미토콘드리아 '대사관점' 4가지 기능!! death signal 말고!!

[문형철]

1. 핵심 리뷰 (가장 직접적으로 관련된 논문)

Spinelli JB, Haigis MC.
“The multifaceted contributions of mitochondria to cellular metabolism.”
Nature Cell Biology 20, 745–754 (2018).
DOI: 10.1038/s41556-018-0124-1
인용 수: 약 1,900회 이상 (매우 높은 영향력)

핵심 내용

• 미토콘드리아는 ATP 생산 외에도
  • 영양소를 분해하고
  • 생합성 전구체를 만들며
  • 산화환원 항상성을 위해 대사물질을 구획화하고
  • 대사 폐기물 관리의 허브(hub for metabolic waste management) 역할을 한다.
• 이 논문은 “대사 산물 처리”라는 측면을 명확히 다루는 대표 리뷰로, 무케르지의 비유와 가장 잘 맞습니다.

https://pmc.ncbi.nlm.nih.gov/articles/PMC6541229/

제목: The multifaceted contributions of mitochondria to cellular metabolism
(미토콘드리아의 세포 대사에 대한 다면적 기여)
저자: Jessica B. Spinelli, Marcia C. Haigis
저널: Nature Cell Biology (2018 Jul; 20(7):745–754)
DOI: 10.1038/s41556-018-0124-1
인용: 약 2,000회 이상 (고인용 리뷰)

미토콘드리아를

단순히 “세포의 발전소(ATP 생산)”로만 보는 고전적 관점을 넘어,

세포 대사의 다면적 허브로 재해석한 리뷰 논문입니다.

미토콘드리아의 대사관점에서 4가지 주요 역할

역할                                                                                          설명

1. 에너지 생산 (Bioenergetics)	피루브산, 지방산, 아미노산을 산화하여 NADH/FADH₂를 만들고, 전자전달계를 통해 ATP 생성
2. 생합성 전구물질 공급 (Biosynthetic hub)	핵산, 지방산, 콜레스테롤, 아미노산, 글루코스, 헴 합성에 필요한 중간체 제공
3. 산화환원 항상성 유지 (Redox balance)	NAD⁺/NADH, NADP⁺/NADPH 균형을 구획화(compartmentalization)를 통해 조절
4. 대사 노폐물 관리 (Waste management hub)	ROS, 암모니아 등 부산물을 처리하거나 재활용

주요 포인트

• 미토콘드리아는 **구획화(compartmentalization)**를 통해 서로 다른 대사 기능을 동시에 수행할 수 있습니다.
• 영양소 결핍, 산화 스트레스, DNA 손상, ER 스트레스 등 세포 스트레스에 대한 적응 반응에도 핵심적으로 관여합니다.
• 정상 생리뿐 아니라 암, 대사 질환 등 질병 상태에서의 중요성도 다룹니다.

한 줄 요약

미토콘드리아는 ATP만 만드는 기관이 아니라, 에너지 생산 + 생합성 재료 공급 + 산화환원 조절 + 노폐물 처리를 동시에 수행하는 세포 대사의 중심 허브라는 것을 체계적으로 정리한 고전적 고인용 리뷰입니다.

전체 구조

• 바깥쪽: 세포질(Cytosol)
• 안쪽 큰 타원: 미토콘드리아 기질(Matrix)
• 이중막: 외막(OMM)과 내막(IMM)
• 내막에 전자전달계(ETC)와 ATP 합성효소가 위치

주요 대사 경로

1. 포도당 → 피루브산 (세포질)

• 포도당이 해당과정(Glycolysis)을 거쳐 PEP → 피루브산이 되며 2 ATP를 만듭니다.
• 피루브산은 두 갈래로 갈 수 있습니다.
  • LDH를 통해 젖산(Lactate)으로 전환 (NADH → NAD⁺ 재생)
  • MPC(미토콘드리아 피루브산 운반체)를 통해 미토콘드리아 안으로 진입

2. 미토콘드리아 안에서의 피루브산 처리

• PDC(피루브산 탈수소효소 복합체)가 피루브산을 아세틸-CoA로 전환합니다.
  • 이 효소는 PDK4에 의해 인산화되면 활성이 억제됩니다.
• 동시에 PC(피루브산 카르복실라제)는 피루브산을 옥살로아세트산으로 만들어 TCA 회로를 보충(anaplerosis)합니다.

3. TCA 회로 (크렙스 회로) 기질 안에서 아세틸-CoA와 옥살로아세트산이 결합해 구연산(Citrate)이 되고, 다음 순서로 돌아갑니다.

• CS → IDH2 → OGDH → SDH → MDH2
• 과정에서 NADH, FADH₂가 대량 생성되어 전자전달계로 들어갑니다.
• 중간에 나오는 중간체들(α-KG, 숙시닐-CoA 등)은 다른 대사와 연결됩니다.

4. 전자전달계와 ATP 합성 (내막)

• Complex I, II, III, IV를 통해 전자가 이동하면서 양성자(H⁺)를 막 밖으로 밀어냅니다.
• 그 농도 기울기를 이용해 ATP 합성효소가 ADP + Pi → ATP를 만듭니다.
• 최종 전자 수용체는 산소(O₂)이며, 물이 생성됩니다.

5. 지방산 산화 (β-oxidation)

• 중성지방(TAG) → 지방산 → ACS로 Fatty Acyl-CoA
• CPT1이 카르니틴과 결합해 아실카르니틴으로 만들어 미토콘드리아 안으로 운반 (CPT2로 다시 Fatty Acyl-CoA 복원)
• 이후 β-산화를 통해 아세틸-CoA를 대량 생산합니다.
• CPT1은 말로닐-CoA에 의해 억제되며, 말로닐-CoA는 ACC(아세틸-CoA 카르복실라제)가 만듭니다.
• AMPK가 ACC를 인산화하면 말로닐-CoA 생성이 줄어들어 지방산 산화가 촉진됩니다.

6. 아미노산 대사

• 글루타민 → GLS → 글루탐산 → GDH → α-케토글루타르산 (TCA 진입)
• 분지쇄 아미노산(이소류신, 류신, 발린)도 BCAT2, BCKDH 등을 통해 TCA 중간체로 들어갑니다.

1. 뉴클레오타이드 합성 (좌측 상단)

미토콘드리아와 세포질이 협력해 퓨린·피리미딘을 만드는 경로입니다.

• 핵심은 **1탄소 대사(one-carbon metabolism)**입니다. 세린(Serine)과 글리신(Glycine)이 메틸렌테트라하이드로폴레이트(5,10-CH₂THF)를 만들고, 이것이 다시 10-Formyl-THF나 포메이트(Formate)로 전환됩니다.
• 주요 효소:
  • 세포질: SHMT1, MTHFD1
  • 미토콘드리아: SHMT2, MTHFD2, MTHFDL1, FTDH
• 피리미딘 합성의 마지막 단계에 있는 DHODH(디하이드로오로테이트 탈수소효소)는 미토콘드리아 내막에 위치하며, 전자전달계와도 연결됩니다.
• 이 경로에서 생성된 1탄소 단위가 퓨린과 피리미딘 합성에 사용됩니다.

2. 포도당·헴 합성 (우측 상단)

글루코네오제네시스와 헴(Heme) 합성이 미토콘드리아에서 시작되거나 연결되는 부분입니다.

• 글루코네오제네시스:
  • 피루브산 → PC(피루브산 카르복실라제) → 옥살로아세트산(Oaa)
  • Oaa → MDH2 → 말산 → 세포질로 나가 MDH1, PCK1을 거쳐 PEP 생성
  • 미토콘드리아 내부에도 PCK2가 존재합니다.
• 헴 합성:
  • 숙시닐-CoA + 글리신 → ALAS(아미노레불린산 합성효소) → δ-아미노레불린산
  • 이후 여러 단계를 거쳐 PPIX(프로토포르피린 IX)가 되고, FECH(페로킬라타제)가 철을 넣어 헴을 완성합니다.
  • 중간체 수송에 ABCB6, SLC25 계열 운반체(25A11, 25A10 등)가 관여합니다.

3. 지방산·콜레스테롤 합성 (좌측 하단)

미토콘드리아에서 나온 구연산(Citrate)이 세포질로 나가 지방산과 콜레스테롤의 원료가 되는 경로입니다.

• 구연산이 세포질로 수송된 뒤 ACLY(ATP 구연산 리아제)에 의해 아세틸-CoA와 옥살로아세트산으로 분해됩니다.
• 아세틸-CoA는 ACC2(아세틸-CoA 카르복실라제 2)에 의해 말로닐-CoA가 되고, 이것이 지방산 합성의 출발점이 됩니다.
• PHD3가 ACC2의 활성을 조절하는 것으로 표시되어 있습니다.
• 일부 경로에서는 환원적 카르복실화(Reductive Citrate Carboxylation)를 통해 α-KG에서 구연산을 만드는 역방향 흐름도 나타납니다.
• 아세틸-CoA는 콜레스테롤, 케톤체 합성에도 사용됩니다.

4. 아미노산 합성 (우측 하단)

글루타민을 중심으로 여러 아미노산과 요소회로 관련 물질이 만들어지는 부분입니다.

• 글루타민 → GLS → 글루탐산 → GDH 또는 아미노기 전이효소 → α-KG
• 글루탐산은 다시:
  • GS로 글루타민 재합성
  • P5CS → PYCR 경로로 프롤린 합성
  • OAT를 통해 오르니틴 생성
• 오르니틴은 ORNT1을 통해 미토콘드리아 안팎을 이동하며 시트룰린과 연결됩니다(요소회로).
• GOT2, GPT2를 통해 아스파르트산, 알라닌도 생성됩니다.

1. 말산-아스파르트산 셔틀 (Malate-Aspartate Shuttle)

가장 중요한 NADH 수송 시스템입니다.

• 세포질에서 NADH가 축적되면, 옥살로아세트산(Oaa)이 MDH1에 의해 말산(Mal)으로 환원됩니다.
• 말산이 Malate-α-KG Antiporter를 통해 미토콘드리아 안으로 들어갑니다.
• 미토콘드리아 안에서 MDH2가 말산을 다시 옥살로아세트산으로 산화시키면서 NADH를 생성합니다.
• 옥살로아세트산은 GOT2에 의해 아스파르트산으로 전환되고, 아스파르트산은 세포질로 나갑니다.
• 결과적으로 세포질의 NADH가 미토콘드리아 안의 NADH로 전환되어 전자전달계에 사용될 수 있습니다.

2. 말산-구연산 셔틀 (Malate-Citrate Shuttle)

지방산 합성이나 세포질 아세틸-CoA 공급과 연결된 셔틀입니다.

• 미토콘드리아 안에서 CS(구연산 합성효소)가 구연산을 만들고, 구연산이 Malate-Citrate Antiporter를 통해 세포질로 나갑니다.
• 세포질에서 ACLY가 구연산을 아세틸-CoA와 옥살로아세트산으로 분해합니다.
• 옥살로아세트산은 MDH1에 의해 말산이 되면서 세포질 NADH를 소비합니다.
• 말산은 다시 미토콘드리아로 돌아가 MDH2에 의해 NADH를 생성합니다.
• 이 셔틀은 특히 지방산·콜레스테롤 합성 시 세포질에 아세틸-CoA를 공급하면서 동시에 redox를 조절하는 역할을 합니다.

3. α-글리세로포스페이트 셔틀 (α-Glycerophosphate Shuttle)

비교적 단순한 NADH 산화 경로입니다.

• 세포질의 NADH가 cGPDH에 의해 디하이드록시아세톤 포스페이트(DHAP)를 글리세롤-3-포스페이트(GAP)로 환원합니다.
• GAP가 미토콘드리아 내막의 mGPDH에 의해 다시 DHAP로 산화되면서 FADH₂를 생성합니다.
• 이 FADH₂는 전자전달계로 직접 전자를 전달합니다.
• 말산-아스파르트산 셔틀보다 효율은 낮지만, 속도가 빠르고 조직에 따라 중요한 역할을 합니다.

4. 폴레이트 셔틀 (Folate Shuttle / 1C Metabolism)

NADPH와 1탄소 단위를 교환하는 경로입니다.

• 미토콘드리아 안에서 MTHFD2, SHMT2가 작용해 NADPH를 생성하고, 10-formyl-THF나 포메이트를 만듭니다.
• 포메이트가 세포질로 나가면, 세포질의 MTHFD1과 SHMT1이 이를 이용해 다시 1탄소 단위를 재구성하면서 NADPH를 소비하거나 생성합니다.
• 이 셔틀은 특히 뉴클레오타이드 합성과 세포질·미토콘드리아 간의 NADPH 균형을 맞추는 데 중요합니다.

정상 세포 (Healthy)

• 포도당과 아미노산을 적당량 섭취합니다.
• 이를 이용해 거대분자(단백질, 핵산 등)와 에너지를 만듭니다.
• 대사 과정에서 생기는 노폐물인
  H₂S(황화수소), NH₃(암모니아), 젖산(Lactate), ROS는
  세포 밖으로 분비된 뒤 빠르게 제거(clearance)됩니다.
• 결과적으로 세포 안팎의 노폐물 농도가 낮게 유지됩니다.

암세포 (Cancer)

• 포도당과 아미노산 섭취량이 크게 증가합니다(increased consumption).
• 대사 속도 자체도 빨라져(increased metabolism) 거대분자와 에너지를 많이 만듭니다.
• 그 결과 노폐물 생성량이 과도하게 늘어납니다(excess metabolic waste).
  • H₂S, NH₃, Lactate, ROS가 정상보다 훨씬 많이 생성됩니다.
• 일부는 재활용되거나 신호 전달에 이용되지만(Recycling & Signaling),
  대부분은 세포 밖으로 분비된 뒤 제대로 제거되지 못하고 축적됩니다(Secretion & Accumulation).
• 이렇게 축적된 노폐물들이 종양 미세환경을 산성화하고, 면역세포 기능을 억제하며, 주변 정상 세포에도 영향을 미칩니다.

B. 황화수소 (Hydrogen Sulfide, H₂S)

• H₂S는 주로 세 효소에 의해 생성됩니다.
  • CBS(시스타티오닌 β-합성효소)
  • CSE(시스타티오닌 γ-리아제)
  • 3MST(3-머캅토피루베이트 황전달효소)
• 장내 미생물도 H₂S를 만듭니다.
• 생성된 H₂S는 미토콘드리아 내막의 SQR(황화물 퀴논 산화환원효소)에 의해 산화됩니다.
• SQR은 H₂S를 산화시키면서 전자를 전자전달계로 전달하고, 그 과정에서 양성자(H⁺)를 펌핑해 ATP 생성에 기여할 수 있습니다.
• 최종적으로 SO₄²⁻(황산염) 형태로 산화되어 노폐물로 배출됩니다.
• 적정량의 H₂S는 전자 공여체로 작용하지만, 과다하면 독성을 나타냅니다.

C. 암모니아 (Ammonia, NH₃)

암모니아는 주로 아미노산과 질소 염기 대사에서 생성됩니다.

• 아미노산 → 글루탐산/α-케토글루타르산 과정에서 NH₃가 나옵니다.
• 미토콘드리아 안에서는:
  • CPS1이 NH₃를 이용해 카르바모일 인산을 만들고
  • OTC가 시트룰린을 합성해 요소회로로 보냅니다.
• 세포질에서 요소회로가 완성되어 요소(Urea)로 배출됩니다.
• 동시에 NH₃는:
  • GS(글루타민 합성효소)를 통해 글루타민으로 재고정
  • 자가포식(Autophagy) 조절
  • 미생물 대사에도 관여
• 과도한 암모니아는 미토콘드리아 기능을 저해하고, 종양에서는 주변 환경을 변화시키는 신호로 작용할 수 있습니다.

