Structural and functional studies of protein complexes involved in energy metabolism

(1)

Structural and functional studies

of protein complexes involved in

energy metabolism

Jens Berndtsson

Structur

al and functional stud

ies of pr otein complex es in volv ed in energy met abol ism

Doctoral Thesis in Biochemistry at Stockholm University, Sweden 2021

Department of Biochemistry and Biophysics

ISBN 978-91-7911-552-4

(2)

(3)

Structural and functional studies of protein

complexes involved in energy metabolism

Jens Berndtsson

Academic dissertation for the Degree of Doctor of Philosophy in Biochemistry at Stockholm University to be publicly defended on Friday 1 October 2021 at 13.00 in Magnélisalen, Kemiska övningslaboratoriet, Svante Arrhenius väg 16 B.

Abstract

Mitochondria are eukaryotic organelles with a multitude of functions including biosynthesis of molecules and cellular regulation. Most prominently though is their role in energy conversion which culminates with the production of ATP, the universal molecular unit of currency. This is done through several metabolic pathways, including the pyruvate dehydrogenase bridging reaction, the citric acid cycle and the oxidative phosphorylation. In the latter pathway, electrons are transferred from electron carriers formed in the previous pathways and shuttled trough a chain of protein complexes (complex I – complex IV) via the mobile electron carriers coenzyme Q and cytochrome c. Collectively this is known as the respiratory chain. This process harnesses energy from the transferred electrons to translocate protons across the mitochondrial inner membrane, forming an electrochemical gradient that the ATP synthase uses to generate ATP. In this thesis we study parts of these metabolic pathways both structurally and functionally, using a combination of cryo-EM, biochemistry and cell biology. In the first project we used cryo-EM to solve the structure of the pyruvate dehydrogenase complex of E. coli, gaining new insight into how the flexible lipoyl-domain interacts with the active site of the core of the complex. We could determine that this interaction is mediated through electrostatic interaction formed between an acidic patch of amino acids of the lipoyl-domain and positively charged amino acids on the core. In the second project we again employed cryo-EM, this time to solve the structure of the yeast respiratory supercomplex, and for the first time we could obtain a near-atomic resolution structure of how complex III and complex IV in yeast interact with each other to form respiratory supercomplexes. Two forms of these higher order assemblies exist in the respiratory chain of yeast (CIII2/CIV and CIII2/CIV2), which assembles very differently compared to the mammalian CI/CIII2/CIV respirasome.

The main interaction point of the yeast supercomplexes occurs between the subunits Cor1 and Cox5a. Through selective point mutations, we were able to disrupt this interaction and effectively hinder supercomplex formation in yeast. Using biochemistry and cell biology on such disrupted cells, we could determine that supercomplexes form to facilitate better diffusion of cytochrome c between the individual complexes of the supercomplex. In the third project we look at how manganese toxicity impacts the respiratory chain in yeast on a molecular level. By combining proteomics, biochemistry and metal analyses, we found that manganese overload causes mismetalation of Coq7, an essential subunit of the coenzyme Q synthesis pathway, which causes a loss of the electron carrier between complex II and complex III. This loss of coenzyme Q could be restored when cells were augmented with Coq7 overexpression, which restored functional respiration and prevented age-related cell death.

Keywords: mitochondria, metabolism, OXPHOS, yeast respiratory supercomplex, cytochrome c, coenzyme Q, pyruvate

dehydrogenase complex.

Stockholm 2021

http://urn.kb.se/resolve?urn=urn:nbn:se:su:diva-195224 ISBN 978-91-7911-552-4

ISBN 978-91-7911-553-1

Department of Biochemistry and Biophysics

(4)

(5)

STRUCTURAL AND FUNCTIONAL STUDIES OF PROTEIN COMPLEXES INVOLVED IN ENERGY METABOLISM

(6)

(7)

Structural and functional

studies of protein complexes

involved in energy metabolism

(8)

©Jens Berndtsson, Stockholm University 2021

ISBN print 978-91-7911-552-4 ISBN PDF 978-91-7911-553-1

(9)

Så ytterst få personer trivs med vad de gör om dagarna, de släpar sig till jobbet med besvär; om man är glad att ha ett jobb så är man baskemej suspekt, det känslomässigt mer korrekta är dystra tankar om att slippa jäkta: ’Man skulle varit vetenskapsplebej.’ När städaren går upp på morgonkvisten så tänker han: ’Åh, fy, nu ska det svabbas.’ För läraren är varje dag en trist en. Han gläds blott åt eleverna, som drabbas. Men vetenskapsplebejen har det bra. Han möter varje dag med ett hurra. ”Var är min rock, nu jävlar ska här labbas!”

(10)

(11)

1

List of publications

I. Škerlová J, Berndtsson J, Nolte H, Ott M, Stenmark P. (2021). Structure of the native pyruvate dehydrogenase

complex reveals the mechanism of substrate insertion.

(manuscript accepted in Nature communications).

II. Rathore S*, Berndtsson J*, Marin-Buera L*, Conrad J, Carroni M, Brzezinski P, Ott M. (2019). Cryo-EM structure

of the yeast respiratory supercomplex. Nat. Struct. Mol

Biol 26, 50-57.

III. Berndtsson J*, Aufschnaiter A*, Rathore S, Marin-Buera, Dawitz H, Diesel J, Kohler V, Barrientos A, Büttner S, Fontanesi F, Ott M. (2020). Respiratory supercomplexes

enhance electron transport by decreasing cytochrome c diffusion distance. EMBO reports 21, e51015.

IV. Diesel J, Berndtsson J, Broeskamp F, Habernig L, Kohler V, Vazquez-Calvo C, Nandy A, Peselj C, Drobysheva S, Pelosi L, Vögtle N, Pierrel F, Ott M, Büttner S. (2021).

Manganese overload disrupts mitochondrial energy metabolism via inhibition of CoQ biosynthesis.

(submitted manuscript).

(12)

(13)

3

Abbreviations

3D three-dimensional ADP adenosine diphosphate ATP adenosine triphosphate

CI - complex I NADH:ubiquinone oxidoreductase CII - complex II succinate dehydrogenase

CIII - complex III coenzyme Q:cytochrome c reductase CIV - complex IV cytochrome c oxidase

CV - complex V ATP synthase CL cardiolipin Cyt. c cytochrome c

Cryo-EM electron cryo-microscopy

E. coli Escherichia coli

FAD/FADH2 flavin adenine dinucleotide

FeS iron sulfur cluster FMN flavin mono nucleotide

IMM inner mitochondrial membrane IMS inter membrane space

Mitoribosome mitochondrial ribosome mtDNA mitochondrial DNA

NAD+_/NADH _{nicotinamide adenine dinucleotide}

NMR nuclear magnetic resonance OMM outer mitochondrial membrane OXPHOS oxidative phosphorylation PTM post-translational modification Q/Q·-_/QH

2 ubiquinone/semiquinone/ubiquinol

RCC respiratory chain complex ROS reactive oxygen species SC supercomplex

S. cerevisiae Saccharomyces cerevisiae

(14)

(15)

5

1. Introduction

Structure and function. Two words that to a biochemist mean so much more than the sum of their letters. Understanding three-dimensional (3D) arrangements of biomolecules at an atomic level can lead to great leaps for our understanding of biological systems. In medicine this has proven to be essential in the literal meaning of the word as many life-saving drugs have been developed with information gained from structural biology1_{. To this end, many}

techniques of solving the structures of biological macromolecules such as proteins, nucleotides and cell structures have been developed over the years.

