English:Aging – Transportability from axolotl to humans
aiMOOCAging – Transportability from axolotl to humans
QR-Code
Introduction
How far can a result from an axolotl, a mouse, or an African turquoise killifish legitimately inform a claim about human aging? This expert colloquium treats that question as a problem of causal transportability, not as a search for superficial similarity. You will work with one case-study target, mTOR complex 1 (mTORC1), because it is evolutionarily conserved, experimentally perturbable, strongly connected to nutrient sensing and aging, and highly context-dependent during growth, repair, immunity, and chronic tissue maintenance.
The central discipline of this course is to separate four questions that are often collapsed into one: orthology, cell state, injury response, and chronic aging. A conserved gene name is not a conserved causal effect. A conserved expression change is not a conserved intervention response. A regenerative blastema is not an aged steady-state tissue. A six-month-old fish and a 70-year-old human are not aligned merely because both are called "old".
The case target is the mTORC1 axis, operationalized by the intervention mTORC1 inhibition and by proximal readouts such as phosphorylation of S6 and 4E-BP1. The long-term outcome is not assumed to be the same in every species. In axolotl limb regeneration, recent work shows that mTOR signaling supports systemic activation and blastema proliferation after amputation, and rapamycin slows regeneration. In genetically heterogeneous mice, late-life rapamycin has repeatedly extended lifespan. In humans, mTOR-pathway inhibitors have altered immune readouts in older adults, but a phase 3 trial of RTB101 did not reduce the prespecified rate of clinically symptomatic respiratory illness even though interferon-induced antiviral genes were upregulated. These contrasts are exactly why a transportability audit must distinguish a proximal pathway response from a functional outcome.
Axolotl amputation and mTOR signaling Rapamycin and lifespan in heterogeneous mice mTOR inhibition trials in older adults
Learning Goals
By the end of the colloquium, you should be able to:
- Construct a causal transportability diagram: Represent where species, age, temperature, life history, cell state, injury context, and measurement batch can change the data-generating process.
- Audit orthology: Distinguish a conserved protein from conserved regulatory wiring, dosage, paralog usage, or drug pharmacology.
- Align cell states: Compare homologous cell states rather than whole tissues or nominal cell-type labels.
- Separate acute injury from chronic aging: Treat regenerative activation and age-associated steady-state decline as different biological regimes.
- Define falsification criteria: State in advance what result would reject transport even if expression is conserved.
- Design minimal human validation: Specify the smallest human experiment that tests target engagement and causal functional direction.
Four Species, Four Biological Clocks
Axolotl, Ambystoma mexicanum. Axolotls are neotenic salamanders with exceptional regenerative capacity. They continue to grow after sexual maturity and can retain limb-regenerative ability well into adulthood, although regeneration can slow and lose fidelity with age. Husbandry temperature is itself a biological variable: the Ambystoma Genetic Stock Center keeps axolotls at approximately 15–18 °C and warns against sustained high temperature. Therefore, "age" in an axolotl experiment cannot be interpreted independently of body size, skeletal maturity, growth rate, temperature, and regenerative history.
Axolotl regeneration and aging review Axolotl husbandry and temperature
African turquoise killifish, Nothobranchius furzeri. This annual fish compresses vertebrate aging into months. Its natural history includes ephemeral ponds, rapid maturation, strong strain differences in lifespan, and embryonic diapause. Temperature is not a nuisance variable: lowering water temperature from 25 °C to 22 °C has been shown to lengthen lifespan and delay locomotor, cognitive, and histological aging phenotypes. A transport claim that ignores thermal history can therefore confuse temperature-dependent aging rate with target-dependent aging biology.
Temperature and killifish aging Killifish as a scalable vertebrate aging model
.png)
Mouse, Mus musculus. The mouse supplies mammalian physiology, controlled genetics, and a large intervention literature. Yet strain, sex, microbiome, housing, ambient temperature, diet, and facility can modify aging phenotypes. Standard mouse housing is commonly below thermoneutrality, so thermal physiology can become a translational variable rather than mere husbandry detail. The National Institute on Aging Interventions Testing Program reduces some design risk by using genetically heterogeneous mice and parallel sites. Mouse evidence is still not a substitute for a human causal test.