D. 활성산소 (ROS)

• 전자전달계(Complex I, II, III)에서 전자가 새어나오면서 **슈퍼옥사이드(O₂⁻)**가 생성됩니다.
• SOD2(미토콘드리아 슈퍼옥사이드 디스뮤타제)가 O₂⁻를 **과산화수소(H₂O₂)**로 전환합니다.
• H₂O₂는 철(Fe²⁺)과 만나 펜톤 반응을 통해 매우 반응성이 높은 **수산화 라디칼(OH·)**을 만들 수 있습니다.
• 또한 NOX4도 ROS 생성에 기여합니다.
• ROS의 이중적 역할:
  • 적정량: 세포 증식, 분화, 이동을 촉진하는 신호 분자로 작용
  • 과다량: DNA, 단백질, 지질을 손상시키는 산화 스트레스를 유발

2. 최신 고영향력 리뷰 (2024)

Suomalainen A, Nunnari J.
“Mitochondria at the crossroads of health and disease.”
Cell 187(11), 2601–2627 (2024).
DOI: 10.1016/j.cell.2024.04.037

핵심 내용

• 미토콘드리아가 이화(catabolic)와 동화(anabolic) 대사의 교차점에 있다고 설명합니다.
• 조직별 에너지·성장·재생 요구에 따라 구조와 기능이 동적으로 조절되며, 대사 신호(metabolic signaling)를 통해 세포와 멀리 떨어진 조직까지 영향을 미칩니다.
• 질병과 노화에서 미토콘드리아 기능의 중심성을 포괄적으로 다룬 최신 리뷰입니다.

https://www.cell.com/cell/fulltext/S0092-8674(24)00463-X

• Mitochondria at the crossroads of health and disease
• 저자: Anu Suomalainen, Jodi Nunnari
• 저널: Cell (2024년 5월 23일), Volume 187, Issue 11, pp. 2601–2627
• DOI: 10.1016/j.cell.2024.04.037
• 형태: Open Access Review (Cell 창간 50주년 기념호)
• 특징: 2012년 같은 저자들이 Cell에 발표한 “Mitochondria: In Sickness and in Health”의 업데이트 리뷰

미토콘드리아는

단순한 ATP 공장이 아니라 이화작용(catabolism)과 동화작용(anabolism)의 교차점에 있으며,

세포와 개체 전체의 생리·병리를 조절하는 중심 기관이다.

조직별 에너지·성장·재생 요구에 따라 구조와 기능이 동적으로 조절되며,

내·외부 스트레스에 반응해 대사 신호를 세포 내부와 다른 조직으로 전달한다.

이 리뷰는 이러한 중심성이

건강, 다양한 질병, 노화에서 어떻게 나타나는지를 종합적으로 정리한다.

주요 내용 포인트

1. 미토콘드리아의 이중적 역할
   에너지 생산뿐 아니라 생합성 전구체(뉴클레오타이드, 헴, 지방산, 아미노산 등) 공급, redox 균형, 대사 신호 전달의 허브 역할을 수행한다. (사용자가 이전에 올린 그림들이 이 논문의 주요 그림으로 추정됨)
2. 미토콘드리아 질병의 광범위성과 조직 특이성
   • 미토콘드리아 단백질의 40% 이상이 인간 질병과 연관됨.
   • 같은 유전자 돌연변이라도 조직·나이에 따라 전혀 다른 표현형을 보임 (예: Complex I 결함 → 성인 시신경 위축 또는 소아 Leigh 증후군).
   • ATP 부족만으로는 설명되지 않으며, 대사 재프로그래밍, 스트레스 반응, 조직 간 신호 전달이 핵심이다.
3. 대사 신호와 전신 조절
   • 미토콘드리아 스트레스 시 FGF21, GDF15 같은 **metabokine(mitokine)**이 분비되어 전신 대사를 조절.
   • Integrated stress response (ISRmt), one-carbon metabolism, redox shuttles 등이 세포 내·외 신호로 작용.
4. 질병과 노화와의 연결
   • 일차성 미토콘드리아 질환뿐 아니라 신경퇴행성 질환, 대사증후군, 심근병증, 암, 노화 등 광범위한 질환에 관여.
   • 조직 특이적 메커니즘(예: mtDNA 유지, 뉴클레오타이드 풀 조절)이 질병의 다양성을 설명.
5. 최신 개념 반영
   2012년 리뷰 이후 축적된 지식(미토콘드리아 역학, 품질관리, 세포 간 미토콘드리아 전달, 대사 재프로그래밍, 조직 간 신호 등)을 통합적으로 업데이트.

이 논문의 의의

미토콘드리아를 “세포의 발전소”에서 세포·조직·개체의 대사와 신호를 통합하는 중심 허브로 재정의하는 포괄적 리뷰입니다. 사용자가 지금까지 올려주신 그림들(에너지 대사, 생합성 경로, redox shuttle, 종양 미세환경의 노폐물 축적, H₂S·암모니아·ROS 처리 등)이 바로 이 논문의 핵심 도식과 잘 맞습니다.

A. 전자현미경 사진 (TEM)

미토콘드리아의 초미세 구조를 보여줍니다.

• OM: 외막 (Outer Membrane)
• IM: 내막 (Inner Membrane)
• IMS: 막간 공간 (Intermembrane Space)
• Cr: 크리스타 (Cristae) — 내막이 안쪽으로 접혀 만들어진 주름. 전자전달계와 ATP 합성효소가 주로 위치
• CrJ: 크리스타 접합부 (Crista Junction) — 크리스타가 내막과 연결되는 좁은 부위
• ER: 소포체 (Endoplasmic Reticulum) — 미토콘드리아와 접촉하고 있는 모습

크리스타 구조가 잘 발달되어 있어 에너지 생산 효율이 높은 전형적인 미토콘드리아 형태입니다.

B. 3차원 재구성 이미지

전자단층촬영(Electron Tomography) 등으로 만든 3D 모델입니다.

• 회색 부분: 크리스타가 겹겹이 배열된 모습
• 노란색 부분: 외막과 크리스타 접합부, 그리고 주변 막 구조
• 크리스타가 단순한 주름이 아니라 복잡한 3차원 공간을 형성하고 있음을 보여줍니다.

C. 형광현미경 이미지 (세포 전체)

• 초록색: 미토콘드리아 네트워크 (길쭉하고 가지를 친 형태)
• 자주색: 세포핵
• 미토콘드리아가 세포질 전체에 걸쳐 연결된 네트워크를 이루고 있는 모습입니다. 분열(fission)과 융합(fusion)을 통해 형태가 계속 변합니다.

D. 고배율 형광현미경 이미지

• 초록색: 미토콘드리아 네트워크 (그물처럼 촘촘하게 연결됨)
• 빨간색: 일부 미토콘드리아 또는 특정 마커 (예: 손상된 미토콘드리아, 특정 단백질 등)
• 파란색: 세포핵
• 여러 세포에서 미토콘드리아가 균일한 네트워크를 형성하고 있는 모습을 보여줍니다.

이 그림은

미토콘드리아가 세포의 다양한 기능에 기여하는 방식(패널 A)과,

그중에서도 1탄소 대사(Folate cycle)를 중심으로 한 대사 연결망(패널 B)을 보여줍니다.

패널 A: 미토콘드리아의 다기능성

미토콘드리아에서 나오는 다양한 산물과 그 기능이 정리되어 있습니다.

왼쪽 (세포 기능 조절)

• mtDNA/mtRNA → 선천면역 활성화
• 지질(Lipids) → 세포막 구성
• 펩타이드 → 스트레스 반응·신호 전달
• 헴(Heme) → 효소 보조인자, 헤모글로빈, 산화적 인산화(Oxphos)
• Ca²⁺ → 소포체(ER)와 함께 칼슘 항상성 조절

아래쪽 (핵심 대사산물)

• Formate → 퓨린 합성, 미토콘드리아 번역
• ATP → 에너지, 뉴클레오타이드, 인산
• NAD⁺/NADH, NADP/NADPH → redox 조절, 영양소 대사, 오탄당 인산 경로, 피리미딘 합성

오른쪽 (신호·보조인자)

• Cu, Fe → Fe-S 클러스터, 보조인자
• ROS → 신호 전달, 줄기세포 분화
• NO, CO₂ → 혈관 긴장도 조절
• H₂S, HCN → 황 대사
• 아미노산 대사산물 → 성장, 신호, 수선, 후성유전, 단백질 번역

패널 B: 폴레이트 회로와 연결된 대사 네트워크

미토콘드리아와 세포질이 협력하는 1탄소 대사를 중심으로, 여러 경로가 어떻게 연결되는지 보여줍니다.

1. Folate cycle (폴레이트 회로)

• 미토콘드리아: SHMT2, MTHFD2, MTHFD1L 등이 작용해 세린/글리신으로부터 1탄소 단위를 만들고 Formate를 생성
• 세포질: SHMT1, MTHFD1, TYMS 등이 작용해 dTMP(피리미딘) 합성과 10-formyl-THF 생성
• 미토콘드리아에서 나온 Formate가 세포질로 나가 퓨린 합성에 사용됨

2. Transsulfuration (트랜스설퍼레이션)

• 세린 + 호모시스테인 → CBS → 시스타티오닌 → 시스테인
• 시스테인은 글루타티온 합성의 원료가 되며, 이 과정에서 H₂S도 생성됨

3. Methyl cycle (메틸 회로)

• 5-methyl-THF → 메티오닌 → SAM (S-아데노실메티오닌)
• SAM은 메틸 공여체로 작용해 후성유전 조절, 크레아틴·포스파티딜콜린 합성 등에 사용됨

4. Purine synthesis (퓨린 합성)

• 10-formyl-THF가 IMP 합성에 사용되고, 이후 AMP, GMP 등으로 전환됨

패널 A: 미토콘드리아와 다른 소기관의 상호작용

미토콘드리아는 단독으로 존재하지 않고 여러 소기관과 물질·신호를 교환합니다.

• 소포체(ER): 지질과 Ca²⁺를 교환. 미토콘드리아-ER 접촉 부위(MAM)에서 칼슘 신호와 지질 전달이 일어남
• 지질 방울(Lipid Droplet): 지질을 주고받음
• 핵(Nucleus): 뉴클레오타이드와 대사산물을 교환
• 리소좀(Lysosome): 대사산물, 철(Fe) 등을 교환. “Kiss and run” 방식으로 일시적으로 접촉
• 퍼옥시좀(Peroxisome): β-산화 관련 대사산물 교환
• 미토콘드리아 유래 소포(MDVs): 미토콘드리아에서 작은 소포가 떨어져 나와 항원이나 손상된 성분을 운반. 일부는 미토콘드리아 항원으로 제시됨

패널 B: 미토콘드리아 분열 (Fission)

미토콘드리아가 둘로 나뉘는 과정입니다.

• DRP1(Dynamin-related protein 1)이 미토콘드리아 표면에 고리처럼 모여 수축을 일으킴
• 액틴(Actin) 섬유가 이 과정을 도와줌
• 분열 부위에서 Ca²⁺ 신호가 관여
• 각 미토콘드리아 안에는 뉴클레오이드(Nucleoid) — mtDNA가 단백질과 결합한 구조 — 가 존재
• 분열을 통해 손상된 부분을 분리하거나, 세포 분열 시 미토콘드리아를 분배함

패널 C: 미토콘드리아 품질 관리 (Quality Control)

손상된 미토콘드리아를 처리하는 여러 경로입니다.

1. 국소적 품질 관리
   • 미토콘드리아 내부 프로테아제가 손상된 단백질을 분해
   • MDV(미토콘드리아 유래 소포)를 통해 손상된 부분을 선택적으로 배출
2. 세포 사멸 관련
   • 심한 손상 시 시토크롬 c(Cyt c)가 방출되어 세포 사멸을 유도
3. 미토파지 (Mitophagy)
   • 손상된 미토콘드리아 전체를 자가포식막으로 감싸 리소좀으로 보내 분해
   • Piecemeal mitophagy: 미토콘드리아의 일부만 선택적으로 제거하는 방식도 존재
4. 단백질 회전과 신호 전달
   • 리소좀에서 분해된 성분이 재활용되거나 신호로 작용

패널 A: 줄기세포 니치 (Stem Cell Niche, SCN)

줄기세포가 머무는 특수 환경입니다.

• 줄기세포: 산화적(Oxidative), 휴지기(quiescent) 상태
  → 미토콘드리아 대사를 주로 사용하며, 증식하지 않고 조용히 유지됨
• 돌봄 세포(Caretakers): 동화적(Anabolic), 해당과정(glycolytic) 상태
  → 주변 세포들이 해당과정을 통해 줄기세포를 지원
  • 뇌실막 세포(Ependymal) → 신경 줄기세포 니치
  • 지방세포(Adipocytes) → 조혈 줄기세포 니치
  • 파네트 세포(Paneth) → 장 줄기세포 니치

패널 B: 증식 세포 (Proliferative Cells)

활발히 분열하는 세포들입니다.

• 대사 상태: 동화적(Anabolic), 해당과정(glycolytic)
• 해당과정을 통해 빠르게 에너지를 얻고, 동시에 세포 구성 물질(핵산, 지질, 단백질)을 대량 생산
• 해당하는 세포:
  • 증식 중인 전구세포(Proliferating progenitors)
  • 세포 질량을 늘리는 세포
  • 암세포

패널 C: 분화 세포 (Differentiation)

최종적으로 분화된 세포들입니다.

• 대사 상태: 이화적(Catabolic), 산화적(oxidative)
• 미토콘드리아 산화적 인산화를 주로 사용
• 특징:
  • 더 이상 증식하지 않음(Nonproliferative)
  • 운명이 결정됨(Committed to fate)
  • 유사분열 후 세포(Postmitotic cells)
• 예시: 신경세포, 성상교세포, 평활근·골격근, 분화된 면역세포

주요 지질 약어

• LPA: Lysophosphatidic acid (리소포스파티드산)
• PA: Phosphatidic acid (포스파티드산)
• PS: Phosphatidylserine (포스파티딜세린)
• PC: Phosphatidylcholine (포스파티딜콜린)
• PE: Phosphatidylethanolamine (포스파티딜에탄올아민)
• CL: Cardiolipin (카디오리핀) — 미토콘드리아 내막에 특징적인 지질
• CoQ: Coenzyme Q (코엔자임 Q)

지질 흐름 경로

1. 소포체(ER)에서의 합성

• LPA → PA가 소포체에서 만들어짐
• PA는 PS, PC로 전환될 수 있음
• VAP(VAMP-associated protein)가 지질 운반을 매개

2. 소포체 ↔ 미토콘드리아 지질 전달

• PRELID/TRIAP1: PA를 미토콘드리아로 운반
• RMDN/MIGA: PS 전달에 관여
• MTCH2: 미토콘드리아 외막에 위치하며 지질 전달 조절
• VPS13: 소포체와 미토콘드리아 사이에서 PC 등을 대량 운반하는 단백질
• Stard7: PC를 미토콘드리아 내막으로 운반

3. 미토콘드리아 내부 전환

• 미토콘드리아로 들어온 PA → CL(카디오리핀)으로 전환 (내막에 풍부)
• PS → PE로 전환 (미토콘드리아에서 PS decarboxylase에 의해)

4. 지질 방울(Lipid Droplet)

• LPA, PA와 연결되며 지질 저장·공급원 역할

핵심 포인트

• 미토콘드리아는 자체적으로 모든 지질을 합성하지 못하고, 소포체에서 만든 지질을 수입해야 합니다.
• 특히 **카디오리핀(CL)**은 미토콘드리아 내막의 구조와 기능(전자전달계, 크리스타 형성)에 필수적입니다.
• PS → PE 전환도 미토콘드리아에서 일어나며, 이는 미토콘드리아 막 조성에 중요합니다.
• VPS13, PRELID/TRIAP1, MTCH2, Stard7 등의 단백질이 이 지질 수송을 담당합니다.