Historically, X-ray crystallography has been the go-to method for high resolution structure solving, allowing atomic or near-atomic resolution. With over 158,000 entries in the PDB-database to date2_,

it is the undisputed number one in its category, however it is not alone. Nuclear magnetic resonance (NMR) is another technique for structurally solving smaller macromolecules at an atomic resolution. And unlike crystallography, NMR works even when their targets are in solution rather than in a crystal, allowing for more dynamic molecules (which often do not crystallize) to be visualized. Light microscopy is probably the most famous way of magnifying small objects, but this method has the disadvantage of being limited by the wavelength of the photon in the visual light spectra. Thus, high-resolution macromolecules cannot be resolved by this technique. Nevertheless, light microscopy has over the years helped to advance science immensely, from Antonie van Leeuwenhoek’s visualization of the first microorganisms in the 17th_{century to super resolution}

(18)

8

Solving higher order macromolecules of great complexity, or with high levels of dynamic parts, is something the aforementioned techniques have struggled with. Electron cryo-microscopy (Cryo-EM) is a technique that like X-ray crystallography uses an elementary particle with very small wavelengths (electrons) to visualize its target, thus permitting atomic resolution. In the past decade, large advances of technology, particularly in detecting electrons, have led to the so-called “resolution revolution”4_of

cryo-EM and led to cryo-cryo-EM becoming a serious challenger to all the above-mentioned techniques for scientific research.

Cryo-EM comes in two types: transmission electron microscopy (TEM), and cryo-tomography. The TEM version is often paired with single particle analysis of target macromolecules5,6_{. Unlike}

crystallography, however, this technique does not require the crystallization process of its target in order to function. In short, cryo-EM traps its target in solution in a layer of vitreous ice by freezing the sample extremely rapidly, avoiding the formation of ice crystals7,8_{. Like NMR, this allows us to observe the dynamics of}

proteins, but it also has a much higher tolerance for extremely big macromolecules to be viewed. Cryo-tomography is a variant of cryo-EM that like light microscopy can be used to view targets as big as organelles like mitochondria9,10_{, but at a substantially higher}

resolution.

Primarily working with yeast as a model organism, the aim of this thesis is to understand the functions of macromolecular assemblies using structural knowledge obtained from cryo-EM studies, thus combining the two words: structure and function.

(19)

9

2. Mitochondria

2.1. Structure and morphology

Mitochondria are organelles that exist in almost all eukaryotic organisms and are amongst the best studied organelles in eukaryotic cells. Traditionally depicted as a “bean shaped” organelle freely floating in the cytosol (Figure 1), the structure of mitochondria is that of a double-membrane bound organelle, which has two lipid bilayers: an outer mitochondrial membrane (OMM), and an inner mitochondrial membrane (IMM). The IMM is usually compartmentalized further into folds, called cristae. This allows for a greater density of IMM within a mitochondrion, which means that more proteins can be housed there11_{. Between the two membranes}

exist an intermembrane space (IMS), and enclosed by the IMM is a compartment called the mitochondrial matrix12_.

This “bean shaped” model of mitochondria does not account for all the shapes that these organelles can adopt. In fact, the morphology of mitochondria is highly dynamic and varies depending on organism, cell type, metabolic state or at which stage of the cell cycle the cell is in13_{. In baker’s yeast: Saccharomyces cerevisiae (S.}

cerevisiae), the mitochondria form a dynamic tubular network that

continually changes its shape through fusion and fission14,15_{. These}

changes are part of cellular regulation and often occur via interaction with other cellular components such as microtubules and membrane nanotubes16,17_{. The abundance of mitochondria, like their}

morphology, also varies depending on organisms and tissue. Mitochondrial levels range from 20-30 per cell in S. cerevisiae18_{, to}

104_{(~35% of the cell volume) in mammalian heart muscle tissue}19_,

(20)

10

Figure 1. Architecture and function of mitochondria. Mitochondria are double-membrane bound organelles separated from the cytosol via an outer- (OMM) and an inner membrane (IMM). Between these membranes lies an intermembrane space (IMS), and on the inside of the inner membrane lies the mitochondrial matrix. Mitochondria execute a multitude of functions, many of which revolve around the catabolic pathways of converting energy from pyruvate to ATP via the pyruvate dehydrogenase reaction, the TCA cycle and the OXPHOS.

2.2. Dual genetic origin and mitochondrial proteins

Due to its dynamic nature, the size of an individual mitochondrion can be hard to measure precisely, but the rough dimensions are between 0.5 µm in diameter, and 1-10 µm in length for a “bean shaped” mitochondrion21_{. This is roughly the same size as a}

bacterium, which is in line with the predominant theory that the mitochondria arose as an organelle through an endosymbiotic event ~2 billion years ago22_{, in which an α-proteobacteria was engulfed by}

a primitive eukaryotic cell and kept around for both cells’ benefit23,24_.

This symbiogenesis between the two organisms can still be observed today as mitochondria have retained part of their DNA (mitochondrial DNA - mtDNA), along with its own replication, transcription, and translational machinery25–30_{. The mtDNA found in}

(21)

11

today’s cells is however comparatively small, and most of the original genes encoded in its genomes have been transferred to the nuclear DNA in continuous processes that have occurred independently in separate organisms. This has led to differences in the mitochondrial genome size between organisms. For instance in yeast, the genome encodes for eight proteins, 24 tRNAs, and two rRNAs, whereas in humans it encodes for 13 proteins, 22 tRNAs, and two rRNAs25_.

Of the >1,000 proteins that are found in mitochondria, the portion of proteins synthesized within mitochondria is small across all species, with the mitochondrially expressed proteins usually being very hydrophobic inner membrane proteins that would be difficult to transport into the mitochondria from the cytosol25_{. The remaining}

proteins are synthesized on cytosolic ribosomes as precursor proteins before being imported into the mitochondria, where they are processed to their mature forms31,32_{. The precursor proteins are}

usually equipped with a mitochondrial targeting sequence (MTS)33_,

that is recognized by the translocase of the outer membrane (TOM) complex, a process that can occur pre-, co-, or post-translationally. From there, the precursor proteins follow different pathways depending on type and destination. Some can get inserted into the OMM, which can happen via the sorting and assembly (SAM) complex34_{. Proteins targeted for the IMM or the mitochondrial}

matrix are transported by the translocase of the inner membrane (TIM) complex before being inserted into the IMM35_{, or with the}

help of the pre-sequence translocase-associated motor (PAM) complex, that mediates import into the matrix31_{. Finally, most of the}

mitochondrial precursor proteins are separated from their MTS, which permits their folding into the mature form.

(22)

12

2.3. Metabolic pathways

Colloquially, mitochondria are known as the powerhouse of the cell36_{. This moniker comes from its role in energy conversion as}

mitochondria house essential catabolic pathways in the creation of the “molecular unit of currency”: adenosine triphosphate (ATP) (Figure 1). ATP can store chemical energy in its phosphoanhydride bonds, and hydrolysis of this bond is a highly exergonic reaction. The energy released is used by a multitude of cellular processes to the extent that an adult human roughly turns over his/hers bodyweight in ATP daily37_{. There are several metabolic pathways}

that take place within mitochondria, and depending on the organism, regulation of metabolism can be very complex with a great variety of cues. If we however look at only the catabolic pathways in mitochondria during energy conversion, there are a few pathways that are more prominent than others. One of the key molecules in these pathways is acetyl-coenzyme A (acetyl-CoA), which is the end product of two of these pathways and the start of the third.

