English:Aging – Energy, inflammation and proteostasis as coupled systems

aiMOOC

Aging – Energy, inflammation and proteostasis as coupled systems

QR-Code

Introduction

Aging – Energy, inflammation and proteostasis as coupled systems is an expert colloquium on a central problem in Biogerontology: mitochondrial quality, innate immune signaling, and protein turnover do not fail independently. They form coupled feedback systems whose direction, strength, and timescale can differ by cell type, sex, tissue, and stage of aging. Your task is therefore not merely to list “hallmarks of aging,” but to discriminate causal models.

You will compare current primary studies, derive competing feedback architectures, specify measurable timescales, and design an intervention that can separate models rather than merely improve a phenotype. You will also test three alternatives that frequently complicate causal interpretation: hormesis, compensatory responses, and long-term tissue damage.

The central systems question is:

When an aged tissue shows mitochondrial stress, inflammatory signaling, and impaired protein homeostasis at the same time, which process is upstream, which is compensatory, and which is a downstream consequence?

Load video

YouTube

YouTube might collect personal data. Privacy Policy

Continue Dismiss

Learning Objectives

By the end of the colloquium, you should be able to:

1. Causal inference in aging: distinguish correlation, mediation, feedback, compensation, and downstream convergence in longitudinal aging data.
2. Mitochondrial quality control: connect mitochondrial calcium handling, mitochondrial DNA release, mitophagy, and bioenergetic stress to inflammatory outputs.
3. Innate immunity: compare cGAS–STING, NF-κB, type I interferon, and cytokine pathways at cell-type resolution.
4. Proteostasis: distinguish proteasomal degradation, chaperone-mediated autophagy, macroautophagy, and mitophagy as related but non-identical turnover systems.
5. Timescale separation: use seconds-to-months measurements to infer ordering among signaling, transcription, organelle turnover, proteome turnover, and tissue damage.
6. Experimental design: choose perturbations for expected information gain rather than apparent therapeutic benefit alone.
7. Bayesian model comparison: quantify how strongly a proposed result would change uncertainty among competing mechanistic models.
8. Hormesis: recognize biphasic dose-response and adaptation as alternatives to simple damage accumulation.

A Coupled-Systems View of Aging

A useful starting point is to treat energy metabolism, inflammation, and proteostasis as three interacting state variables rather than three isolated chapters. Mitochondria supply ATP, buffer calcium, generate metabolites and reactive oxygen species, and contain mitochondrial DNA. Protein-quality systems determine whether damaged respiratory-chain proteins, misfolded cytosolic proteins, and dysfunctional organelles are repaired or removed. Innate immune sensors respond to misplaced nucleic acids and stress signals, while inflammatory mediators can in turn alter mitochondrial function, translation, lysosomal activity, and tissue repair.

The observable state of an aged cell is therefore compatible with several causal histories. High mitophagy can mean successful quality control, but it can also mean that damage production is so high that the cell must increase clearance. Low inflammatory output can mean low upstream damage, effective immune suppression, or loss of immune competence. Accumulation of insoluble proteins can arise from slower degradation, increased synthesis of aggregation-prone proteins, altered solubility, or selective survival of cells with different proteomes.

For causal inference, you should separate at least four quantities:

1. Damage input: the rate at which dysfunctional proteins, mitochondria, DNA lesions, or membrane defects are generated.
2. Clearance capacity: the maximum and realized flux through proteasomes, lysosomes, CMA, macroautophagy, and mitophagy.
3. Immune gain: how strongly a given danger signal is converted into NF-κB, STING, interferon, or cytokine output.
4. Tissue consequence: the cumulative effect on synapses, extracellular matrix, vascular function, regeneration, and organ performance.

Timescales as a Causal Tool

A feedback loop can only be inferred if the measurements resolve the processes that are supposed to cause one another. Approximate design windows are:

Process	Useful measurement window	Example readout	Causal use
Cytosolic calcium oscillations	Seconds to minutes	Live-cell calcium imaging	Tests whether altered mitochondrial buffering precedes transcriptional inflammation
NF-κB activation and STING pathway phosphorylation	Minutes to hours	Nuclear NF-κB, phospho-STING, phospho-TBK1, phospho-IRF3	Resolves early immune gain
Interferon-stimulated genes and cytokine secretion	Hours to one or two days	Single-cell RNA, multiplex cytokines	Measures transcriptional and secretory propagation
Mitophagic flux	Hours to days	mito-QC, mt-Keima, lysosomal trapping controls	Distinguishes mitochondrial delivery to lysosomes from static abundance
Proteasomal and lysosomal protein turnover	Days to weeks	Pulse-chase proteomics, activity assays, CMA reporters	Tests whether protein-quality changes precede organelle failure
Whole-proteome replacement	Weeks to months	Stable-isotope labeling and mass spectrometry	Captures slow remodeling and long-lived proteins
Tissue remodeling and functional decline	Weeks to months in mouse studies, often years in humans	Histology, behavior, organ function, frailty measures	Separates transient molecular rescue from durable tissue benefit

These windows are experimental design guides, not universal constants. The exact kinetics vary with pathway, tissue, perturbation strength, age, and species.

