Mammalian protein	Drosophila homolog	C. elegans homolog	Yeast homolog	Function
DRP1	DRP1	DRP1	Dnm1p	Fission
MFF	Tango11	MFF‐1, MFF‐2	Mdv1p*	Fission
FIS1	FIS1	FIS‐1, FIS‐2	Fis1p	Fission
MiD49/MIEF2	—	—	—	Fission
MiD51/MIEF1	—	—	—	Fission
MFN1/2	FZO	FZO	Fzo1p	Fusion
OPA1	DMEL	EAT‐3	Mgm1p	Fusion
MIRO1/2	MIRO	MIRO‐1	Gem1p	Trafficking
Milton1/2 (OIP106/GRIF‐1)	Milton	TRAK‐1	—	Trafficking

Figure 2. Mechanism of mitochondrial fission and fusion.

(A) Mitochondrial fission is initiated by endoplasmic reticulum (ER, blue)‐dependent or endoplasmic reticulum‐independent mechanisms. Receptor proteins such as mitochondrial fission factor (MFF), mitochondrial fission 1 protein (FIS1), mitochondrial dynamics protein 49 (MiD49), and mitochondrial dynamics protein 51 (MiD51) recruit the cytosolic scission factor dynamin‐related protein 1 (DRP1). Once localized to the mitochondria, DRP1 multimerizes to form spirals on the outer mitochondrial membrane (OMM), hydrolyzes GTP, and causes mitochondrial constriction. Dynamin‐2 is then recruited and orchestrates sequential constriction leading to mitochondrial division. These events are followed by disassembly of the fission proteins that translocate to the mitochondria and a decrease in ER‐mitochondrial tethering. (B) Mitochondrial fusion begins with tethering of two separate mitochondrion via outer membrane tethering mitofusins (MFN1/2). Successful tethering and OMM fusion are followed by the fusion of inner mitochondrial membrane (IMM) mediated by the protein OPA1 resulting in fusion of two separate mitochondrion.

Several proteins involved in mitochondrial fission are known to be dysregulated with age, which likely contributes to the altered mitochondrial network architecture seen in old organisms. Firstly, aged mice demonstrate reduced DRP1 activity and altered mitochondrial morphology in several tissues including neurons, skeletal muscle, and oocytes 37, 38. Aged skeletal muscle in mice shows increased mitofusin 2‐to‐DRP1 ratio and longer intermyofibrillar mitochondria 28. Furthermore, aged human endothelial cells (HUVECs) in vitro show a similar downregulation of both DRP1 and FIS1 expression as well as elongated mitochondrial networks 39. Recently, it was found that the RNA‐binding protein Pumilio2 (PUM2) binds Mff mRNA and prevents its translation. Coupled with the finding that PUM2 levels increase with age in humans, mice, and nematodes, this study demonstrates a post‐transcriptional mechanism by which mitochondrial fission and mitophagy are impaired with age 40. Importantly, it has also been shown that induction of Drp1p expression in Drosophila midlife prolongs both lifespan and health span via improved mitochondrial respiration and autophagy. Furthermore, in both Caenorhabditis elegans and flies, induction of mitochondrial fragmentation specifically in the intestine is known to increase lifespan suggesting that maintaining mitochondrial fission might have pro‐longevity effects 27, 41.

These findings in different aging models have indicated that proteins involved in mitochondrial network fission decline with age and that restoring their expression might result in improved mitochondrial function, morphology, and organismal lifespan. However, paradoxical findings in yeast have shown that deleting the Drp1 ortholog Dmn1p to retard mitochondrial fission significantly delays yeast aging without impairing growth rate or fertility 42. Conversely, findings from mouse skeletal muscle have shown that increased mitochondrial fission is correlated with impaired insulin signaling and mitochondrial dysfunction 43.

Together, these data show that promoting mitochondrial fission can increase or decrease mitochondrial fitness and function depending on the tissue or even organism. Even with these seemingly conflicting findings, one thing remains clear; processes regulating mitochondrial fission alter with age. Nonetheless, important questions remain regarding the link between aging and changes in mitochondrial fission. For example, which factors cause altered expression of mitochondrial fission proteins during aging? Are these factors genetic, epigenetic, nutritional, or environmental? Is altered mitochondrial fission the primary cause of mitochondrial dysfunction in aged cells or tissues? Can mitochondrial fission be specifically modulated by altering specific pathways and, if so, might proteins involved in mitochondrial fission be key targets for promoting healthy aging? Future studies designed to investigate the convergence of these questions will help us better understand the role of dysregulated mitochondrial fission in age‐related mitochondrial dynamics dysfunction.

Mitochondrial fusion

Mitochondrial fusion is achieved by membrane‐bound dynamin‐related proteins. Mitofusin 1 (MFN1) and MFN2 mediate the fusion of the outer membrane, whereas fusion between the IMM is mediated by optic atrophy 1 (OPA1, Table 1, Fig 2B) 50. Loss of either OMM or IMM fusion proteins results in a hyperfragmented mitochondrial network.

Similar to fission, mitochondrial fusion enables manipulation of mitochondrial morphology to meet specific cellular demands. Metabolically active cells often have interconnected mitochondrial networks 44. Fused mitochondrial networks allow efficient exchange of metabolites, transmission of membrane potential from mitochondria in one part of the cell to the other, increased ATP production, and decreased ROS production. Additionally, fusion reduces endoplasmic reticulum (ER)‐mitochondrial Ca²⁺ transfer to prevent Ca²⁺ induced cell death 4, 45, 46, 47. Importantly, mitochondrial fusion helps mitigate mitochondrial stress by fusing the contents of partially damaged mitochondria with healthy mitochondria as a form of complementation.

Mitochondrial fusion and fission both allow mitochondria to adapt to acute changes in nutrient availability, which is important for metabolic flexibility. Fusion specifically is known to increase mitochondrial production of ATP 48. Dietary restriction (DR) increases mitochondrial fusion by increasing cyclic AMP (cAMP) levels, activating protein kinase A (PKA), and inhibiting DRP1 via PKA‐mediated phosphorylation 43. This change in mitochondrial dynamics is a response to increased cellular ATP levels. This suggests that a switch to fused mitochondrial network aims to increase metabolic efficiency. Indeed, in obese individuals, impaired glucose oxidation and skeletal muscle metabolic efficiency are linked to decreased Mfn2 expression and interventions that improve glucose oxidation increase Mfn2 expression 49.

In addition to physically combining two individual mitochondria, fusion proteins play an important role in another key process of mitochondrial dynamics: mitochondrial trafficking 51. Proper distribution of mitochondrial mass within a cell is critical for ATP production. Furthermore, mitochondria need to be mobilized to defined subcellular destinations depending on ATP requirement to fuel specific functions such as cellular repair and renewal of senescent organelles 52. This trafficking of mitochondria is accomplished by their mobilization along the cell's microtubular apparatus, a process in which MFNs play a key role 53. MFNs bind the mitochondrial rho GTPase 1/2 (MIRO1/2) proteins on the OMM and help tether the MIRO proteins to Milton/TRAK which in turn interface with the kinesin or dynein motors attached to the microtubules (Table 1) 54. MFNs thus help tether the MIRO‐Milton‐kinesin/dynein complex and enable mitochondrial trafficking 55. This role of MFNs is distinct from their role in fusing mitochondria and is not shared by fission proteins.

Of the two MFNs, MFN1 is the primary isoform expressed in liver and pancreas, whereas MFN2 is more abundant in skeletal muscle, brain, and brown adipose tissue 56, 57. Both MFNs are expressed in heart and adrenal glands. Being the primary isoform expressed in skeletal muscle, MFN2 function has been extensively studied in aged and diabetic skeletal muscle. MFN2 expression is directly proportional to skeletal muscle insulin sensitivity and decreases with age, indicating decreased fusion is consistent with age‐related dysfunction in mitochondrial dynamics and metabolic function 49, 58. Indeed, it has also been shown that in cultured cells, mitochondrial fusion promotes metabolic efficiency 48. Consistent with these findings in cultured cells, mitochondrial dynamics follow a diurnal pattern controlled by the circadian regulator brain and muscle arnt‐like protein‐1 (Bmal1), which regulates mitochondrial fusion by modulating MFN1 and MFN2 expression to impact mitochondrial metabolism during feeding‐to‐fasting transitions. Increased fission during fasting‐to‐feeding transition increases metabolic efficiency and reduces oxidative stress. Additionally, the C. elegans Bmal1 ortholog, aha‐1, also modulates oxidative metabolism and promotes longevity via fusion of mitochondrial networks 2.

Recently, a study investigating diverse longevity paradigms in C. elegans has shown that mitochondrial fusion increases and facilitates lifespan extension in various known pro‐longevity interventions such as insulin/ IGF‐1‐like signaling (IIS) inactivation, caloric restriction, germline depletion, electron transport chain (ETC) dysfunction leading to mitochondrial unfolded protein response (UPRmt), mechanistic target of rapamycin (mTOR) inactivation, sirtuin overexpression, AMP‐activated protein kinase (AMPK) overexpression, impaired glycolysis, and exercise 59. Furthermore, lifespan extension for all these longevity pathway mutants with increased mitochondrial fusion was significantly reduced by eat‐3 RNAi, the C. elegans homolog for mammalian fusion protein OPA1. Interestingly, eat‐3 RNAi by itself did not alter the lifespan of wild‐type animals. Additionally, RNAi of drp‐1 or overexpression of the mitofusin ortholog FZO‐1, both of which increase mitochondrial fusion, does not increase lifespan in C. elegans 25.