Beginning with glucose catabolism, outside of mitochondria, glucose is converted to pyruvate via glycolysis38_{. After being}

imported into the mitochondria, pyruvate undergoes an oxidative decarboxylation reaction catalyzed by the pyruvate dehydrogenase complex (PDHc) that converts it to acetyl-CoA39_:

Pyruvate + CoA-SH + NAD+_→

Acetyl-CoA + CO2 + H+ + NADH

The PDHc is a multienzyme complex composed of three enzymatic subunits that exist in multiple copies in the complex, the exact number depends on the organism. These three catalytic subunits are the pyruvate dehydrogenase (E1), the dihydrolipoamide acetyltransferase (E2), and the dihydrolipoamide dehydrogenase (E3) (Figure 2). In addition to the three catalytic subunits, higher

(23)

13

eukaryotes have a fourth protein called the E3-binding protein (E3BP) which recruits the E3 subunit to the complex, and two regulatory enzymes: pyruvate dehydrogenase kinase, and pyruvate dehydrogenase phosphatase39,40_{. For both mammals and yeast, the}

E2 subunit forms an icosahedral core consisting of 60 copies and forms the center of the entire complex. In Escherichia coli (E. coli) this core only consists of 24 copies that form an octahedral cubic core41_{. From this core, the E2 subunit extends flexible arms that carry}

a binding domain to the peripheral subunits E1 and E3, as well as one or several lipoyl domains onto which a lipoate cofactor is covalently attached. The reaction of the PDHc starts with the decarboxylation of pyruvate by the E1 subunit with the help of the cofactor thiamine diphosphate (TPP), which then transfers the resultant acetyl group to an oxidized lipoic moiety of the E2 lipoyl domain. This lioyl domain can then swing over to the catalytic site of the E2 core, where the acetyl group is transferred over to a thiol group of the cofactor Coenzyme A. This reduces the lipoic moiety, which is again oxidized by the E3 subunit via its cofactor FAD, to make it ready to accept a new acetyl group from another decarboxylated pyruvate42_.

The second pathway to generate acetyl-CoA is β-oxidation.

β-oxidation is part of the fatty acid catabolism, and received its name from the oxidation of the β-carbon of the fatty acid it is processing. Unlike the oxidative decarboxylation of pyruvate, the reactions carried out during β-oxidation are not carried out by a single protein complex, but by several enzymes43_:

Cn-acyl-CoA + FAD + NAD+ + H2O + CoA-SH →

(24)

14

Figure 2. Reaction scheme for the pyruvate dehydrogenase complex. The oxidative decarboxylation of pyruvate to acetyl-CoA is a five-step process: 1) The decarboxylation of pyruvate using the TPP cofactor on the pyruvate dehydrogenase subunit (E1). 2) The oxidation of the attached lipoate moiety of the dihydrolipoamide acetyltransferase (E2) by the acetyl group and hydroxyl ion from reaction 1. 3) Acetylation of CoA to produce acetyl-CoA. 4) Reduction of FAD cofactor of the dihydrolipoamide dehydrogenase (E3) resulting in the fully

oxidized form of the lipoic moiety of E2. 5) Dehydrogenation, where the FADH2

transfers hydrogen to NAD+_{forming NADH. Figure adapted from Nelson and}

Cox44_.

β-oxidation is a cyclic four-step process that cleaves off two carbons from an acyl-CoA fatty acid, to form the two carbon acetyl-CoA. In the first step acyl-CoA dehydrogenase uses FAD as a cofactor to form a double bond between the second and third carbon of the acyl-CoA. This then gets hydrated by enoyl CoA hydratase in the second step, forming a hydroxyl group on the third carbon. In the third step, 3-hydroxyacyl-CoA dehydrogenase then oxidizes this hydroxyl group using NAD+_{as a cofactor, converting it into a keto group.}

Lastly, the fourth step is catalyzed by β-ketothiolase, which cleaves the molecule between the second and third carbon using a thiol group of another CoA molecule, forming an acetyl-CoA and an acyl-CoA molecule that is two carbons shorter than before45_.

(25)

15

Acetyl-CoA generated from the PDHc or the β-oxidation is further processed in the tricarboxylic acid cycle (TCA cycle). The TCA cycle is a cyclic eight step reaction that catalyzes the oxidation of acetyl-CoA, yielding reduced electron carriers NADH and FADH246:

Acetyl-CoA + 3 NAD+_{+ FAD + GDP + Pi →}

2 CO2 + GTP + CoA-SH + 3 NADH + FADH2

The TCA cycle is also an important anabolic pathway as several of its intermediates are precursors in biosynthesis of various molecules. Citrate, for example, is a precursor for fatty acids and steroids, while the amino acid glutamate is produced from α-ketoglutarate, which can then be converted to glutamine, proline and arginine, as well as purines47_{. Oxaloacetate is another molecule in the TCA that acts as}

a precursor for a multitude of amino acids as well as pyrimidines48_.

The reduced electron carriers obtained from these metabolic pathways will eventually be utilized in oxidative phosphorylation (OXPHOS) which will be covered extensively in chapter 3. In brief, the main purpose of OXPHOS is to use the energy from the oxidation of these electron carriers to drive the synthesis of ATP in the IMM.

2.4. Biomolecule synthesis

Another essential purpose of the mitochondria is to act as a hub for the synthesis of certain biomolecule proteins. ATP and the precursors for amino acids and nucleotides found in the TCA cycle have already been mentioned, but in addition to these there are several essential biomolecules produced entirely or partially in the mitochondria49_.

Iron-sulfur clusters (FeS) are one of these essential biomolecules that are partially produced in the mitochondria. These consist of iron and inorganic sulfide, and are one of the oldest known cofactors in

(26)

16

biology50_{(Figure 3a). These molecules play key roles in several}

cellular processes including nitrogen fixation, protein synthesis, DNA maintenance, and respiration where FeS clusters serve as a platform for electron transfer50_{. The first steps of FeS synthesis are}

highly conserved in eukaryotes and occur via the mitochondrial ISC pathways on the scaffolding protein ISCU51_{. In fact, the only known}

eukaryotes without mitochondria, Monocercomonoides sp., have adopted a reduced form of the bacterial SUF pathway to compensate for the lack of the ISC pathway found in mitochondria52,53_.

Figure 3. Biomolecules produced in the mitochondria. A) Iron sulfur cluster,

rhombic (above), and cubic (below), from PDB: 5XTH54_{. B) Heme b with iron}

atom in orange, from PDB: 6YMX55_{. C) Cardiolipin between complex III and}

complex IV, from PDB: 6YMX.

Another biomolecule that partially stems from the mitochondria is the cofactor heme (Figure 3b). Heme is an essential iron containing redox cofactor that consists of an iron ion that is coordinated within a porphyrin molecule56_{. Perhaps best known for its role in mediating}

gas transport in the hemoglobin of red blood cells, heme, like FeS, is also a key molecule for the electron transfer in respiration, and different types of hemes are present in the various proteins in OXPHOS57_.