Current Primary Studies

The studies below are selected because together they provide orthogonal leverage on mitochondria, immune signaling, and protein turnover. They should not be read as a single linear story.

Study 1: Microglial cGAS–STING and Age-Related Neuroinflammation

Gulen et al., 2023, Nature identified cGAS–STING signaling as a driver of age-associated inflammation and neurodegenerative phenotypes. In aged microglia, perturbed mitochondria were associated with cytosolic mitochondrial DNA, providing a plausible endogenous ligand for cGAS. Genetic and pharmacological manipulations supported a causal role for cGAS–STING in aged inflammatory states. The STING inhibitor H-151 reduced inflammatory phenotypes in senescent human cells and in aged mice and improved several functional outcomes.

The cell-type point is crucial: microglial cGAS activity can propagate inflammatory effects to neighboring cells, so the inflammatory state of neurons cannot be interpreted only from neuronal measurements. This study strongly supports a mitochondrial-DNA-to-innate-immunity route, but downstream STING inhibition alone does not prove that mitochondrial leakage is the only upstream cause.

Model leverage: strong for a leak-first model and for a microglial amplifier; weaker for deciding whether the original mitochondrial defect is caused by proteostasis failure, calcium dysregulation, senescence, or another process.

Study 2: Mitophagy as an Age-Dependent Compensatory Response

Jiménez-Loygorri et al., 2024, Nature Communications used mito-QC reporters to compare mitochondrial turnover in young and old mice. A simple “mitophagy always declines with age” model was not supported. Mitophagy was increased in several old tissues, including parts of the brain, retina, kidney, and liver, while remaining relatively unchanged in others. Acute lysosomal trapping experiments supported increased flux rather than merely increased reporter accumulation. At the same time, markers of general macroautophagic impairment and protein accumulation were present.

In aged retina, the investigators reported a large increase in cytosolic mitochondrial DNA together with cGAS–STING–IRF3 activation and interferon-stimulated genes. Enhancing mitophagy with urolithin A over eight weeks reduced inflammatory and extracellular-matrix programs and improved selected neurological and visual outcomes in aged mice.

The key systems interpretation is that increased flux can be compensatory yet insufficient. If mitochondrial damage production rises faster than clearance capacity, both mitophagy and damaged-mitochondria burden can increase simultaneously.

Model leverage: strong against a universal mitophagy-decline model; strong for tissue-specific compensation; consistent with mitochondrial clearance affecting inflammatory signaling, but urolithin A is not a perfectly pathway-specific perturbation.

Study 3: Mitochondrial Calcium Uptake and Macrophage Inflammatory Gain

Seegren et al., 2023, Nature Aging connected reduced mitochondrial calcium uptake to age-associated macrophage inflammation. Human blood transcriptomic analysis and mouse macrophage experiments implicated age-related reductions in components of mitochondrial calcium uptake, including MCU and MICU1. Reduced mitochondrial calcium buffering amplified cytosolic calcium oscillations and promoted NF-κB activation.

This mechanism is important because it offers a competing route to inflammation that does not require mitochondrial DNA to be the initiating signal. A macrophage can become hyper-responsive because its calcium-handling system changes the gain of inflammatory signaling.

Model leverage: strong for an immune-gain-first model in macrophages; it predicts very early calcium abnormalities before slower transcriptional and tissue-level changes.

Study 4: Protein Turnover Across the Aging Brain, Heart, and Liver

Rao, Upadhyay and Savas, 2024, Molecular Systems Biology used in vivo stable-isotope labeling and proteomics to follow protein turnover across aging. The brain showed distinctive and dynamic turnover changes compared with heart and liver, including sex-linked differences and turnover defects in insoluble protein fractions. Proteasome activity also fluctuated with age and was connected to the turnover of proteolytic machinery.

The study used multi-month labeling windows, illustrating a timescale that is fundamentally different from minutes-to-hours signaling assays. A causal model in which proteostasis failure is upstream must therefore predict not only acute immune markers but also slower changes in protein replacement, solubility, and degradation capacity.

Model leverage: strong for slow, tissue-specific protein-turnover remodeling; it does not by itself establish that altered turnover is the upstream cause of mitochondrial inflammation.

Study 5: Cell-Type- and Sex-Specific Chaperone-Mediated Autophagy

Khawaja et al., 2025, Nature Aging used KFERQ-Dendra reporter mice and single-cell transcriptomic resources to examine chaperone-mediated autophagy across tissues. Many organs and cell types showed age-related CMA decline, often more strongly in males, but the effect was not uniform. Hepatocytes and Kupffer cells, for example, did not show identical aging trajectories.

This result warns against treating “the liver” or “the immune system” as a single compartment. If two neighboring cell populations differ in CMA decline, their mitochondrial stress and inflammatory outputs may diverge even under the same systemic environment.

Model leverage: strong for proteostasis heterogeneity and sex-by-cell-type interactions; it predicts that bulk-tissue averages can hide causal subpopulations.