3. 관련 최신 연구·리뷰 (2024–2025)

논문저널 (연도)주요 내용

Ryu et al. Cellular ATP demand creates metabolically distinct subpopulations of mitochondria	Nature (2024)	세포의 ATP 수요에 따라 미토콘드리아가 기능적으로 서로 다른 하위집단으로 나뉨. 대사 처리 특화 가능성 시사
Liu et al. Mitochondrial genetics, signalling and stress responses	Nature Cell Biology (2025)	미토콘드리아가 에너지 생성뿐 아니라 신호전달·다양한 합성 경로의 허브임을 강조
Picard 관련 논의 (Scientific American 등, 2025)	리뷰·논평	미토콘드리아를 “powerhouse”가 아닌 motherboard / CEO of the cell로 재정의. 대사·신호·세포 운명 조절 역할 강조

요약 및 추천

• 가장 직접적으로 “대사 산물 처리·폐기물 관리”를 다룬 논문: Spinelli & Haigis (2018, Nat Cell Biol)
• 최신이면서 포괄적인 고영향력 리뷰: Suomalainen & Nunnari (2024, Cell)
• 기능적 다양성·하위집단을 다룬 최신 연구: Ryu et al. (2024, Nature)

ReviewVolume 187, Issue 11p2601-2627May 23, 2024Open access

Download Full Issue

Mitochondria at the crossroads of health and disease

Anu Suomalainen1,2,3 Send email to anu.wartiovaara@helsinki.fi ∙ Jodi Nunnari4 Send email to jnunnari@altoslabs.com

Affiliations & Notes

Article Info

DOI: 10.1016/j.cell.2024.04.037 External LinkAlso available on ScienceDirect External Link

Copyright: © 2024 The Authors. Published by Elsevier Inc.

User License: Creative Commons Attribution (CC BY 4.0) | Elsevier's open access license policy

Linked Articles (1)

• Download PDF
• Cite

• Share

• Set Alert
• Get Rights
• Reprints

Download Full Issue

Previous articleNext article

Show Outline

Summary

Mitochondria reside at the crossroads of catabolic and anabolic metabolism—the essence of life. How their structure and function are dynamically tuned in response to tissue-specific needs for energy, growth repair, and renewal is being increasingly understood. Mitochondria respond to intrinsic and extrinsic stresses and can alter cell and organismal function by inducing metabolic signaling within cells and to distal cells and tissues. Here, we review how the centrality of mitochondrial functions manifests in health and a broad spectrum of diseases and aging.

Introduction

Mitochondria are efficient, adaptable bioenergetic machines that power eukaryotic cells biosynthetically. Mitochondria display incredible functional and compositional diversity even within an organism, despite being derived from a common bacterial ancestor, predicted to be of alpha-proteobacterial origin. The endosymbiotic event that gave rise to mitochondria was pivotal in the evolution of eukaryotic cells. Endosymbiosis fundamentally altered cellular organization—mitochondria became one of only two endomembrane systems in addition to the endoplasmic reticulum (ER) in non-photosynthetic eukaryotes that contain a genome. Direct and intimate interactions between these two major endomembrane systems evolved into critical organizational and regulatory hubs in cells.

Mitochondria lost most of the bacterial ancestral genome, and many of these genes were transferred to the nucleus: current eukaryotic genomes carry over 600 genes of alpha-proteobacterial origin.1 The vast majority (99%) of the mitochondrial proteome in humans—over 1,100 proteins—is encoded by the nuclear genome. The small 16.6-kb plasmid-like mitochondrial genome (mtDNA) encodes only 13 proteins devoted to producing the chemical energy molecule ATP via oxidative phosphorylation. This reduced size of mtDNA enables its dramatic amplification in cells. The mitochondrial-nuclear partnership and the dispersed and efficient mitochondrial compartment configuration enabled cells to boost their energy production and dramatically increase nuclear gene expression‎ capacity by powering transcription and protein translation—the highest consumers of cellular ATP. This allowed eukaryotic cells to evolve into relatively large and functionally diverse entities.2 Loss of the ancestral genome also drove the evolution of new pathways for mitochondrial protein and lipid biogenesis and mitochondrial behavior to integrate their function with the cell state.

The centrality of mitochondrial function for eukaryotic cells is reflected in how their dysfunctions manifest in a spectrum of diseases that span almost all medical fields. Remarkably, over 40% of mitochondrial proteins are linked to human diseases, underscoring the centrality of mitochondria to human health.3 Mitochondrial diseases present at any age, in any organ system, in a highly tissue-specific or multisystemic manner and display any inheritance pattern.4 How this essential organelle causes such diverse diseases is still largely a mystery. For example, mutations of complex I of the respiratory chain, the first enzyme of the ATP-producing oxidative phosphorylation system, can cause isolated optic atrophy in young adults5 or subacute necrotic brain disease in children (Leigh disease). A defect in mitochondrial translation can cause strokes early in life, hearing loss, and diabetes or hypertrophic cardiac disease.4 mtDNA depletion syndrome can manifest either in the skeletal muscle or as a brain-liver disease but not elsewhere.6,7 These examples indicate that the underlying mechanisms of mitochondrial diseases extend far beyond deficits in ATP production. The physiological complexity of mitochondrial diseases also highlights the importance of understanding the multifaceted nature of mitochondria in a cellular context. True leaps in knowledge have been made possible by transformative technologies, such as metabolomics and proteomics, illuminating the significant effects of mitochondria and their diseases on cellular and systemic metabolic homeostasis.

Structurally, mitochondria harbor two distinct membranes: the outer and inner membranes (OM and IM, respectively) and two functionally diverse compartments: the intermembrane space (IMS) and the matrix (Figure 1). The OM is a smooth lipid-enriched surface containing ancestral beta-barrel proteins, including voltage-dependent anion-selective channels (VDACs)/porins, which permeabilize the membrane to molecules up to 10 kDa in size. The OM also contains alpha-helical membrane proteins, primarily with only one transmembrane domain, such as MFF and MIRO,8 which are receptors for the mitochondrial division dynamin and motility machinery, respectively, underscoring the role of the OM in the control of mitochondrial behavior and cellular communication.9 The IMS is the smallest compartment, housing only 5% of the mitochondrial proteome. However, the IMS is critical as it facilitates diverse pathways for protein and lipid biogenesis, metal homeostasis, redox regulation and oxidative folding, apoptosis, and mitochondrial dynamics and structure.10 The IM is the most protein-dense membrane in eukaryotic cells, highly impermeant and shaped into structures called cristae. The IM is recognized as the functional core of the organelle—it houses the electron transport chain, also called the respiratory chain, with respiratory enzyme complexes (RCs; in humans, complex I–IV) and a turbine-like ATP synthase machine (RC V) that dimerizes to shape and drive the formation of cristae.11,12 The mitochondrial matrix houses a vast number of biosynthetic reactions as well as the mitochondrial genetic system. It is the home of the tricarboxylic acid cycle (TCA) that oxidizes the carbon backbones of nutrients to NADH- and FADH2-reducing equivalents, which are used by the RCs to generate chemiosmotic energy for ATP synthesis, as well as by many other biosynthetic pathways. This compartmentalization is essential for overall organelle and cellular function.

Figure 1 Mitochondrial morphology: Compartmentalized, interactive, dynamic, and mobile

Show full captionFigure viewer

The discovery of molecular players and mechanisms underlying primary mitochondrial diseases is essential for understanding common diseases linked to mitochondrial alterations, such as Parkinson’s disease, retinopathy, and diabetes. The progression of different types of cancers is also affected by mitochondrial metabolism. Many of these diseases are age dependent, consistent with the observations that mitochondrial functions are altered with aging in a tissue-specific manner. Such alterations could result from a chronic shift toward mitochondrial biosynthetic repair functions at the cost of ATP production. This impacts the cell state through changes in transcription and translation and cell-extrinsic factors, such as the extracellular matrix. As such, mitochondria have been recognized as contributors to most age-related degenerative diseases. Furthermore, altered metabolism can also cause adverse drug effects. Here, we provide an update on advances in understanding the mechanistic basis of mitochondrial functions and structure. We highlight their crucial roles in regulating the balance of anabolism and catabolism in cells and organisms and how this is relevant for pathophysiology.

Mitochondria as metabolic hubs

Metabolic programs are critical in determining cell state by providing the necessary energy and molecular precursors for various cellular functions and influencing gene expression‎ and signaling pathways. Mitochondria are at the heart of setting such metabolic programs as they either house or are connected to all intermediary metabolic pathways, dynamically changing their metabolic mode based on nutrient, stress, and differentiation status of cells.

Multifunctional metabolic organelles

The best-known cellular role of mitochondria is the production of ATP via oxidative phosphorylation. RCs I, III, and IV form the electron transport chain, transferring reducing equivalents from glycolysis and the TCA cycle to the final electron acceptor, inhaled oxygen, which produces CO2 and H2O.12 The RCs can function as solitary units, but they also assemble into defined supercomplexes, thought to increase the efficiency of respiration, although these are not essential for life.13,14 Supercomplexes also confer stability to respiratory complexes, which may be important during stress. Electron transfer by RCs is coupled to proton transport, which generates a tremendous electrochemical potential (ΔΨmv) of 150–200 mV across the IM. From the viewpoint of a molecule, this ΔΨmv is equivalent to the field strength of a lightning bolt and is used to power the rotary turbine-shaped engine, complex V, ATP synthase, to catalyze the formation of ATP from ADP and Pi.

Mitochondrial ΔΨmv functions in many non-ATP-related aspects of mitochondrial biology. For example, ΔΨmv powers and regulates mitochondrial ion homeostasis and the production of reactive oxygen species (ROS), which act as physiological and pathological signaling molecules.15 ΔΨmv is also essential for mitochondrial fusion and protein import. In this context, ΔΨmv has been proposed to function as a sentinel for mitochondrial health. Mitochondrial dysfunction within the network is detected by a decrease in ΔΨmv. Decreases in ΔΨmv impede the import of nuclear-encoded mitochondrial proteins that, in turn, initiate cellular stress responses. Examples of such pathways are the mitochondrial integrated stress response (ISRmt), which metabolically buffers the cell to promote recovery, and PINK1-mediated mitophagy, which initiates the degradation of mitochondria that are beyond repair. Loss of ΔΨmv also occurs in cell death via the opening of the mitochondrial permeability transition pore (mtPTP) and BCL associated X (BAX)-mediated cytochrome c release during apoptosis. Under conditions where oxidative phosphorylation is disrupted, such as during hypoxia, ATP synthase can run in the reverse direction to maintain ΔΨm at the expense of the hydrolysis of ATP supplied by glycolysis, underscoring its regulation and importance.

Mitochondrial metabolic plasticity

Although mitochondria retain some ancestral functions and responses, such as the ability to respond to viral infections and temperature,16 there are significant changes in their metabolic roles and plasticity. They synthesize iron-sulfur (Fe/S) clusters, an essential cofactor for RC enzymes, DNA repair, and redox-sensitive enzymes.17 Mitochondria take part in synthesizing the indispensable heme-cofactor, utilized in oxygen binding, redox reactions, drug detoxification, and management of ROS, as well as the regulation of transcription and translation. Regulatory steps of cholesterol biosynthesis occur in mitochondria. Mitochondria also produce central regulatory metabolites such as acetyl-CoA, hydroxy-ketoglutarate, 2-oxoglutarate, succinate, and fumarate that modify the epigenome to tune nuclear gene expression‎ and function as signaling organelles, mediating the metabolic state to the whole organism (Figure 2A).

Figure 2 Multifaceted mitochondrial contributions to cellular homeostasis

Show full captionFigure viewer

On the cellular level, mitochondria respond directly and via signaling pathways to nutrient state—a critical determinant of cell state. During fasting, mitochondria transition to catabolic oxidative burners of all available fuels: glucose, lipids, amino acids, and ketone bodies. When nutrients are plentiful, ATP is produced via non-oxygen-dependent glycolysis, which facilitates the transformation of mitochondria to an anabolic mode. In the anabolic mode, mitochondria utilize nutrient carbons to drive the production of amino acids, nucleotides, and other metabolites via the TCA and folate cycles to promote growth and repair (Figure 2B).18 This mode prevails in the proliferating state of many cell types to support building-block synthesis, including cancer cells (known as the Warburg effect), different tissue progenitor cells, and primed pluripotent stem cells. Switching metabolic programs to upregulate mitochondrial oxidative phosphorylation promotes differentiation and functional specification of pluripotent cells. The control of cellular fate by mitochondrial metabolism is a burgeoning area in mitochondrial biology.

Pyruvate is a critical regulatory metabolite at the crossroads of cytoplasmic glycolysis and mitochondrial oxidative phosphorylation, which is important, for example, for stem cell quiescence.19 Its utilization by mitochondria is tightly controlled by the activities of the mitochondrial pyruvate carrier, as well as by kinases and a phosphatase, which respectively inactivate and activate pyruvate dehydrogenase, which converts pyruvate to the hub metabolite acetyl-CoA to fuel the TCA cycle.20 In some somatic stem cells, however, the primary mitochondrial mode is fatty acid oxidation (FAO), which maintains quiescence or is required for activation to proliferate.21 Why quiescent stem cells are strongly dependent on FAO is currently unclear. FAO could provide cells with a redox state required for maintaining quiescence or protect them against the build-up of toxic metabolic intermediates. An increased reliance on glycolysis reduces the self-renewal capacity of adult stem cells,22 relevant for an aging-related altered metabolic program. Understanding how various regulatory enzymes direct carbon fluxes in different cell types under physiological and pathophysiological states is essential for new intervention targets to buffer disease states and age-related changes.

Mechanistically, mitochondrial metabolic plasticity is linked to structural plasticity.23 Mitochondrial shape and distribution result from the concerted actions of the mitochondrial division, fusion, and motility machines. In glycolytic pluripotent embryonic stem cells, mitochondria are fragmented and underdeveloped in their internal cristae structure. As stem cells differentiate, mitochondria undergo concerted changes: they switch to oxidative phosphorylation for energy production and smaller elongated mitochondria and form networks with extensive cristae. These changes in mitochondrial structure are partly determined by the relative rates of mitochondrial division and fusion, catalyzed by dynamin-related GTPases DRP1 in the cytosol and MFN1/2 and OPA1, in the mitochondrial OMs and IMs, respectively.23 The activities of these membrane-shaping proteins are required in cell fate transitions and integrated with metabolism.24 The organization and localization of biosynthetic enzymes within and near mitochondria are also regulated by the cellular metabolic state. For example, inside the organelle, enzymes in the coenzyme Q (CoQ) biosynthetic pathway organize into precisely positioned supramolecular metabolon.25 Similarly, de novo purine synthesis pathway enzymes cluster on the OM to form a purinosome metabolon that channels mitochondrial-derived formate into purine synthesis.25,26 Thus, dynamic changes in mitochondrial structure are not simply products of metabolic changes. Understanding the mechanistic links between metabolism and mitochondrial behavior is an important active research area.

Beyond identifying dynamic changes in mitochondria, the field has started to describe distinct “mitotypes” within a cell. Mitochondrial sub-populations exist, for example, in brown fat cells, where mitochondria-contacting lipid droplets are specialized for lipid storage, whereas unassociated mitochondria are active in oxidative phosphorylation.27 Indeed, mitochondrial specialization may more broadly be mediated by contacts with other cellular organelles, such as ER. Mitochondrial contact sites mediate the exchange of lipids, ions, and metabolites and regulate mitochondrial dynamics, mtDNA replication, and quality control (Figure 3). They are formed in a manner responsive to cell and metabolic state and likely fine-tune the output of mitochondria in ways we do not yet understand. The diversity of mitochondrial composition, organization, and physiology at the cellular level highlights their essential roles in the evolution of eukaryotic organisms.

Figure 3 Mitochondrial interactions with organelles

Show full captionFigure viewer

In multicellular organisms, mitochondria have evolved to adapt their functions to the specific needs of different cell types. However, how mitochondria match their biosynthetic functions with the metabolic requirements of the tissue has yet to be fully understood, and how these decisions contribute to specific organismal vulnerabilities is another open question. For example, it is a mystery how the heart, which is highly dependent on ATP, maintains metabolic homeostasis in most mitochondrial diseases and how highly anabolic cells, such as oligodendrocytes, prioritize glucose carbon flux to promote their specific biosynthetic phospholipid pathways.