Similar to C. elegans, studies in Drosophila show a shift toward increased mitochondrial elongation in midlife along with the accumulation of mitofusin proteins during fly aging 27. However, to what extent mitochondrial fusion facilitates lifespan extension in flies under different longevity paradigms still needs to be comprehensively investigated. Converse to findings in C. elegans, reduction in mitofusin levels in aging flies via ubiquitous or neuron‐specific overexpression of parkin, an E3 ubiquitin ligase, extends both mean lifespan and maximum lifespan 60. Together, these studies currently suggest that although mitochondrial fusion is a conserved mechanism to increase metabolic efficiency, maintain membrane potential, and promote exchange of metabolites and is required for the lifespan extension via multiple interventions in C. elegans, the effects on longevity might not be consistent in Drosophila. Furthermore, in C. elegans, while fusion is required for lifespan extension under different pro‐longevity paradigms, it does not seem solely sufficient for promoting longevity.

Mitochondrial trafficking

In addition to changes in mitochondrial shape, mitochondrial dynamics include another active process called mitochondrial trafficking. As discussed previously, the movement of mitochondria across cells is key to various homeostatic processes and is accomplished with the coordination of mitofusins, tethering proteins such as MIRO and Milton, and the cell's microtubular apparatus 61. In general, mitochondria move in the anterograde direction (away from cell's center) using kinesin motors and move in the retrograde direction (toward the cell's center) using the dynein motors 62.

Both anterograde and retrograde mitochondrial trafficking have been very well characterized in neurons, more than any other cell type because of their unique cellular morphology. Neurons extend their axons and dendrites for centimeters, whereas most cells are only measured in micrometers. Therefore, neurons have been a key model to study the need for mitochondrial distribution and energy supply to distant cellular regions 63. Whereas mitochondria in most cell types form an interconnected network, axonal mitochondria in the neuron often get trafficked there and then exist as discrete organelles 64. Additionally, it has been previously shown that mitochondria display three distinct stages where their size increases, is maintained, and then decreases in neurons during adulthood 65.

Earlier hypotheses proposed that mitochondrial biogenesis in neurons occurs in the cell body, followed by anterograde transport of these “young” mitochondria to distal regions and subsequent return of “aged” mitochondria for degradation via retrograde transport 66. This implies that firstly, two distinct mitochondrial populations exist in the neuron in separate regions and that secondly, retrograde trafficking increases in neurons with age as dysfunctional mitochondria accumulate. The results from studies testing these hypotheses ranged from a clear demonstration of dysfunctional mitochondria in distal regions of the neurons to a modest difference in morphology between central and distal populations, to no apparent differences in mitochondrial populations 67, 68, 69. While these results are confounding, it is apparent that both anterograde and retrograde mitochondrial trafficking decrease with age.

A study measuring mitochondrial transport in distal regions of anterior lateral motor (ALM) neurons in C. elegans found that both anterograde and retrograde mitochondrial trafficking progressively decline in WT animals from the first day of adulthood to old age and this decline is blunted in long‐lived C. elegans strains such as eat‐2 and daf‐2 mutants 65. Consistent with decline of mitochondrial motility in C. elegans, Vagnoni et al used Drosophila wing nerve system to show that the same phenomenon is observed in flies. In flies, mitochondrial transport is regulated by cyclic AMP/protein kinase A pathway 70. Vagnoni et al showed that the level of catalytic subunit of PKA decreases in flies leading to age‐dependent impairment of mitochondrial transport. They further demonstrated that this age‐related decline can be reversed by cAMP supplementation or PKA upregulation. Given that cAMP is an important second messenger in intracellular signaling and its supplementation has been shown to extend lifespan in both flies and mice, it could be that part of this pro‐longevity effect is through maintenance of mitochondrial transport in aging cells 71, 72.

Mitochondrial trafficking and morphology in part are interconnected through MFN1/2s dual role in both mitochondrial movement and fusion. Live imaging of neurons cultured from Mfn2 knockout mice shows that axonal mitochondria spend more time paused and undergo slower bi‐directional movements 51. Since both OMM mitofusins interact with MIRO and Milton complex, loss of either Mfn1 or Mfn2 alters mitochondrial movement, whereas knockdown of the IMM fusion factor Opa1 has no effect on mitochondrial motility. Consistent with findings that MFN2 expression decreases in aged muscle, mitochondrial motility decreases in muscle as well suggesting that decreased mitofusin expression accompanies decreased mitochondrial motility 49, 73.

In addition to a decrease in MFN expression, previous studies have shown that gross mitochondrial and cellular changes can initiate changes in mitochondrial trafficking. Firstly, mitochondrial membrane potential is key for efficient mitochondrial movement because mitochondrial depolarizing agents tend to inhibit mitochondrial transport 74. Supporting the theory of transport of dysfunctional mitochondria back to cell body, depolarized mitochondria have been shown to preferentially move retrogradely and mitochondria with high membrane potential have been shown to move preferentially anterogradely 74. Secondly, changes in cellular ATP/ADP concentrations have been shown to affect mitochondrial movement because of the requirement for the ATP for the active transport process 75. Thirdly, high cellular Ca²⁺ levels have been shown to decrease mitochondrial motility as a way to locally contribute to Ca²⁺ buffering 76. Expectedly, mitochondrial membrane potential, cellular ATP/ADP ratios, and Ca²⁺ all change as cells age indicating that mitochondrial motility might be affected by a number of different processes.

Similarly, increased ROS production with age decreases mitochondrial trafficking through a p38α‐dependent pathway. Debattisti et al recently demonstrated that ROS generated from intracellular and extracellular sources can retard mitochondrial movement inside cells. Increased ROS requires p38α kinase activity in cells to inhibit the mitochondrial adaptor complex formed by proteins MIRO and TRAK, which link mitochondria to motors that allow their transport 77. The study shows that the cellular milieu senses increase in ROS levels and employs a specific signaling pathway to decrease mitochondrial motility. Precisely how p38α senses and is activated in response to ROS is unknown and so is the specific mechanism by which it inhibits the association of the mitochondrial adaptor complex, why a decrease in mitochondrial motility is required in response to increased ROS is not fully understood as well, but what is interesting is that this mechanism does not require a gross mitochondrial defect. This effect is independent of mitochondrial permeability transition pore (PTP) opening, changes in membrane potential, and changes in Ca²⁺ homeostasis.

In addition to MFN‐modulated mitochondrial transport, DRP1‐dependent mitochondrial transport has been observed during immune synapse formation, suggesting fission proteins and other canonical dynamins might also play a role in mitochondrial movement. Baixauli et al reported that in human T cells DRP1 binding to OMM and subsequent mitochondrial fission coordinate mitochondrial interaction with microtubule organization centers and move the mitochondria toward the immune synapse 78. Studies in Drosophila support DRP1's role in mitochondrial movement toward neuromuscular junctions and neuronal axons 79, 80. Age‐dependent decreases in DRP1 level might thus not only reduce fission but also directly or indirectly affect mitochondrial movement.

It is evident that mitochondrial trafficking decreases with age, but because mitochondrial trafficking has been shown to be affected by several broad changes in cellular milieu, which change with age, separating causal effectors from epiphenomenon is difficult. The decrease in mitochondrial motility is likely due to a combination of several changes that occur with age, but specific signaling mechanisms still remain elusive. In the last decade, several of the mitochondrial motility effectors have been determined and additional regulators are still being identified. In a recent study, a novel regulator of mitochondrial motility was identified. It was demonstrated that ubiquitination of several residues in a mitochondrial localized protein called syntaphilin (SNPH) is required for its proper association with tubulin, thus tethering the mitochondria to the microtubule and maintaining them in a stationary state 81. While proteins such as MFN1/2, MIRO, TRAK, and SNPH emerged as central regulators of mitochondrial trafficking, further studies are needed to fully characterize proteins, post‐translational modifications, and signaling pathways that regulate the mitochondrial trafficking process. This would enable identification and targeting of specific proteins that mediate this process to inhibit the age‐related decline in mitochondrial movement.

Mitochondrial dynamics: is timing key?

Age onset of swollen, fragmented mitochondria has been reported across organisms, suggesting mitochondrial fragmentation is pro‐aging 25. Additionally, the fusion of mitochondrial networks tends to correlate with lifespan extension, as mentioned earlier. While it might thus be appealing to categorize fission as “bad” and fusion as “good”, many questions require resolution for that classification to be warranted. If high mitochondrial fission rates decrease organismal lifespan, why are decreases in fission machinery components associated with aging? Additionally, why does increasing fission in Drosophila midlife prolong both lifespan and health span 27? Furthermore, if fusion is essentially pro‐aging then why does chronically driving fusion via FZO‐1 overexpression or drp‐1 deletion in C. elegans not increase lifespan 25?

