Scheffler Immo E. Mitochondria

Подождите немного. Документ загружается.

THE DYNAMICS OF MITOCHONDRIAL MORPHOLOGY 41

membranes will cause mitochondria to grow, and eventually there must be a

signal for such a mitochondrion to divide, although insertion of lipids and

proteins into the inner membrane alone may simply increase the number of

cristae. Mitochondrial ﬁ ssion appears a priori to be an essential event in pro-

liferating cells, since mitochondria have to multiply and ultimately be divided

between daughter cells. A surprising ﬁ nding is that mitochondria and peroxi-

somes share a signiﬁ cant number of components of their division machinery

(60) . Some spectacular electron micrographs (24) showing bilaminar septa

during progressive stages of mitochondrial divisions leave one with little doubt

about the overall mechanism, but they raise new questions about the control

of this process which remain largely unanswered. Since an inner and an outer

membrane are involved, the process is not a simple pinching - in - half of a mem-

brane vesicle, and it may be more analogous to the division of a bacterium,

where the plasma membrane and the cell wall represent separate structures

surrounding the cytosol. Microscopic observations suggest that ﬁ ssion is pre-

ceded by stretching, but constrictions may appear at one site transiently and

disappear without the initiation of ﬁ ssion (34) .

There is an alternative view of the division of mitochondria, based on

observations with a select number of organisms. In the unicellular red alga

Cyanidioschyzon merolae a structure referred to as an MD ring (mitochon-

drion dividing apparatus) was observed. This ring - like structure was found on

the outside of mitochondria, and it was interpreted to function in a manner

analogous to the contractile ring in cytokinesis (61) . The authors claimed

that cytochalasin B inhibited “ mitochondriokinesis, ” and they deduced the

presence of actin ﬁ laments in the ring. However, these ﬁ laments could not be

stained with anti - actin antibodies or with phalloidin. Because such rings have

not been observed in any other organism, the generality and signiﬁ cance of

these observations must remain in doubt.

The ﬁ ssion mechanism in yeast requires a complex of at least three major

proteins that is localized on the outer mitochondrial membrane, with major

domains exposed to the cytosol. Fis1p (hFis1p in humans) is an integral mem-

brane protein with a single transmembrane domain. Its C - terminus is anchored

in the outer membrane, and its N - terminal domain in the cytosol contains an

array of six - helices forming a tetratricopeptide - like fold with a binding pocket

for interactions with either of two closely related components, Mdv1p and

Caf4p. Structural studies suggest that these can form homo - oligomeric struc-

tures. Fis1p is distributed uniformly on mitochondria and required for the

association of Mdv1p (or Caf4p) on the mitochondrial surface. Together, Fis1p -

Mdv1p (in yeast) assist in the incorporation of the dynamin - related GTPase,

Dnm1p, in a ﬁ ssion complex that can be visualized by punctate staining on

mitochondria. The assembly pathway and its regulation remain to be eluci-

dated in detail. A signiﬁ cant amount of information has, however, been accu-

mulated on dynamin and dynamin - related proteins (DRP1 or DLP1 in humans)

(62 – 64) . Dnp1 behaves like a soluble protein when puriﬁ ed, and it contains a

GTPase domain at the N - terminal and a C - terminal GTPase effector domain.

42 STRUCTURE AND MORPHOLOGY. INTEGRATION INTO THE CELL

As ﬁ rst recognized in the study of endocytosis, the protein assembles at the

location of the constriction where a membrane vesicle buds from its parent

membrane. A plausible and favored model suggests that the dynamin or

dynamin - like proteins aggregate in a ring - like structure at the site of the mem-

brane constriction and ultimately pinch off the vesicle. A tubular structure like

a mitochondrion could be pinched into two pieces at such a site. GTP hydro-

lysis represents a likely driving force, but a detailed mechanism is still elusive.

For example, it is not clear whether constriction is achieved by a contraction

of a dynamin ring due to conformational changes, or whether the ring con-

tracts with the concomitant loss of dynamin monomers (62) . Further specula-

tions in the case of mitochondria include the formation of an initial, local

constriction with the help of the inner - membrane protein Mdm33p, or a

stretching of mitochondria by molecular motors interacting with cytoskeletal

elements (59) . Clearly, the mechanism must include the outer and inner mem-

branes in a concerted fashion, and there must be signals for when and where

to assemble the ﬁ ssion apparatus.