Study 6: Minority MOMP, Mitochondrial DNA Release, and the Senescence-Associated Secretory Phenotype

Victorelli et al., 2023, Nature showed that a subset of mitochondria in senescent cells can undergo minority mitochondrial outer membrane permeabilization. BAX and BAK macropores enable mitochondrial DNA release, activating cGAS–STING and contributing to the senescence-associated secretory phenotype. Blocking this route reduced inflammatory outputs in senescent human fibroblasts and improved selected aging phenotypes in mice.

This study adds a mechanistically distinct mitochondrial-DNA release process to the discussion. Cytosolic mitochondrial DNA can therefore arise not only from generic organelle “damage” but from regulated membrane-permeabilization events linked to the apoptotic machinery.

Model leverage: strong for a senescence-linked leak mechanism and for distinguishing mtDNA release from a simple loss of mitochondrial abundance.

Cross-Study Comparison

Primary study	Main system and cell type	Key timescale	Causal perturbation or measurement	What it supports	Main interpretive limitation
Gulen et al. 2023	Aged brain, especially microglia; senescent human cells	Hours to months	cGAS–STING genetics and H-151 inhibition	mtDNA-sensitive innate immunity can drive aged inflammatory states	Downstream pathway inhibition does not identify the unique upstream mitochondrial lesion
Jiménez-Loygorri et al. 2024	Retina, brain, kidney, liver and other tissues	Sixteen-hour flux challenge to eight-week intervention	mito-QC flux, cytosolic mtDNA, urolithin A	Mitophagy can rise with age and still be insufficient; mitochondrial clearance and inflammation are coupled	Urolithin A is broader than a single mechanistic switch
Seegren et al. 2023	Human blood signatures and mouse macrophages	Seconds to hours for calcium and NF-κB	Mitochondrial calcium uptake manipulation	Reduced mitochondrial calcium buffering can raise inflammatory gain	Macrophage mechanism may not generalize to all tissues or immune cell types
Rao et al. 2024	Brain, heart and liver proteomes	Months	Stable-isotope turnover proteomics and proteasome perturbation	Protein turnover is dynamic, tissue-specific and linked to proteolytic machinery	Slow turnover associations do not alone establish direction toward mitochondrial inflammation
Khawaja et al. 2025	Multiple organs, hepatocytes, Kupffer cells and other cell types	Age comparison over the life course	CMA reporter and single-cell transcriptomic integration	CMA decline is cell-type- and sex-specific	Reporter and transcriptomic signatures must still be connected to downstream mitochondrial and immune causality
Victorelli et al. 2023	Senescent human fibroblasts and aged mice	Hours to chronic aging phenotypes	BAX and BAK-linked mtDNA release and MOMP inhibition	Senescence can create a direct mtDNA–cGAS–STING route	Senescent-cell mechanisms need not dominate in every non-senescent tissue

The studies agree that aged inflammatory phenotypes are not adequately explained by a single universal decline. They differ in which process is positioned upstream, in the cells studied, and in the temporal resolution. Those differences are precisely what allow model discrimination.

Competing Feedback Models

Model A: Mitochondrial Leak First, Clearance Limited

Core loop: mitochondrial injury increases cytosolic mitochondrial DNA; cGAS–STING and related pathways raise inflammatory output; inflammatory mediators and tissue stress produce additional mitochondrial injury; mitophagy attempts to remove the damaged organelles.

A simple qualitative loop is:

Mitochondrial damage ↑ → mtDNA release ↑ → cGAS–STING ↑ → cytokines and SASP ↑ → tissue and mitochondrial stress ↑ → further mtDNA release.

Mitophagy forms a negative-feedback arm:

Mitochondrial damage ↑ → mitophagy ↑ → damaged-mitochondria burden ↓.

The decisive feature is whether this negative feedback has enough capacity. If damage input exceeds clearance capacity, mitophagy can be high while damaged mitochondria and inflammation also remain high.

Cell-type prediction: microglia, senescent fibroblasts, and retinal cells with high cytosolic mtDNA should show early STING-pathway activation. Neighboring neurons can show secondary damage without being the initial source of mtDNA signaling.

Timescale prediction: cytosolic mtDNA and STING phosphorylation should change before broad tissue remodeling; mitophagic-flux changes should be visible within hours to days; durable functional rescue requires weeks or longer.

Model B: Immune Gain First Through Mitochondrial Calcium Handling

Core loop: reduced mitochondrial calcium uptake increases cytosolic calcium oscillations; this raises NF-κB responsiveness; cytokines create secondary mitochondrial stress; the resulting damage can later produce mtDNA release and STING activation.

MCU or MICU function ↓ → mitochondrial calcium buffering ↓ → cytosolic calcium oscillation amplitude ↑ → NF-κB gain ↑ → cytokines ↑ → secondary mitochondrial stress ↑.

This model does not deny mtDNA–cGAS–STING signaling. It places that pathway later in the causal sequence for at least some myeloid cells.

Cell-type prediction: macrophages should show rapid calcium and NF-κB abnormalities even when cytosolic mtDNA is initially unchanged. Microglia may share some myeloid logic but should not be assumed to be identical without direct measurement.

Timescale prediction: calcium changes appear in seconds to minutes, NF-κB in minutes to hours, cytokines later, and mitochondrial structural damage or mtDNA leakage still later.