The answer to these questions might lie in the timing. A recent study investigating the effect of perturbing mitochondrial fission and fusion on C. elegans lifespan and health span showed that loss of fission decreased neuronal function and performance across several behavioral tests primarily during development (L4 stage) and in aged animals (day 7 and day 11 adults), but not during early adulthood (day 3 adults). Loss of fusion, however, produced more consistent and progressive neurological and behavioral defects all across development and adulthood (L4, day 3, day 7, and day 11 adults) 82. Complementary findings from Rana et al showing lifespan extension in Drosophila only upon midlife induction of mitochondrial fission, not in early adulthood, and lifespan shortening with midlife upregulation of dMFN highlight that mitochondrial fission plays a key role in early development and around midlife 27.

Consistent with studies in Drosophila, research in both mice and C. elegans has shown that increasing mitochondrial fission and mitophagy in middle‐aged animals correlates with lifespan extension. PUM2, an RNA‐binding protein, accumulates in skeletal muscle and brain tissue of both mice and humans and inhibits Mff RNA translation, thereby causing defects in mitochondrial fission and mitophagy 40. Depletion of PUM2 in old mice or the C. elegans homolog PUF‐8 in middle‐aged (adult day 8) worms improves mitochondrial homeostasis, mitochondrial fission, and promotes mitophagy. Furthermore, levels of PUM2 or PUF‐8 negatively correlate with lifespan. Thus, studies in various animal models exemplify the need of mitophagy at middle and old age. Given that a major role of mitochondrial fission is to facilitate clearance of defective mitochondria via mitophagy, the fluctuating importance of fission during development and midlife may be a reflection of changes in the cellular demand for mitophagy, which is already known to vary with age and is known to play a key role during development 83, 84, 85.

At the same time, fusion might play a more consistent, homeostatic role in cellular functioning as opposed to fission 82. This could explain why mutations in genes associated with mitochondrial fusion, MFN2 and OPA1, are associated with diseases such as Charcot–Marie–Tooth disease and optic atrophy, respectively, while diseases caused by mutations in DRP1 are much less known and occur far less frequently 86, 87, 88. Furthermore, new studies suggest that the roles of fission and fusion proteins might be more interlinked than previously thought. Human Fis1 has recently been shown to negatively regulate fusion by binding to MFN1, MFN2, and OPA1 and inhibiting their GTPase activity 89. This is a departure from the previous notion of Fis1 being a mitochondrial membrane anchor for DRP1 recruitment 90. In fact, emerging studies have shown Fis1 is largely dispensable for DRP1 localization to the mitochondria 91, 92. Fission and fusion proteins and pathways might thus be more interlinked than previously thought and might act via interconnected pathways to maintain mitochondrial network homeostasis.

Overall, these findings might suggest that neither mitochondrial fusion nor fission is necessarily “good” or “bad”, but that mitochondrial fusion is uniformly more critical for cellular health, whereas the importance of mitochondrial fission is connected to the temporal requirement of increased mitophagy. Consequently, mitochondrial fusion and fission are critical for specific cellular needs and hence are affected by different intracellular and extracellular signals. Furthermore, aging adds another dimension along which these signals and mitochondrial dynamics change. We will now review emerging research on factors that influence how mitochondrial shape, form, and function change with age and how that affects organismal lifespan.

Factors affecting mitochondrial dynamics with age

Mitochondrial plasticity

Many recent studies have attempted to decipher the relative importance of mitochondrial fusion and fission in aging by perturbing these processes or by testing for their requirement in various lifespan‐extending interventions. The varied results of these studies have highlighted the fact that neither process is universally pro‐ or anti‐aging. The requirements for mitochondrial fusion and fission vary by cell type 93, and the effects of perturbing these processes are often different depending upon the technique used and the organism in which it is applied.

One theme emerging from these studies is that the balance between fusion and fission may be more important than the modulation of either process in isolation. Despite the requirement of fusion for lifespan‐extending interventions 59, most studies show that driving fusion or fission alone is not sufficient to increase lifespan 25 and sometimes negatively affects health and longevity. Deletion of drp‐1, fzo‐1, or eat‐3 in C. elegans reduces median lifespan and induces motor and sensory impairments, yet pathologies associated with inhibiting one process can often be restored by inhibiting the other 82. Yeast cells deficient for fission and fusion (dnm1;mgm1 double knockout, Table 1) actually have healthy‐looking mitochondria with a wild‐type morphology, although these yeast do have a shortened lifespan 94. Alternatively, in C. elegans, initial evidence suggests simultaneous deletion of fusion and fission factors can prolong lifespan. Mitochondrial networks in the C. elegans body wall muscle cells are fused in young animals and gradually fragment with age. While deleting drp‐1 or fzo‐1 individually in C. elegans results in abnormal mitochondrial networks in the body wall muscle, simultaneous deletion of drp‐1 and fzo‐1 results in the maintenance of a fused mitochondrial network similar to that of young, wild‐type worms 25. Additionally, these double mutants have an extended lifespan compared with wild‐type worms or either of the single mutants. The mechanisms underlying this longevity effect require further investigation, including how it effects other mitochondrial stress responses such as the UPR^mt and interactions between mitochondria and other organelles.

While studies in worms and yeast may have conflicting results regarding the longevity effects of an adynamic mitochondrial network, they both highlight the importance of mitochondrial plasticity. In both organisms, although the morphology of the mitochondria of the double mutants resembles that of wild‐type mitochondria, they respond very differently to stress and changes in their environment. In worms, intermittent fasting (IF) extends lifespan and remodels the mitochondrial network as they are switched between fasted and fed states. The drp‐1;fzo‐1 double mutant does not live longer in response to IF, as these worms are unable to remodel their mitochondrial networks in response to the feeding and fasting phases 25. In yeast, the double mutant responds poorly to various stress tests 94. Together, these results indicate that eliminating mitochondrial dynamics through simultaneous disruption of fusion and fission can result in normalized mitochondrial morphology and possibly increased lifespan, but it also results in an inability to respond to stress and changes in nutrient availability.

Work by Song et al suggests that this requirement for mitochondrial network homeostasis is conserved in mammals. Inhibiting fission or fusion in mouse hearts by deletion of Drp1 or Mfn2, respectively, leads to two distinct phenotypes of mitochondrial disruption, cardiac pathology, and rapid lethality. On the other hand, promoting fission or fusion by overexpressing the corresponding gene leads to changes in mitochondrial morphology but does not lead to cardiac pathology. These findings led them to hypothesize that cardiac pathology did not entirely result from the fragmented or fused mitochondrial networks, but from the disruption of the fission/fusion balance that results from complete ablation of one of these processes. To test this, the authors knocked out Drp1, Mfn1, and Mfn2 simultaneously in adult mouse hearts. They found that the triple knockout had a slower progression of cardiac pathology and an extended lifespan compared with either the Drp1 KO or the Mfn1/Mfn2 double KO 95. This finding is supported by a separate study in mice lacking the Mff fission gene. These mice die prematurely around week 13 due to cardiomyopathy. Similarly, mutating the fusion gene, Mfn1, resulted in perinatal lethality. However, in a cohort of 10 Mff/Mfn1 double mutants, all mice were alive and healthy after being followed for over a year 96.

These studies demonstrate that maintaining a balance between fusion and fission may be more beneficial for an organism than the inhibition or promotion of either of these processes individually. Deleting the fusion and fission machinery simultaneously locks the mitochondrial network into an adynamic state which, by visualization, resembles a younger, healthier network. While this static mitochondrial network may be better equipped to maintain metabolic homeostasis throughout an organism's life, it may be less able to respond to stress and changes in nutrient availability.

Mitochondrial nutrient sensing

Remodeling of mitochondrial networks through fission and fusion is essential for organisms to respond to stress and changes in nutrient and energy availability. However, the mechanisms linking nutrient‐sensing pathways to the regulation of mitochondrial fission and fusion machinery remain largely unclear. Emerging evidence suggests that nutrient‐responsive regulators such as sirtuins, mTOR, AMPK, and insulin may all regulate mitochondrial dynamics as part of their attempt to restore homeostasis upon shifts in cellular conditions. Regulation of mitochondrial dynamics through these pathways provides a mechanism by which the mitochondrial network can respond to energy fluctuations. Importantly, these pathways are all also implicated in aging. Therefore, the regulation of mitochondria by these nutrient sensors might be compromised with age and/or targeted to promote increased lifespan or health span.

Sirtuins

Sirtuins are a family of NAD‐dependent deacetylases, which are essential for maintaining metabolic homeostasis 97. Sirtuins act in response to changes in the NAD⁺:NADH ratio and were first linked to longevity in yeast 98. Further studies indicated that sirtuins may be important for mediating the extended lifespan that results from dietary restriction 99. Although the effect of sirtuins on longevity in higher organisms has been disputed, SIRT1 and SIRT6 have been shown to promote longevity in mice 100, 101, and it is clear that this family of enzymes regulates many of the metabolic and stress response processes that are either dysregulated with age or targeted by lifespan‐extending interventions 102. Of the 7 mammalian sirtuins, 3 of them (SIRT3, SIRT4, and SIRT5) primarily localize to the mitochondria. Emerging evidence shows that, among other things, these mitochondrial sirtuins function to regulate mitochondrial dynamics in response to cellular stress and senescence.

SIRT3 is responsible for deacetylating many mitochondrial proteins. Higher levels of SIRT3 have been associated with longevity in humans, and lower levels have been associated with diabetes in humans and heart failure in mice 103, 104. Hyperacetylation of the inner mitochondrial membrane fusion protein OPA1 reduces its activity and has been associated with a variety of cardiac pathologies. Samant et al found that SIRT3 activates OPA1 through deacetylation, therefore promoting mitochondrial fusion 105.