3.3.1.2 Fusion Collisions brought about by the movement of the organ-

elles may lead to fusions, especially when at least one tip makes contact with

TABLE 3.1 Genes/Proteins Involved in Mitochondrial Fusion and Fission (59)

Yeast

Gene/Protein

Human

Homologue

Localization in

Mitochondria Function

Fzo1 Mnf1/2 Outer membrane (IMP) Fusion

Mgm1 OPA1 IMS or inner membrane

(IMP?)

Fusion

Ugo1 Outer membrane (IMP) Fusion

Mdm30 Cytosol/outer membrane Fzo1 degradation

Pcp1 hPARL Inner membrane (IMP) Mgm1p processing

Dnm1 Drp1/DLP1 Cytosol/outer membrane

(peripheral)

Fission

Fis1 hFis1 Outer membrane (IMP) Fission

Mdv1 Cytosol/outer membrane

(peripheral)

Fission

Caf1 Cytosol/outer membrane

(peripheral)

Fission

Mmm1 Outer membrane - inner

membrane spanning

Tubulation

Mdm10 Outer membrane (IMP) Tubulation

Mdm12 Outer membrane (IMP) Tubulation

Mdm31 Inner membrane (IMP) Tubulation

Mdm32 Inner membrane (IMP) Tubulation

Mdm33 Inner membrane (IMP) Inner membrane ﬁ ssion?

Gem1 Miro - 1/2 Outer membrane (IMP) Ca

signaling?

THE DYNAMICS OF MITOCHONDRIAL MORPHOLOGY 43

Figure 3.11 GFP targeted to the matrix in wild - type yeast cells and in various mutants:

fusion mutant ( fzo1 Δ ), ﬁ ssion mutant ( dnm1 Δ ), tabulation mutant ( mmm1 Δ ); GFP

ﬂ uorescence images were superimposed on images from a differential interference

contrast microscope. (From the review of Okamoto and Shaw (59) .) See color plates.

Figure 3.12 Schematic representation of fusion proteins in the outer (OM) and inner

(IM) membrane. (From the review of Okomoto and Shaw (59) .)

another, or with the side. Contact sites and the creation of contact zones are

inferred from observations with the electron microscope, but unfortunately

such images represent a static situation that cannot be further resolved in time.

In contrast to the secretory pathway, the membrane fusions occur between

identical organelles, although some asymmetry between docking vesicle and

target membrane may be transiently created when a mitochondrial tip encoun-

ters a side. Needless to say, fusion involves the coordinate but presumably

sequential fusion of biochemically distinct outer and inner membranes. It is a

unique process that excludes all other intracellular membranes. The same

mechanisms responsible for the dynamic nature of cristae will operate subse-

quently to rearrange the cristae in the fused organelle.

Mitochondrial fusion is a developmentally programmed mechanism in sper-

matogenesis in fruit ﬂ y males. The analysis of a mutation causing male sterility

led to the discovery of a gene, fuzzy onion (Fzo), named after the appearance

Mdm30p

Fzo1p

Ugo1p

Mgm1p

Pcp1p

IMS

MATRIX

O M

I M

44 STRUCTURE AND MORPHOLOGY. INTEGRATION INTO THE CELL

of a large structure formed when mitochondria fail to fuse normally during

spermatogenesis. Advances in genomics greatly aided in the search for homol-

ogous genes in yeast (Fzo) and humans (Mfn1 and Mfn2 — mitofusin). When

the gene was found in yeast, the characterization of the Fzo protein and its

function led to some rapid advances and greatly stimulated the search for

additional genes required for the fusion pathway. The Fzo1 protein is an

integral outer - membrane protein, with an evolutionarily conserved GTPase

domain on the cytoplasmic side. Two transmembrane segments are connected

by a short linker exposed to the intermembrane space. The presence of a

temperature - sensitive allele in yeast prevents mitochondrial fusions at the

nonpermissive temperature, and the mitochondria become increasingly frag-

mented. A second GTPase essential for fusion was discovered unexpectedly

by studies of a gene required for mitochondrial genome maintenance in yeast.