Mouse housing temperature and translational physiology
Human, Homo sapiens. Human aging is long, heterogeneous, exposure-rich, and ethically constrained. Chronological age is only one coordinate. Comorbidity, medication, sex, ancestry, physical activity, immune history, tissue source, and donor-specific biology can all alter mTORC1 activity and the response to inhibition.
The Target: mTORC1 as a Stress Test for Transportability
mTORC1 integrates amino acids, growth factors, energy status, oxygen, and cellular stress. Its downstream effects include protein synthesis, autophagy control, metabolism, and growth. Because those functions are useful during development and repair but can contribute to chronic age-associated dysregulation, the sign of the desirable intervention effect can depend on context.
The experimental target should therefore be defined at three levels:
- Molecular target: MTOR kinase within the mTORC1 complex, with RPTOR and MLST8 as core complex components.
- Proximal mechanism: Reduced phosphorylation of S6 and 4E-BP1 after a calibrated mTORC1 perturbation.
- Functional consequence: Context-specific outcomes such as blastema proliferation after injury, proteostasis or resilience during chronic aging, tissue function, and clinically meaningful human endpoints.
A study that confirms only the first two levels has not yet shown that the third level transports.
Transportability Diagram
A transportability diagram is a causal diagram augmented with selection variables that mark mechanisms or distributions that differ between a source system and the target system. Here, the target population is human tissue or humans; axolotl, killifish, and mouse are source systems. The diagram below is intentionally explicit about where transport can fail.
S_species
┌──────────┼───────────────┐
v v v
Orthology Cell-state map Life history
| ^ |
| | v
| Biological age <---+
| ^ ^
| | |
| Temperature Sex/strain
| |
v v
Perturbation --> mTORC1 activity --> Cell-state response --> Functional outcome
| ^ ^ ^
| | | |
+---------- Injury context ------------+--------------------+
+---------- Chronic-aging context -----+--------------------+
S_batch --> measured expression
S_batch --> measured phosphoprotein
S_batch --> inferred cell state
Human target: test whether the perturbation-to-outcome edge remains valid
after conditioning or matching on the variables that differ across species.
The diagram contains four separable transport layers.
Layer 1: Orthology
Question: Is the molecular object truly comparable? Begin with sequence and synteny, but do not stop there. For MTOR itself, deep conservation supports a strong molecular starting point. However, a pathway is more than one conserved kinase. Upstream receptors, paralogs, feedback loops, regulatory motifs, expression dosage, complex stoichiometry, and drug metabolism can differ across lineages.
An orthology audit should require: reciprocal sequence support or a curated orthology database; conserved protein domains; syntenic support where available; explicit handling of teleost gene duplication; confirmation that the assayed antibody epitope or mass-spectrometry peptide is conserved; and a pharmacology check demonstrating comparable target engagement rather than assuming equal drug concentration produces equal inhibition.
Transport rule: A one-to-one ortholog licenses comparison of the molecular target, not the causal effect of perturbing it.
Layer 2: Cell State
Question: Are you comparing homologous biological states? A fibroblast in an uninjured aged human dermis, an axolotl blastema progenitor, a killifish fin cell after injury, and a mouse wound fibroblast may all express related markers while occupying different dynamical states.
.svg)
Use single-cell or single-nucleus data to align states by shared gene modules, transcription-factor activity, chromatin accessibility, cell-cycle state, metabolic program, and spatial context. Do not force alignment solely by labels such as "fibroblast" or "macrophage". A robust analysis should first identify within-species states and then perform cross-species integration on verified ortholog sets. It should preserve species-specific states instead of algorithmically mixing them away.
Transport rule: Compare the same causal role in a tissue process, not merely the nearest transcriptomic cluster.
Layer 3: Injury Response
Question: Is the experiment measuring repair or aging? Axolotl limb regeneration is a strongly induced program involving wound epithelium, nerves, immune cells, blastema formation, and proliferation. mTOR activity can be necessary for this acute response. Therefore, an mTORC1 inhibitor can impair regeneration while still having beneficial effects in a different context such as chronic nutrient-sensing dysregulation.
The axolotl literature on senescent cells offers a second warning. Senescent cells can appear transiently during salamander regeneration and are efficiently cleared; their presence in injury is not equivalent to chronic senescent-cell accumulation in mammalian aging.