Recent studies on SIRT4 suggest that it drives mitochondrial fusion through negative regulation of fission proteins and positive regulation of fusion proteins. SIRT4 overexpression in lung cancer‐derived cell lines results in more fused mitochondria and reduced mitochondrial fission via downregulation of DRP1 expression and inhibition of Ser⁶¹⁶ phosphorylation 106. Further evidence for SIRT4 regulation of mitochondrial dynamics comes from another in vitro study in cell lines in which SIRT4 overexpression was observed to drive mitochondrial fusion and was associated with an enrichment of L‐OPA1, the fusion‐promoting long form of OPA1, relative to S‐OPA1 107. The authors found that SIRT4 directly interacts with OPA1, but the particular mechanism remains unclear.

Finally, research suggests that SIRT5 also functions to promote mitochondrial fusion and elongation during starvation and times of energy stress. In SIRT5‐overexpressing cells, there is a decrease in mitophagy, an increase in elongated mitochondria, and an increase in levels of MFN2 and OPA1 108. On the other hand, SIRT5 knockout promotes fission and the mitochondria in these cells appear more fragmented. SIRT5 knockout results in higher levels of DRP1 and lower levels of phosphorylation of the Ser⁶³⁷ site, which is responsible for cytosolic retention of DRP1, resulting in more DRP1 available for recruitment to mitochondrial membranes. There are also higher levels of MiD51 and FIS1, which are responsible for recruiting DRP1 to the outer mitochondrial membrane 109.

Although sirtuins have previously been implicated in the regulation of mitochondria biogenesis, these studies indicate an exciting new mode of regulation through the regulation of fusion and fission proteins. Moving forward, it will be important to further determine whether the effect of sirtuins on mitochondrial dynamics is mediated through direct or indirect interactions with fusion and fission proteins.

Mechanistic target of rapamycin (mTOR)

mTOR is a kinase that promotes a variety of processes involved in cellular growth and metabolism when nutrients are abundant. Inhibition of mTOR complex 1 (mTORC1) or components involved in mTORC1 signaling results in lifespan extension in all organisms tested 110. Previous work on mTOR's effect on mitochondria has been broadly centered around mTOR's ability to inhibit mitophagy and promote biogenesis with little focus on its role in influencing mitochondrial dynamics 111. New research has revealed a possible mechanism for its regulation of mitochondrial fission.

Morita et al showed that mTORC1 drives mitochondrial fission through 4E‐BP1 translational control of the IMM fission protein mitochondrial fission process protein 1 (MTFP1). Treating mouse embryonic fibroblasts (MEFs) with mTOR inhibitors resulted in diminished translation of MTFP1, changes in the localization and phosphorylation of DRP1, and, ultimately, mitochondrial hyperfusion, branching, and circularization 24. Since mTOR is inactivated under low‐energy conditions, the idea that mTOR inhibition drives mitochondrial fusion is consistent with the hypothesis that fused mitochondrial networks are more efficient and protective of cell death during starvation. Additionally, mitochondrial fission is essential both for mitochondrial biogenesis and for mitophagy of damaged mitochondria. Therefore, when nutrients are abundant, one could speculate that mTORC1 promotes fission to permit mitochondrial turnover.

Seemingly, at odds with the idea that mTOR inhibition drives fusion, an older study suggests that mTOR stimulation by insulin promotes fusion. In mouse hearts and a human cardiomyocyte cell line, insulin increased mitochondrial fusion by increasing levels of OPA1 in an mTOR‐dependent manner 112. More research is needed in order to understand how mTOR regulates fusion and fission, but both of these studies demonstrate the ability of mTOR to respond to shifts in nutrient availability by remodeling mitochondrial networks.

AMP‐activated protein kinase (AMPK)

AMPK is an age‐related fuel sensor that is activated in low‐energy conditions. Pharmacological or genetic activation of AMPK promotes longevity, and AMPK is known to control both mitochondrial biogenesis and mitophagy 113. There is accumulating evidence for AMPK's ability to directly regulate mitochondrial fusion and fission factors, but some of the results regarding the overall outcome on mitochondrial morphology seem to be contradictory and may be context‐dependent 25, 114, 115, 116, 117.

In a mouse model of Huntington's disease that has more fragmented mitochondria at baseline, treatment with the drug metformin activates AMPK and reduces the number of fragmented and swollen mitochondria 114. Treatment of human and rat endothelial cells with an AMPK activator (AICAR or salicylate) reduced the degree of palmitate‐induced mitochondrial fission by increasing phosphorylation at the inhibitory Ser⁶³⁷ site on DRP1. Treating these same cells with an siRNA specific to AMPKα1/2 reduced Ser⁶³⁷ DRP1 phosphorylation, confirming that the ability of salicylate and AICAR to reduce mitochondrial fission is dependent on AMPK 115. MFF is also a substrate of AMPK, with AMPK activation resulting in phosphorylation at Ser¹²⁹ 117. Finally, in C. elegans, chronic AMPK activation resulted in a more fused mitochondrial network in the body wall muscle 25. These studies correlating AMPK activation and fusion seem to align with the idea that a fused mitochondrial network produces less ROS and is more efficient at producing ATP. Under stressful or low‐energy conditions known to activate AMPK, mitochondrial fission is reduced, and fusion is increased to promote mitochondrial efficiency.

However, a recent study by Toyama et al suggested that AMPK activation can directly promote fission. Human U2OS osteosarcoma cells and MEFs were treated with an ETC inhibitor (rotenone or antimycin A) or an AMPK activator (AICAR or A76992). All treatments caused AMPK activation and mitochondrial fragmentation. The authors found that AMPK phosphorylation of MFF induces DRP1 localization to the mitochondria 116. This result seems to conflict with previous studies which suggest that AMPK promotes fusion. However, there are situations in which one could rationalize AMPK‐promoting fission. In a scenario in which mitochondria were damaged and their ability to produce ATP was reduced, the AMP:ATP ratio would increase and activate AMPK. In this scenario, one cellular priority would be to remove the damaged mitochondria through fission and mitophagy. Therefore, instead of driving fusion to promote ATP production, it might be more important for AMPK to drive fission‐mediated mitophagy.

The discrepancies between the effects of AMPK activation on mitochondrial dynamics may be explained, in part, by the level and duration of AMPK activation. In the studies, which showed that AMPK activation drives mitochondrial fusion, AMPK was activated either chronically by genetic manipulation or for at least 24 h with a drug treatment. Alternatively, in the Toyama et al study, which suggested that AMPK activation drives fission, the cells were only exposed to the electron transport chain inhibitors or AMPK activators for 1 h before mitochondrial networks were measured. Perhaps, in coordination with other signaling pathways, AMPK is able to distinguish between acute and chronic stress to promote either fusion or fission.

Energy sensors such as AMPK signal through multiple pathways in order to reprogram metabolism and maintain cellular and organismal homeostasis. In addition to regulating mitochondrial fission and fusion factors directly, AMPK has the opportunity to control mitochondrial dynamics indirectly through its downstream targets. In C. elegans, activated AMPK was found to promote longevity via a cell‐nonautonomous signaling cascade involving the cyclic AMP‐responsive element‐binding protein (CREB)‐regulated transcriptional coactivator (CRTC)‐1 118. Further research on the mechanisms of this signaling pathway in aging revealed that CRTC‐1 was able to regulate mitochondrial dynamics in a cell‐nonautonomous fashion. Remarkably, mutating the AMPK phosphorylation sites on CRTC‐1 only in neurons blocks lifespan extension by constitutively active AMPK and causes mitochondrial fragmentation in muscle cells 119. This research demonstrates the ability of nutrient sensors to regulate mitochondrial dynamics through their downstream targets and in a cell‐nonautonomous manner to control organismal metabolism and aging.

Insulin/insulin‐like growth factor signaling (IIS)

The IIS pathway incorporates information about nutrient status relayed through hormonal signals to control metabolism and growth. Downregulation of various components of this pathway has been associated with increased lifespan in worms, flies, and mice 120. Similar to sirtuins, AMPK, and mTOR, research suggests that the IIS nutrient‐sensing pathway is able to influence mitochondrial dynamics. As previously mentioned in the mTOR section, insulin treatment in rat cardiomyocytes and in mice increased OPA1 protein levels, promoted mitochondrial fusion, and improved mitochondrial function. Further experiments revealed that this occurred through the IRS‐PI3K‐Akt‐mTOR‐NFκB signaling cascade, and in silico analysis revealed a possible NFκB binding site in the Opa1 promoter 112.

However, alterations in mitochondrial dynamics may not be strictly downstream of the IIS pathway. In a human‐derived diabetes‐susceptible cell line, overexpression of mitochondrial fission proteins led to decreased levels of insulin‐induced IRS1 and Akt phosphorylation. Accordingly, overexpression of mitochondrial fusion proteins or pharmacological inhibition of fission with Mdivi‐1 led to higher levels of insulin‐induced IRS1 and Akt phosphorylation 121. This study suggests that the state of the mitochondrial network influences the response to nutrient signals such as insulin and that excessive fission promotes insulin insensitivity. In addition, reduced IIS in C. elegans can still robustly increase lifespan in animals lacking either FZO‐1 or DRP‐1 25.