The Mgm1p is a GTPase related to the family of dynamins; the corresponding

mammalian gene is OPA1, initially named for a mutation in OPA1 that causes

autosomal dominant optic atrophy (65) . It is an integral inner - membrane

protein with its functional domain localized in the intermembrane space.

Another dynamin - like GTPase has already been introduced in the context of

ﬁ ssion (see above). A third protein, Ugo1p, so far identiﬁ ed only in fungi, spans

the outer membrane and has functional domains on the inside to interact with

Mgm1p, and on the outside to interact with Fzo1p. Fzo1p, Ugo1p, and Mgm1p

thus form a fusion complex connecting the inner and outer membranes (53,

59) . A recent development identiﬁ es two functional isoforms of Mgm1p pro-

duced by alternative topogenesis. The sorting is controlled by a novel protein,

Ups1p (in yeast) and PRELI (in humans) (66) .

The most dramatic progress in the study of mitochondrial fusion was

achieved recently when Meeusen et al. (67) developed an in vitro mitochon-

drial fusion assay. Effectively, two populations of mitochondria were labeled

with two different chromophores in the matrix, mixed, concentrated, and

brought into contact by centrifugation, and fusion could be observed by the

mixing of the chromophores in the fused mitochondria. For the ﬁ rst time,

conditions in this in vitro system could be chosen to restrict the fusion only to

the outer membranes. Such a step required functional Fzo1 proteins on both

membranes, suggesting that cytoplasmic domains of the Fzo1p bind to each

other in a homotypic interaction. In support, X - ray crystallographic data reveal

that the C - terminal domains of Mfn1 (the human homologue) form a dimeric,

antiparallel coiled - coil structure (68) . Endogenous levels of GTP are sufﬁ cient,

and the GTPase activity of Fzo1 is required, since fusion did not take place

when temperature - sensitive Fzo1 proteins were present. Most intriguingly,

outer - membrane fusion requires the proton gradient component of the inner

membrane ( Δ pH), but not the electrochemical component ( ΔΨ ). Thus, when

mitochondria are brought close together physically, the Fzo1 proteins act as

tethers, similar to the SNARE proteins in the secretory pathway. In vivo this

initial interaction is likely brought about by an interaction with cytoskeletal

elements and ATP - dependent molecular motors. The ﬁ nal steps in fusion and

THE DYNAMICS OF MITOCHONDRIAL MORPHOLOGY 45

the use of the proton gradient are still mysterious. To achieve inner - membrane

fusion, two additional requirements must be met: A sufﬁ ciently large electro-

chemical gradient ( ΔΨ ) must be present across the inner membrane, and ele-

vated levels of GTP must be available for hydrolysis. This observation clearly

implicates the second GTPase (Mgm1p/OPA1) localized in the IMS. It is also

consistent with a role of Mgm1/Opa1 in maintenance of cristae morphology

and ATP synthase assembly (69, 70) .

A major challenge for the future will be to understand the control of ﬁ ssion

and fusion. In wild - type cells these processes must be balanced, except when

the number of mitochondria increases in preparation for cell division. At one

level, one expects control over the synthesis and turnover of the major proteins

introduced above. Undoubtedly, other participating proteins will be added to

the picture, some of them universal and some species - speciﬁ c. Their assembly

into fusion and ﬁ ssion complexes is also likely to be under some control mecha-

nism. For example, Mgm1p exists in two forms, a “ precursor ” and a mature

form that are both required for fusion. After import and cleavage of the target-

ing signal, the precursor is processed by the inner - membrane rhomboid prote-

ase Pep1. The exposure of the cleavage site in turn is dependent on matrix ATP

levels. As usual, our understanding is more complete in yeast compared to

mammalian cells, where it has been suggested that different variants of OPA1

are created by variation in splicing instead of proteolytic processing (53, 71) .