Model C: Proteostasis First

Core loop: slower or selective failure of proteasomal, CMA, or lysosomal turnover permits damaged proteins and organelle-quality-control components to accumulate. Mitochondrial function then deteriorates, triggering inflammatory signaling that further burdens protein synthesis and degradation.

Protein-quality capacity ↓ → damaged or long-lived proteins ↑ → mitochondrial quality control ↓ → bioenergetic and membrane stress ↑ → inflammatory signaling ↑ → proteome remodeling and damage ↑.

This model predicts that turnover defects can precede mtDNA leakage, but the relevant signal may be slow and cell-type specific. CMA decline in one cell type and preserved CMA in another could create divergent trajectories within the same tissue.

Cell-type prediction: hepatocytes, neurons, and selected stromal populations with reduced protein-quality capacity should show early turnover or solubility signatures. Kupffer cells or other populations with preserved CMA may not follow the same trajectory.

Timescale prediction: changes in protein half-lives and insoluble fractions emerge over days to months, potentially before chronic tissue damage but more slowly than calcium or STING phosphorylation.

Model D: Hormetic Compensation and Adaptive Overdrive

Core loop: mild stress transiently increases mitochondrial or proteostatic quality control, producing an adaptive state. The same pathway becomes maladaptive at higher dose, greater duration, or older physiological state.

Mild stress → adaptive signaling ↑ → mitophagy and proteostasis ↑ → resilience ↑.

But:

Persistent or severe stress → energetic cost and incomplete clearance ↑ → inflammatory signaling and tissue damage ↑.

This model explains why a marker such as high mitophagy cannot be labeled “good” or “bad” without dose, duration, flux, and outcome data.

Cell-type prediction: cells with reserve capacity should show a transient stress response followed by lower damage, whereas cells near a bioenergetic or proteostatic limit may fail to recover.

Timescale prediction: a beneficial hormetic response should show an early perturbation, a recovery or overshoot phase over hours to days, and improved later function. Chronic injury should instead show incomplete recovery and cumulative tissue damage.

Model-Discriminating Measurements

A strong experiment measures variables that are ordered differently by the models. A weak experiment measures only a final common pathway.

Measurement	Model A prediction	Model B prediction	Model C prediction	Model D prediction
Cytosolic calcium in myeloid cells	May change secondarily	Earliest abnormality	Secondary or variable	Transient, dose-dependent perturbation
Cytosolic mtDNA	Early causal signal	Later secondary signal	Later after mitochondrial QC failure	May rise briefly then normalize at adaptive doses
Phospho-STING and phospho-IRF3	Early and mtDNA-linked	Later or parallel	Secondary to mitochondrial decline	Biphasic or transient
NF-κB nuclear translocation	Downstream of danger and cytokine amplification	Early, calcium-linked	Secondary	Transient if adaptation succeeds
Mitophagic flux	Compensatory and potentially insufficient	Secondary to inflammatory stress	May decline after turnover machinery fails	Increased during successful adaptation
CMA and proteasome flux	May remain intact early	May remain intact early	Earliest persistent deficit	Transiently induced at adaptive dose
Long-term tissue histology and function	Improves if leak is truly upstream and safely reduced	Improves if immune-gain correction prevents secondary damage	Requires restoration of slow turnover	Improves only within a hormetic window

Selected Intervention for Maximum Model Discrimination

For this colloquium, the intervention with the greatest expected ability to discriminate the four models is a brief, inducible, cell-type-restricted enhancement of mitophagy in aged microglia, combined with dense early sampling and long-term follow-up.

This is intentionally different from choosing the intervention most likely to be immediately therapeutic. A STING inhibitor acts at a convergence point shared by several models; a reduction in cytokines after STING inhibition therefore provides limited information about what was upstream. Urolithin A has useful in vivo evidence but affects more than one molecular process. An MCU rescue is highly informative for the calcium-gain model but less directly separates mitochondrial leak from proteostasis-first explanations outside macrophages. By contrast, a short microglia-restricted mitophagy pulse directly changes mitochondrial clearance while leaving the initial immune signaling machinery unblocked.

A practical implementation could use a validated inducible microglial driver such as a Tmem119-based system coupled to a genetically encoded mitophagy actuator. The exact actuator must be validated for specificity and for absence of direct STING, NF-κB, or proteasome effects. The intervention should be calibrated as a short pulse rather than constitutive overexpression, because constitutive manipulation would confound acute causality with developmental or compensatory remodeling.

Proposed Sampling Schedule

Use aged mice with young reference animals and include both sexes. Sample the same mechanistic chain at:

1. Baseline: establish cell-state and tissue-state distributions.
2. Thirty minutes to two hours: calcium dynamics, immediate signaling, mitochondrial membrane state.
3. Six hours: cytosolic mtDNA, phospho-STING, phospho-TBK1, phospho-IRF3, NF-κB.
4. Twenty-four hours: cytokines, interferon-stimulated genes, mitophagic flux, early proteostasis responses.
5. Seventy-two hours: persistence or recovery, single-cell state transitions, compensatory autophagy.
6. Two weeks: proteome turnover direction, synaptic and glial remodeling.
7. Eight weeks: tissue damage, behavior, organ function, adverse effects, and durability.