Despite the unclear directionality of the interaction between IIS and mitochondrial dynamics, a growing body of work has demonstrated the importance of this intersection in pathological states such as type 2 diabetes and obesity 122. Mitochondrial fission is increased in the skeletal muscle of mouse models of obesity 43, and an array of studies support the idea that mitochondrial fission is associated with insulin resistance. Tipping the balance toward fission by knocking out MFN2 caused mice to be more glucose‐intolerant and more prone to insulin resistance with age or upon feeding of a high‐fat diet 123. Multiple studies suggest that the mechanistic link between changes in mitochondrial dynamics and altered insulin signaling may be due to changes in ROS production 121, 123.

While it is becoming clear that mitochondrial dynamics are able to regulate insulin sensitivity, the consequences of fusion and fission may vary by cell type. Contrary to the idea that fission promotes insulin resistance, downregulation of DRP1 expression in mouse pancreatic beta cells resulted in reduced glucose‐stimulated insulin secretion 124. This study suggests that fission, rather than simply disrupting insulin signaling, is actually required for beta‐cell function and insulin secretion.

Finally, studies in the nutrient‐sensing proopiomelanocortin (POMC) neurons highlight a role for mitochondrial dynamics in cell‐nonautonomous regulation of insulin signaling. MFN1 knockout in POMC neurons in mice resulted in impaired glucose‐stimulated insulin secretion due to increased ROS production, highlighting an important role for fusion 125. Another study corroborated this finding by showing that adult‐onset deletion of DRP1 in POMC neurons in mice improves glucose homeostasis 126.

Overall, these studies demonstrate an important, and possibly bi‐directional, role for mitochondrial dynamics in insulin/insulin‐like signaling. Activation of the IIS pathway can remodel mitochondria. On the other hand, the state of the mitochondrial network can affect the magnitude of the cell's response to glucose or insulin. Although these studies have mostly been performed in the context of type 2 diabetes and obesity, they have broad implications for age‐related insulin resistance and metabolic disorders.

Crosstalk between nutrient sensors

Our current understanding of the effects of sirtuins, AMPK, mTOR, and insulin signaling on mitochondrial dynamics comes from only a handful of studies, all of which were performed in very different contexts. These nutrient sensors have previously been implicated in the regulation of mitochondrial metabolism, but exciting new research has demonstrated that they do this in part through direct regulation of fission and fusion machinery. However, the results of some studies seem to contradict each other, and this is further complicated by the fact that there is a lot of crosstalk between these nutrient‐sensing pathways. For example, AMPK has long been acknowledged as an upstream inhibitor of mTORC1, but some recent studies suggest that S6 kinase, a downstream target of mTORC1, may also be able to regulate AMPK 127. Additionally, insulin signaling activates mTORC1 through Akt‐mediated inhibition of TSC1/2, and mTORC1 can participate in a negative feedback loop to block insulin signaling 128. Therefore, some of the changes in mitochondrial dynamics due to AMPK activation could be due to mTOR inhibition and changes due to insulin signaling could be mediated by mTOR activation. It may be futile to define any of these sensors as pro‐fusion or pro‐fission, as their effect on mitochondrial dynamics may be influenced by multiple pathways, which provide additional information about the cellular state. Moving forward, we need studies specifically designed to measure the effect of sirtuins, AMPK, mTOR, and insulin signaling on mitochondrial dynamics while also accounting for the crosstalk between these age‐related cellular energy sensors. Finally, we should investigate whether some of the observed deregulation of fusion and fission with age is caused by the age‐related decline of the aforementioned nutrient‐sensing pathways.

Inter‐organelle communication

The cellular milieu is extremely different from textbook images depicting isolated organelles floating around in a pool of cytoplasm. An increasing number of studies have pointed out that cytoplasm has hydrogel‐like properties, where different organelles including mitochondria, lysosomes, peroxisomes, and the endoplasmic reticulum/Golgi apparatus exist in close proximity with each other and with many other cytoplasmic structures not surrounded by investing membranes 129. This cellular environment facilitates the function of organelles as units performing specific tasks and the exchange of biochemical information required for these processes 130.

In this cellular environment, mitochondria are further known to exist as a complex reticulum engaging in physical interactions, invaginations, transient contacts, vesicular transport, molecular exchange, and close proximal localization with various other organelles 131. Unsurprisingly, mitochondrial function and dynamics are both influenced by these connections with various organelles. Here, we will discuss examples of mitochondrial interactions with the ER, lysosomes, and peroxisomes to elucidate how their communication influences mitochondrial dynamics with age.

Mitochondria‐ER interactions

ER forms close physical contacts with the mitochondrial network and plays an integral role in mitochondrial fission 132. ER tubules associate with and define mitochondrial fission sites 133. While initial studies found that ER tubules wrap around mitochondria to mark the sites for DRP1 recruitment and constriction, new studies have revealed that the ER tubules can mediate DRP1‐independent constriction of mitochondria 134. Actin polymerization through ER‐bound inverted formin 2 (INF2) results in force generation via myosin and initial constriction of mitochondria, which further enhances DRP1 oligomerization and DRP1‐dependent constriction 134. Charcot–Marie–Tooth disease (CMTD), most commonly caused by mutations in MFN2, can also result exclusively from mutations in ER‐bound INF2 135 (Fig 3). The hallmark of CTMD is the presence of dysfunctional and fragmented mitochondria. CMTD‐associated mutations in IFN2 map to the autoinhibitory domain indicating that the mutant proteins are unresponsive to autoinhibition, driving increased mitochondrial constriction, DRP1 localization, and mitochondrial fragmentation 135. Given that mutations in ER‐specific proteins, such as INF2, can directly cause dysfunctional mitochondrial dynamics, it is likely that age‐related changes in ER functions contribute to mitochondrial dynamics dysregulation as well.

Figure 3. Mitochondria‐ER crosstalk.

Mitochondria and ER exist in close proximity with each other forming mitochondria‐associated membranes (MAMs). MAMs are enriched with mitochondria‐specific proteins such as voltage‐dependent anion channel (VDAC) and mitofusin 1/2 (MFN), ER‐specific proteins such as sarco/endoplasmic reticulum Ca²⁺‐ATPase (SERCA), inositol 1,4,5‐trisphosphate (IP3R), calnexin, and calreticulin. Tethering proteins such as glucose‐regulated protein 75 (GRP75) stabilize MAM structures, while proteins such as FUNDC1 translocate to MAMs. MAMs allow efficient Ca²⁺ homeostatic exchange between ER and mitochondria through IP3R VDAC mitochondria calcium uniporter (MCU) and permeability transition pore (PTP) and SERCA. In addition to MAMs, ER forms direct contacts with mitochondria and mark sites for mitochondrial fission and ER‐specific inverted formin 2 (INF2) can cause actin‐dependent constriction of mitochondria to initiate mitochondrial fission. Several proteins involved in mitochondrial fission such as mitochondrial fission factor (MFF) and mitochondrial fission 1 protein (FIS1) mediate dynamin‐related protein 1 (DRP1) localization on ER suggesting that additional unidentified pathways by which ER controls mitochondrial dynamics exist.

In addition to ER‐specific proteins regulating mitochondrial dynamics, a subpopulation of the tail‐anchored proteins MFF and FIS1 which are receptors for DRP1 recruitment on mitochondria also localize to the ER 136. MFF1 and FIS1 subsequently assemble DRP1 oligomers on ER, from where they can be transferred to the mitochondria or peroxisomes. Additionally, fractions of ER‐localized MFF can also be transferred to the mitochondria or peroxisomes suggesting that ER can function as a platform for MFF‐mediated DRP1 oligomerization 136.

Why these fission proteins need to be shared with the ER and what role their subsequent transfer to mitochondria plays in fission remains unanswered. Because MFF and DRP1 punctae show both fast and slow transfer dynamics to the mitochondria from the ER, one possibility is that by sequestering a fraction of these fission proteins, ER might normalize mitochondrial fission events if fission rates are high. Conversely, ER might function to increase DRP1‐independent constriction of the mitochondria if fission rates are low. Indeed, reduced mitochondria‐ER contacts have been reported in aged cells, suggesting that with age ER regulation of mitochondrial dynamics might decline and contribute to age‐associated imbalance in mitochondrial dynamics 137. Furthermore, in addition to fission proteins, MFN2 but not MFN1 was shown to localize to ER 123. Studies suggest that ER‐localized MFN2 regulates ER‐mitochondria contact either via homotypic interactions with MFN2 or heterotypic interactions with MFN1 138. Age‐associated decreases in MFN2 might also reduce ER‐mitochondria contacts resulting in dysregulation of several processes that ER and mitochondria regulate cooperatively, such as Ca²⁺ buffering and mitophagy.

Cellular Ca²⁺ buffering is coordinated by ER and mitochondria. ER and mitochondria exchange calcium fluxes through ER‐Inositol 1,4,5‐trisphosphate receptor (IP3R) Ca²⁺ release followed by mitochondrial calcium uniporter (MCU)‐dependent Ca²⁺ uptake. Exchange also occurs conversely by mitochondria permeability transition pore (PTP) Ca²⁺ release followed by ER sarco/endoplasmic reticulum Ca²⁺‐ATPase (SERCA) Ca²⁺ uptake 139, 140 (Fig 3). It is known that this Ca²⁺ exchange dynamic declines with age. Aged cells show decreased mitochondrial Ca²⁺ uptake secondary to ER release 141. Since ER‐derived cytoplasmic Ca²⁺ in the cytoplasm can regulate the activity of dynamin and promote mitochondrial translocation of DRP1 and enhanced fission 142, these results suggest that decreased Ca²⁺ sequestration in aged cells might further contribute to mitochondrial fragmentation.