(27)

17

Being a double membrane-bound organelle, mitochondria have a constant need for lipids in their membranes, both for structural and functional purposes. To that end, several pathways of lipid synthesis are also present in mitochondria. Most prominent amongst these organelle-specific phospholipids is cardiolipin (CL) (Figure 3c), which can constitute ~15 % of the lipids in the IMM58_,_{and is the}

signature lipid of the mitochondria. Cariolipin is synthesized in the mitochondria by the enzyme cardiolipin synthase (Crd1 in yeast, CLS in mammals), which catalyzes one of the last steps of its pathway, whereby two phosphatidylglycerols are fused together to

form the mature lipid59,60_._{Another lipid that is important for}

mitochondrial function and is found in large amounts in both the OMM and the IMM is phosphatidylethanolamine (PE), which is produced in the mitochondria via decarboxylation of phosphatidylserine (PS) by the enzyme Psd1. It has however been shown that yeast can also produce PE into the mitochondria from another pathway that uses choline as a precursor61_.

2.5. Role in cellular regulation

Apart from housing metabolic pathways and being a hub for the synthesis of biomolecules, mitochondria also play a prominent role in cellular regulation and the regulation of the cell cycle itself. This occurs predominantly during mitosis where mitochondrial biogenesis either needs to increase, or is fragmented and segregated prior to cytokinesis62–64_{. Mitochondria are also heavily involved in}

the cellular response to toxic insults, when a cell death program (apoptosis) can be induced via the release of the mobile electron carrier cytochrome c (chapter 5.1). Mitochondria have additionally been linked to aging, with aging phenotypes correlating with increased mtDNA mutations, and down-regulated OXPHOS65,66_{. It}

is however still unknown exactly how aging and mitochondria are connected. Another regulatory function that mitochondria fulfill is their role in calcium (Ca2+_{) homeostasis. Ca}2+_{is the most common}

(28)

18

divalent cation in the cell and is commonly used for various signal transduction pathways in the cell67_{. Mitochondria are together with}

the endoplasmic reticulum (ER) the major storage sites of calcium in the cells and the interplay between these two organelles acts as cytosolic buffers of Ca2+_{, and their interplay is important to}

re-establish calcium homeostasis after the burst of its release in signaling67,68_.

2.6. Mitochondrial dysfunction and diseases

Because of the importance of mitochondria, mitochondrial malfunctions can have catastrophic ramifications for the cell. In humans, defective mitochondrial function has been linked to several types of diseases and disorders69_{. Due to the prominent role of}

OXPHOS in mitochondria, many of these are linked to problems of the respiratory chain complexes70–72_{. Such disruptions of the}

respiratory chain will often manifest as various conditions under the term mitochondrial diseases, and range from muscular diseases, to endocrine system disruption, to neurodegenerative diseases such as Alzheimer's disease, and Parkinson's disease73–75_{. Mutations in the}

mtDNA have also been associated with cancer. In these cases the mtDNA variants can either act as “inducers”, which to various degrees contributes to tumorigenesis, or as “adapters” which allows cancer cells to better adapt to their surroundings76_{. The OXPHOS}

machinery is also a major source of reactive oxygen species (ROS) in cell, which modulate a number of cellular malfunctions including apoptosis77_{. Other than mutations in the OXPHOS subunits, the}

system can also be affected by external factors such as manganese poisoning, which causes oxidative stress often leading to neurodegenerative diseases78,79_{. Exactly how manganese causes}

(29)

19

3. Oxidative phosphorylation

3.1. The electron transport chain and ATP synthesis

In 1961 Peter Mitchell postulated that the ATP production in the mitochondria is coupled to both an electron and hydrogen transfer via a chemiosmosis type mechanism80_{. His work laid the foundation}

for the modern OXPHOS theory, which is the combination of proton pumping by the electron transport chain and the subsequent ATP synthesis.

Figure 4. Electron transfer in the respiratory chain and ATP synthesis. Electrons from NADH or succinate are transferred via complex I (green) or complex II (orange) respectively to coenzyme Q. Complex III (blue) uses electrons from coenzyme Q to reduce cytochrome c (yellow) which then transfers electrons to CIV (red) which in turn reduces molecular oxygen to water. Complex V (pink) uses the proton motive force that is created by complex I, complex III and complex IV to drive the synthesis of ATP.

The electron transport chain in mammals is made up of four multi-enzyme complexes: NADH:ubiquinone oxidoreductase (complex I - CI), succinate dehydrogenase (complex II - CII), coenzyme Q:cytochrome c reductase (complex III - CIII) and cytochrome c oxidase (complex IV - CIV). These four complexes obtained their names from their participation in a shuttling of electrons from electron donors (mostly NADH and succinate from the matrix) all the way to molecular oxygen, which is then reduced to water in the

(30)

20

matrix (Figure 4). Three of these complexes, CI, CIII and CIV, use the energy gained from the electron transfer to translocate protons from the matrix into the IMS. This establishes an electrochemical gradient across the IMM called the proton motive force, which is the source of energy that the ATP synthase (complex V - CV) uses to generate ATP. In order to transfer electrons between the complexes, two mobile electron carriers are used, namely coenzyme Q and cytochrome (cyt. c). Coenzyme Q is also called ubiquinol (QH2)

when it is reduced and ubiquinone (Q) in its oxidized state. A third state called semiquinone (SQ·-_{) also exists, but is usually very}

transient as it is a source of reactive oxygen species (ROS)81,82_. 3.2. Complex I - NADH:ubiquinone oxidoreductase

NADH + H+_{→ NAD}+_{+ 2H}+_{+ 2e}-

Q + 2H+_{+ 2e}-_{→ QH} 2

4H+

Matrix → 4H+IMS

Mammalian CI is the biggest and most complex of the four multi-enzyme complexes of the respiratory chain. In humans it consists of 45 subunits divided into four functional modules: N, Q, Pp and Pd83– 85_{. Structurally, CI is L-shaped, with a large intermembrane portion}

(Pp and Pd) with a matrix-side arm (N and Q) protruding from it (Figure 5a). The reaction starts in the N-module where NADH donates two electrons to the flavin mononucleotide (FMN) cofactor. These electrons then enter into a chain of eight iron-sulfur clusters spanning both the N- and Q-module, translocating the electrons to the quinone binding site in the Pp-module in the IMM83_{. The energy}

gained from the reduction of the ubiquinone also induces conformational changes in both the Pp- and Pd-modules, which opens four proton channels, each allowing one proton to be translocated from the matrix across the IMM into the IMS. CI is a considerable source of reactive oxygen species (ROS) in

(31)

21

mitochondria, which derive from the reduced FMN cofactor interaction with molecular oxygen in the matrix86_.

Figure 5. NADH dehydrogenases in mammals and yeast. A) Reaction scheme of mammalian CI. NADH is oxidized by the FMN cofactor and electrons are transferred to a chain of eight iron-sulfur (FeS) clusters before reducing coenzyme Q. Four protons are transferred across the IMM via CI as a result of this reaction. B) Reaction scheme of yeast NADH dehydrogenases: Nde1, Nde2 and Ndi1. NADH is oxidized by the FAD cofactor of the dehydrogenases and electrons are transferred directly to coenzyme Q. Dashed red arrows annotate electron pathways and blue arrows indicate proton pathways.