Recurrent turnover of senescent cells during salamander regeneration
Transport rule: Never transport an acute injury effect directly into a chronic-aging claim without a separate uninjured aging arm.
Layer 4: Chronic Aging
Question: Does the target alter the trajectory of age-associated decline in the absence of acute injury? Chronic aging experiments should use uninjured or steady-state tissue and measure longitudinal or age-stratified changes in functional resilience, proteostasis, inflammatory signaling, stem-cell function, and tissue performance.
The key comparison is not "young versus old in every species" but matched biological states across the life course. Chronological age is an input, not the alignment variable.
Audit 1: Age Alignment
A defensible age-alignment strategy uses multiple anchors instead of a single age conversion.
| Alignment axis | Axolotl | Killifish | Mouse | Human |
|---|---|---|---|---|
| Chronological position | Months or years, reported exactly | Weeks post-hatching, with strain | Months, with strain and sex | Years |
| Life-history stage | Juvenile, sexually mature adult, older adult | Post-maturation, midlife, late life | Young adult, middle-aged, old | Young adult, middle-aged, older adult |
| Functional age | Regeneration speed, body size, locomotion | Swimming, cognition, frailty-like measures | Grip, activity, frailty, organ function | Physical function, organ-specific measures |
| Molecular age | Tissue-specific transcriptomic or epigenetic state when available | Transcriptomic state and aging markers | Multi-omic or epigenetic measures | Tissue-specific omics and validated biomarkers |
Do not use percentage of maximum lifespan as the only normalization. Maximum lifespan is affected by husbandry, strain, environment, and censoring. Prefer a latent biological-age model that combines developmental stage, functional decline, and molecular state. Pre-register which anchors are primary and which are sensitivity analyses.
Audit 2: Temperature
Temperature has different causal meanings across the four systems.
Axolotl and killifish are ectotherms. Water temperature directly alters metabolic rate, development, repair, and aging. Record continuous tank temperature, not just the room set point. Report thermal variance, acclimation period, feeding regime, density, dissolved oxygen, and photoperiod.
Mouse and human are endotherms. Core temperature is buffered, but ambient temperature can still alter metabolism and stress physiology in mice, while fever, circadian rhythm, and clinical status matter in humans. Therefore, "temperature matched" does not mean setting all organisms to the same Celsius value. It means modeling the species-specific pathway by which thermal environment changes physiology.
Audit decision: If species is confounded with temperature regime, a species difference cannot be interpreted as genetic or mechanistic without an explicit temperature sensitivity analysis.
Audit 3: Life History
Life history changes what aging biology has been selected to do.
Axolotl: neoteny, continued growth, and extreme regenerative capacity. Killifish: annual life cycle, rapid maturation, ephemeral habitat, strain-specific longevity, and diapause. Mouse: fast reproductive mammal with laboratory-adapted strains and large genetic effects on lifespan. Human: long post-reproductive lifespan, highly heterogeneous exposures, medicine, social environment, and chronic disease burden.
The transportability diagram should therefore place life history upstream of biological age, tissue state, and functional outcome. Life history is not a descriptive footnote; it is a possible effect modifier.
Audit 4: Batch and Measurement Confounding
Cross-species omics are especially vulnerable to a design in which species equals batch. If all axolotl samples are processed on Monday, all mouse samples on Tuesday, and all human samples in a separate laboratory, technical variation is perfectly aliased with species.
A minimum batch-control plan should include randomized sample order within species and age; multiple biological replicates in every processing batch; shared reference samples or spike-ins where technically possible; matched library chemistry and sequencing depth; identical or explicitly modeled tissue dissociation windows; blinded sample identifiers; and pseudobulk analyses that treat the animal or human donor, not the cell, as the biological replicate.
Do not use batch correction to erase real biology. A cross-species integration method should be tested with negative controls: known species-specific genes or states should remain distinguishable when they ought to.