Mitophagy is another key process where ER‐mitochondria proximity plays an important role. A mitophagy receptor protein, FUN14 domain‐containing 1 (FUNDC1), specifically localizes to mitochondria‐associated membranes (MAMs) by associating with the ER membrane protein calnexin 143. During mitophagy, FUNDC1 interacts with DRP1 and recruits it to MAMs to initiate mitochondrial fission thereby integrating mitochondrial fission for efficient mitophagy 144, which is positively regulated by the mammalian autophagy regulator Unc‐51‐like kinase (ULK1) 145. Age‐related decreases in ER contacts could thus also decrease MAM associations resulting in a decrease in mitophagy, which has been reported with age 85. A decrease in mitophagy would then further contribute to age‐related mitochondrial dysfunction.

In addition to processes regulated in unison by ER and mitochondria, ER‐specific unfolded protein response (ER‐UPR) can cause changes in mitochondrial dynamics as well 146. ER‐UPR is triggered by unfolded proteins and increased protein misfolding, both of which increase with age 147. PERK (protein kinase RNA‐like endoplasmic reticulum kinase) is activated in response to ER‐UPR and activates eIF2α to initiate translation attenuation to combat ER‐UPR 148. A recent study shows that PERK‐dependent eIF2α activation also remodels mitochondrial morphology by a process called stress‐induced mitochondrial hyperfusion (SIMH) 149. Increased mitochondrial fusion increases ATP production and metabolic efficiency to combat ER‐UPR. It is known that cellular ability to combat stress declines with age 150. Consistent with this, PERK mRNA has been shown to significantly decline in the hippocampus of aged rats 151. Furthermore, eIF2α protein levels, phosphorylation, and activity downstream of PERK are known to decrease in various tissues from aged rats 152. Together, these results indicate that with age, cells’ ability to increase mitochondrial fusion to tackle stress decreases with age, potentially contributing to age‐dependent mitochondrial fragmentation.

Stress caused by the UPR orchestrates a series of cellular events that culminate in changes in both mitochondrial dynamics and lysosomal function. ER stress‐induced PERK activation results in increased lysosomal amino‐acid sensing and autophagy via the transcription factor ATF4 153, while mitochondrial biogenesis is induced by the ATF6 target gene PGC‐1α 154. Together, these studies highlight that crosstalk between ER–mitochondria–lysosome axis and their malleable response to stress via altering dynamics and function is key to tackle cellular UPR stress 155.

Mitochondria and ER exist thus in close association. They share several proteins from the mitochondrial fission/fusion machinery, coordinately regulate Ca²⁺ signaling, mitophagy, and stress responses, and bi‐directionally affect each other's function. While MFF, DRP1, and MFN2 colocalize to both organelles, exactly what mitochondrial dynamic processes they facilitate under which conditions remain a subject of speculation. What is clear is that this shared machinery does allow ER to exert apparent effects on mitochondrial morphology and as ER‐mitochondria association changes with age, so do these effects. Furthermore, processes requiring ER involvement which directly or indirectly affect mitochondrial shape such as Ca²⁺ buffering, mitophagy, and ER‐UPR are all dysregulated with age indicating that ER dysfunction can itself contribute to mitochondrial dynamic dysregulation. Consistent with the importance of ER‐mitochondria encounter structures, ERMES (mitochondria‐ER contact sites, MAM equivalents) are also decreased in senescent yeast cells suggesting that maintenance of ER‐mitochondria proximity might be effective in conservation of organelle function 137. Whether shared mitochondrial and ER functions and dynamics co‐evolved and whether their interactions can be altered to decrease age‐related organellar dysregulation await further scientific attention.

Mitochondria–lysosome interactions

Similar to mitochondria being synonymous with “powerhouses”, lysosomes have been over‐simplified as the cell's “recycle bin”. Recent research has challenged this concept, and we now know that lysosomes perform many essential roles in nutrient sensing, metabolic adaptation, protein quality control, and mTOR and AMPK signaling cascades 156, 157, 158, 159, 160. Accordingly, lysosomal activity and signaling have important implications on several organelles, including mitochondria. Many lysosomal storage diseases caused by mutations in lysosomal hydrolases exhibit aberrant mitochondrial networks 161. Furthermore, alterations in mitochondria and lysosomes are often seen concurrently in neurodegenerative diseases, suggestive of close functional links between the two organelles 7, 162, 163. Recently, close physical interactions between mitochondria and lysosomes have also been defined, as a key determinant of mitochondrial dynamics.

Links between mitochondrial dynamics and lysosomal function started emerging with observations in yeast which showed that reduced vacuolar acidification, often observed in aged cells, also triggered aberrant mitochondrial dynamics 164. Hughes et al observed increased mitochondrial fragmentation as budding yeast cells age. This fragmentation concomitantly occurs with a reduction in vacuolar pH (vacuoles being the yeast equivalent to lysosomes). Increasing vacuolar pH by expression of the subunit A of yeast V‐ATPase V1 domain (Vma‐1) or vacuolar ATPase assembly integral membrane protein (Vph‐2) prevented age‐dependent mitochondrial fragmentation and also increased yeast lifespan 164. This rescue of mitochondrial fragmentation was specific to V‐ATPase activity because treatment with concanamycin A, a specific inhibitor of vacuolar V‐ATPase, prevented rescue of mitochondrial fragmentation. Furthermore, DR prevented mitochondrial fragmentation with age by preserving vacuolar acidity in aged yeast. This study suggested that part of DR's effect of remodeling mitochondrial dynamics to extend lifespan could be intrinsically wired through lysosomal function.

This wiring of mitochondrial dynamics and lysosomal function is possibly explained through the mitochondrial transcription factor A (TFAM)–transcription factor EB (TFEB) link. TFAM is a nuclear‐encoded mitochondrial transcription factor, which binds to and regulates mtDNA transcription, maintenance, replication, and repair 165. Its activity is essential in ensuring mtDNA replication for mitochondrial biogenesis and transmission of mtDNA to daughter mitochondria 166. Similarly, TFEB controls lysosomal biogenesis through the coordinated lysosomal expression and regulation (CLEAR) gene network 167, 168. During lysosomal stress or autophagy‐inducing conditions, cytosolic TFEB translocates to the nucleus and activates genes promoting lysosomal biogenesis and function 169. Recently, it was shown that in TFAM‐deficient cells, mitochondrial morphology is highly punctate and fragmented due to gross mitochondrial dysfunction 170. Sensing increased need for mitophagy to replace damaged mitochondria, TFEB translocates to the nucleus in TFAM^−/− cells and not only upregulates lysosomal biogenesis genes and increases lysosomal number 170 but also stimulates mitochondria biogenesis by activating the mitochondrial master‐regulator peroxisome proliferator‐activated receptor‐gamma coactivator 1‐α (PGC1‐α) α a known direct target of TFEB 171, 172.

The biogenesis of mitochondria and lysosomes is thus interconnected. Indeed HLH30, the C. elegans homolog of TFEB has been shown to regulate lifespan in various longevity paradigms such as germline‐deficient glp‐1 mutants, dietary‐restricted eat‐2 mutants, insulin receptor daf‐2 mutants, and TOR pathway rsks‐1 mutants, several of which require mitochondrial dynamics remodeling, potentially through master regulators such as TFEB 173.

In addition to linked biogenesis networks, evidence has emerged that direct physical contacts between lysosomes and mitochondria regulate mitochondrial dynamics. Using electron microscopy and live‐cell imaging, Wong et al discovered that mitochondria–lysosome contacts form dynamically in healthy cells 174. These contacts are distinct from mitophagy‐induced contacts of damaged mitochondria and lysosomes. Contacts sides are tethered by lysosomal RAB7 GTPase and untethering is mediated by recruitment of TBC1D15, which drives RAB7 GTP hydrolysis, via FIS1. Two main findings from this study implicate a direct role of lysosomal contacts in mitochondrial dynamics. Firstly, expression of a constitutively active form of RAB7 GTPase dramatically increased lysosome–mitochondria contacts but significantly decreased mitochondrial fission suggesting that whether or not lysosomes and mitochondria associate has a direct consequence on mitochondrial fission rate. Secondly, the involvement of mitochondrial fission protein FIS1 in recruitment of the untethering protein TBC1D15 implicates that mitochondrial fission proteins might also act as regulatory proteins monitoring contacts with other organelles. Interestingly, the respective tether proteins on mitochondria have not been identified yet, but it is highly likely that mitochondrial fission or fusion proteins are involved in the interaction with the lysosome. Because mitochondrial dynamics are known to change with age, age‐dependent decrease in lysosomal function and biogenesis could be implicated in reduced lysosomal–mitochondria contacts, which would normally act to decrease fission rates.

Future studies are required to fully investigate the role of lysosomal function and lysosomal–mitochondria contacts in mitochondrial dynamics, cellular metabolism, and age‐related organellar dysfunction. Interestingly, C. elegans HLH30 chemical agonists have been shown to extend lifespan in worms by maintaining lysosomal acidity and function 175. It would be critical to use these models of improved lysosomal function to investigate corresponding effects on mitochondrial dynamics, metabolism, mitophagy, and age‐related mitochondrial dysfunction.