In S. cerevisiae the start of the electron transport chain is similar to that of mammals yet also very different. Baker’s yeast does not have a canonical CI as found in higher eukaryotes, instead the oxidation of NADH is mediated by three NADH dehydrogenases: Nde1, Nde2 and Ndi1, where Nde1/2 face the IMS, and Ndi1 faces the matrix87,88

(Figure 5b). As these three enzymes are the only way for yeast to oxidize NADH, it is important that NADH in both of these compartments can be utilized, as a balance between NADH and NAD+_{needs to exist in the cell. The active site in these NADH}

dehydrogenases contains a flavin adenine dinucleotide (FAD) rather than the FMN as found in CI. The electrons from NADH are also not transported through a chain of FeS clusters, but rather immediately donated to coenzyme Q. The difference in size between CI in

(32)

22

mammals and the NADH dehydrogenases in yeast is vast, with the CI being approximately 1 MDa, whereas the three dehydrogenases are ~60 kDa each. The NADH dehydrogenases in yeast also do not perform any proton translocating across the IMM.

3.3. Complex II - succinate dehydrogenase

Succinate → Fumarate + 2e-_{+ 2H}+

Q + 2H+_{+ 2e}-_{→ QH} 2

Another entry point for electrons into the respiratory chain is via succinate and CII, which directly links the TCA cycle to the respiratory chain. CII is a much smaller protein complex than CI, and contains only four, nuclear-encoded, subunits: Sdh1, Sdh2, Sdh3 and Sdh4 (Figure 6a). Like the NADH dehydrogenases, the active site of CII contains a FAD cofactor in the Sdh1-subunit, which accepts electrons from succinate to yield fumarate (Figure 6b). Like CI, the electrons then enter a chain of iron-sulfur (FeS) clusters bound to Sdh2, before reducing coenzyme Q. Unlike CI, CII does not mediate the translocation of protons across the IMM, but is like CI a source of ROS89,90_.

Sdh4 also contains the cofactor heme b. However, it has long been debated whether heme b has a role in the electron transfer or if it is just a structural factor for the stabilization of the CII heterotetramer91,92_{. A recent study, in which heme b was}

computationally modeled as part of the electron transfer, indicated that heme b would not affect the speed of electron transfer nor that it would take part of the electron transfer as a redox cofactor93_.

(33)

23

3.4. Complex III - coenzyme Q:cytochrome c reductase

2QH2 → Q + 4H+IMS + 4e-

Q + 2H+

Matrix +2e- → QH2

2 cyt c3+_{+ 2e}-_{→ 2 cyt c}2+

CIII is an obligate homodimer of roughly 500 kDa, containing between 4-11 subunits depending on species (Figure 7a, Table 1). The assembly of CIII in yeast is an interesting field of research since it involves coordinating the expression of subunits from two genetical origins: mitochondria and the nucleus. The synthesis of cytochrome b (Cytb), which is the only subunit of mitochondrial origin, is particularly interesting as it is co-translationally inserted into the IMM. This event requires the coordination between the mitochondrial ribosome (mitoribosome) and the Oxa1/Mba1/Mrx15 insertion machinery in order to be successful94–97_{. The translation of}

Cytb is controlled by three translational activators: Cbs1 and Cbs2 which interact with the mRNAs 5’ end as well as the mitoribosome98_,

and the Cbp3-Cbp6 complex99,100_{. The Cbp3-Cbp6 complex controls}

Cytb expression via a translational feedback loop in which it binds to the mitoribosome and induces synthesis of Cytb. Next it binds to the newly synthesized Cytb and stabilizes it until the other assembly factors of CIII can interact with it. Once the subunits Qcr7 and Qcr8 bind to Cytb, the Cbp3-Cbp6 complex can disassociate from it and

Figure 6. Complex II structure and

function. A) Subunits of CII: Sdh1

(yellow), Sdh2 (blue), Sdh3/4 (orange), PDB: 1NEK89_{. B) Reaction}

scheme of CII. Succinate is oxidized by cofactor FAD which transfers the electrons to a series of iron-sulfur (FeS) clusters to coenzyme Q. Dashed red arrows annotate electron pathways.

(34)

24

return to the mitoribosome to induce another round of Cytb synthesis, leaving an intermediate Cytb-Qcr7-Qcr8 complex unto which the other subunit of CIII can assemble101,102_{. Another}

interesting point about the CIII assembly is when the dimerization of the complex occurs. In a recent study, it was shown that the dimerization begins early in the assembly when two copies of Cor1 and Cor2 form a dimer. This Cor1-Cor2-module can then interact with the Cytb-Qcr7-Qcr8 intermediate and induce dimerization for the rest of CIII103_.

Figure 7. CIII subunits and the Q-cycle. A) Catalytic subunits Cytb (red), Cyt1 (yellow), and Rip1 (orange) in CIII. Blue subunits are the seven structural subunits of S. cerevisiae. PDB: 6YMX55_{. B) The Q-cycle of CIII. One electron from}

oxidized QH2 is transferred to cytochrome c via Rip1 and Cyt1, realizing protons

to the IMS at the same time. The other electron is transferred to a Q at the Qi

binding site via two hemes b and takes up protons from the matrix side at the same time. Dashed red arrows annotate electron pathways and blue arrows indicate proton pathways.

Yeast CIII has three catalytic subunits that form the redox center of the complex. These are: Cytb which contains two heme b cofactors (heme bL and heme bH), Rip1 which has a rhombic FeS cluster, and

Cyt1, which carries a heme c1 cofactor. The oxidation of QH2 and

subsequent reduction of cyt. c is catalyzed by these three subunits, a reaction sometimes referred to as the Q-cycle, and is a two-step reaction that involves the oxidation of two electrons of QH2 and

happens twice per reaction104_{(Figure 7b). It begins with the binding}

(35)

25

transferred to heme c1 in Cyt1 via the FeS cluster in Rip1, which then

undergoes a conformational change in order to facilitate this transfer105_{. The energy released from this electron transfer is also}

used to facilitate the release of two protons to the IMS. From heme

c1 in Cyt1, the electron reduces heme c in cytochrome c which binds

to Cyt1 in the IMS via electrostatic interactions106,107_{. The second}

electron from QH2 is transferred to the heme bL of Cytb. From there

it is transferred to heme bH before it is used to reduce a Q in the Qi

site of Cytb. This overall reaction occurs twice, as the first round of the second electron only reduces the Q to a semiquinone (SQ·). The second round of QH2 oxidation provides the second electron, which

fully reduces the semiquinone at the Qi site and releases it into the

IMM.

Table 1. The subunits of CIII and CIV in S. cerevisiae and H. sapiens.

Complex III Complex IV

S. cerevisiae H. sapiens S. cerevisiae H. sapiens

Cor1 UQCRC1 Cox1 MT-CO1

Cor2 UQCRC2 Cox2 MT-CO2

Cytb Cytochrome b Cox3 MT-CO3

Cyt1 CYC1 Cox4 COX5B

Rip UQCRFS1 Cox5a COX4I1

Qcr6 UQCRH Cox5b -

Qcr7 UQCRB Cox6 COX5A

Qcr8 UQCRQ Cox7 COX7A2

Qcr9 UQCR10 Cox8 COX7C

Qcr10 UQCR11 Cox9 COX6C

- Su9 Cox12 COX6B1

Cox13 COXA1

Cox26 -

- COX7B

(36)

26

3.5. Complex IV - cytochrome c oxidase

4cyt c2+_{→ 4cyt c}3+_{+ 4e}-

O2 + 4H+ + 4e- → 2H2O

4H+

Matrix → 4H+IMS

The last complex in the electron transport chain is CIV. CIV consists of three mitochondrially encoded proteins (Cox1, Cox2 and Cox3), and up to 11 more subunits encoded in the nucleus (Figure 8a, Table 1). Like in the case of CIII, this dual genetic origin of CIV presents a biogenesis challenge, and CIV has at least 30 factors involved in its assembly108,109_{. An interesting aspect of the CIV assembly is that}

Cox6 forms a complex with Atp9 of the ATP synthase (CIV) called Atco110_{, implying a coordination between the CIV and the CV}

assembly as well. Often depicted as a homodimer in literature, the evidence for CIV being an obligate homodimer like CIII is only shown in crystallization experiments. In cryo-EM structures, CIV is more often found either in supercomplexes together with CIII and CI (chapter 4), or when analyzed through Blue-Native page acrylamide gels, as a monomer of roughly 400 kDa111_.