At each point, preserve cell-type identity. Microglia should be separated from infiltrating macrophages, neurons, astrocytes, oligodendrocytes, vascular cells, and peripheral immune populations. Bulk brain measurements alone cannot tell whether an apparent rescue reflects true state change within microglia or a change in cell composition.

Core Readouts

1. Mitophagy: mito-QC or mt-Keima flux with appropriate lysosomal controls.
2. Mitochondrial DNA: cytosolic mtDNA by fractionation plus quantitative PCR or digital PCR, with contamination controls.
3. cGAS–STING signaling: phospho-STING, phospho-TBK1, phospho-IRF3, and interferon-stimulated transcription.
4. NF-κB signaling: nuclear localization or live reporters where feasible.
5. Calcium signaling: cytosolic and mitochondrial calcium imaging in purified or acute-slice myeloid cells.
6. Proteostasis: CMA reporter activity, proteasome activity, insoluble-protein burden, and pulse-chase proteomics.
7. Cell state: single-cell RNA or multiome profiling with explicit microglia, macrophage, neuron, and glial annotations.
8. Tissue consequence: synaptic density, neuronal injury, extracellular-matrix remodeling, functional testing, and longitudinal adverse-event measures.

Quantifying Decision Uncertainty

Model discrimination should be expressed as a change in uncertainty, not merely as a significant P value. One useful teaching framework is Bayesian model comparison.

Suppose the prior probabilities are assigned as a transparent design assumption:

Model	Prior probability
Model A: mitochondrial leak first	0.35
Model B: immune gain first	0.30
Model C: proteostasis first	0.25
Model D: hormetic compensation	0.10

These numbers are not empirical estimates from the cited papers. They are an illustrative prior for deciding which experiment is most informative. The prior Shannon entropy is approximately 1.88 bits.

Now consider one possible result: after a microglia-restricted mitophagy pulse, cytosolic mtDNA falls within six hours, STING and IRF3 signaling falls by six to twenty-four hours, calcium-oscillation abnormalities remain initially unchanged, and proteasome or CMA measures do not change until later. Under an illustrative likelihood model, the posterior might become:

Model	Illustrative posterior probability
Model A	0.72
Model B	0.08
Model C	0.15
Model D	0.05

The posterior entropy would be approximately 1.26 bits, a reduction of about 0.62 bits. That reduction is the information gained under the assumed priors and likelihoods.

For comparison, one could assign illustrative expected information gains of about 0.18 bits to downstream STING inhibition, 0.28 bits to a pleiotropic mitophagy-enhancing drug, 0.45 bits to macrophage MCU rescue, and 0.62 bits to the cell-restricted mitophagy pulse. These are decision-analysis examples, not measured properties of the interventions. Their purpose is to force explicit assumptions that can be sensitivity-tested.

Sensitivity Analysis

You should not report a single expected-information-gain number without checking whether the conclusion survives reasonable changes in priors and assay reliability. Repeat the calculation under:

1. More equal priors across all models.
2. A strong prior for the calcium-gain model.
3. Lower specificity of the mitophagy actuator.
4. Measurement error in cytosolic mtDNA.
5. Cell-composition shifts that mimic reduced inflammatory expression.
6. A delayed rather than immediate relationship between mitophagy and mtDNA release.

If the preferred experiment changes under small prior adjustments, the decision is fragile. If it remains preferred over a broad prior range, the design is more robust.

Alternative Outcomes and What They Mean

Outcome favoring Model A: mitophagy rises first, cytosolic mtDNA falls next, and STING signaling falls before cytokines or tissue phenotypes change. Early calcium abnormalities remain largely unchanged.

Outcome favoring Model B: enhanced mitophagy lowers mitochondrial damage markers, but cytosolic calcium oscillations and NF-κB activation remain high and continue to drive cytokines. A subsequent MCU-restoring perturbation produces the early rescue.

Outcome favoring Model C: acute mitophagy changes have little effect on inflammation, whereas restoration of CMA or proteasomal turnover over days to weeks precedes mitochondrial and inflammatory normalization.

Outcome favoring Model D: low-intensity, short-duration mitophagy activation produces a transient stress response followed by improved resilience, whereas stronger or prolonged activation causes ATP stress, inflammatory rebound, or tissue injury.

A single “rescue” is therefore insufficient. The temporal order of multiple readouts is the discriminating evidence.

Hormesis, Compensation, and Long-Term Damage

Hormesis

Hormesis predicts a non-monotonic relation between stress dose and outcome. A low dose of mitochondrial or proteotoxic stress can induce quality-control pathways and improve later resilience, while a high or persistent dose overwhelms the same systems. To test hormesis rather than merely assert it, you need at least three perturbation intensities, repeated time points, and a delayed challenge or functional outcome.

A convincing hormetic result would show:

1. A transient early stress signal.
2. Induction of protective turnover or stress-response machinery.
3. Recovery below baseline damage or improved resistance to a later challenge.
4. Loss of benefit or emergence of harm at a higher dose or longer duration.

Without the dose-by-time interaction, “hormesis” can become an unfalsifiable explanation.