Interestingly, adjacent cells with distinct characteristic steady-state metabolic profiles can share metabolites. For example, the stem cell niche has anabolic “caretaker” cells that provide metabolites for stem cells. The caretakers are biosynthetic and use glucose as their main fuel, while the stem cells are considered oxidative (Figure 4).28 Mitochondrial functions also depend on and dictate cellular differentiation, organ maturation state, and organismal sex, age, and environment.

Figure 4 Metabolic modes of cells in their proliferative and non-dividing states

Show full captionFigure viewer

The context is also important: a genetic defect in a mitochondrial ribosome (mitoribosome) subunit often causes a severe cardiac disease that is typically lethal in early life. However, if the child survives the first few years, the heart condition can stabilize, negating a mortality risk.

These examples highlight the gaps in knowledge of mitochondrial tissue-, age-, and context-specific roles that are highly relevant for understanding and mitigating disease. Metabolic signaling at the level of the organism and its intimate link to dietary metabolism are also critical. For example, depletion of a mitochondrial protease in both muscle and heart or high-fat feeding can suppress the severe cardiac phenotypes resulting from the loss of the same protease in the heart.29

Emerging knowledge indicates that even when mitochondrial disease phenotypes are apparent in a single tissue, they broadly affect the organism through systemic metabolism and hormone-like signaling.30,31,32 For example, in muscle mitochondrial disease, muscle-derived “metabokines” travel systemically to modify glucose uptake and mitochondrial biogenesis in a specific region of the brain.33 There is a need to determine how mitochondrial-controlled metabolism changes with different stages of human diseases and/or the maturation of postnatal tissue and aging. In addition, a more detailed understanding of the organism-wide manifestation of mitochondrial diseases and their underlying molecular and cellular mechanisms is critical for a better understanding of basic biology and physiology. High-resolution precision tools for spatial, cell-specific analyses in intact tissues currently being developed provide increasing insights into these questions.

Mitochondrial roles in folate cycle and anabolic responses

Mitochondria have essential functions in the main anabolic biosynthesis pathway—the folate-driven one-carbon (1C) cycle34 (Figure 2B). 1C-metabolic pathway enzymes are distributed between mitochondria and the cytoplasm, with some folate-modifying enzymes residing in the nucleus. Mammals derive folate, vitamin B9, from their diet with a minor component provided by the gut microbiota. Folate is the main carrier of 1C-units to growth-related biosynthesis pathways and is a perfect candidate to signal food availability at the cellular level. Folic acid is converted into active tetrahydrofolate forms (THFs) in the liver and released into circulation. THFs enter cells via specific folate receptors, where they are polyglutaminylated via mitochondrial or cytoplasmic folylpoly-gamma-glutamine-synthetase enzymes, resulting in a pool of stable compartment-specific THFs in postmitotic cells. Different folate forms are highly specific and rate-limiting for the enzymes and pathways they serve. For example, methylene-THF drives the methyl cycle, and mitochondrial-derived formate feeds the cellular purine synthesis via 10-formyl-THF and formyl-methionine generation, required for mitochondrial translation initiation, to facilitate organellar protein synthesis.35 The mitochondrial folate cycle produces NADPH and NADH that, together with mitochondrial aspartate, are required for pyrimidine synthesis.34

NAD+ and glutathione: Redox-regulators of mitochondrial and cellular metabolism

Mitochondria also produce reduction-oxidation (redox) regulators—metabolites that direct metabolic flux and biosynthesis pathways, serve as antioxidants, modify proteins, and maintain redox homeostasis. The main regulators are NAD forms and glutathiones: NAD+, NADH, NADP, NADPH, GSH, and GSSG. The redox-active pathways that produce these regulators are linked, as GSSG is recycled back to GSH by NADPH-dependent enzyme glutathione reductase.36 NAD+ is a critical nucleotide cofactor for the TCA cycle, which produces NADH as the main carrier of nutrient-derived electrons from the cycle to complex I of the respiratory chain, which regenerates NAD+. Recently, the mitochondrial NAD+ transporter SLC25A51 was described,37,38 emphasizing the compartment-specific functions of cytoplasmic and mitochondrial NAD+ pools. The different forms of vitamin B3 (nicotinic acid, nicotinamide, nicotinamide mononucleotide, and nicotinamide riboside) are NAD+ precursors driving synthesis through tissue-specific enzymatic pathways. NAD metabolites broadly regulate over 200 cellular biosynthetic pathways and the direction of their flux. They utilize NAD+ as an electron acceptor, and NAD+ and its forms also function as cofactors. For example, NAD+ is a cofactor for the NAD+-dependent protein lysine deacetylases that regulate various cellular functions, including nutrient sensing, energy metabolism, and stress responses.39 NAD+ pools have wide-ranging effects on cellular functions and homeostasis. Mitochondria are the main regulators of cellular and systemic NAD+/NADH ratio, and through this function, mitochondrial status broadly influences cell and organismal health.

Our understanding of the relevance of NAD+ metabolism for human health and diseases is increasing. In mice and humans with mitochondrial muscle disease, NAD+ pools and 1C metabolism are altered, and NAD+ deficiency is observed in the muscle and blood as a secondary consequence of disease.40 Importantly, restoration of NAD+ pools with NAD+ precursors in these patients corrected NAD+ levels and defects in mitochondrial structure and function and slowed disease progression.40 The patients showed remarkable whole-body benefits including reduced liver fat and increased muscle mass, strength, and oxidative phosphorylation activity. This restoration in function shifted metabolism toward healthy controls but did not correct the causative genetic defect or the associated transcriptional stress responses. In addition to mitochondrial muscle disease, NAD+ deficiency and the consequential coupled metabolic alterations are increasingly being linked to various degenerative, aging-related diseases of humans, including neurodegeneration, cancer cachexia, and cardiomyopathy.41,42,43 These observations highlight the need to eval‎uate whether restored NAD+ levels may act as a buffer to treat a broad range of human diseases and/or increase human health span.

Glutathione and its oxidized and reduced forms, GSSG and GSH, represent the second major redox modifier and antioxidant.44 While glutathione is not synthesized in mitochondria, the mitochondrial folate cycle is closely linked to the transsulfuration pathway, responsible for producing cysteine, hydrogen sulfide (H2S), and glutathione. Like NAD+:NADH, the GSH:GSSG ratio is a critical measure of the cellular redox state, which controls key cellular processes, including cell proliferation, differentiation, and cell death. The half-life of glutathione is short—only hours—but its cellular pools reach millimolar concentrations, and up to 15% of glutathione exists in mitochondria, sequestered by SLC25A39- and SLC25A40-dependent transport.45,46,47 GSSG is recycled back to its reduced form through the NADPH-dependent enzyme glutathione reductase. Glutaredoxin enzymes harness the reducing power of GSH to regulate redox homeostasis and act as a defense system against oxidative stress. The GSH/glutaredoxin system is also required for de novo dNTP synthesis by ribonucleotide reductase to maintain nucleotide pools and mtDNA replication in postmitotic cells. Ferroptosis is a form of iron-dependent, mitochondrial-linked cell death associated with decreased GSH biosynthesis and the consequent inactivation of the glutathione peroxidase, GPX4, resulting in lipid peroxidation and cell death.48 In this manner, dNTP production competes with GPX4-dependent lipid detoxification to use GSH, linking GSH metabolism to ferroptosis sensitivity. Increased flux toward GSH synthesis also produces H2S, a potent complex IV inhibitor that is degraded by mitochondria, providing a mitochondrial feedback loop.49

Glutathione usage is high in the nervous system and decreased levels of GSH are associated with the progression of neurodegenerative disorders. In neurons, the production capacity of GSH is limited and is significantly supported by astrocytes, which fuel neuronal GSH synthesis by providing cysteine and glutamate. As such, GSH metabolism is considered a therapeutic target of neurodegenerative disorders: the glutathione precursor N-acetyl-L-cysteine is clinically used to rescue liver function in response to acetaminophen intoxication, but its delivery to the brain is controversial.47

Reactive oxygen species

ROS are common side products of oxidative phosphorylation. Their regulation is important because of their potential to cause oxidative damage to DNA, proteins, and lipids. They also act as signaling molecules to regulate cell state such as, for example, muscle growth, stem cell differentiation, and erythrocyte maturation.50,51 Mitochondria possess multiple mechanisms for generating ROS. Mitochondrial complexes I and III are predominant ROS producers, generating O2−, which reacts with H2O to produce H2O2.52 Complex I can also generate O2− by reverse electron transfer under conditions where the pool of reduced coenzyme-Q-pool and proton motive force are relatively high. Under these conditions, at complex I, electrons from CoQH2 are transferred back to NAD+, with an associated increase in succinate. Conditions that favor reverse electron transfer include ischemia-reperfusion. Recent data suggest that succinate acts as a signal for the reduction state of the Q-pool and controls superoxide production from complex I, performing as a redox-active metabolite. Succinate can also be secreted and regulate ATP production, oxygen sensing, and hypoxia signaling.53 These functions highlight the roles of systemic metabolite signaling initiated by mitochondrial metabolism. ROS signaling and ROS sensitivity are highly tissue-specific, as exemplified by the actions of the antioxidant MitoQ, a modified form of CoQ10, which accumulates in mitochondria in a manner dependent on membrane potential. MitoQ doses that mitigate defects in erythrocyte precursors are toxic for neural stem cells in mice with increased mtDNA mutagenesis.51 Furthermore, antioxidants inhibit physical training-related muscle growth. Thus, like many other signals, there is a defined range of efficacy beyond which lies toxicity that can manifest in a tissue-specific manner.

Building and shaping mitochondria

The composition and regulation of the mitochondrial proteome and lipidome are highly tissue- and context-specific to tailor function to need.54 For example, electron transport chain proteins are more abundant in the heart and skeletal muscle to meet the high energetic needs for muscle contraction. Cutting-edge mass spectrometry approaches have been used to determine mitochondrial protein and lipid compositions, representing an incredible advancement for the field. These mitochondrial atlases have been extensively mined using genetics, multi-omics, and bioinformatics to give deep insights into mitochondrial biology.3,55,56,57

Diverse and dynamic mitochondrial proteomes and lipidomes

The human mitochondrial proteome amounts to about 7% of the total cellular proteome, containing proteins with an incredible range of abundances (over six orders of magnitude) and half-lives (from minutes to months), providing important clues for their function and regulation. Mass-spectrometry-based approaches have resolved the composition of mitochondrial sub-compartments and how proteins interact within them.3,54 Most mitochondrial proteins are localized to the IM and matrix and dedicated to metabolism and functioning. For example, they are directly involved in oxidative phosphorylation or indirectly by mediating the expression‎ of the handful of mitochondrial genes essential for respiration. A full annotation of how mitochondrial proteomes differ between proliferative and postmitotic cells, between different tissues and cell types and even between mitochondria within a cell, is still needed. Such atlases could help identify additional mitochondrial protein functions that are linked to disease.

Like the proteome, the lipidome is complex, membrane-specific, and dynamic, with over 200 species, including the predominant mitochondrial glycerolipids, phosphatidylcholine (PC) and phosphatidylethanolamine (PE), and the unique signature lipid, cardiolipin (CL), inherited from its bacterial origins.58 Mitochondria also contain small amounts of sphingolipids and sterols, except mitochondria in steroid-hormone-synthesizing cells. In the liver, the mitochondrial lipidome oscillates in a diurnal, likely circadian-directed manner in an opposite phase to the nuclear lipidome, highlighting that cellular lipid fluxes are critical regulators of cellular and organismal homeostasis.57 Understanding mitochondrial lipid regulation in different cell types and how they more broadly impact cellular and organismal lipid homeostasis will reveal new biology and therapeutics. Indeed, peptides that stabilize mitochondrial CLs are already in clinical trials for mitochondrial diseases.

Nuclear-encoded mitochondrial protein biogenesis

Nuclear-encoded mitochondrial protein biogenesis is initiated close to the mitochondrial OM on cytosolic ribosomes, which can also be membrane-tethered to enable efficient protein import, sorting, and assembly.59 Molecular chaperones play a critical role in the post-translational import of mitochondrial preproteins by maintaining an unfolded state and guiding them to the mitochondrial OM import machinery. The cytosolic ubiquitin-proteosome system (UPS) prevents the accumulation of mitochondrial precursor proteins in the cytosol and acts as an important homeostatic regulator of mitochondrial biogenesis. In the OM, the translocase of the outer membrane (TOM) complex is the gateway for protein import and contains the pore-forming beta-barrel protein Tom40, which does the work of protein import. In humans, MTCH1/2 also function in the OM as direct protein insertases for single and rare multi-pass alpha-helical membrane proteins synthesized in the cytosol.60 Beyond the TOM complex, the routes and translocation mechanisms for proteins vary based on their sub-mitochondrial destinations and their diverse structural, chemical, and folding features.61,62 For example, β-barrel proteins are inserted from the IMS by the concerted actions of the small translocase of the inner membrane (TIM) proteins and the sorting and assembly machinery (SAM) complex in the OM. Mitochondrial protein import across the IM depends on the TIM23 translocase that forms a channel that allows proteins to be pulled into the matrix by the mitochondrial Hsp70 motor complex. The IM ΔΨmv is also required for precursor import and acts to aid import in part via electrophoretic effects on the relatively basic mitochondrial import signals. In this context, import signals vary in terms of their charge properties. Thus, conditions that change membrane potential can affect protein import efficiency in a substrate-selective manner over a large dynamic range. How membrane potential fine-tunes the mitochondrial proteome in different contexts is not fully understood and is important for fully elucidating mitochondrial plasticity and disease mechanisms.

Under conditions in which the capacity of import and UPS turnover mechanisms is exceeded, cytosolic precursor proteins accumulate and import can arrest—a condition termed mitochondrial precursor overaccumulation stress (mPOS).63 Multiple cellular stress pathways are activated by mPOS, including pathways that upregulate proteostasis and the mitochondrial ISRmt and mitoCPR (mitochondrial compromised protein import response), mitophagy, and cell death.63 Recently, a large E3 ligase complex was identified, termed silencing factor of the integrated stress response (SIFI), that resolves import stress by degrading both cytosolic precursors and the key transducers of the ISRmt, DELE1, and the eIF2-α kinase heme-regulated inhibitor (HRI).64 SIFI analysis indicates its function in concluding stress response signaling is necessary for cellular survival. In this context, SIFI is mutated in ataxia and early-onset dementia, suggesting that chronic stress signaling is a causative factor of these diseases and potentially any disease associated with defective import. Consistently, genetic defects in mitochondrial protein import result in neuropathies, myopathies, neurodegenerative diseases, cancer, and cardiovascular disorders.4

Mitochondrial-encoded protein biogenesis

In humans, the 16.6-kb circular mtDNA contains 37 genes: 22 tRNAs and two ribosomal RNAs for mitochondrial translation of the 13 structural proteins encoded by mtDNA, which are core components of the respiratory chain protein complexes I, III, IV, and V.65 mtDNA is transcribed as a polycistronic transcript by mitochondrial RNA polymerase, and mRNAs are typically flanked by tRNA genes recognized by specific processing enzymes, which release and process the ends of the mRNAs and tRNAs.66 The transcription machinery proceeds through the first transcripts-tRNA-Phe, 12S rRNA, tRNA-Val, and 16S rRNA—and after completion of this operon, transcription frequently terminates instead of proceeding through the entire polycistronic transcript. This ensures the appropriate stoichiometry and abundance of mitoribosome structural RNA components to enable translation of the 13 mitochondrially encoded proteins.66 The entire process of mitochondrial gene expression‎ is interconnected, giving rise to RNA granule condensates that can be visualized using super-resolution microscopy.67,68 Most of the mitochondrially encoded proteins are highly hydrophobic IM proteins. Consequently, they are translated and inserted into the IM by mitoribosomes associated with the ancient Oxa1 insertase complex, which promotes the biogenesis of some nuclear-encoded IM proteins.69

Significant recent advances in cryo-electron microscopy have captured mitoribosome structures, including from humans.70,71 Human structures reveal that in addition to mitochondrial-encoded 16S and 12S rRNAs, the ribosome possesses a third integral RNA component: tRNA-Val, encoded in mtDNA and present in equal stoichiometry. Within the mitoribosome, tRNA-Val forms extensive interactions with mitoribosomal proteins as a structural replacement for the bacterial 5S rRNA. Mitoribosome structures have also provided many mechanistic insights into tRNA recognition, polypeptide synthesis, the roles of ribosome modifications, and the co-translational mode of insertion of mitochondrial-encoded proteins in the IM.72 The mitoribosome is modified by metabolites that are in other contexts known to represent different metabolic states. The role of such modifications to mitoribosomal functions remain to be studied. Understanding ribosome structures also allows explanations for clinical manifestations of mitochondrial diseases caused by mitoribosomal defects.