Figure 8. Catalytic subunit and electron transfer of CIV. A) Catalytic subunits

of CIV: Cox1 (yellow), Cox2 (blue) and Cox3 (orange). PDB: 6YMX55_{. B) CIV}

reaction schematics. Electrons from cytochrome c is transferred to CuA and then

to heme a before being transferred to the catalytic site of CuB and heme a3. In the

catalytic site, the electrons are transferred to molecular oxygen, which is then reduced to water with the help of protons. This also energizes additional protons to be translocated across the IMM. Dashed red arrows annotate electron pathways and blue arrows indicate proton pathways.

(37)

27

The three mitochondrially encoded subunits are highly conserved throughout the kingdoms of life, with some bacteria only having homologues to these subunits and no accessory subunits at all112_{. In}

yeast and mammals, mitochondrial CIV is a member of the heme-copper oxidase family of oxidases (A-type oxidases), and reduces molecular oxygen to water with an active site containing a heme a3,

and a copper ion, CuB (Figure 8b). These are located in the Cox1

subunit and form a so-called binuclear site (BNS). The Cox2 subunit also has a copper atom (CuA), and it is the starting point of CIV

electron transfer. Reduced cyt. c associates and dissociates to the IMS side of Cox2 via electrostatic interactions113_{, and this allows}

electrons to transfer from cyt. c to CuA. The electrons then transfer

to the BNS via a heme a, allowing the reduction of molecular oxygen to water. In parallel with the electron transfer, CIV also houses a proton transfer mechanism in which protons are pumped from the matrix to the IMS (Figure 8b). This is done through two proton pathways called the D- and K-pathway, named for the conservative residues D92, and K319 in yeast Cox1114,115_{. A third proton pathway,}

H-pathway, has also been proposed, but in yeast it has been shown to not be active116_{. Cox3 does not take part in either electron transfer}

or proton pumping, but serves as a stabilizer for Cox1 and Cox2 in order to prevent suicide inactivation of CIV117_.

3.6. Complex V - ATP synthase

ADP + Pi + nH+IMS → ATP + nH+Matrix

Although not a part of the electron transport chain, the ATP synthase is often called CV because of its involvement with the respiratory chain complexes to perform OXPHOS. Like CIII and CIV, CV is a complex with subunits encoded both in the nucleus and the mitochondria, and is thus subjected to similar coordination of expression between the two organelles, and the subunit Atp9 forms, as mentioned, a complex with Cox6 of CIV110_{. In yeast, CV is a}

(38)

28

conserved multiprotein enzyme of roughly 600 kDa called the F1Fo

ATP synthase, for its two major modules: F1 and Fo118 (Figure 9a).

The catalytic head domain of CV (F1) is the matrix protruding part

of the complex where the ATP is synthesized from ADP and inorganic phosphate (Pi). This site is made up of several subunits,

subunit b, which connects to subunit a of the Fo module via a

stator-like arm. F1 also contains three copies each of altering α- and

β-subunits at the end of the γ-subunit, which forms a central stalk in the complex together with the δ- and ε-subunit. The β-subunit can bind either ATP or ADP and Pi, and how γ-subunit is oriented will

determine which of these substrates will bind119_{. When ADP and P} i

bind, the reaction can happen and ATP can form. Apart from subunit a, the Fo subunit consists of a c-ring composed of multiple copies of

subunit c (Atp9 in yeast). Subunit c forms a rotating barrel that can interact with subunit a, which has two half channels which are traversable for protons. It is this proton channel in subunit a that allows CV to use the proton motive force that has been built up by the respiratory chain to gain the necessary energy to rotate the c-ring120_{, and the central stalk, which in turn infers conformational}

changes in the β-subunit, which allows for ATP to be generated (Figure 9b).

CV is assembled into multimers, and in BlueNative page analyses of yeast mitochondria, it will often show as a dimer of roughly 1 MDa121_{. The CV dimer forms through connections via additional}

mitochondrial ATP synthase subunits (subunit e-k)122,123_{, which}

provokes the two ATP synthase monomers to be situated together at a V-shaped angle, which varies depending on species. This angle also has an effect on the IMM, and often these CV dimers are found in the curved part of the cristae, influencing their morphology124–127_.

Recently, cryo-electron tomography studies of higher order multimers have shown that complexes of CV can take even more drastic shapes, and have a significant impact on the shaping of the IMM128_.

(39)

29

Figure 9. Subunits and reaction of CV. A) Subunits of CIV, Fo domain contains

subunit C (green) and subunit A (teal). F1 domain contains subunit B (pink), the

α-subunit (blue), β-subunit (yellow), γ-subunit (red), ε-subunit (brown) and δ-subunit (orange). PDB: 5ARE129_{. B) Reaction of ATP synthase, protons from the}

IMS enter into the subunit A half channels and force a rotation of the c-ring which in turn turns the central stalk and allows the β-subunit to make ATP. Blue arrows indicate proton pathways.

(40)

(41)

31

4. Respiratory supercomplexes

4.1. Models of OXPHOS components

The study of OXPHOS and its components have been ongoing for the past century, and how these respiratory complexes (RCCs) interact with each other has been a topic of debate for as long a time. One of the earliest models that gained attraction was the “solid-state model”, in which the RCCs and the electron carriers cyt. c and coenzyme Q, were all components of a large rigid structure called the oxysome130_{. This model did not allow any diffusion of individual}

RCCs nor the redox components. When it became evident that this was the case, another model, “random-collision model”, was introduced131,132_{. Unlike the solid-state model, the random-collision}

model relies entirely on a lateral diffusion of RCCs and electron carriers in the IMM, and it is only by pure chance that these interact with each other when they need to. At the turn of the millennium however, BlueNative page analyses allowed the discovery that RCCs form higher order complexes with each other133–135_{. These}

were named respiratory supercomplexes (SC) or respirasomes, and were found in various organisms ranging from yeast and mammals to bacteria. With this discovery, a new model for how RCCs interacted were popularized: the “plasticity model”136,137_{. It can be}

seen as a combination of the solid-state and random-collision models, and proposes that individual RCCs and SCs exist in an equilibrium, and adapt dynamically depending on which conditions the cells are under. This model is still changing, especially with regards to the electron carriers coenzyme Q and cytochrome c, and how they interact with, and diffuse between, RCCs (chapter 5).

(42)

32

4.2. Structure and assembly of respiratory supercomplexes

In terms of new structures solved, SCs along with ribosomes are amongst the macromolecules in the cell that have benefitted the most from the recent developments in cryo-EM. Since 2016, when the first near-atomic resolution structure of the mammalian CI was published, there have been more than a dozen structures at near-atomic resolution from various species (Table 2).

Table 2. Published structures of near-atomic resolution respiratory

supercomplexes with at least two different RCCs.