Jeyaraju et al. (72) have investigated the human orthologue called PARL

(presenilin - associated rhomboid - like protein) in animal cells. PARL has an

unrelated, vertebrate - speciﬁ c N - terminal domain. The protein is an integral

inner - membrane protein with the N - terminal on the matrix side. After import

and processing, a second cleavage by PARL activity (in trans) occurs. The

highly conserved sequence of this domain, along with the presence of potential

Ser/Thr phosphorylation sites, suggested to these authors to test alanine sub-

stitutions at these residues. To summarize a much more detailed analysis, the

authors provided evidence for the involvement of the PARL protease in the

control of mitochondrial morphology. The non - phosphorylated, cleavable form

of the enzyme was able to induce mitochondrial fragmentation when tran-

siently overexpressed in HEK 293 cells. In contrast, aspartate substitutions at

these positions to mimic phosphorylation prevented this process.

In another recent study it was shown that the F - box protein Mdm30 binds

to Fzo1 and mediates its proteolysis by a proteasome - independent mechanism

(73) . This represents a novel mechanism for Fzo1 degradation in yeast in addi-

tion to the mechanism involving ubiquitinylation and proteasome - dependent

turnover.

Fusion events involving mitochondria that are developmentally pro-

grammed have been characterized in many systems (74) . A later section will

detail some salient insights into this phenomenon in the context of cell

differentiation.

One should also consider the behavior of the mitochondria and the mito-

chondrial genomes when two cells fuse, as in mating of haploid yeast cells. This

46 STRUCTURE AND MORPHOLOGY. INTEGRATION INTO THE CELL

aspect will be considered in more detail below. The observation of constant

fusions and ﬁ ssions of mitochondria in a mammalian somatic cell has implica-

tions for the geneticist. In the absence of additional information (see below)

the dynamic behavior leads one to conclude that mitochondrial genomes in a

heteroplasmic cell are continuously mixed and redistributed, with the result

that a given mitochondrion contains both types of genomes at any time. It

may serve as a mechanism to “ homogenize ” mitochondrial genomes among

the many mitochondria in a cell. The implications for our understanding of

human mitochondrial diseases will be discussed in a later chapter. In some

organisms it may also serve to promote recombination between mitochondrial

genomes.

3.3.2 Distribution During Cell Division

When a cell contains a large number of mitochondria more or less evenly dis-

tributed throughout the cytoplasm, and doubling in number and volume during

the cell cycle, no signiﬁ cant conceptual problem exists for distributing these

mitochondria evenly into the daughter cells during cytokinesis. Furthermore,

if mitochondria become attached evenly to the newly formed microtubules of

the mitotic half spindles, their segregation is easy to visualize. The problem

during the cell cycle, therefore, is to assure a doubling of mitochondrial mass

and number and to maintain a constant ratio of nuclear to mitochondrial

genes. Cell differentiation during development, along with mitochondrial

proliferation and turnover in terminally differentiated cells like muscle and

neurons, poses a different problem.

A special problem arises when cell division occurs not by the classical

process of cytokinesis in which a larger cell is partitioned into two cells either

by a contractile ring, as in animal cells, or by the assembly of a phragmoplast

in plant cells. In the budding yeast Saccharomyces cerevisiae a bud is started

and enlarged during the cell cycle, and all the constituents for a new viable

cell have to be transported through a narrow connection between the bud and

the mother cell. How this is achieved for the nuclear chromosomes has been

a subject of study for some time, but the problem of equipping the bud with

mitochondria has been illuminated by pioneering studies starting in the labo-

ratory of M. Yaffe (see reference 50 for an expert review). Taking a genetic

approach, the Yaffe laboratory has identiﬁ ed yeast ts mutants that are defec-

tive in the transfer of mitochondria and nuclei into growing daughter buds at

the nonpermissive temperature. Others have contributed over the years, and

the subject has started to merge with the general topic of mitochondrial mor-

phology and dynamics in yeast and other eukaryotes (52, 59) . A pathway that

may be unique to fungi has been referred to as the tabulation pathway.