The nuclear-encoded and mitochondrial-encoded translation products converge on the biogenesis of the I, III, IV, and V respiratory complexes. How this is coordinated is an area of active research. In yeast, cytosolic translation controls mitochondrial translation to facilitate balanced mitochondrial and cytosolic protein synthesis.73 In contrast, in human cells, imbalance caused by inhibiting one system does not affect the other in the short term.74 In humans, the nuclear-encoded respiratory complex assembly component MITRAC15 also functions in the translational regulation of mitochondrial-encoded subunits of complexes I and IV. Miscommunication between the mitochondrial and nuclear genomes can cause activation of stress pathways such as the ISRmt. Thus, more work is needed to identify additional coordination mechanisms and how they contribute to disease and aging.75

Mitochondrial translation defects

Since 2010, next-generation exome sequencing (NGS) approaches have revealed pathological mutations in nuclear genes that cause a spectrum of different mitochondrial translation and respiratory complex assembly defects on a cellular level. Mitoribosome subunit protein defects are a common cause of infantile hypertrophic cardiomyopathies, highlighting the importance of mitochondrial biogenesis in the heart after birth. In contrast, phenotypes associated with mitochondrial aminoacyl-tRNA synthetases (ARS2s), also essential for mitochondrial translation, manifest in a highly variable manner in different organ systems at different ages. ARS2s can cause neonatal cardiomyopathy (AARS2), cerebral white matter disease (DARS2, MARS2), encephalopathy and developmental delay (e.g., EARS2, AARS2, FARS2), hearing loss, and ovarian dysfunction (HARS2, LARS2).76 Based on amino acid abundance, mitochondrial tRNA synthetases can redundantly conjugate multiple similar amino acids. Such events are known to occur, for example, for mammalian AARS2, which can replace serine with alanine tRNA-Ala, which has deleterious consequences for the protein function.77 In contrast to cytosolic ARS2s, only some mitochondrial ARS2s possess an editing domain to release mischarged amino acids, shifting quality control of mitochondrial protein synthesis to post-translational protease-dependent mechanisms. Examples of variants in ARS2s that cause cellular malfunction only in combination with specific mtDNA variants exist, demonstrating the complexities arising from the dual genetic mitochondrial biogenesis system.78 These interactions are challenging to identify because of the inherent variability of mtDNA and the common occurrence of nuclear gene variants of unknown significance. Surprisingly, defects affecting mitochondrial translation rarely manifest a respiratory chain deficiency in cultured cells from patients, while defects in the translation products—respiratory chain enzyme subunits—often show a cellular phenotype.

Complex I: A megacomplex challenging to assemble

Complex I of the respiratory chain is a beautiful example of how nuclear and mitochondrially encoded proteins coordinate for assembly. Seven of the 45 subunits of complex I are encoded by mtDNA with the rest of the structural components and at least 14 assembly factors encoded by nuclear genes.79,80,81 These are separately translated on mitochondrial and cytoplasmic ribosomes and assembled as modules. The membrane arm of complex I contains all the mtDNA-encoded subunits. A peripheral matrix-localized arm, composed of 17 nuclear-encoded subunits, has two functional parts: a matrix proximal NADH-oxidizing N-module and a membrane adjacent ubiquinone-reducing-Q module. The two arms associate together to form an L-shaped complex with seven Fe/S clusters that form an “iron wire” through the peripheral arm, which accepts electrons from NADH and transfers them to complex III via the electron carrier CoQ. Complex I coordinately pumps four hydrogen atoms via its membrane arm to the IMS. This modular assembly of complex I allows separate control of module turnover and assembly. The N-module, the primary site of ROS production, is degraded at a significantly higher rate than the rest of the enzyme by the matrix-localized ClpXP protease, which is required for optimal complex I function in cells.82 A similar modular strategy could exist for complex V as well, where de-capping the ATP synthase reveals the membrane-embedded c-ring, forming a leak channel and loss of ΔΨmv.83 For complex I, such a modular mechanism might reduce ROS in transient metabolic stress, and for complex V, it might resolve protein import stress. However, the mechanisms require further research to reveal their potential physiological roles.

Complex I deficiencies are a common cause of Leigh syndrome, a fatal infantile-onset disease affecting the brain stem and basal ganglia. Indeed, almost any gene mutation affecting a nuclear-encoded respiratory chain subunit or assembly factor can cause Leigh syndrome.5 One reported pathological mechanism in mice with complex I accessory subunit (NDUFS4) inactivation involves aberrant hypoxia signaling. This promotes hypervascularization in the affected brain region, flooding the region with too much oxygen and resulting in hyperoxic damage.84 In the NDUFS4-knockout mice, microglial activation and peripheral lymphocyte invasion also contribute to the local damage.85

All complexes have elaborate, stepwise assembly mechanisms, coordinated by specific complex-specific assembly factors, many of which were initially identified from the genetic characterization of mitochondrial diseases86,87 Their composition may change based on metabolic state (complex II)88 or tissue-specific functions (complex IV).89 Their replacement in the IM requires de novo synthesis and assembly. However, such turnover occurs slowly in non-dividing cells: the respiratory chain proteins show exceptionally long half-lives to several weeks.90

Mitochondrial lipid biogenesis

Like the mitochondrial proteome, the organellar lipidome is made from a combination of extra-mitochondrial and mitochondrial-derived components (Figure 5).58 The mitochondrial IM possesses essential de novo biosynthetic pathways for the glycerolipids PE and CL, as well as for the ancient electron carrier lipid CoQ. The PE and CL pathways are fueled primarily by transporting ER-synthesized precursors, phosphatidic acid (PA) and phosphatidylserine (PS)91 to the OM. From there, they are shuttled to the IM by the highly conserved human PRELID3b-TRIAP1 and PRELID1-TRIAP1 complexes localized in the IMS, as initially described in yeast.92 PE is an abundant lipid in OM and IM, whereas CL is highly IM enriched. Together, they play crucial roles in IM structure and organization, being essential regulators of mitochondrial fusion, respiratory complexes integrity, and crista morphogenesis.91 CL is also in the OM in relatively small amounts, where it recruits the division machinery and BCL proteins to regulate mitochondrial dynamics and apoptosis, respectively.93 PC is the major phospholipid in mitochondria required for organelle integrity and must be acquired from the ER or other organellar membranes. PC is incorporated into the OM by multiple lipid transport proteins and transferred to IM by the intramitochondrial lipid transfer protein STARD7. In cultured cells, STARD7 is cleaved by the IM peptidase presenilin associated rhomboid-like (PARL), enabling it to operate as an export shuttle for CoQ to the cytosol, where CoQ protects against ferroptosis.94

Figure 5 Mitochondrial lipid metabolism

Show full captionFigure viewer

Mitochondria also lack biosynthetic pathways for phosphatidylinositol (PI), sphingolipids, and cholesterol, which are made in the ER. Although present in small amounts, sphingolipids and cholesterol are linked to mitochondrial dynamics, energy metabolism, oxidative stress, and cell death pathways.95,96,97 Accumulation of sphingolipids in mitochondria has been implicated in mitochondrial-associated pathologies, such as muscle wasting. Therefore, de novo sphingolipid synthesis has been suggested as a therapeutic candidate for human diseases.98 However, the mechanistic bases for the effects of sphingolipids and sterols in mitochondria must be better understood.

Mitochondrial lipid transport at membrane contact sites

Mitochondria can also acquire lipids from the ER and other organelles via outer-membrane lipid transporters that create membrane contact sites as part of a tethering complex that holds together two organelles.99 At these sites, mitochondria primarily partner with the ER and also other organelles, such as lysosomes/vacuoles, endosomes, and lipid droplets. The molecular basis and regulation of lipid transport to and from mitochondria at membrane contact sites is an active area of research, as dysregulation of mitochondrial contact sites has been implicated in a wide range of human pathologies, including neurodegeneration.

Work in yeast revealed a conserved role of the ER-localized Vps13 family of proteins in lipid transport at mitochondria contacts.100 Characterization of Vps13 and Vps13-like chorein domain proteins, such as ATG2, which functions in autophagosome biogenesis, indicates that they form bridge-like structures that deliver relatively large volumes of lipids to facilitate rapid membrane expansion.101 The expanded human family (VPS13A-D) is associated with neurological disorders, highlighting the significance of these proteins in maintaining cellular function. They have a differential localization at ER contacts: VPS13A and VPS13D are localized at ER-mitochondrial contact sites, where they likely mediate relatively non-specific bulk lipid transport of ER lipids necessary for mitochondrial growth and integrity.100,102 When bulk lipids are transferred to membranes, lipid scramblases that exchange lipids across the bilayer are necessary to prevent lipid asymmetry. At the mitochondrial OM, the MTCH1/2 protein insertases and VDACs may also function as phospholipid scramblases, contributing to the import of phospholipids into mitochondria.99,103,104

In contrast to the VPS13 family, the yeast ER-mitochondria encounter structure (ERMES), which transports phospholipids between the ER and mitochondria, is not conserved in metazoans. Expansion of the VPS13 family and metazoan-specific families of lipid transporters have likely evolved to replace ERMES at ER-mitochondrial contact sites. In particular, the mitoguardin (MIGA) and regulator of microtubule dynamics (RMDN) families represent new classes of mitochondrial OM-localized lipid transporters that partner with ER-localized vesicle-associated membrane protein (VAMP)-associated proteins (VAPs) to form contact sites. The MIGA and RMDN families likely convergently evolved TPR/helical domains to transport the precursors PS and PA for PE and CL synthesis, respectively.105,106 Their alpha-helical TPR-like domains form cup-shaped structures lined with hydrophobic amino acids to solubilize and shuttle a relatively low volume of lipids as compared to the VPS13 transporters. This raises the possibility that they have evolved to fine-tune mitochondrial lipid composition for specific functions.107 Indeed, the expansion of mitochondrial lipid transporters in humans suggests a mechanism for the diversification necessary to match mitochondrial functions to cell-type needs.

Mitochondrial contact sites are complex in their composition and function as distinct contact site complexes can co-localize and act in a combinatorial manner. By controlling lipid flux, they form a larger connected network encompassing all cellular membranes, which can reconfigure to compensate for each other redundantly to control lipid homeostasis (Figure 5). Contact sites can also regulate and integrate cellular activities by competing for a common pool of components, such as ER-localized VAP tethers.102,103,104 In this context, in adipocytes, MIGA2 interacts with mitochondria, ER, and lipid droplets and facilitates lipid storage by linking de novo lipogenesis in mitochondria to triacylglycerol (TAG) production in the ER.

The dual localization of a lipid transporter to lipid droplets and mitochondria highlights the role of mitochondria as a decision point for fat anabolism versus catabolism and phospholipid homeostasis.104,108 Distinct but functionally redundant enzymes localized to the ER or the mitochondrial OM esterify fatty acids on the lipid glycerol 3-phosphate backbone to create lysophosphatidic acid (LPA) and PA. Flux, specifically through the mitochondrial esterification arm of the pathway, is harnessed to stimulate mitochondrial fusion via the production of LPA, underscoring how metabolism is mechanistically linked to the regulation of mitochondrial behavior. PA generated at either mitochondria or the ER can be transported into mitochondria for CL synthesis or used at the ER to create phospholipids or diacylglycerols (DAGs) and TAGs via LIPIN and DGAT1/2, respectively, for lipid storage.109 Conversely, DAGs mobilized from lipid droplets by triglyceride lipases, such as adipose triglyceride lipase, could also be utilized for mitochondrial lipid biogenesis. How these flux decisions are made is not fully known, but their impact on cellular function highlight the importance of understanding the spatial organization and regulation of lipid metabolism in the context of the contact site network.

Mitochondrial contact sites as organizers of mitochondrial behavior and signaling

Mitochondrial contact sites play diverse roles beyond lipid transport. Specialized non-lipid transporting contacts form domains that facilitate and regulate cellular Ca2+ buffering via a direct conduit that transfers Ca2+ from ER stores to the mitochondrial matrix. These contacts are thought to be composed of a tethering complex of the ER inositol-1,4,5 trisphosphate receptor/Ca2+ channel, the cytosolic adaptor glucose-regulated protein 75, the OM-based voltage-dependent anion channel VDAC1, and IM-based mitochondrial calcium uniporter channel MCU.110,111 In this manner, ER-mitochondrial contacts modulate the generation of ATP in response to cell signaling by the mitochondrial Ca2+-regulated dehydrogenases, such as pyruvate dehydrogenase, isocitrate dehydrogenase, and α-ketoglutarate dehydrogenase, or promote cell death in the face of cellular stress by activating the mtPTP.112

Lipid-transporting membrane contacts also form specialized domains that broadly regulate mitochondrial and cell function. These domains are hot spots for the localization and assembly of mitochondrial dynamics and distribution machinery on the OM. Consequently, mitochondrial division, fusion, motility, and anchoring events are enriched at ER-mitochondrial contact sites.108,113 The formation of mitochondrial-derived vesicles/compartments (MDVs/MDCs) is a process conserved from yeast to humans to aid in mitochondrial quality control and requires the mitochondrial motility machinery. Thus, MDV/MDC formation also likely occurs at ER-mitochondrial contacts. Mitochondrial nucleoids—assemblies of mtDNA molecules and proteins—are positioned in the matrix near ER-mitochondria contact sites, and this positioning is required for mtDNA replication and distribution.114,115 The mechanism of mtDNA replication licensing by the ER is not understood, but Ca2+ and lipid fluxes may be involved given the role of ER-mitochondria contacts in regulating these molecular events (Figure 3).

Additional key structures localized inside mitochondria are positioned adjacent to ER-mitochondria contacts, such as the CoQ metabolon and the mitochondrial contact site and cristae organizing system (MICOS), which is responsible for the organization of the IM.116,117 The mitochondrial IM typically has a greater surface area than the OM, and although continuous, it is laterally differentiated and shaped into different structural and functional domains. MICOS is one of the key conserved components, in addition to the ATP synthase dimer, which directly shapes and organizes the IM; the mitochondrial IM fusion dynamin, OPA1, which interacts with MICOS; and the mitochondrial lipids, PE and CL. MICOS is a multi-subunit complex formed by two independent sub-complexes localized in IM and IMS.118,119 The complex is enriched at membrane junctions connecting the cristae lamella, which houses the RCs, to the smooth inner boundary membrane, where it likely plays a direct role in shaping and/or stabilizing the highly curved membranes at these sites. Mutations in MICOS subunits are associated with abnormal IM morphology and are linked to severe human mitochondrial diseases. Additionally, MICOS forms a physical bridge to OM via interactions with the SAM complex. In this context, ER-mitochondria contacts co-localize components required for de novo nuclear and mitochondrial-encoded protein biosynthesis, lipid import, and mtDNA replication to organize a “unit of mitochondrial inheritance,” ensuring the appropriate distribution of the “relaxed” mitochondrial genetic system.