Compensatory Responses

The 2024 mitophagy study is especially important for this alternative. Increased mitophagy in an aged tissue does not prove that mitochondrial quality is improved. Let D be damage input and C be clearance. If both D and C increase but D increases more, the tissue can show high clearance flux and high residual damage at the same time.

To distinguish compensation from successful maintenance, measure both:

1. Flux: how rapidly damaged material enters and passes through the clearance pathway.
2. Burden: how much damaged material remains.
3. Capacity reserve: whether flux can increase further after a challenge.
4. Functional outcome: whether organelle and tissue performance improve.

This logic also applies to proteasome activity, inflammatory cytokines, antioxidant enzymes, and unfolded-protein responses.

Long-Term Tissue Damage as a Competing Explanation

An acute molecular improvement can coexist with long-term tissue harm. For example, broad suppression of innate immune signaling might lower inflammatory transcripts while impairing host defense or debris clearance. Excessive mitophagy might remove damaged mitochondria but create an energetic deficit if biogenesis cannot replace them. Chronic proteostasis activation could impose metabolic costs or alter normal signaling proteins.

Therefore, every short-term intervention should be paired with delayed measurements of tissue architecture and function. In brain studies, include synaptic integrity, neuronal loss, glial states, cognition or sensorimotor function, and vascular or extracellular-matrix changes. In liver or muscle, use organ-specific functional assays rather than cytokines alone.

Long-term damage can also be the upstream driver. Once extracellular matrix, vascular supply, or innervation is chronically altered, mitochondrial stress and proteostasis defects can become secondary consequences. This possibility is especially important in late-life intervention studies.

Cell-Type Resolution

A systems model that uses only “tissue inflammation” as a variable is underspecified. At minimum, distinguish the following compartments:

Cell type	Mechanism of special interest	High-value measurement	Major confound
Microglia	mtDNA sensing through cGAS–STING and paracrine neurotoxicity	Cytosolic mtDNA, STING pathway, cytokines, single-cell state	Infiltrating macrophages can resemble activated microglia
Peripheral macrophages	Mitochondrial calcium uptake and NF-κB gain	MCU and MICU abundance, calcium oscillations, NF-κB dynamics	Ex vivo culture can reset metabolic state
Neurons	High energetic demand and long-lived proteins	Respiratory reserve, turnover proteomics, synaptic function	Neuronal injury can be secondary to glial inflammation
Senescent fibroblasts	Minority MOMP, mtDNA release and SASP	BAX and BAK activation, cytosolic mtDNA, SASP	Senescence induction method can change the phenotype
Hepatocytes	CMA and metabolic proteostasis	KFERQ reporter, lysosomal competence, turnover	Bulk liver averages mix parenchymal and immune cells
Kupffer cells	Tissue-resident immune proteostasis	CMA reporter and inflammatory state	Sex and age interactions can differ from hepatocytes
Retinal cells	Mitophagy, mtDNA release and inflammatory signaling	mito-QC, cytosolic mtDNA, visual function	Multiple retinal layers have distinct turnover demands

Cell-type resolution should be combined with spatial information when possible. A small population of highly inflammatory cells can drive tissue-level cytokines, while the majority of cells show a different molecular state.

Experimental Design for an Expert Colloquium

A defensible experiment should include randomization, blinded outcome assessment where feasible, sex as a biological variable, young and old reference groups, and explicit exclusion criteria. The unit of replication must match the question: many cells from one animal are not independent animals.

For mechanistic timing, use repeated or staggered cohorts rather than relying on a single terminal time point. For single-cell data, distinguish biological replicates from cells and control for cell-composition changes. For mitochondrial DNA fractionation, include markers that reveal nuclear or mitochondrial contamination. For autophagy and mitophagy, measure flux rather than static abundance whenever possible.

Pre-register the model predictions before seeing the intervention outcome. A particularly useful preregistration table states, for each model, the expected sign and earliest time of change for calcium, cytosolic mtDNA, STING, NF-κB, mitophagy, CMA, proteasome activity, cytokines, and tissue function.

Interpretation Traps

1. Flux versus abundance: more autophagosomes or mitolysosomes can indicate more flux or a downstream block.
2. Cause versus compensation: an age-associated increase can be protective compensation rather than a driver of pathology.
3. Downstream convergence: blocking STING or NF-κB can improve inflammation under several different upstream models.
4. Cell composition: bulk RNA changes can reflect different numbers of microglia, macrophages, neurons, or stromal cells rather than within-cell regulation.
5. Temporal aliasing: a six-week endpoint can miss a six-hour causal sequence and a two-day compensatory rebound.
6. Survivorship: cells most damaged by the intervention may disappear, making the remaining population appear healthier.
7. Sex averaging: combining males and females can erase opposite or unequal CMA and turnover trajectories.
8. Tissue generalization: a macrophage result cannot automatically be applied to neurons, retina, liver, or muscle.
9. Pharmacological pleiotropy: a drug that improves several pathways may be therapeutic but poor for discriminating mechanisms.
10. Acute rescue versus durable benefit: lower cytokines today do not guarantee preserved tissue function months later.