Four-Species Experimental Matrix
| Species | Context A: injury response | Context B: chronic aging | mTORC1 perturbation | Primary proximal readout | Functional readout | Main transport risk |
|---|---|---|---|---|---|---|
| Axolotl | Limb amputation and blastema | Uninjured limb or skin across biological age | Rapamycin plus an orthogonal genetic perturbation where feasible | p-S6 and p-4E-BP1 | Regeneration speed and tissue restoration | Regeneration is not chronic aging |
| Killifish | Standardized fin or tissue injury where justified | Longitudinal uninjured aging in a defined strain | Calibrated pharmacology or genetic reduction of mTORC1 signaling | p-S6 and pathway transcriptional modules | Swimming, tissue function, survival or resilience | Temperature and strain strongly shift aging rate |
| Mouse | Standardized wound or regeneration model | Genetically heterogeneous or clearly defined aging cohort | Rapamycin and target-specific validation | p-S6 and p-4E-BP1 | Tissue function, frailty, survival where appropriate | Sex, site, microbiome, ambient temperature |
| Human | Ex vivo wound model or clinically justified acute-response assay | Uninjured primary tissue from young and older donors | Ex vivo rapamycin plus RPTOR CRISPR interference or another orthogonal target perturbation | p-S6 and p-4E-BP1 | Wound closure, proteostasis, matrix or immune function | Donor heterogeneity and disease or medication confounding |
Falsification: What Would Prove That Transport Fails?
A strong translational program states its failure criterion before seeing the data.
Falsifying result: MTOR and core mTORC1 components show conserved orthology and similar age-associated expression across axolotl, killifish, mouse, and human samples. A calibrated intervention also suppresses p-S6 to the same target-engagement range in all four systems. However, after matching the homologous cell state, biological-age stratum, and injury-versus-chronic context, the functional causal effect in human tissue has the opposite sign or is absent. For example, chronic mTORC1 inhibition improves a pre-registered resilience endpoint in aged killifish and mouse tissue but reproducibly worsens or does not improve that same homologous endpoint in older human tissue, despite confirmed on-target p-S6 suppression by both pharmacological and genetic perturbations.
That result falsifies the transport claim. Conserved expression would not rescue it, because transportability concerns the perturbation-to-outcome mechanism, not only the distribution of a marker.
A second falsifier is context inversion. If mTORC1 inhibition slows axolotl regeneration but improves a chronic-aging phenotype, then an axolotl injury result cannot be transported as a generic anti-aging effect. The correct conclusion is that the edge is context-specific.
The 2021 human RTB101 trials illustrate the same logic at another level: interferon-induced antiviral genes were upregulated, yet the phase 3 trial did not reduce clinically symptomatic respiratory illness. A proximal molecular response can therefore be real while the higher-level functional endpoint fails to follow.
Minimal Human Validation Package
The minimal package should be small enough to run early but strong enough to test the causal edge.
- Human material: Use primary human cells or ex vivo tissue from pre-specified younger and older donor strata, balanced as far as feasible for sex and major clinical covariates; keep donor as the unit of replication.
- Two perturbations: Use a calibrated mTORC1 inhibitor and an orthogonal target-specific perturbation such as RPTOR CRISPR interference so that a drug-specific off-target effect cannot masquerade as transport failure.
- Proximal confirmation: Require a pre-specified reduction in p-S6 and p-4E-BP1 before interpreting downstream biology.
- State confirmation: Verify that the assayed human cells map to the intended homologous state and are not merely the same nominal cell type.
- One functional endpoint per context: In an injury arm, use a repair endpoint such as wound closure or regenerative proliferation; in a chronic-aging arm, use a steady-state resilience endpoint such as proteostasis, matrix homeostasis, or immune function.
- Pre-registered falsifier: Reject transport if both human perturbations achieve target engagement but fail to reproduce the pre-specified direction of the functional effect under the matched context.
A practical minimal design is a paired donor experiment: each donor contributes control and perturbed samples, reducing inter-individual noise. The analysis should estimate donor-level treatment effects and test whether age modifies the effect. Single-cell measurements are descriptive unless the donor remains the replication unit.
Decision Gate for Translation
A transport claim advances only if all four gates pass.
| Gate | Pass condition | Failure meaning |
|---|---|---|
| Orthology | Molecular target and assay epitope are verified | You may be measuring a non-equivalent molecule |
| State | Homologous cell state is demonstrated | Same label does not mean same biology |
| Context | Injury and chronic aging are analyzed separately | Acute repair may dominate the effect |
| Human causal check | On-target perturbation reproduces the functional direction | Transport is rejected or narrowed |
The scientifically useful outcome is not always "transport succeeds". A narrow failure can reveal where evolution changed pathway wiring, where a regenerative program is unique, or where a biomarker is decoupled from function.