Year Variant Organism PDB(s) /

EMD(s)

Reference

2016 CI/CIII2/CIV O. aries 5J4Z, 5J7Y Letts et al.138

CI/CIII2/CIV S. scrofa 5GPN Gu et al.139

CI/CIII2/CIV S. scrofa 5GUP Wu et al.140

CI/CIII2/CIV B. taurus 5LUF Sousa et al.141

2017 CI/CIII2/CIV

CI2/CIII4/CIV2

H. sapiens 5XTH

5XTI

Guo et al.54

2018 CIII2/CIV2 M. smegmatis 6ADQ Gong et al.142

CIII2/CIV2 M. smegmatis 6HWH Wiseman et al.143

ACIII1/CIV F. johnsoniae EMD-7286 Sun et al.144

2019 CIII2/CIV S. cerevisiae 6GIQ Rathore et al.145

CIII2/CIV2 S. cerevisiae 6HU9 Hartley et al.146

2020 CIII2/CIV

CIII2/CIV2

S. cerevisiae 6T15

6T0B

Hartley et al.147

CIII2/CIV S. cerevisiae 6YMX Berndtsson et al.55

2021 CIII2/CIV S. cerevisiae EMD-23414 Moe et al.148

CIII2/CIV V. radiata 7JRP Maldonado et al.149

CIII2/CIV R. capsulatus 6XKW

6XKX 6XKZ Steimle et al.150 CIII2/CIV2 C. glutamicum - Moe et al.151

(43)

33

In yeast, there are two forms of SCs, one with a CIII homodimer bound to one copy of CIV: CIII2/CIV (Figure 10a), and another

where the CIII homodimer is flanked by two copies of CIV: CIII2/CIV2133,145,146. Neither the NADH dehydrogenases nor CII

have ever been shown to be part of any SCs in S. cerevisiae, but other species of yeast such as Yarrovia. lipolytica does contain CI, and have a CI/CIII2/CIV supercomplex152. Arguably, the most notable

feature of all SCs is the composition of the interface surface between the RCCs. In baker’s yeast, there are two major connections: one is a protein-lipid-protein interaction that goes from Qcr8 (CIII) and Cox5 (CIV) via a cardiolipin molecule, and the other connection is a series of protein-protein interactions between Cor1 (CIII) and Cox5 (CIV). The latter interface is strong enough that if it is disrupted, the entire SC ability to form is impaired55_{. There is also a}

minor protein-protein interaction on the IMS side between Qcr6 (CIII) and Cox5 (CIV). Cox5, which is part of the interface in all cases, is also interesting as it exists in two isoforms: Cox5a and Cox5b, the latter isoform being more prevalent in hypoxic conditions, whereas the former is more prevalent in normoxic conditions153–155_{. Structural studies of yeast SC where Cox5b is}

present show no changes in the RCCs interface or activity, but its presence does induce the binding of another protein: Rcf2147_.

Mammalian SCs are more complex than yeast as they also contain CI (Figure 10b). These SCs contain interaction sites between CI-CIII, CI-CIV and CIII-CIV. When compared to yeast SCs one big difference is obvious, and that is that the way CIII and CIV interact has been drastically changed. Instead of relying on COX4I1 (Cox5 in yeast) to be the platform of interaction, the mammalian CIV instead forms weak bonds to CIII via COX7A2 (Cox7 in yeast) to UQCRC1, UQCR11 and UQCRB (Cor1, Qcr10 and Qcr7 in yeast)138–141_{. Interestingly, the same loop of Cor1 (UQCRC1 in}

humans) that binds Cox5 in yeast is in mammalian cells used for the primary interaction between CIII and CI. Mammalian SCs that

(44)

34

contain CI, CIII and CIV are sometimes called respirasomes and exist in different stoichiometry, the CI/CIII2/CIV being the most

common, but larger complexes have also been observed54_{. Smaller}

SCs in mammals have also been observed. The CIII2/CIV complex

is of particular note as it consists of the same subunits as the yeast SC. However, it has been shown that this SC cannot form without another protein SCAF1 (supercomplex assembly factor 1)156,157_{. It is}

however not clear why the interface between CIII and CIV appears in that case, or what it looks like.

Figure 10. Respiratory supercomplexes in yeast and mammals. A) The CIII2/CIV supercomplex of S. cerevisiae. PDB: 6YMX55. B) The CI/CIII2/CIV

respirasome of mammals. PDB: 5XTH54_.

Assembly of yeast SCs is another interesting topic that is not fully understood. Most of the research in this field has been done in mammalian systems, and it has long been assumed that the individual RCCs must finish their assembly first before the SCs can assemble158,159_{. However, a recent study has shown that this might}

not be the case as defective CIII impaired both CI and CIV biogenesis, supporting a cooperative assembly model where the RCCs are not formed first and then assembled into SCs, but rather that SCs are important for overall RCC biogenesis160_{. Another recent}

study shows that only a membrane arm of CI is needed to induce SCs formation161_{. Lipids that are associated with RCCs and SCs,}

(45)

35

particularly CL, have also been suggested to be important for SCs assembly. And whilst some studies suggest that CL is essential for SCs formation162–164_{, other studies show that SCs form without it}55_,

most likely using other lipids in its place. There are several proteins that associate with SCs, in mammals SCAF1 (also COX7A2L, a homologue to COX7A) has already been mentioned as a factor that stabilizes CIII2/CIV supercomplexes and potentially also

respiraosmes165_{, possible by replacing COX7A in CIV}138_{. HIGD1A}

and HIGD2A (Rcf1 in yeast) along with the yeast Rcf2 were long thought to be assembly factors of SCs, but more recent studies are more inclined to single them out as modulators of CIV activity166_.

Baker’s yeast also has a protein called Coi1 (Cytochrome c oxidase interacting protein) that has been suggested to play a part in SCs assembly167_{. However, since knockout strains of Coi1 decreased the}

levels of individual RCCs, further studies are required to fully understand how Coi1 affects SCs assembly.

4.3. Why does supercomplexes exist?

The discovery of SCs naturally led to the two questions of how they look, and what their function is. The fact that SCs are present in such a diverse group of organisms indicates that they serve a universal purpose. Cryo-EM has allowed researchers to begin answering questions about SCs structure, but the question of the functional significance has been a topic of debate ever since its inception. An early suggestion regarding SCs function was that they would increase overall OXPHOS activity by means of channeling substrates, either coenzyme Q or cyt. c, between its RCCs168,169_.

Substrate channeling has however met strong opposition for various reasons. Structures of SCs have shown that there are no special compartments for the redox factors to be sequestered into, and that both coenzyme Q and cyt. c can freely diffuse into their respective general pools138–141_{. This was confirmed by experiments where}

(46)

36

alternative oxidase which was not a part of any SC170_{, and the fact}

that coenzyme Q trapped within CI/CIII2 SCs shows reduced CI

activity171_{. The fact that CII is not part of mammalian or yeast SCs}

also argues against coenzyme Q substrate channeling. Crystal structures of cyt. c binding to CIII and CIV, superimposed on structures of SCs, also reveals that the distance between these two sites is too great (~100 Å in mammals, ~70 Å in yeast) to facilitate direct electron transfer between the RCCs145_{. It should be noted that}

SCs could still form to decrease that distance and thus facilitate better diffusion between CIII and CIV (chapter 5.1). Another point that has been put forward relating to SCs and its catalytic process is that SCs might help to decrease the level of ROS that is formed during catalysis172–174_{. The experiments suggesting such scenarios}

are however not conclusive as they targeted RCCs assembly and the increase in ROS production observed could be due to that rather than the lack of SCs. SCs formation could also be the result of micro optimization to its environment. Considering that the IMM is amongst the most protein-packed membranes in the cell with only ~20% lipids11_{, having all those proteins efficiently packed in this}

membrane will undoubtedly lead to more OXPHOS activity. SCs could therefore be a way to avoid aggregation due to overcrowding175_.