Load video

YouTube

YouTube might collect personal data. Privacy Policy

Continue Dismiss

Primary Literature for the Colloquium

1. Gulen et al. 2023, cGAS–STING drives ageing-related inflammation and neurodegeneration: Microglial mitochondrial DNA sensing and STING inhibition.
2. Jiménez-Loygorri et al. 2024, Mitophagy curtails cytosolic mtDNA-dependent activation of cGAS/STING inflammation during aging: Tissue-resolved mitophagy, cytosolic mtDNA and urolithin A.
3. Seegren et al. 2023, Reduced mitochondrial calcium uptake in macrophages is a major driver of inflammaging: Mitochondrial calcium buffering and NF-κB gain.
4. Rao, Upadhyay and Savas 2024, Derailed protein turnover in the aging mammalian brain: In vivo stable-isotope protein-turnover analysis.
5. Khawaja et al. 2025, Sex-specific and cell-type-specific changes in chaperone-mediated autophagy across tissues during aging: CMA heterogeneity across cell types and sexes.
6. Victorelli et al. 2023, Apoptotic stress causes mtDNA release during senescence and drives the SASP: Minority MOMP, mitochondrial DNA release and senescence-associated inflammation.

Interactive Tasks

Quiz: Test Your Knowledge

Which variable is best suited to resolve the earliest prediction of the immune-gain-first model? (Cytosolic calcium oscillations) (!Whole-tissue fibrosis) (!Long-term memory performance) (!Insoluble proteome abundance)

Which study most directly linked aged microglial mitochondrial DNA sensing to cGAS–STING activation? (Gulen and colleagues) (!Rao and colleagues) (!Khawaja and colleagues) (!Seegren and colleagues)

What did the 2024 mito-QC study show about mitophagy across old mouse tissues? (It was increased in several tissues and unchanged in others) (!It declined uniformly in every tissue) (!It disappeared completely from the retina) (!It was measurable only in skeletal muscle)

Which method was central to the 2024 protein-turnover study? (Stable-isotope labeling and proteomics) (!Single-cell calcium imaging) (!STING inhibition alone) (!Bacterial challenge assays)

What is a central lesson of the 2025 CMA study? (Aging effects on CMA depend on cell type and sex) (!CMA changes identically in every organ) (!CMA is unrelated to lysosomal function) (!Only neurons perform CMA)

Why is a brief cell-restricted mitophagy pulse highly informative for model discrimination? (It perturbs mitochondrial clearance without directly blocking the shared immune endpoint) (!It guarantees a therapeutic benefit) (!It removes the need for time-course measurements) (!It makes cell-type identity irrelevant)

Why does downstream STING inhibition have limited ability to identify the unique upstream cause? (Several upstream models can converge on STING signaling) (!STING is found only in mitochondria) (!STING changes only over decades) (!STING cannot be measured experimentally)

Which pattern is most consistent with hormesis? (A low dose improves later resilience while a higher dose causes harm) (!Every dose produces the same response) (!Only late tissue damage is measured) (!The response is independent of exposure duration)

What can increased mitophagic flux in old tissue mean? (It can be a compensatory response to increased mitochondrial damage) (!It proves that mitochondrial damage is absent) (!It proves that all autophagy is increased) (!It excludes inflammatory signaling)

Which evidence best tests whether an acute molecular rescue is durable? (Long-term tissue structure and functional outcomes) (!A single cytokine measurement) (!One early phosphoprotein value) (!Static mitochondrial abundance alone)

Memory Game

cGAS	Cytosolic DNA sensor that can initiate innate immune signaling
STING	Adaptor that transmits signals from cytosolic DNA sensing
MCU	Pore-forming component of the mitochondrial calcium uniporter
CMA	Selective lysosomal degradation pathway using chaperone-mediated cargo recognition
Mitophagy	Selective turnover of mitochondria through lysosomal degradation
Proteasome	Multisubunit complex that degrades many ubiquitin-tagged proteins
SASP	Secretory inflammatory program associated with many senescent cells

Drag and Drop

Match the correct terms.	Topic
MCU buffering	Early cytosolic calcium dynamics
cGAS activation	Cytosolic mitochondrial DNA
Chaperone-mediated autophagy	KFERQ-bearing protein cargo
Stable-isotope incorporation	Protein turnover measurement
Mitolysosome formation	Mitophagic flux

...

Crossword Puzzle

Mitochondria	Which organelles couple energy production, calcium handling and mitochondrial DNA stress signals?
Inflammation	What process includes sustained cytokine and innate immune activation in aging tissues?
Proteostasis	What term describes maintenance of protein synthesis, folding, trafficking and degradation?
Mitophagy	What selective pathway delivers mitochondria to lysosomal degradation?
Interferon	Which cytokine family is strongly associated with cGAS–STING–IRF3 signaling?
Proteasome	Which multisubunit complex degrades many ubiquitin-tagged proteins?

Cloze Text

Complete the text.