Interactive Tasks
Quiz: Test Your Knowledge
What is the primary reason conserved MTOR expression is insufficient to establish transportability? (Causal response to perturbation can differ despite similar expression) (!Expression always determines protein activity) (!Orthologous genes cannot be compared across vertebrates) (!Human cells do not express MTOR)
Which comparison best separates injury response from chronic aging? (An injured arm and an uninjured age-matched arm) (!Two injury time points only) (!Two doses in the same wounded tissue) (!Young injured tissue in one species only)
Why must killifish water temperature be included in an aging audit? (It can change aging rate and lifespan) (!It changes the species genome sequence) (!It eliminates all strain effects) (!It makes fish endothermic)
What does a selection variable represent in a transportability diagram? (A mechanism or distribution that differs between source and target) (!A gene selected by natural selection only) (!A sequencing read filtered from analysis) (!A randomization seed)
What is the biological replicate in a human single-cell validation study? (The human donor) (!Each sequenced cell) (!Each transcript) (!Each UMAP coordinate)
Which readout most directly demonstrates mTORC1 target engagement? (Reduced phosphorylation of S6) (!Increased body size) (!Longer sequencing reads) (!Higher water temperature)
What is the main risk when every species is processed in a different sequencing batch? (Species becomes confounded with batch) (!Orthology becomes impossible) (!All cells become identical) (!Chronological age is randomized)
Which observation most strongly falsifies transport? (Opposite human functional effect despite matched state and target engagement) (!Conserved MTOR sequence) (!Similar pathway expression across species) (!Comparable sequencing depth)
Why is percentage of maximum lifespan insufficient as the only age alignment? (Maximum lifespan depends on strain and environment) (!It cannot be calculated in animals) (!It is identical for all vertebrates) (!It measures only temperature)
What is the best interpretation if rapamycin slows axolotl regeneration but improves a chronic aging endpoint elsewhere? (The causal effect is context dependent) (!The axolotl result is invalid) (!The human result must be false) (!Orthology has been disproven)
Memory Game
| Orthology | Evolutionary correspondence of genes or proteins across species |
| Blastema | Regeneration-associated proliferative cell population after limb injury |
| Transportability | Valid transfer of a causal result from a source system to a target system |
| Pseudobulk | Aggregation of single-cell measurements at the biological-replicate level |
| Diapause | Developmental dormancy used by annual killifish embryos |
| Falsifier | Pre-specified observation that rejects a transport claim |
Drag and Drop
| Match the correct terms. | Topic |
|---|---|
| Sequence and synteny audit | Orthology layer |
| Homologous transcriptomic state | Cell-state layer |
| Amputation and wound response | Injury context |
| Uninjured age-associated decline | Chronic aging context |
| Randomized processing across groups | Batch-control strategy |
...
Crossword Puzzle
| Orthology | What term describes evolutionary correspondence between genes in different species? |
| Blastema | What regeneration structure forms after axolotl limb amputation? |
| Rapamycin | Which classic drug inhibits mTORC1 signaling? |
| Diapause | What dormant developmental state occurs in annual killifish embryos? |
| Pseudobulk | What analysis aggregates single-cell counts by biological replicate? |
| Falsifier | What pre-specified result rejects a transport claim? |
LearningApps
Cloze Text
Open-Ended Tasks
Easy
- Transportability vocabulary: Create a one-page glossary that distinguishes orthology, cell state, injury response, chronic aging, target engagement, functional endpoint, and transportability.
- Life-history comparison: Draw a four-column life-history map for axolotl, killifish, mouse, and human and mark where simple chronological age matching becomes misleading.
- Temperature audit: Build a checklist showing how water temperature, ambient temperature, core temperature, acclimation, and thermal history should be recorded across the four models.
- Evidence map: Produce a figure that separates evidence for mTORC1 expression, pathway activity, perturbation, and functional outcomes in each species.
Standard
- Selection diagram: Redraw the course transportability diagram and justify every arrow from species, age, temperature, life history, injury, and batch to the measured outcomes.
- Cell-state alignment: Design a cross-species single-cell workflow that begins with within-species clustering and then aligns homologous states using verified orthologs and pathway modules.