(47)

37

5. Mobile electron carriers

5.1. Cytochrome c

Cyt. c is often depicted, including in the better part of this thesis, in line with the plasticity model of RCCs, as a water-soluble mobile electron carrier that transports electrons from CIII to CIV via free diffusion in the IMS. This type of diffusion model is called 3D-diffusion (Figure 11a), in that it means that the cyt. c is free to diffuse in all dimensions and exist as a general pool of cyt. c in the IMS. In yeast, cyt. c binds to CIII (Cyt1), and CIV (Cox2), via electrostatic interactions where both these subunits have acidic patches of amino acids which can interact with the positively charged cyt. c. Structurally there are no restrictions on cyt. c diffusion between CIII and CIV145,146_{, and cyt. c binding to the RCCs and its maximum}

efficiency in terms of electron transport is determined only by the ionic strength of its surroundings176_{. A more recent study on the}

kinetics advantages of SCs showed that indeed within the parameters of a 3D-diffusion model, cyt. c diffusion could be the rate-limiting step of electron transfer in the respiratory chain and thus the raison d’être for SCs177_.

The 3D-diffusion model for cyt. c has however been questioned in light of both old kinetics studies178–182_{, as well as recent structural}

evidence for alternative binding sites of cyt. c along a negatively charged “path” between the cyt. c binding sites on CIII and CIV in yeast148_{. Furthermore, NMR studies combined with isothermal}

titration calorimetry experiments of cyt. c binding with plant CIII and CIV report that both RCCs can bind cyt. c at two sites: a distal site that cannot facilitate direct electron transfer between cyt. c and CIII, and a proximal site which can mediate that electron transfer183,184_{. In light of these discoveries a rivaling hypothesis has}

(48)

38

3D but rather in 2D, so that it stays and binds to either the RCCs185,186

or lipids in the membrane187,188_{, and diffuse laterally along the}

complexes or the IMM (Figure 11b). As previously mentioned, cyt.

c binds RCCs electrostatically, and the ionic strength of the

environment dictates their ability to interact as well as which type of diffusion will be prominent176_{. The exact ionic strength of the IMM}

is not known, but the general ionic strength in cells is usually equivalent to 100-150 mM NaCl, which means that cyt. c binding wi not be particularly strong under these conditions176_{. A recent study}

on ionic strength in E. coli however has challenged this, as the ratio between small cations and anions becomes warped due to small anions becoming sequestered into larger complexes189_{, and the de}

facto ionic strength would be closer to an equivalent of 20 mM NaCl.

This would affect the ionic strength such as the Debye screening length, which is the distance at which electrostatic molecules “feel” each other. This would be increased from 0.8 nm to 2.2 nm189_{. This}

would favor 2D-diffusion of cyt. c186_{. Some organisms, such as the}

bacteria Mycobacterium smegmatis, have cyt. c that does not diffuse but rather is an integral part of their SCs142,143_{, and other bacteria}

such as Rhodobacter sphaeroides have cyt. c versions that are bound to the membrane149,188–190_{. A recent study with yeast cyt. c fused to}

the transmembrane domain of Cyb2 and the linker of R. sphaeroides cyt. cγ showed that electron transfer was possible, but that such cells

suffered from increased cell death via apoptosis193_{. It should,}

however, be mentioned that such a construct is not limited to 2D-diffusion, as the linker segment allows for a limited type of 3D-diffusion to occur as well.

(49)

39

Figure 11. Different types of cytochrome c diffusion. A) Yeast supercomlex with cyt. c diffusion in 3D in the IMS. B) Yeast supercomlex with cyt. c diffusion in 2D along the IMM.

Although cyt. c has primarily been presented within this thesis as the electron carrier between CIII and CIV in the respiratory chain, it should be noted that cyt. c also has other functions within the cell. Inside of the mitochondria, cyt. c also serves as an electron acceptor from the Mia40/Erv1 pathway of protein import194–196_{. In short,}

proteins with thiol groups that get imported into the mitochondria need to be oxidized so that the disulfide bridges can form. This is done by the Mia40/Erv1 pathway where electrons from the thiol groups are transferred, first onto Mia40 and then to Erv1. From Erv1 the electrons can either move to cyt. c, to molecular oxygen, or under anaerobic conditions to the soluble fumarate reductase Osm1197_{. A}

recent study shows that the interaction between Erv1 and cyt. c is not an electrostatic binding, as in the case with RCCs, but rather that a collision type interaction between the two proteins conveys electron transfer198_{. Whether this is compatible with a model of 3D- or 2D}

diffusion of cyt. c , or both, or neither, was not addressed and remains unknown.

Cyt. c also has roles outside of the mitochondria, the most prominent of which is its role in programmed cell death, apoptosis199,200_{. This}

(50)

40

such as yeast201,202_{, presumably as an altruistic sacrifice as yeast}

forms communities in the form of biofilms203_{. When released into}

the cytosol, cyt. c becomes toxic to cells, as it forms apoptosomes together with apoptotic protease activating factors (APAF1), which induces cascades of caspases in the cytosol, which in turn induces apoptosis. Furthermore, cyt. c is also thought to interact with other proteins to induce apoptotic pathways as well as block pro-survival pathways204–208_{. Another interesting place that cyt. c outside of the}

mitochondria has been tied to is the nucleus, where it has been proposed to act in response to DNA damage by binding to histone chaperons and affecting nucleosome assembly activity209,210_.

However, such a mechanism would require cyt. c to be able to exit the mitochondria in such a way that it does not trigger apoptosis, as well as be translocated to the nucleus. How this is done is still unknown.

5.2. Coenzyme Q synthesis

The other mobile electron carrier in the respiratory chain is coenzyme Q. Coenzyme Q is an interesting molecule for many reasons; it plays an important part in the respiration and it functions as an antioxidant around various cell membranes211_{. It also has a very}

fascinating synthesis pathway. In yeast coenzyme Q is synthesized as coenzyme Q6, where the number six refers to the number of

carbon atoms in its isoprene side chain synthesized by Coq1212_{. The}

other part of coenzyme Q is its head group which stems from 4-hydrobenzonate or para-aminobenzoic acid and is attached to the lipid side chain by Coq2213_{. Next, a series of modifications to the}

head group occur212_{. These modifications are done by seven}

enzymes, Coq3-9, collectively called the CoQ-synthome214_{. This}

assembly is believed to associate into a larger complex consisting of several sub-modules215,216_{. Exactly how the CoQ-synthome is}

structurally arranged is not known, but it is thought to be associated with the IMM due to Coq2 having transmembrane helecis217_{. Coq3,}

(51)

41

Coq4, Coq5, Coq6 and Coq9 perform the first steps of the head group modification, resulting in a DMQ molecule. The full CoQ-synthome is completed when Coq7 associates to it212,218_{. Coq7 is a}

metalloprotein with a diiron center in its active site, thought to be important for its function and structure219_{. It performs the}

penultimate step of the maturation of coenzyme Q, catalyzing the hydroxylation of the C5 of DMQ to DMeQ ,which then can be matured further into coenzyme Q6 and be distributed by Coq10 and

(52)