Mitochondrial DNA in the cytosol can activate . The adaptor downstream of cGAS is . In macrophages, reduced mitochondrial calcium uptake can increase cytosolic calcium . Selective lysosomal turnover of mitochondria is called . Selective lysosomal degradation of KFERQ-bearing proteins is associated with . The 2024 brain study measured protein replacement using stable-isotope . An age-related increase in a clearance pathway can reflect . A low-dose benefit with high-dose harm is consistent with . Bayesian model comparison can quantify uncertainty using information measured in . Durable causal interpretation requires long-term tissue and functional .

Open-Ended Tasks

Easy

1. Causal loop map: Draw the four feedback models and mark every positive and negative feedback edge with the measurement that could test it.
2. Timescale matrix: Build a table from seconds to months and place calcium, NF-κB, STING, cytokines, mitophagy, protein turnover and tissue damage on the earliest plausible measurement window.
3. Primary-study abstract critique: Choose one of the six primary studies and identify its perturbation, cell type, causal claim, strongest evidence and one unresolved alternative explanation.
4. Cell-type annotation: Create a one-page visual that contrasts microglia, macrophages, neurons, senescent fibroblasts, hepatocytes and Kupffer cells in this coupled system.

Standard

1. Bayesian entropy calculation: Recalculate the prior and posterior entropy from the worked example and test how the information gain changes when the four priors are made equal.
2. Sampling design: Design a seven-time-point experiment that can distinguish early signaling from slow proteome remodeling while preserving biological replication.
3. Scientific explainer video: Produce a five-minute video explaining why increased mitophagy can coexist with increased mitochondrial damage and inflammation.
4. Replication proposal: Select one key finding from Gulen, Seegren, Jiménez-Loygorri, Rao, Khawaja or Victorelli and propose an independent replication in a different cell type or tissue.

Advanced

1. Pulse perturbation experiment: Design a microglia-restricted inducible mitophagy pulse with specificity controls, flux measurements and prespecified outcome orderings for all four models.
2. Expected information gain: Construct a likelihood matrix for four candidate interventions and calculate which experiment maximizes expected reduction in model uncertainty under multiple priors.
3. Hormesis surface: Design a dose-by-duration experiment that can distinguish beneficial adaptation, simple toxicity and compensatory overdrive, including delayed challenge testing.
4. Long-term tissue study: Build a study that follows acute molecular rescue through eight weeks or longer and tests whether reduced inflammation predicts preserved tissue architecture and function.

Learning Assessment

1. Causal model defense: Defend one feedback model using at least three primary studies, then specify the observation that would most strongly falsify your preferred model.
2. Temporal inference: Given a dataset in which calcium changes at two minutes, NF-κB at one hour, cytosolic mtDNA at six hours and STING at twelve hours, explain which models gain or lose support and why.
3. Cell-type transfer: Explain why a macrophage MCU result cannot automatically be generalized to neurons, and propose the minimum measurements needed for a valid cross-cell-type test.
4. Compensation analysis: Interpret a tissue with increased mitophagic flux, increased damaged-mitochondria burden and reduced respiratory reserve without assuming that any single marker is beneficial or harmful.
5. Intervention comparison: Compare STING inhibition, MCU restoration, CMA enhancement, urolithin A and cell-restricted mitophagy activation for mechanistic specificity, expected timing and model-discrimination value.
6. Uncertainty audit: Recalculate the model ranking under at least three plausible prior distributions and report whether the preferred experiment is robust or prior-sensitive.
7. Long-term consequence: Propose evidence that would distinguish a transient anti-inflammatory effect from prevention of cumulative tissue damage.

Evidence of Learning

Evidence of learning should include both mechanistic knowledge and the ability to reason under uncertainty.

Knowledge evidence

1. You can explain how mitochondrial calcium handling, mitochondrial DNA release, cGAS–STING, NF-κB, mitophagy, CMA and proteasomal turnover can interact.
2. You can identify which conclusions are supported by each primary study and which remain model-dependent.
3. You can distinguish cell-type-specific findings from tissue-wide generalizations.

Skills evidence

1. You can build competing causal feedback models rather than a single narrative.
2. You can align measurements with seconds, hours, days, weeks and months.
3. You can design perturbations that target upstream variables while preserving downstream readouts.
4. You can quantify prior and posterior uncertainty and perform sensitivity analysis.
5. You can identify hormesis, compensation, survivorship and cell-composition change as alternative explanations.

Product evidence

1. A causal-loop diagram with testable edges.
2. A cell-type-resolved measurement plan.
3. A Bayesian decision table with explicit priors and likelihood assumptions.
4. A pre-registered prediction matrix for at least four models.
5. A long-term follow-up plan linking molecular changes to tissue function.

Transfer evidence

1. You can apply the same framework to muscle aging, immune aging, neurodegeneration, retinal aging, or liver aging without assuming identical mechanisms.
2. You can distinguish a therapy that improves a shared endpoint from an experiment that discriminates causal models.

OERs on the Topic

Useful related open resources include Mitochondrion, Inflammaging, Autophagy, Proteostasis, Cellular senescence, cGAS–STING pathway, Nuclear factor kappa B, and Systems biology.

Linked Learning Areas

Aging – Energy, inflammation and proteostasis as coupled systems Aging Mitochondria Inflammaging Proteostasis Autophagy Cellular senescence Mitochondrial DNA Innate immunity Systems biology Bayesian inference