ER-mitochondria contacts are also hot spots for assembling the Bax pore complex, which initiates apoptotic cell death by releasing cytochrome c from the IMS and likely immune responses by releasing mtDNA and mtRNA from the matrix.120 The mitochondrial antiviral signaling protein (MAVS) assembles in the OM at ER-mitochondria contacts to act as a central hub for signal transduction initiated by retinoic acid inducible gene (RIG-I)-like receptors, which mediate the activation of NF-κB and interferon regulating factors (IRFs) and the induction of interferons (IFNs) in response to viral and cytosolic mtDNA/mtRNA.121 The molecular basis for the multifaceted functional nature of ER-mitochondria contacts remains to be investigated. Future work will focus on how different types of molecular contacts specify their regulatory functions, how lipid fluxes contribute to their domain organization and function, and whether combinations of contact site complexes mediate different functional outputs.

Maintenance of the mitochondrial genome

Almost all the proteins required for the maintenance of mtDNA, the regulation of mtDNA of copy number, and the expression‎ of mtDNA-encoded genes are encoded by nuclear DNA. This feature is exploited to regulate the expression‎ of mtDNA transcripts to match metabolic needs. Mitochondrial gene expression‎ levels vary significantly in humans at both the tissue and individual levels, the basis of which is still not fully understood but likely to be important to meet tissue-specific energy demands, balanced with tissue-specific requirements of mitochondrial metabolite production.

MtDNA nucleoids include proteins that function in mtDNA synthesis, nucleotide metabolism, and metabolic signaling.122 The most abundant protein of the human nucleoid is TFAM, the mitochondrial transcription factor A, which, in addition to activating mtDNA transcription, also binds mtDNA in a histone-like manner, supercoiling it.123,124,125 The core proteins of the mtDNA replisome include DNA polymerase gamma (catalytic subunit POLG1 and processivity subunit POLG2), replicative hexameric helicase Twinkle, and single-stranded DNA binding protein (mtSSB), which are the minimal set of proteins able to synthesize DNA in vitro.126 Although mitochondria are related to bacteria, mtDNA replication and transcription enzymes originate from bacteriophages.127 Phage polymerases might have been selected because of their high processivity, which is likely beneficial to maintaining the high mtDNA copy number in cells. The viral origin of POLG has relevance for medicine: the enzyme is sensitive to nucleoside analogue inhibitors of reverse transcriptases used to target viral polymerases to, for example, treat HIV.128 These drugs cause reversible mtDNA depletion in certain tissues, depending on the specific nucleoside analogue. The tissue specificity of reverse transcriptase inhibitors suggests that nucleotides for mtDNA replication are rate limiting and their levels differ between organs.

MtDNA copy number per cell varies depending on the cell type and is highly regulated during cell fate transitions. For example, during differentiation of myotubes or neurons, mtDNA copy number dramatically increases, and in non-dividing cells in vivo, the half-life of mtDNA increases from hours to days in proliferating cells to several weeks or even months. Expression‎ levels of Twinkle and TFAM can directly regulate mtDNA copy number by regulating replication frequency or by altering mtDNA half-life and/or accessibility of mtDNA to replication machinery, respectively.129,130 The regulation of cellular nucleotide pools also plays a role in copy number; for example, the activity of the nucleotide kinase NME6 may limit nucleotide availability for mtDNA replication.131,132 However, which factors limit the specific mtDNA amount for different cell types in postmitotic and proliferating cells is still not fully understood. A recent study using 300,000 human population samples identified several nuclear loci that affect mtDNA copy number and the selection of heteroplasmic variants133 in agreement with mouse genetics.134

In humans, mtDNA is inherited uniparentally via the maternal lineage: paternal mitochondria are actively degraded by mitophagy.135,136 However, a recent report suggested a different mechanism in which the nuclear localization of a spermatoid cell-specific form of TFAM destabilized sperm mtDNA leading to its loss during sperm maturation.137 The human oocyte contains a reservoir of 200,000 copies of mtDNA, distributed after fertilization during division to cells forming the early embryo—mtDNA replication only resumes at the blastocyst stage. During oocyte development, there is an mtDNA “bottleneck,” which selects against pathogenic variants, especially protein-coding variants of mtDNA.138 The basis of this selection is suggested to occur during oogenesis at the level of individual mitochondria with a functional and, thus, replicative advantage.139

Given the maternal uniparental mode of mtDNA inheritance in humans, mitochondrial replacement therapy has emerged as a tool to prevent mtDNA-transmitted diseases.140 The approach transfers an oocyte nucleus from a mother, who carries mutant mtDNA, to an enucleated oocyte from a healthy mitochondrial donor, then to be fertilized by sperm. Currently, this method is the only allowed germline genetic modification. It is only approved in specific countries, and is strictly restricted to families with recurrent diseases caused by mtDNA mutations.

Tuning mitochondria

The plastic nature of the mitochondrial lipidome and proteome is harnessed to match mitochondrial function to cell-type- and cell-state-dependent metabolic needs. This plasticity is a product of transcriptional, translational, and post-translational regulation, involving both antegrade nucleus to mitochondria, and retrograde mitochondria to nucleus signals.

Transcriptional control of mitochondria

The antegrade transcriptional control of mitochondrial composition involves a network of upstream sensors and downstream transcription factors that sense and relay cellular energy status, nutrient availability, and the organism’s physiological state, enabling short- and long-term adaptive responses to adjust mitochondrial function and abundance.141 A key regulator of the mitochondrial biogenesis network is peroxisome proliferator-activated receptor gamma coactivator 1-alpha protein (PGC-1α). This transcriptional coactivator interacts with various transcription factors, such as, for example, the peroxisome proliferator-activated receptor (PPAR) family, to control the expression‎ of genes involved in mitochondrial FAO.142 PGC-1α activates the expression‎ of nuclear respiratory factor 1/2, which is essential for the increased expression‎ of genes involved in mitochondrial biogenesis, such as the mtDNA copy-number regulator TFAM. The roles of the PGC-1α network extend beyond mitochondria—the network coordinates metabolic programs, such as gluconeogenesis, glucose transport, and lipogenesis.

Nutrient sensors such as the energy sensor AMPK and mTORC1 are key cell growth and metabolism regulators. They act upstream of transcription and factors that regulate mitochondrial composition, behavior, and function. AMPK senses energy status, activated by many mitochondrial stressors that alter the ATP-to-AMP ratio.143 When activated, AMPK phosphorylates its multiple substrates, including PGC-1α, to shift cellular metabolic status to a catabolic program by, for example, increasing mitophagy flux. AMPK roles in autophagy are, however, context dependent.144 The mTORC1 kinase is localized on the metabolic lysosome organelle, where it is regulated by metabolites, including amino acids and growth factors, to coordinate cellular metabolic status with growth, as well as lysosome biogenesis and autophagy via the transcription factor EB (TFEB). mTORC1 activation induces mitochondria to shift to an anabolic repair mode. In mitochondrial replisome stress, mTORC1 activation stalls autophagy and induces 1C metabolism. mTORC1 also directly phosphorylates key lipid metabolic enzymes, such as LIPIN, a rate-limiting phospholipid biosynthetic enzyme that consumes the growth-signaling lipid, PA, for DAG production and fat storage. Inhibition of mTORC1 activates a catabolic state by stimulating autophagy and mitophagy. Inhibitors of mTORC1, such as rapamycin, have shown efficacy in mitigating mitochondrial muscle disease and complex I deficiency in mouse models.145,146 However, rapamycin was highly deleterious in mice with astrocyte-targeted mtDNA depletion.147 These data show that mTORC1 has tissue-specific, opposing effects in different mitochondrial diseases. Also, rapamycin has opposing effects on muscle function in young and developing mice versus old animals. These examples indicate that while inhibition of mTORC1 shows promise as an approach to treating age-related diseases, the physiological context is critical.

Translational control of mitochondria

The mitochondrial translatome is emerging as a mechanism for facile context-specific regulation of mitochondrial function. Localized mRNA synthesis of mitochondrial proteins likely plays a role in mitochondrial biogenesis and stress responses, particularly in highly polarized cells such as neurons. In neurons, mitochondrial biogenesis occurs in axons via a pool of local mRNAs encoding mitochondrial proteins. Localization of mRNAs in axons occurs by their ability to “hitchhike” on organelles, such as endosomes, which ultimately form contact sites with mitochondria for local mitochondrial biogenesis.148 Defects in translation-dependent local mitochondrial biogenesis may contribute to neurodegeneration. For example, in neurons, SYNJ2BP, the mitochondrial anchor for synaptojanin 2, also binds mRNA-encoding PINK1, an ubiquitin kinase that is essential for mitophagy.149 ER-mitochondria contact site proteins bind mRNAs that encode mitochondrial proteins, independent of cytosolic translation.149,150 Such mitochondrial tethered mRNAs have also been shown to play a role in cellular recovery from translational stress, suggesting that the proximity of mRNA-encoding mitochondrial proteins fast-track their synthesis and import. One example of an mRNA-binding contact-site protein is the PA transporter, RMDN3, which suggests a mechanism by which lipid and protein biogenesis may be coordinated. The mitochondrial translatome is also regulated by the highly conserved cytosolic protein clustered mitochondria homolog (CLUH), which functions to increase the stability and translation of a subset of nuclear-encoded mRNAs that encode mitochondrial-targeted proteins.151 CLUH functions are essential at birth when the glycolytic metabolism of fetal mitochondria transitions to oxidative metabolism. CLUH senses nutrient status and the cell cycle and is highly regulated by phosphorylation status by nutrient sensors such as mTORC1, which inhibits CLUH.152

Post-translational control of mitochondrial functions

At the post-translational level, the action of over 20 mitochondrial proteases plays a major role. They tune the proteome to match mitochondrial function to cell type and state dependent metabolic needs, in addition to their functions in quality control. For example, modulation of the mitochondrial lipidome by the IMS-based AAA protease, YME1L, is a regulatory post-translational mechanism in cells. The half-lives of the intramitochondrial lipid transfer proteins STARD7, PRELID1, and PRELID3B, and the PRELID partner TRIAP1 are determined by YME1L, which also degrades around 40 additional key mitochondrial proteins, including regulatory subunits of protein translocases in the IM.92,153 During hypoxia, mTORC1 is inhibited, which activates its substrate LIPIN through dephosphorylation. The resulting decrease in the PA pool reduces PC biosynthesis and ultimately causes a reduction in both PS and PE levels in the IM. A decrease in PE in the IM causes activation of YME1L, which in turn decreases the steady-state levels of intramitochondrial lipid transporters and import components.153 Thus, in response to hypoxia and starvation, mitochondrial lipids are altered by a post-translational pathway triggered by TORC1 inhibition that acutely arrests mitochondrial lipid and protein biogenesis to adjust mitochondrial function to altered metabolic homeostasis.

Mitochondrial quality control and repair

Mitochondrial dysfunctions are signaled in a retrograde manner to elicit corresponding stress responses that progress in stages. These range from remodeled metabolic pathways that promote repair to removal of damaged proteins by microautophagy, mitophagy,154 and mitochondrial-derived vesicle-mediated pathways.155,156,157 When the capacity of these pathways is exceeded, the cells die by apoptosis, ferroptosis, or pyroptosis, depending on a signal and cell/tissue context. Disease models in vivo and in vitro have revealed a spectrum of mitochondrial stress responses, reflecting the multitude of mitochondrial functions that are sensed by the cell and responded. The responses with relevance to mammalian stress and diseases include ISRmt conserved in mouse and human tissues,30,40,158,159,160,33 mitochondrial unfolded protein response (UPRmt) characterized largely in C. elegans and mammalian cell culture models,161,162 and mPOS.63,163 Also, lipid-driven responses have been reported in different model systems from yeast to worms, but the relevance of these for mammalian cells to tissues remains to be studied.

Mitochondrial proteases

Cells have evolved numerous redundant pathways for the turnover of their components. Several of these mechanisms target mitochondria to reduce the organellar amount, change the mitochondrial proteome to match a new metabolic state, or eliminate and recycle damaged organellar components.164 Mitochondrial proteases are the first-line responders for mitochondrial quality control. AAA proteases are localized to the IMS, IM, and matrix and are associated with import and assembly machinery where they monitor, extract, and degrade misfolded, misassembled, and damaged proteins: YME1L in the IM has its active site localized to the IMS, and that of the AFG3L2/paraplegin protease faces the matrix. The soluble ClpXP and LonP reside in the matrix. The IM-embedded AAA proteases are associated with stomatin/prohibitin/flotillin/HflKC (SPFH) proteins, which create microdomains to organize the proteases together with other components dedicated to mitochondrial complex assembly and quality control.165 Like the IM, quality control of the OM also involves mechanisms to recognize, extract, and remove defective and mistargeted proteins and precursor proteins that may impede the TOM machinery.

While some mitochondrial proteases, such as YME1L, are constitutively active, the mitochondrial metalloprotease OMA1 is mostly dormant in unstressed cells.166 It gets activated by loss of ΔΨmv, heat, and toxins and plays a unique quality-control pro-survival role.48,167 Activated OMA1 processes several substrates, including itself, as a negative feedback loop. One of these is the membrane isoform of the mitochondrial IM fusion dynamin OPA1. OMA1 cleavage causes OPA1 inactivation and, consequently, mitochondrial fragmentation, which promotes mitophagy.168 OPA1 is also constitutively processed by YME1L, which generates shorter OPA1 isoforms required for fusion. In this context, OPA1 also acts as an anti-apoptotic factor via its ability to promote crista remodeling, independent of its role in IM fusion.169 Given its pro-survival functions, overexpression‎/activation of OPA1 mitigates mitochondrial disease and neurodegeneration in pre-clinical mouse models, but also may promote tumorigenesis. Consequently, OPA1 inhibitors have been proposed as candidates for cancer therapy. The therapeutic window for OPA1 inhibitors and activators is still being elucidated. OPA1 mutations cause optic atrophy and various signs of neurodegeneration,170 and care should be taken to avoid such adverse events.

Vesicle-mediated mitochondrial turnover and mitophagy

Mitochondrial protein turnover is also mediated by the compartmentalization of proteins to budding membranous vesicular structures (MDVs), which may be equivalent to structures positive for outer mitochondria membrane (SPOTs) and mitochondrial-derived compartments (MDCs).171,172,173,174 Generation of these vesicles is likely conserved, mediated by the pulling forces of cytoskeletal-associated motors attached to the OM via the MIRO family of proteins at ER-mitochondrial contact sites. Multiple functions have been attributed to these vesicular compartments. These include plasma membrane mitochondrial antigen presentation, degradation-mediated quality control—which may be selective to OM stress—and signaling via their fusion to organelles such as peroxisomes (Figure 3).157 Their formation is targeted by parasites to escape cellular defense systems.174 The cell-type-specific roles of these vesicles in homeostasis and disease states, such as neurodegeneration, is an active, underexplored area of mitochondrial biology.

The best-understood quality-control pathway for mitochondria is selective macroautophagy or mitophagy, which directs whole or parts of mitochondria (piecemeal mitophagy) to lysosomes for degradation.175,176 The mechanism depends on the Pink1 kinase,177 whose import is arrested at the OM when membrane potential is reduced. At the OM, Pink1 phosphorylates the E3 ubiquitin ligase parkin and ubiquitin, creating a phosphoubiquitin cloud, marking mitochondria for mitophagy178 in cultured cells. Additional ubiquitin-independent mitophagy pathways are initiated by OM receptor proteins, such as NIX, BNIP3, and FUNDC1.179 Pink1-dependent mitophagy became a compelling candidate to explain Parkinson’s pathogenesis, as genetic defects in Pink1 and parkin cause early-onset Parkinson’s disease; however, Pink1 or parkin knockout mice show no dopaminergic neuronal loss, nor do they have detectable defects in mitophagy.180,181 An intriguing alternative mechanism for parkin/Pink1-related Parkinson’s disease may involve MDVs; Pink1 and parkin repress MDV-mediated mitochondrial antigen presentation and potentially release the cytosolic mtDNA and RNA to decrease inflammation.182,183,184 Therefore, Pink1 and parkin deficiency could cause immunity activation on cells that present antigens, including dopaminergic neurons.