Mitochondria–peroxisome interactions

An exciting area of inter‐organelle communication in the regulation of aging is the mitochondria and peroxisome connection. Peroxisomes are ubiquitous, multifunctional, and dynamic organelles that are indispensable for organismal health and viability. They perform key functions related to lipid and ROS metabolism that are essential for cellular homeostasis and signaling. Yet their role in complex diseases, such as aging, has largely been undermined due to the common misconception these functions are ancillary or redundant to those of mitochondria. Like mitochondria, peroxisomes are dynamic organelles that modulate their morphology, abundance, and function to meet fluctuating energetic demands or in response to environmental stimuli. Their size and appearance often vary among different tissues, ranging from 0.1 to 1 um in diameter. Intriguingly, the regulation of peroxisome membrane dynamics is intimately tied to mitochondrial dynamics with shared fission and biogenesis mechanisms 176. The mechanisms of peroxisome fission are discussed in detail elsewhere 177. Here, we will briefly focus on the shared mechanisms of peroxisome and mitochondrial fission and summarize the evidence of their physical or functional interaction with mitochondria. We will highlight the studies implicating how peroxisome and mitochondrial dysfunction impinge on each other during aging.

Peroxisome membranes are remodeled by a coordinated cycle of elongation and fission. The mechanism of membrane scission largely parallels that of mitochondrial fission and requires DRP1 178. DRP1 ring structures are found at sites of peroxisome membrane constriction, where GTP hydrolysis is required for DRP1 polymerization and membrane scission 179. The C‐terminal‐anchored membrane adaptor proteins FIS1 and MFF are also found on peroxisome membranes and interact with DRP1 and PEX11 to regulate peroxisome fission 180, 181, 182 (Fig 4). Membrane constriction is necessary for Drp1‐mediated fission, but the role of the ER or actin assembly at sites of peroxisome membrane division is currently unclear. PEX11β, a member of the highly conserved PEX11 protein family known to regulate peroxisome size and number 183, accumulates at the constriction sites, where it functions in the early fission phase. PEX11β initiates the formation and elongation of nose‐like protrusions from the spherical membrane. It has been reported that PEX11β itself is also able to generate sites of constriction along these protrusions, and contributes to the recruitment and activity of DRP1 184, 185. Although PEX11β does not have intrinsic membrane scission activity, loss of Pex11β results in abnormal peroxisome morphology that appears enlarged and elongated. A study in yeast Saccharomyces cerevisiae also suggested that PEX11 interacts with the ERMES complex component MDM34 to establish contact sites between peroxisomes and mitochondria 186. Manipulation of Pex11 in Drosophila and C. elegans (prx‐11) was linked to lifespan regulation in a study where knockdown of several Pex genes involved in peroxisome biogenesis led to reduced levels of endogenous ROS and increased lifespan 187.

Figure 4. Peroxisomes are dynamic organelles that share fission machinery with mitochondria.

Peroxisomes modulate their morphology, abundance, and function through coordinated cycles of growth, elongation, and fission. The regulation of peroxisome membrane dynamics is intimately tied to mitochondrial dynamics, as the mechanism of peroxisome membrane fission largely parallels that of mitochondrial fission. The protein machinery controlling mitochondrial fission, namely DRP1, FIS1, and MFF, is also required for peroxisome membrane fission. FIS1 and MFF are C‐terminal‐anchored membrane adaptor proteins that facilitate recruitment of and interaction with DRP1 and PEX11 at sites of fission. PEX11 is required for the initial phase of peroxisome fission, where it initiates the formation and elongation of nose‐like protrusions from the spherical membrane. Pex11 has also been reported to generate sites of constriction along these protrusions, and contributes to the recruitment and activity of DRP1. Final scission of peroxisome membranes is catalyzed by assembly of DRP1 ring structures and GTP hydrolysis.

Since regulation of mitochondrial and peroxisome dynamics are intimately coupled through their overlapping mechanisms of membrane fission, it is important to consider that manipulation of these factors will affect the morphology, and likely function, of both organelles. Studies that implicate a role for mitochondrial fission in aging by targeting DRP1 could, in fact, be due to peroxisome fission. Limited work has been done to delineate the contribution of mitochondrial and peroxisome fission to aging. In yeast S. cerevisiae, inhibiting fission via deletion of Fis1 (present on both mitochondria and peroxisomes) increases chronological lifespan, and there is evidence to suggest that inhibition of peroxisome fission is a major contributor to these longevity effects 188, 189. In cells containing a Fis1 deletion background, the authors reintroduced a Fis1‐Pex15 fusion construct that was directed specifically to peroxisomes. This effectively rescued peroxisome fission, but not mitochondrial fission, yet was sufficient to abolish longevity of the Fis1 deletion mutant 188. Moreover, mitochondrial shape and function can be influenced by peroxisomal ROS and metabolites. Localized oxidative damage to peroxisomes and increased intraperoxisomal ROS production causes mitochondrial fragmentation and can lead to cell death 190, 191. Early observations reported the peroxisomal ROS‐detoxifying enzyme, catalase, to decrease in old rodent livers and human cells 192, and loss of catalase causing an accelerated aging phenotype in C. elegans 193.

Several recent studies have continued to elucidate how peroxisomes play a critical role in the aging process, through independent or interacting functions with mitochondria [25, 188, 193, 194,195]. In work defining how manipulation of mitochondrial dynamics by DR and constitutively activated AMPK promotes longevity in C. elegans, interplay between mitochondrial and peroxisome metabolism was shown to be essential 25. In this study, direct manipulation of mitochondrial dynamics by simultaneous inhibition of fusion and fission induced longevity and revealed that peroxisome dynamics and very‐long‐chain fatty acid metabolism were key players in this mechanism. This seems to hold true in yeast as well, where the supply of acetyl‐CoA by peroxisomal β‐oxidation helped prevent mitochondrial network fragmentation and contributed to DR‐induced extension of chronological lifespan in yeast 196. In C. elegans, peroxisome matrix protein import was necessary for the longevity‐promoting effects of DR and constitutively activated AMPK, as RNAi knockdown of PRX‐5 (homologous to mammalian PEX5, involved in recognition of the C‐terminal peroxisome targeting sequence) abolished DR‐ and AMPK‐mediated longevity. RNA‐seq analysis also revealed that DR increases global peroxisome gene expression 25. This is interesting because with normal aging, it was reported that peroxisome proteins are among the most significantly downregulated in the C. elegans proteome 197, suggesting that these longevity‐promoting interventions may work in part by reversing age‐associated decline of peroxisome function. Even more promising is the likely conserved role of peroxisomes in aging. A study of calorie restriction (CR) in Rhesus monkeys similarly found that even short‐term CR increases peroxisome gene and protein expression in the liver 198.

Communication between mitochondria and peroxisomes is facilitated through physical membrane contacts and vesicular trafficking. Currently, little is known about how these specific mechanisms of inter‐organelle communication are involved in aging pathways, but it is an exciting area of future exploration. As previously mentioned, the peroxisome membrane elongation and fission protein PEX11 may be involved in forming membrane contact sites with mitochondria through its interaction with MDM34 186. The yeast mitofusin FZO1 and PEX34 were also identified as potential mitochondria–peroxisome tether proteins in a recent split Venus‐based screen of contact sites in yeast 199. Finally, a better understanding of how mitochondria‐derived vesicles (MDVs) function in age‐related processes is warranted, as MDVs have been shown to mediate vesicular transport between mitochondria and peroxisomes 200, 201. Given rising evidence for mitochondria–peroxisome interactions at the level of lipid metabolism, ROS signaling, and protein exchange, it should be speculated that mechanisms of inter‐organelle communication via direct membrane contact sites and MDVs have critical implications for aging and ultimately could be targeted to improve cellular health.

Inter‐species interactions: microbiome–mitochondria communication

Mitochondria communicate not only with other organelles but also with the microbiome via a bi‐directional signaling axis. Microbiome modulates host aging through various direct and indirect signaling mechanisms 202. Recently, more studies have started investigating how bi‐directional communication between gut microbiota and host–mitochondria can alter mitochondrial dynamics and affect longevity. Key effectors of mitochondrial function and longevity, such as AMPK and PGC1‐α, are modulated by metabolites produced by the gut microbiome 203. A major flow of metabolic signals from the gut microbiome to the host is through short‐chain fatty acids (SCFAs) and molecules comprised of SCFAs 204. SCFAs are the primary metabolites produced by the gut microbiota through the fermentation of nondigestible carbohydrates (NDCs) in humans 205. Butyrate and acetate, the two major SCFAs, can be converted into acetyl‐CoA to fuel mitochondrial tricarboxylic acid cycle and have been shown to affect mitochondrial biogenesis, metabolism, and dynamics 206.

Butyrate increases the expression of the mitochondrial master regulator, PGC1‐α α, to induce mitochondrial biogenesis and oxidative metabolism in cells and in skeletal muscle of C57BL/6 mice 207. Furthermore, butyrate induced the expression of genes involved in both mitochondrial fission (Drp‐1, Fis‐1) and mitophagy (Pink‐1, Lc3, Pten), in vitro, suggesting that SCFAs produced primarily from gut microbiome can fine‐tune mitochondrial function, biogenesis, and dynamics 208.