- Batch-confounding simulation: Create a small simulated dataset in which species is perfectly confounded with batch, then redesign the experiment so that batch and species can be separated.
- Falsification protocol: Write a pre-registration paragraph stating one result that supports transport, one that narrows transport to a context, and one that falsifies it.
Advanced
- Orthology stress test: Audit MTOR, RPTOR, and two upstream pathway genes across all four species, documenting paralogs, domain conservation, assay compatibility, and uncertainties.
- Cross-species perturbation study: Design a matched-dose experiment that calibrates mTORC1 inhibition by p-S6 suppression rather than nominal drug concentration and compares both injury and chronic-aging contexts.
- Human validation package: Draft a minimal ex vivo human study with donor stratification, paired perturbations, target-engagement thresholds, one functional endpoint, and a pre-specified rejection rule.
- Transport formula critique: Translate the biological selection diagram into a formal causal query and identify which variables would need measurement in the human target population for transport to be defensible.
Learning Assessment
- Causal graph defense: Present your transportability diagram orally and defend which selection nodes can be adjusted for, which require direct human measurement, and which make transport non-identifiable.
- Cross-species protocol review: Critique a hypothetical study that compares old axolotl, killifish, mouse, and human bulk RNA-seq without temperature control or cell-state alignment, and propose a repair plan.
- Sign-reversal analysis: Explain how you would interpret conserved MTOR expression, conserved p-S6 suppression, and an opposite human functional effect after mTORC1 inhibition.
- Age-alignment synthesis: Build a multivariate age-alignment model and explain why life-history stage and functional age may be more informative than percentage of lifespan.
- Human evidence threshold: Defend whether a molecular human biomarker response is sufficient for translation when a clinical or tissue-level functional endpoint does not change.
- Reproducibility audit: Specify how randomization, blinding, biological replication, balanced batches, orthogonal perturbations, and negative controls reduce false transport claims.
Evidence of Learning
Evidence of learning includes a technically correct selection diagram; a species-by-species orthology table; a cell-state alignment plan; an age and temperature audit; a life-history confounder analysis; a balanced batch design; a pre-registered transport falsifier; a minimal human validation protocol; and the ability to explain why proximal pathway conservation does not guarantee conservation of a functional causal effect.
A strong portfolio also demonstrates transfer: you should be able to replace mTORC1 with another aging target and repeat the same logic without reusing species labels as a shortcut for mechanism. The key skill is to identify which causal edges are assumed to remain stable and which must be tested in humans.
OERs on the Topic
Useful open resources include:
- Axolotl: Regeneration biology, neoteny, and experimental model context.
- Nothobranchius furzeri: Annual life history and aging model.
- Mouse: Mammalian model-organism context.
- Aging: Concepts, mechanisms, and hallmarks.
- Mechanistic target of rapamycin: mTOR complexes, regulation, and downstream signaling.
- Single-cell RNA sequencing: Cell-state measurement and integration.
- Causal inference: Directed acyclic graphs and transportability concepts.
Selected Scientific Sources
- Adrenergic signaling coordinates distant and local responses to amputation in axolotl: Includes evidence that mTOR signaling supports axolotl systemic activation and regeneration after amputation.
- Recurrent turnover of senescent cells during regeneration of a complex structure: Demonstrates transient senescence and clearance during salamander regeneration.
- Temperature affects longevity and age-related locomotor and cognitive decay in Nothobranchius furzeri: Establishes temperature as a causal modifier of killifish aging rate.
- The African Turquoise Killifish: A Scalable Vertebrate Model for Aging and Other Complex Phenotypes: Reviews killifish life history and experimental strengths.
- Rapamycin fed late in life extends lifespan in genetically heterogeneous mice: Reports lifespan extension in genetically heterogeneous mice.
- Targeting the biology of ageing with mTOR inhibitors to improve immune function in older adults: Shows why proximal molecular responses and clinical outcomes must be distinguished.
- Transporting results in an observational epidemiology setting: Introduces selection diagrams and transportability logic applicable to source and target populations.
- Multi-species atlas resolves an axolotl limb development and regeneration paradox: Demonstrates the value of multi-species single-cell atlases for identifying homologous cell states.
Linked Learning Areas