The original mdm1 - 1 allele, along with additional mdm1 alleles isolated

later, deﬁ nes a gene encoding a protein (Mdm1p) with several intriguing

properties. The protein is essential for viability. With some mutant proteins the

problem of mitochondrial distribution can be separated from nuclear distribu-

THE DYNAMICS OF MITOCHONDRIAL MORPHOLOGY 47

tion. The protein sequence has some similarity to intermediate ﬁ lament pro-

teins of vertebrates, and it forms 10 - nm ﬁ laments in vitro . Filament formation

fails at the nonpermissive temperature in vitro with the protein encoded by

the original mdm1 - 1 (ts) allele. On the other hand, indirect immunoﬂ uores-

cence microscopy with anti - Mdm1p antibodies reveals punctate staining

distributed throughout the cytoplasm of wild - type cells. With mutant alleles

affecting mitochondrial inheritance, the mitochondrial morphology in the

mother cells is also abnormal. Instead of the usual reticulated, tubular wild -

type structure, one ﬁ nds small, round mitochondria in clusters. These observa-

tions invite speculation that a cytoplasmic ﬁ lamentous network or scaffold

exists which is responsible for the maintenance of the mitochondrial morphol-

ogy of yeast discussed above, and that it also plays a major role in guiding

presumably threadlike mitochondrial structures through the connection into

the bud. It should be mentioned that fungi do not have the typical intermedi-

ate ﬁ lament proteins of higher eukaryotes. The yeast ﬁ laments made from

Mdm1p may represent an alternate solution to a problem that reached an

evolutionary dead end, and they were replaced by proper intermediate ﬁ la-

ments in more advanced organisms. The mdm1 mutation does not appear to

affect actin - and tubulin - based cytoskeletal functions.

Mdm1p may not be the only protein involved in this peculiar yeast cyto-

skeletal structure. Additional mutants isolated in the Yaffe lab have led to the

characterization of other MDM genes ( m itochondrial d istribution and m or-

phology), of which MDM14 encodes a 35 - kDa protein that appears to interact

with the Mdm1p and is distributed in the cytoplasm in a similar fashion. Not

unexpectedly, on second thought, some mdm mutants are affected in proteins

in the outer mitochondrial membrane ( mdm10, mdm12 ) (50, 75) . (See Figure

3.13 .) Clearly, association with a cytoskeletal element implies the existence of

a protein domain on the mitochondrial surface that must at least bind to the

scaffold, and perhaps even have motor - like properties. A related gene product

in the outer membrane is the Mmm1p ( m aintenance of m itochondrial m or-

phology) identiﬁ ed by Burgess et al. (76) . Depletion of these proteins has

pronounced effects on mitochondrial morphology, and null mutations yield a

temperature - sensitive phenotype. As pointed out by Berger and Yaffe (50) ,

the fact that these null mutants grow at the permissive temperature on glucose

(albeit slowly) suggests that a bypass or backup pathway must exist to assure

mitochondrial transmission to daughter buds. Nevertheless, mdm10 and mdm12

mutants lose mitochondrial DNA and generate respiration - deﬁ cient cells at

higher - than - normal rates. New members have been added to the MMM and

MDM groups of genes (52, 59, 77, 78) . The corresponding proteins have been

localized in the inner membrane (Mdm31p and Mdm32p), or in (or on) the

outer membrane (Mdm10p, Mdm12p, Mmm1p, Mmm1p). The Mmm1p is most

unusual in spanning both the inner and outer membrane, with its N - terminus

in the matrix and its C - terminus on the cytoplasmic side. It acts like a rivet for

these two membranes, and at the same time it forms a complex with the

Mdm10p and Mdm12p (the MMM complex) and transiently interacts with

48 STRUCTURE AND MORPHOLOGY. INTEGRATION INTO THE CELL

Mmm2p. Based on phenotypes exhibited by mutants lacking the MMM

complex, the complex is implicated in at least three signiﬁ cant processes:

attachment of mitochondria to actin, formation of tubular mitochondria, and

the the anchoring of the mtDNA nucleoids to the inner membrane. The ﬁ rst

is likely to be relevant for moving mitochondria around in the cytosol and into

the bud during cell division. The second could also involve attachment to a

cytoskeletal element (still undeﬁ ned; Mdm1p?), or it could include the con-

struction of an external scaffold that constrains mitochondria to a tubular

shape. In its absence the mitochondria turn into spheres. Finally, the N -

terminal domain of Mmm1p in the matrix has been colocalized, together with

other proteins (e.g., Mgm101p) and mtDNA, in a large structure referred to

as the mtDNA nucleod. It is the site of both mtDNA replication and transcrip-

Figure 3.13 (A) Mitochondria in a budding, wild type yeast cell. Note the movement

of one elongated mitochondrion into the bud. (B) Abnormal, giant spherical mito-

chondria in yeast cells with the mdm10 mutation (50, 75) . (Photographs provided by

M. Yaffe, University of California, San Diego.)

THE DYNAMICS OF MITOCHONDRIAL MORPHOLOGY 49

tion, to be discussed in a later chapter. In the present context it is important

to recognize that daughter cells must inherit mitochondria with mtDNA.