Acetate, the second major SCFA, is a key component of the bacterial colanic acid (CA) polysaccharide, which has recently been found to modulate host mitochondrial dynamics to extend lifespan in different species. Han et al conducted a genome‐wide search to identify E. coli genes that affected C. elegans lifespan 41. E. coli are the main gut‐dwellers for the nematode C. elegans 209. They identified 5 key bacterial genes, deletion of which extended C. elegans lifespan. These 5 genes inhibited RcsA, the central activator of CA biosynthesis. Thus, their deletion led to CA overproduction in C. elegans gut. CA was shown to extend C. elegans lifespan in part by inducing increased mitochondrial fission in the gut. These results provided a direct link between host mitochondrial dynamics, microbiome communication, and longevity.

Another study investigating the environment–microbiome–mitochondria axis corroborated that the microbiome directly influences host mitochondrial dynamics, metabolism, and physiological outcomes. Lin et al found that E. coli grown in nutritionally scarce medium and used as food for WT C. elegans strains induced twofold higher fat accumulation in the worms as compared to E. coli grown in nutrition‐rich medium 210. Consistent with higher fat accumulation, C. elegans fed E. coli grown in scarce medium were more resistant to starvation. Furthermore, both these metabolic adaptations were induced by a deprivation of a single amino, led to a decrease in host nuclear hormone receptor (NHR) signaling, and were mediated by increased mitochondrial fragmentation in the gut 210. Although the direct consequences of these mitochondrial changes on C. elegans lifespan were not observed in this particular study, it is possible that known pro‐longevity interventions such as DR might do so via microbial modulation of host metabolism.

In fact, this exact phenomenon has been observed for the lifespan‐extending drug metformin. Metformin treatment significantly increases C. elegans lifespan in an AMPK‐dependent manner 211. Interestingly, this effect is mediated by metformin's effect on the gut bacteria and not the host itself. Metformin suppresses folate metabolism specifically in E. coli and decreases methionine content of the bacteria 212. This suppression of methionine content is key to metformin‐induced longevity in C. elegans. On similar lines, several studies have shown that DR and exercise in different species significantly alter the host microbiome 213, 214, 215, 216. It might be possible then that in addition to metformin, several other pro‐longevity interventions might exert their effect through microbial modulation of host mitochondrial dynamics and metabolism. In a recent study, the treatment of C. elegans with the biguanide phenformin, which indirectly activates AMPK, resulted in the maintenance of youthful mitochondrial network morphology and reduction in mitochondrial content 25. Lifespan‐extending effects of both metformin and phenformin have been studied in C. elegans and rodents 113. It is likely that similar to metformin, effects of phenformin might be mediated at least in part through alteration of microbiome metabolism. Additionally, age‐associated changes in the microbiome, occurring across all species, might be a yet unrecognized cause for age‐associated changes in mitochondrial dynamics 217. The complexity of the microbiotic component therefore should not be ignored when investigating the effect of mitochondrial dynamics alterations in aging.

The gut–microbiota communication may be bi‐directional, and mitochondria could in principle communicate with their ancient counterparts. Indeed, a recent study demonstrated that mutations in mtDNA altered the composition of gut microbiota in mice 218. A mutation in the mitochondrial Atp8 gene changed the levels of oxidative phosphorylation and glycolysis, ultimately altering the metabolite composition of host cells, which appeared to shift the microbial composition in the gut 218. It is thus becoming evident that a bi‐directional communication axis between mitochondrial function, including mitochondrial dynamics, and the microbiome exists. However, whether or not changes in mitochondrial dynamics can induce changes in host microbiome, and to what extent, warrants further investigation.

Concluding remarks

Although aging is a conserved biological process, we are only beginning to identify the precise molecular mechanisms underlying it. The task is made challenging because several complex and interconnected molecular pathways regulate cellular and organismal aging. Among these pathways, ones that bring about mitochondrial dysfunction are thought to be central to the aging process. As cells age, mitochondrial dysfunction manifests in several forms, including—but not limited to—mtDNA mutations, increased ROS production, and loss of membrane potential. Recently, we have started to appreciate that many of these mitochondrial processes are regulated by something quite ordinary, their shape.

Mitochondria possess the ancestral prokaryotic ability to divide, thereby not only altering their morphology but also impacting several cellular processes including aging. While the specific molecular mechanisms regulating mitochondrial fission and fusion have been identified, precisely how these changes in mitochondrial morphology contribute to aging remains unclear (see Box 1). We have recently begun to identify how the precise molecular mediators of fission, fusion, and trafficking, such as DRP1, MFNs, and MIRO, alter with age, thereby affecting mitochondrial dynamics. Furthermore, studies have emerged showing that these molecular mediators can be manipulated to understand the effect of a particular mitochondrial dynamic state on cellular homeostasis and longevity.

Gathered from emerging research, the studies discussed in this review highlight that not one specific mitochondrial morphology state is pro‐ or anti‐aging and that context is key. While initial findings in yeast cells causally linked fragmented mitochondria to aging, the exact opposite was observed in the fruit fly, i.e., fragmented mitochondria prevented aging. Furthermore, studies from the nematode C. elegans have shown that inhibiting either mitochondrial fission or fusion reduces median lifespan and is detrimental to the mechanisms involving DR‐mediated longevity 25, 82. Perhaps, rather than promoting either mitochondrial fission or fusion, processes that maintain mitochondrial network plasticity and allow mitochondria to function in a “youthful” state might prevent age‐associated mitochondrial dysfunction. Because both fission and fusion are central to two key mitochondrial homeostatic processes—biogenesis and metabolic efficiency, restoration of both these processes might result in improved mitochondrial health and increased lifespan.

As our understanding of mitochondrial fusion and fission evolves, it is becoming increasingly important to have tools to measure the morphology and function of mitochondrial networks. Methods for characterizing mitochondrial morphology or quantifying the degree of fusion or fission in a mitochondrial network are under‐developed and subject to bias or effects of overexpression‐based reporters. Additionally, many studies, which report the morphology of mitochondria and the state of the network, do not pair these observations with data on the metabolic function of these networks. A study from Byrne et al highlights the fact that morphology does not always correlate with function. Upon disrupting fusion and fission factors in C. elegans, the authors characterized mitochondrial morphology in different tissues and measured muscle and neuronal function of each mutant. They found that mitochondrial morphology was significantly different from wild type in the drp‐1 mutant, but these worms had no neuronal deficits as measured by a light touch assay. Conversely, fzo‐1 and eat‐3 mutants displayed deficits in their neuronal sensing before any change in mitochondrial morphology was detected 82. Future research on mitochondrial dynamics and their role in healthy aging will be strengthened by more robust characterization of mitochondrial networks and pairing these observations with data on cellular metabolism and tissue functionality.

While it has long been perceived that mitochondrial dysfunction precedes age‐related cellular decline, studies now show that this hypothesis is not unequivocally true. Homeostatic cellular processes such as autophagy might require specific changes in mitochondrial dynamics with age to enhance clearance of damaged organelles. Similarly, compromised age‐dependent Ca²⁺ and ROS homeostasis might require mitochondrial network restructuring to counteract these changes instead of emerging from dysfunctional mitochondria. Because multiple cellular processes are dysregulated with age, it is difficult to define causal versus casual changes that affect mitochondrial network structure.

Nevertheless, an increasing number of studies highlighting specific molecular pathways and age‐related factors that affect mitochondrial dynamics support the idea that altered mitochondrial dynamics are not just a passive consequence of aging. Key molecules involved in mitochondrial dynamics have been identified that can be effectively targeted to increase or decrease organismal health span and lifespan providing evidence for a causal role in healthy aging. Future studies testing conserved mitochondrial dynamic mediators that can be effectively targeted to promote healthy aging will be key for developing effective molecular gerontology therapeutics.

Conflict of interest

The authors declare that they have no conflict of interest.

Box 1: In need of answers.

1. Which factors cause altered expression of mitochondrial fission and fusion proteins during aging? Are these factors genetic, epigenetic, nutritional, or environmental?

2. Is increased mitochondrial fission with age tied to the increased cellular demand for mitophagy to remove accumulated damaged mitochondria?

3. Is maintaining mitochondrial dynamic plasticity important over promoting either fission or fusion?

4. Is some of the observed deregulation of mitochondrial dynamics with age caused by the age‐related decline in efficiency with which nutrient‐sensing pathways are regulated by AMPK, TOR, and sirtuins?

5. Why are key mitochondrial fission and fusion proteins localized to both ER and mitochondria? Are they mis‐localized or part of an ER‐regulated mitochondrial dynamic process?

6. How do mitochondria–lysosome contact sites alter mitochondrial morphology? Why do lysosomal disorders often present abnormal mitochondrial morphology?

7. Are shared peroxisomal fission proteins and metabolic pathways dysregulated in an age‐dependent manner similar to mitochondria? Can they be specifically targeted to mitigate age‐dependent mitochondrial dysfunction?

8. What pro‐ and anti‐aging effects of nutrition and environment are mediated through microbiome and mitochondria crosstalk?

9. What conserved effectors of mitochondrial dynamics can be successfully targeted in higher organisms to increase lifespan and promote healthy aging?

EMBO Reports (2019) 20: e48395