Mitophagy occurs in vivo in normal physiology, often in a cell-autonomous manner, and as such can be activated or stalled in adjacent cells of a tissue. This is observed in mouse models of mitochondrial disease in which mitophagy appears mosaic: in the muscle fibers with activated ISRmt and attenuated disease pathology, mitophagy is upregulated, whereas in those with an advanced disease stage, mitophagy is undetectable.185 Loss of mitophagy also co-occurs with increased mTORC1 activity in mitochondrial muscle disease; consistently, mTORC1 inhibition with rapamycin activates mitophagy,185 indicating that mitophagy defects contribute to mitochondrial muscle disease progression. These data emphasize the importance of single-cell analysis in disease models associated with mitophagy.

Mitochondrial integrated stress response

ISRmt is a major mammalian mitochondrial retrograde pathway.30,33,158,159,160 The signaling pathway involves the stress protease OMA1 and the mitochondrial protein DAP3 binding cell death enhancer 1 (DELE1).186 Under stress, OMA1 processes DELE1 during its import into mitochondria, leading to its accumulation in the cytosol. Cytosolic DELE1 interacts with the HRI kinase, which phosphorylates the alpha subunit of eukaryotic initiation factor 2 (eIF2α) in response to various cell stresses, including heme deficiency, oxidative stress, and viral infection.186 Phosphorylation of eIF-2-alpha can result in the translation of the activating transcription factors ATF4, ATF5, and ATF3, depending on the cellular context and stress type, which induces the ISRmt transcriptional program. This program has redox and metabolic arms33 that metabolically rewire cells to promote repair and survival at a cellular and organismal level dependent on the type of tissue and dysfunction. Whether DELE1 and OMA1 are the only initiators of ISRmt, how they contribute to the stage-wise progression of ISRmt,33 and which human diseases they are relevant for remains to be determined. Analysis of SIFI, which mediates the degradation of DELE1 and its target ISR kinase, HRI, indicates that chronic activation of DELE1/HRI is detrimental to cell survival and is likely a causative factor in neurodegeneration diseases.64 Thus, ISRmt can be beneficial or harmful depending on the disease context.

ATF transcription factors orchestrate the transcriptional arm of ISRmt, which is similar but not fully overlapping with the conventional ISR responses initiated by the other eIF2α protein kinases, GCN2, PKR, and PERK, which sense different types of stress. Indeed, recent work indicates that HRI selectively drives mitophagy in proliferating cell lines, dependent on the mitochondrial localization of phospho-EIF2α.187 In proliferative cells, ATF4 is the main transcriptional effector of all ISR pathways. At the same time, ATF5 is the main driver of the ISRmt response in postmitotic tissues, such as muscle, heart, and astrocytes, in response to mtDNA expression‎ defects.29,143 In contrast, ATF3 is induced at the terminal stage of the ISRmt.29,141,142,143 These findings indicate that the ATFs act in ISRmt in a context-specific manner. In mammals, ISRmt is robustly induced by mtDNA expression‎ defects (e.g., deletions, mt-tRNA mutations, depletion, translation machinery)141,142,143 and UCP1 overexpression‎.144,145 However, human disease mutations that directly affect OXPHOS proteins or assembly factors activate inflammatory pathways instead of the ISRmt,147 consistent with signal specificity of ISRmt activation.

The metabolic arm of the ISRmt shifts cells to a reparative metabolic state by shifting energy metabolism to glycolysis to direct glucose-derived carbons to serine biosynthesis and the folate cycle and 1C metabolism and by upregulating the mitochondrial folate cycle158,159,188 (Figure 2B). 1C metabolism is highly conserved and present in mitochondrial ancestors, indicating the utility of its dynamic metabolic pathway structure that can flexibly produce growth components to cells under nutrient replete conditions, tailored in multicellular organisms to prioritize the needs of specific cell types. The resulting alteration in the methyl cycle causes imbalanced nucleotide, amino acid, and glutathione pools and decreased carbon flux toward phospholipid and creatine synthesis, major consumers of 1C SAM units. Upregulating nucleotide precursors is physiologically relevant in tissues or cells with mtDNA replication defects, as it aids the mtDNA replisome in postmitotic cells with low dNTP pools. When ISRmt is chronically induced in disease, the imbalanced nucleotide pools can contribute to the accumulation of multiple mtDNA deletions.

De novo serine synthesis is a major component of chronic ISRmt response and a major source of 1C units, contributing to nucleotide, glutathione, and phospholipid synthesis.189 Serine is synthesized de novo from the glycolytic intermediate 3-phosphoglycerate via the action of 3-phosphoglycerate dehydrogenase (PHGDH), the rate-limiting enzyme. In mitochondrial disease models that manifest in the heart or muscle, serine 1C units derived from glucose carbons rapidly and fully label glutathione, indicating their importance in redox balance and antioxidant capacity in these tissues.158 Even though diet and protein degradation are sources for serine, de novo serine synthesis is an essential part of ISRmt in mitochondrial disease: exogenous serine supplementation cannot replace it.159 This suggests that the spatial localization of serine synthesis, potentially on mitochondria, during mitochondrial stress is important. The metabolic program in proliferating cancer cells closely mimics ISRmt, with induction of glucose uptake, mitochondrial folate cycle, serine synthesis, mTORC1 activation, and ATF involvement.190 De novo serine synthesis and the resulting anabolic shift is a growth-promoting factor in many cancers and an important target of therapy.153 In these cells, 1C metabolism provides nucleotides for the replicating cells and phospholipids for their membrane synthesis. mTORC1 activation is an important component of ISRmt of the muscle and heart as well.145 Indeed, rapamycin treatment downregulates ISRmt and its metabolic consequences in mice with mitochondrial muscle disease.145

Characterization of ISRmt has enlightened how systemic signaling occurs in mitochondrial diseases. Activation of ISRmt results in the secretion of metabokines (mitokines) FGF21 and GDF15 into blood circulation.30,191 These hormone-like proteins impinge on systemic metabolism. FGF21 upregulates glucose uptake and modifies lipid metabolism, inducing lipolysis in the liver and white adipose tissue (WAT) and facilitates fat browning in mitochondrial mutant mice.30,188 The endocrine role extends to the brain: muscle-derived FGF21 leads to local glucose uptake and mitochondrial biogenesis in the hippocampal dorsal-CA2 region.33 FGF21 has also a local autocrine/paracrine role: in chronic stress it is required for the second stage of ISRmt33 to induce serine synthesis and the transsulfuration pathway. The role of GDF15 in mitochondrial disease physiology still needs to be fully understood. However, like FGF21, GDF15 also targets the central nervous system, modulating feeding behavior.192 This raises the question of whether these mitochondrial endocrine stress-responsive factors have protective and/or pathological effects distal from the original affected tissue. GDF15 has been described to have anti-inflammatory roles by, for example, suppressing cytoplasmic release of mitochondrial nucleic acids.193 This is especially intriguing because innate immunity and inflammation are increasingly reported as contributors to mitochondrial diseases.

Mitochondria in innate immunity and inflammation

While tissue damage causes mitochondrial antigen release to extracellular sites and inflammation, a virus infection can trigger mitochondria to release its components into cytoplasm.194 Amongst these components are elements of the mitochondrial genetic system—mitochondrial DNA, RNA, and mitochondrial-derived formyl peptides (mDAMPs)—that can act as potent pro-inflammatory agonists of the innate immune system. These mDAMPs activate cellular pattern recognition receptors (PRRs), such as the double-stranded DNA-sensor cyclic GMP-AMP synthase (cGAS)/stimulator of IFN genes (STING) pathway and dsRNA sensor RIG-I.195 RIG-I binds viral and mtRNA at different sites to induce activation, suggesting that mitochondrial-derived RNA can shape the early response to the infection. The DNA- and RNA-PRR activation pathways induce a downstream cascade of an innate immunity response, upregulating nuclear transcription programs and a type-I IFN response, leading to cytokine release to counteract the infection.195 Under physiological conditions, quality control systems, such as mitophagy, have evolved to safeguard against the release of mDAMPs, but when the capacity of these quality-control systems is exceeded, or when they are defective, inflammation can drive pathogenesis. In acute metabolic stress, for example, when pyrimidines are depleted and/or in the event of genotoxic stress, the release of mDAMPs and the associated activation of innate immunity may also play protective roles in organisms.

The mechanisms of mtDNA and mtRNA release from mitochondria are still only partially understood and based mostly on studies in cultured cells. Two pore-based mechanisms have been described, formed by the oligomerization of BAX/BAK or VDAC in the OM.120,121,196,197 More recently, an endosomal-mediated pathway has been implicated in removing defective nucleoids and their cytosolic release.198,199 Impaired cristae morphogenesis and mitochondrial lipid biogenesis increases cytosolic mtDNA release, more broadly indicating that mechanisms that maintain mitochondrial integrity safeguard against the release of mDAMPs.118 The release of mDAMPS has also been linked to defective mitophagy and defective lysosomal function via the lysosomal DNA TLR9 sensor. The induction of mtDNA breaks also activates RIG-I,121 but how and what molecules ultimately activate RIG-I are unknown. mtDNA nucleoid proteins contributing to mtDNA processing or synthesis are attracting interest as active contributors in this pathway. The pathway is relevant in mitochondrial disease: in cells from patients with mitochondrial recessive ataxia syndrome (MIRAS) and in mice carrying a homologous MIRAS point mutation in POLG1, the virus infection-induced mDAMP release is delayed, slowing down RIG-I activation and the IFN-I response. This results in increased viral replication (TBEV, HSV1, SARS-CoV-2) in early infection, followed by an overactivated, delayed proinflammatory response that causes liver damage and acute loss of GABAergic neurons.200 Such findings may explain the epileptic form of MIRAS and the hepatotoxic response of MIRAS patients to the epilepsy drug valproate. Similar aberrant mDAMP-related activation will likely emerge when studied in organismal disease models. It seems likely that mDAMP-mediated inflammation will be highly relevant for mitochondrial and age-linked diseases, but much work is needed to understand the underlying mechanisms of mitochondria in innate immunity as well as disease- and tissue-selective effects.

Tissue-specific mechanisms of mitochondrial diseases

Dysfunctions of the enzymes regulating the mitochondrial replisome and nucleotide pools are an important cause of human mitochondrial diseases and an excellent example of their tissue-specific manifestations, mDAMPs being a new contributor to such outcomes. Phenotypes of mitochondrial DNA depletion syndromes (MDSs) typically manifest early and their manifestations are restricted to specific tissues, such as in the skeletal muscle (TK2, MGME1) or both in the liver and brain (POLG, Twinkle, dGUOK).201 Ribonucleotide reductase dysfunction affects all nucleotide pools and causes a multisystemic MDS.201 Adult-manifesting disorders caused by the same nuclear mutations cause mtDNA instability and the accumulation of mtDNA deletions and typically cause mitochondrial muscle disease and/or symptoms in the nervous system.4 The most common cause of these disorders is mutations in DNA polymerase gamma, POLG1. There is an exceptional number of non-pathogenic and disease-causing amino acid variants in humans,202 suggesting active evolution of the protein. Also, the manifestations of POLG1 are highly variable, the most common being early- or juvenile-onset epilepsy-hepatopathy, adult-onset spinocerebellar ataxia, peripheral neuropathy, or middle-age-onset Parkinson’s disease, only partially explained by mDAMPs (see previous paragraph).203,204,205,206 POLG disease mechanisms may also be relevant for common, sporadically occurring diseases with shared phenotypes. The reason why the replisome and nucleotide pool diseases manifest typically in postmitotic cells and not in proliferative cells that are especially dependent on mtDNA replication is likely explained by several-hundred-fold lower dNTP levels in non-dividing cell types compared to proliferating cells or embryonic cells. This knowledge has been used for therapy: peroral nucleoside treatment started early in disease stops disease progression and prevents lethality of children with TK2 mutations.207 In some patients, the disease process has even been reversed. These findings emphasize the curative potential of precision metabolite therapies.

The variability of mitochondrial diseases is reflected in their broad manifestations in the nervous system (Table 1)—encompassing the whole field of neurological diseases. Some genotype-phenotype correlations exist, but the underlying pathways of pathogenesis vary and affect different nervous system cell types. Indeed, even a “typical” mitochondrial brain disease does not exist. Also, clear indications of external or endogenous modifying factors, such as viral infections, exist as a single homozygous nuclear mutation of one gene can cause highly variable phenotypes.200 Therefore, no single treatment is likely to benefit all mitochondrial brain diseases, and even one disease may require several different targets as a treatment strategy.

ConditionLeigh syndromeAlpers-HuttenlocherMELASMitochondrial ataxia w/o epilepsyParkinsonismSensory organ defects

most common defect	subunits and assembly factors of OXPHOS complexes	mtDNA replisome POLG1, Twinkle	mitochondrial translation tRNALeu(UUR)	mtDNA replisome frataxin	mtDNA replisome Pink1, Park2	OXPHOS subunits, mt-translation, mt-replisome
potential molecular mechanisms	hypoxia/hyperoxia, energy crisis, lactacidosis, oxidative stress, microglial activation, macrophage, or lymphocyte invasion to CNS	toxic astrocyte reactivation, microgliosis, inflammation	lactic acidosis, one-carbon metabolic remodeling, redox imbalance, nitric oxide metabolism, L-arginine deficiency	poor antiviral defense, inflammation, astrocyte reactivation, stalled autophagy	mt-antigen presentation, organellar turnover, metabolic remodeling, oxidative stress	ionic imbalance, lack of inhibition of sensory signals, glutamate toxicity, metabolic remodeling, energy crisis protein/organellar turnover
typical models for mechanistic studies	NDUFS4-knockout mice	Twinkle-cKO mice, GFAP promoter	patient findings, human samples, cell lines	MIRAS knock-in mice, Fxn-cKO- prion protein promoter; human samples, cell cultures	parkin S65A knockin mouse	OPA1 mutations expressed in retinal pigment cells
main affected brain area, patients	symmetric hypervascularized necrotic lesions in the brain stem	laminar necrosis, spongiotic lesions of cerebral cortex	metabolic strokes in the cerebral cortex, often occipital	Symmetric lesions in the cerebellum, thalamus	loss of neurons in substantia nigra, autonomous nervous system dysfunction	atrophy of retina, optic nerve, ocular muscles, inner ear, or sensory nerves
typical life stage of manifestation	before 2 years	2–4 years	before age 20, but may occur any age	FRDA: 5–15; MIRAS: 10–45	after 30 years	all ages

Table 1

Highly variable nervous system manifestations of mitochondrial diseases

Abbreviations: mt, mitochondrial; OXPHOS, oxidative phosphorylation; cKO, conditional knockout; MELAS, mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes; CNS, central nervous system. Key references are cited here.183,184,200,208,209

• Open table in a new tab

Some mitochondrial diseases with otherwise nervous system symptoms are linked to reproductive system manifestations. Adult-onset dominant POLG1-related diseases show parkinsonism, male infertility, and early menopause in women.204 The recessive defect of Twinkle helicase, the companion of POLG complex in the mtDNA replication fork, causes childhood-onset cerebellar ataxia and ovarian atrophy, while male fertility is unaffected. The connection of Parkinson’s disease/ataxia and germ cell dysfunction is unusual in the field of neurology. Deciphering the link between mtDNA maintenance mechanisms and neural disease and germline stem cell maintenance is an important line of future research.

The main focus of neurodegenerative disease research has been neuron centric. However, activation of astrocytes can also be a primary driver of mitochondrial brain disease. The postnatal knockout of Twinkle helicase, leading to slow, progressive mtDNA depletion, causes pathological reactivation of astrocytes, spongiotic encephalopathy, and progressive neurological disease.210 However, a similar Twinkle knockout in neurons manifests late, followed by rapid disease progression and death. The astrocytic reactivation results in structural defects of the primary cilia of these cells, which suggests that mitochondrial spongiotic encephalopathies constitute a new disease group: metabolic ciliopathies.211 The finding raises questions about links between mitochondrial metabolism and nervous system growth signaling—a novel field of investigation.