Therefore, dysfunction of the MMM complex results not only in abnormal

mitochondrial shape and inheritance, but also in severe mtDNA instability.

The overall potential complexity of the processes under consideration is

illustrated by the failure of mitochondrial inheritance in the mdm2 mutant. In

this mutant the OLE1 gene is affected, encoding a fatty acid desaturase (79) .

Mitochondrial membrane ﬂ uidity evidently affects mitochondrial transmis-

sion, but it may be too simple - minded to think of a mitochondrion squeezing

through a tiny pore connecting mother cell and bud. A much more exhaustive

screen of 768 yeast mutants ( ∼ 2/3 of all essential genes) has uncovered

119 genes required for maintenance of mitochondrial morphology (56) .

They encode proteins involved in ergosterol biosynthesis, mitochondrial

protein import, actin - dependent transport processes, vesicular trafﬁ cking, and

Figure 3.13 (Continued)

50 STRUCTURE AND MORPHOLOGY. INTEGRATION INTO THE CELL

ubiquitin/26S proteasome - dependent protein degradation. It becomes quite

apparent that the physical as well as functional integration of mitochondria

into a cell are intimately linked, and many details still elude us.

Meiosis and sporulation in yeast represent a special kind of cell division in

which most of the emphasis is usually placed on the behavior of the chromo-

somes. What about mitochondria? One study has deﬁ ned morphological tran-

sitions in the early stages followed by a dramatic fragmentation of a single

large mitochondrion into small spherical structures that were distributed to

the spores (80) . Mechanisms for such events remain largely speculative at this

time. If, as discussed above, fusion and ﬁ ssion occur normally at equal rates,

one may simply have to inhibit fusions to promote the breakup of a large

mitochondrion into smaller units. The phosphorylation or dephosphorylation

of an essential docking protein may be all that is required, but identifying it

is another problem.

Before meiosis can occur in yeast cells, one must have a diploid cell formed

from the fusion of two haploid cells. When respiration - deﬁ cient yeast mutants

were ﬁ rst investigated by means of mating experiments, and observations were

made on the stability of phenotypes during mitotic divisions of the zygote, a

puzzling observation was made. Zygotes containing heteroplasmic mtDNA

populations generated homoplasmic progeny within approximately 20 genera-

tions, much faster than expected on the basis of stochastic models assuming

random segregation (see references 81 and 82 for a recent review). This obser-

vation is even more puzzling in light of the discussion about the dynamic

nature of the large mitochondrial structure and the implied constant mixing

of mtDNAs. A study by Nunnari et al. (35) has reinvestigated this problem

with a most elegant and ingenious approach. Mitochondria of one haploid cell

were labeled with GFP; mitochondrial proteins of the other mating type were

labeled with the vital dye TMR - CH

Cl, which not only localized in mitochon-

dria because of the membrane potential, but eventually becomes covalently

attached to mitochondrial proteins. When such cells were allowed to mate, the

fusion and complete mixing of the two mitochondrial protein populations

could be directly observed by wide - ﬁ eld ﬂ uorescence microscopy: These obser-

vations were fully consistent with observations described above on the dynamic

behavior of yeast mitochondria. The more provocative result was obtained

when mtDNA from one mating partner was labeled selectively with bromo-

deoxyuridine (by a not totally trivial procedure). Its distribution in the zygote

could subsequently be determined immunochemically after ﬁ xation and per-

meabilization of the cells. The results showed unequivocally that while mito-

chondrial proteins from the two mitochondrial populations mixed readily, the

corresponding mtDNA populations remained segregated in the two lobes of

the zygote derived from the haploid cells. This result was a visual conﬁ rmation

of previous genetic experiments which had inferred the unequal distribution

of mtDNA in the zygote from an analysis of mtDNA in zygotic progeny. In

such progeny the type of mtDNA (genetically marked) depended on the posi-