Mitochondrial mechanisms of aging

Mitochondrial roles in aging have attracted attention since Denham Harman presented his mitochondrial theory of aging.212 He proposed that mtDNA mutations accumulate during aging, which increases free radical species that further increase mtDNA mutagenesis, and that this Muller’s ratchet is a determinant in the aging process. Since then, ROS has been considered an important contributor to aging, but experimental data to support this idea remains scarce. “Mitochondrial dysfunction” is considered a central node in the network of aging hallmarks, affecting other key cellular processes and contributing to age-related pathologies. The critical question is: what features of mitochondria are altered and cause age-associated changes? We posit that the core contributions of mitochondria to aging are multifaceted and lie in their ability to promote metabolic state transitions in response to extrinsic and intrinsic signaling and stresses and in their roles in inflammation and stem cell maintenance.

Mitochondria and stem cell maintenance

Progeroid syndromes are rare genetic disorders characterized by clinical features resembling premature aging.213 Initially, these syndromes were thought to be caused exclusively by mutations in components of the nuclear envelope (e.g., Hutchinson-Gilford progeria syndrome) and in DNA-repair pathways (e.g., Werner syndrome). The discovery that mutant proteins (e.g., MTX2 and TOMM7) in mitochondria can give rise to similar progeroid endophenotypes underscores their interconnectivity with the nucleus and raises the possibility of a unifying aging mechanism. Targeted inactivation of the exonuclease-domain of DNA polymerase gamma causes accumulation of mtDNA mutations in the mtDNA “mutator” mouse. mtDNA mutator mice show progressive aging-like signs from six months of age, with osteoporosis, gray hair, thin skin, and progressive fatal anemia214,215 that are typical for mouse progeria models. These mice do not, however, support Harman’s mitochondrial theory of aging. MtDNA mutation frequency does not increase exponentially. Furthermore, only the hematopoietic progenitor cells show signs of oxidative damage, leading to early capture and degradation of newly generated erythrocytes by the spleen macrophages, with major disturbance of systemic iron homeostasis.50 Also, mitochondrial disease patients with high mtDNA heteroplasmy level show no signs of progeria. Recent evidence points to an alternative mechanism to explain mtDNA mutator progeria: the exonuclease-deficient POLG1 enzyme makes the polymerase faster in mtDNA synthesis.216 This causes imbalanced nucleotide pools, especially in the stem and progenitor cells, causing nucleotide depletion in the nucleus, cell cycle stalling in S-phase, genomic DNA replication stress, and double-strand DNA breaks.216 Genomic instability in stem cells/progenitors is a typical cause of mouse progeria, implying that mtDNA mutator shares progeria mechanisms with other progeria models affecting nuclear DNA integrity. The findings open again the question of whether mtDNA mutations contribute to aging. To conclude, stem cell/progenitor genomic instability is a common feature for mouse progeria models, including the mtDNA mutator mice. More recently, mutations in the mitochondrial proteins MTX2 and TOMM7, which function in OM protein biogenesis, were also linked to human progeroid syndromes.217,218

Mitochondria also contribute to stem cell fates: in asymmetric stem cell division, different populations of mitochondria are differentially segregated to the daughter cells. The newly created organelles remain in the quiescent stem cells while the “old” mitochondria are segregated into daughter cells that commit to a progenitor fate.219,220 These separate mitochondrial populations have different proteomes that facilitate distinct metabolic programs: the progenitors have active oxidative phosphorylation, while quiescent stem cells are anabolic, with immature oxidative phosphorylation machinery.220 ROS is an important signal in stem cell differentiation,51 which may be one of the mechanisms in how actively oxygen-using “old” mitochondria drive progenitor fate. In this context, mitochondria and their metabolic changes may contribute to stem cell aging-related exhaustion.

Epigenome and mitochondrial functions

Changes to the epigenome are another hallmark of aging and contribute to the aging process. The mitochondrial TCA cycle and the 1C pathway provide essential cofactors for histone and genome modifications. The main methyl donor SAM is used for methylation of both targets, while the TCA metabolite alpha-ketoglutarate (α-KG) serves in demethylation as a co-substrate for 5-methylcytosine hydroxylases such as the ten-eleven translocation (TET) family. This raises an interesting possibility that the Horvath clock of biological aging, which relies on changes in DNA methylation at specific loci,221 may reflect changes in mitochondrial catabolic and anabolic metabolic modes. α-KG can also be metabolized to the oncometabolite 2-hydroxyglutarate that inhibits TET enzymes.222 Histone acetylation and de-acetylation are also directly regulated by mitochondrial metabolism. Acetylation uses acetyl-CoA, produced by catabolism of pyruvate and amino acids in the mitochondria, and can also be synthesized in the nucleus. An abundant histone deacetylase enzyme class is the sirtuins, which are activated by increased NAD+/NADH ratio and regulated by mitochondrial oxidative metabolism.223 These examples indicate that mitochondrial functions are essential for histone and genome modifications. Furthermore, dynamic cycling of mitochondrial anabolic and catabolic states as a function of stresses and aging has widespread effects on the epigenome. The plastic nature of mitochondrial metabolism suggests that interventions that target disease- and aging-related epigenetic pathways can potentially promote healthspan.

Mitochondria, proteostasis, and aging

Dietary restriction can extend lifespan in different species, including mammals.224 Fasting increases the NAD+/NADH ratio; thus, NAD metabolism has been a target to increase healthspan.225 Chronically decreased NAD+ recycling represses mitochondrial oxidative phosphorylation, ATP production, and mitochondrial membrane potential, which leads to the accumulation of cytosolic mitochondrial precursors, significantly impacting stress and proteostasis pathways, another aging hallmark. Data are still primarily based on model organisms, but their relevance to humans is increasing.

The decline of function of one organelle can cause a feedforward chain of dysfunctions. For example, the decline of mitochondrial membrane potential can drive lysosome deacidification, leading to defects in macroautophagy and stalled autophagic flux, a typical finding in age-linked neurodegenerative diseases.226 Such declined recycling of cellular components is associated with increased apoptosis and chronic inflammation.

Intriguingly, mitochondrial roles in metabolism and their ability to modify cell fate and functions are the common denominator between all the “hallmarks” of aging.227 Mitochondria can send and receive signals from distal cells and tissues to tune whole organism metabolism via “mitohormesis”—a restorative metabolic mode.228 This underscores the exceptional importance of this organelle in systemic homeostasis. Detailed information on mitochondrial roles in metabolic dynamics in different tissues during aging is required to be able to understand degenerative processes. Furthermore, knowledge of mitochondrial functions in providing metabolites for epigenomic reprogramming in cellular rejuvenation229 has the potential to give valuable insight into the mechanisms of aging-related decline.

Such knowledge will pave the way to determining whether metabolic reprogramming could modify or delay age-linked changes to cellular, tissue, and organismal functions.

Open questions and future directions

The amazing development of transformative technologies and AI-based approaches will ultimately enable the resolution of complete mitochondrial atomic structures. Mass-spectrometry-based proteomics of purified complexes at high throughput and of chemical cross-linked proteins,230,231,232,233 together with complexome profiling, have begun to solve complete mitochondrial interactomes. This information can be harnessed using generative AI-based protein structure predictions to create high-resolution models of unsolved mitochondrial complexes in complex with metabolites and lipids. Such structures and those solved using single particle cryoelectron microscopy can also be imaged in cells using correlative cryo-focused ion beam scanning electron microscopy (FIB-SEM) and/or super-resolution light microscopy to ultimately reveal higher organizational features of mitochondria and their emergent properties. More sensitive and specific metabolite biosensors could give more insights into the metabolism dynamics at the organellar, single-cell, and tissue levels and extend our metabolic understanding from in situ to in vivo. In vivo, metabolic flux analyses uncover metabolic requirements of different cell types and tissues and dynamic changes at the organismal level. Further, the development of spatial metabolomics by imaging mass spectrometry will capture the metabolism of functional units within tissues with increasing resolution.183,184 Finally, a suite of innovative approaches will reveal the interaction interface between the metabolome and higher-order protein, RNA, and DNA biomolecules,234,235 including deep networks.

The multifaceted role of mitochondria as signaling organelles is becoming increasingly recognized. Mitochondria signal cellular state using metabolites by binding to proteins and other macromolecules to shift metabolic flux to match extrinsic or intrinsic stresses and modify cellular functions to match cell-type specific needs. Only a fraction of this complex network is currently understood. How metabolons spatially organize inside or outside of mitochondria and how organellar interactions contribute to this in different metabolic states is also an emerging topic. Which enzymatic pathways organize into metabolons to potentiate metabolic flux? Do mitochondria generate metabolite gradients with specific functions, like regulating growth factors? How do concentrations of mitochondria-dependent gasotransmitters, such as H2S, CO2, NO, and HCN, fine-tune oxidative phosphorylation and anabolic biosynthesis pathways by acting on the activity of complex IV? Moreover, the unexpected roles of mitochondria as modifiers of primary ciliary and growth-factor signaling are emerging in the nervous system. Additional knowledge is required to understand how metabolism is compartmentalized between different, functionally linked cells and how mitochondria influence and are influenced by the extracellular matrix to modulate cell function. These examples highlight the intricate multifaceted and interlinked modes of mitochondrial communication within cells, tissues, and the whole organism.

Mitochondria are central for nutrient sensing and metabolism, implying that nutrition is an important factor in mitochondrial dysfunctions and diseases. Despite the apparent logic, this interaction needs to be studied more, specifically because diet can be a therapeutic tool in mitochondrial diseases. In fatty acid dehydrogenase deficiencies, a diet deficient for a specific length of fatty acids delays disease progression. In the case of pyruvate dehydrogenase deficiency, a high-fat low-carbohydrate diet is essential, indicating that the diet depends on the disease in question. However, nutrition has remarkable effects even in diseases that do not directly affect lipid or glucose metabolism. In adults with mitochondrial myopathy, a ketogenic diet induces remarkable mitochondrial biogenesis in most muscle fibers but causes lysis of those severely deficient for respiratory chain activity. In this case, nutrition functions as a “metabolic knife,” removing the most dysfunctional fibers that could not metabolize fatty/keto acids.236 Nutrition also causes mitochondrial disease. In areas of drought in sub-Saharan Africa, the main food source is a cyanide-containing cassava root. The high cyanide exposure causes spinocerebellar ataxia, called konzo, via inhibiting complex IV of the respiratory chain.237 These examples emphasize the intimate crosstalk between nutrition and mitochondrial functions. Therefore, systematic studies of personalized diets for specific metabolic diseases will likely be important tools for future medicine.

The centrality of mitochondria in cellular metabolic homeostasis, the tissue- and cell-specific use of metabolites, and mitochondrial roles in innate immunity create exceptional complexity in human mitochondrial disease and common diseases that involve altered mitochondrial functions. Diseases are systemic; therefore, knowledge at the systemic and molecular level is required to gain pre-clinical evidence as the basis of therapy trials. To achieve this, information based on cell culture systems and/or non-mammalian model organisms must be more comprehensive, although innovative therapeutic approaches have already emerged from mouse models engineered to mimic human diseases closely. The remarkable potential of targeted, tailored metabolic modifier therapies is being understood: the treatable target may be a secondary indirect metabolic alteration far from the disease’s primary cause. Furthermore, modifier genes that alter age-of-onset and/or the severity of symptoms have been identified based on ample clinical data. However, few factors and mechanisms underlying these buffering effects have been identified. Such buffers could be attractive therapy targets, as they offer potential to delay disease manifestation. In addition to metabolic tools, correcting the primary gene defect would be optimal. Viral-mediated gene therapy and gene-editing tools have been developed for nuclear genes, and even mtDNA has been successfully edited by approaches using cytidine deaminases, TALENs, or zinc-finger nucleases.238 CRISPR-Cas9 editing has not been successful as human mitochondrial DNA resists transformation.238 The delivery of these tools in sufficient amounts to mitochondria that are present in hundreds even in a single cell and delivering them to specific tissues, especially in the nervous system, still remain major challenges.

In conclusion, mitochondria are at the center of metabolism plasticity and have evolved to become multifaceted signaling organelles. Through their signaling mechanisms, they integrate nutrient availability with cellular and organismal status to match needs at the cellular, tissue, and organismal levels for fuel, growth, and repair. We anticipate that the next decade of mitochondrial biology will unlock the mysteries of health and longevity.

Acknowledgments

The authors would like to thank Eija Jokitalo, Ilya Belevich, Juan Cruz Landoni, and Takayuki Mito for the images in Figure 1. During the writing of this review, A.S. was supported by Jane and Aatos Erkko Foundation, Sigrid Jusélius Foundation, The Finnish Foundation for Cardiovascular Research, Novo Nordisk Foundation, The PolG Foundation, Chan Zuckerberg Initiative, Academy of Finland, Business Finland, University of Helsinki, and Helsinki University Hospital. J.N. was supported by Altos Labs.

Declaration of interests

A.S. is a founder, board member, and shareholder in NADMED Ltd. and included in a patent and is a scientific advisory board member of Khondrion; A.S.’s lab has been funded by NGM-Bio and Capacity Bio while writing the review. J.N. is an employee and a shareholder in Altos Labs Inc.

References

1.

Gabaldon, T. ∙ Huynen, M.A.

Lineage-specific gene loss following mitochondrial endosymbiosis and its potential for function prediction in eukaryotes

Bioinformatics. 2005; 21

ii144–150. Suppl 2

Crossref

Scopus (28)

PubMed

Google Scholar

2.

Lane, N. ∙ Martin, W.

The energetics of genome complexity

Nature. 2010; 467:929-934

Crossref

Scopus (840)

PubMed

Google Scholar

3.

Morgenstern, M. ∙ Peikert, C.D. ∙ Lubbert, P. ...

Quantitative high-confidence human mitochondrial proteome and its dynamics in cellular context

Cell Metab. 2021; 33:2464-2483.e2418

Full Text

Full Text (PDF)

Scopus (112)

PubMed

Google Scholar

4.

Gorman, G.S. ∙ Chinnery, P.F. ∙ DiMauro, S. ...

Mitochondrial diseases

Nat. Rev. Dis. Primers. 2016; 2, 16080

Crossref

Scopus (974)

PubMed

Google Scholar

5.

Morris, A.A. ∙ Leonard, J.V. ∙ Brown, G.K. ...

Deficiency of respiratory chain complex I is a common cause of Leigh disease

Ann. Neurol. 1996; 40:25-30

Crossref

Scopus (123)

PubMed

Google Scholar

6.

Saada, A. ∙ Shaag, A. ∙ Mandel, H. ...

Mutant mitochondrial thymidine kinase in mitochondrial DNA depletion myopathy

Nat. Genet. 2001; 29:342-344

Crossref

Scopus (513)

PubMed

Google Scholar

7.

Mandel, H. ∙ Szargel, R. ∙ Labay, V. ...

The deoxyguanosine kinase gene is mutated in individuals with depleted hepatocerebral mitochondrial DNA

Nat. Genet. 2001; 29:337-341

Crossref

Scopus (484)

PubMed

Google Scholar

8.

Eberhardt, E.L. ∙ Ludlam, A.V. ∙ Tan, Z. ...

Miro: A molecular switch at the center of mitochondrial regulation

Protein Sci. 2020; 29:1269-1284

Crossref

Scopus (38)

PubMed

Google Scholar

9.

Konig, T. ∙ Nolte, H. ∙ Aaltonen, M.J. ...

MIROs and DRP1 drive mitochondrial-derived vesicle biogenesis and promote quality control

Nat. Cell Biol. 2021; 23:1271-1286

Crossref

Scopus (109)

PubMed

Google Scholar

10.

Chatzi, A. ∙ Manganas, P. ∙ Tokatlidis, K.

Oxidative folding in the mitochondrial intermembrane space: A regulated process important for cell physiology and disease

Biochim. Biophys. Acta. 2016; 1863:1298-1306

Crossref

Scopus (27)

PubMed

Google Scholar

11.

Walker, J.E.

The ATP synthase: the understood, the uncertain and the unknown

Biochem. Soc. Trans. 2013; 41:1-16

Crossref

Scopus (450)

PubMed

Google Scholar

12.

Wikstrom, M. ∙ Pecorilla, C. ∙ Sharma, V.