Open Research Questions: Why Our Roof Rats Matter to Science

[G1] Do blonde (Rab38-null) roof rats show the bleeding tendency the Rab38 gene predicts?

Where it stands: The gene is settled. Rab38 is the rat Ruby locus, the same gene mutated in chocolate mice and fawn-hooded rats, and it builds the platelet dense granules that help blood clot. Our blonde is PCR-confirmed Rab38-null, which makes it a cleaner Rab38 model than the fawn-hooded rat. What is open is whether our blonde roof rat actually shows the bleeding phenotype, which no one has measured in Rattus rattus. This is the genetic root of the blonde bleeding caution we put on the supplement pages.

The test: Compare bleeding time and platelet dense-granule counts in blonde (Rab38-null) roof rats and their pigmented siblings raised on the same diet.

What you would see if it holds: Blonde rats bleed a little longer and carry fewer or emptier platelet dense granules, matching the Rab38 (Ruby) storage-pool defect, which would turn the blonde bleeding caution from a precaution into measured evidence.

What would falsify it: Equal bleeding time and normal dense granules, which would mean the blonde coat does not carry the Rab38 clotting corollary in this species.

Sources: Oiso N et al. 2004, Mammalian Genome 15:307-314^[1]; Ninkovic I et al. 2008, Journal of Thrombosis and Haemostasis 6:2143-2151^[2]; Ambrosio AL, Boyle JA, Di Pietro SM 2012, Blood 120:4072-4081^[3].

[G2] Is the colony’s black coat the constitutively active MC1R allele, and does PCR confirm E94K in our black rats?

Our rats are uniquely suited to answer this.

Where it stands: Black in wild R. rattus is the Mc1r E94K allele, and the homologous mouse E92K receptor is constitutively active. Our blonde is PCR-confirmed for Rab38, but our black is not yet sequenced for E94K, so confirming it is the first thing to nail down.

The test: Sequence Mc1r in the colony’s black rats and check for the E94K (G280A) substitution that Kambe found in wild black R. rattus.

What you would see if it holds: Our black rats carry E94K, confirming a natural gain-of-function (always-on) MC1R, the genetic mirror of the loss-of-function redhead that the pigment-genetics field studies almost entirely from the loss-of-function end.

What would falsify it: No E94K, which would mean a different gene drives black in this line. Melanism is not always MC1R; in the Okinawan Rattus tanezumi it is not.

Sources: Kambe Y et al. 2011, Zoological Science 28:560-567^[4]; Benned-Jensen T, Mokrosinski J, Rosenkilde MM 2011, PLoS ONE 6:e24644^[5]; Kambe Y et al. 2012, Genes & Genetic Systems 87:29-38^[6]; Sasamori S, Yoshida MA, Suzuki H, et al. 2017, Zoological Science 34:513-522^[7].

[G3] Does the always-on MC1R in black roof rats put them at the anti-inflammatory, Nrf2-protected end of the pigment-immunity axis?

Where it stands: MC1R activation is anti-inflammatory and drives Nrf2 neuroprotection, and loss-of-function redheads are more vulnerable (they lose more dopaminergic neurons). A constitutively active MC1R mammal would be a natural always-on version, which has not been studied. Two related angles need a careful hand rather than a dismissal: the redhead pain and analgesia phenotype is a loss-of-function effect, so it points the opposite way from our gain-of-function black and we would not expect it here; and the darker-equals-bolder behaviour idea is plausible but hard to pin on MC1R itself, because the colour-and-behaviour link in animals usually runs through shared melanocortin ligands rather than the receptor, so the observation stands as real but unexplained.

The test: Challenge black (gain-of-function MC1R) and agouti roof rats with a standard inflammatory or oxidative stressor and compare inflammatory markers, Nrf2-pathway activity, and neuron survival.

What you would see if it holds: Black rats show a blunted inflammatory response and stronger Nrf2 antioxidant defense, the gain-of-function mirror of the redhead, who sits at the vulnerable loss-of-function end.

What would falsify it: No difference, or black rats fare worse, which would mean the coat allele does not carry the predicted protective tone in a live animal.

Sources: Chen X, Chen H, Cai W, et al. 2017, Annals of Neurology 81:395-406^[8]; Mogil JS, Wilson SG, Chesler EJ, et al. 2003, Proceedings of the National Academy of Sciences 100:4867-4872^[9]; Ducrest AL, Keller L, Roulin A 2008, Trends in Ecology & Evolution 23:502-510^[10].

[G4] Is the coat-defined allelic series (agouti, blonde, black, and the melanistic-blonde double) a unique MC1R-by-Rab38 testbed found nowhere else?

Where it stands: The colony carries three named pigment-genetic nodes plus the double in one tame, breedable line. That allelic series, with a live MC1R-by-Rab38 interaction, is a rare research and teaching resource. Brian’s firsthand observation that melanistic blondes run darker fits the rescue idea; the eye-protection consequence is a further untested hypothesis worth naming.

The test: Breed the crosses and score pigmentation and eye phenotype across agouti (wild-type), blonde (Rab38-null), black (MC1R-gain), and the melanistic Rab38-null double.

What you would see if it holds: The melanistic Rab38-null rats are darker than the non-melanistic Rab38-null rats, showing MC1R gain partly rescuing the Rab38 melanosome-trafficking deficit (more eumelanin despite the dilution defect), with a possible eye-protection benefit for the light-sensitive blonde.

What would falsify it: Melanism does not darken the Rab38-null coat, which would mean the two genes act independently here and there is no epistasis to study.

Sources: Kambe Y et al. 2011, Zoological Science 28:560-567^[4]; Oiso N et al. 2004, Mammalian Genome 15:307-314^[1].

[G5] Does the colony’s white tail tip have a heritable pigment basis, and does it travel with belly lightness or tameness?

Where it stands: This is the genuinely open one. No candidate gene for a white tail tip exists anywhere in the pet-rat or ship-rat literature, the marking descends from Roofy’s line, and Brian sees a weak link to tameness and maybe slightly lower fertility. Mechanism unknown.

The test: Track the white-tail-tip marking across litters, score co-segregation with belly lightness and with how readily each animal accepts handling, and look for a candidate pigment or neural-crest locus.

What you would see if it holds: The marking is heritable and co-segregates with belly lightness and tameness, which would fit the domestication-syndrome prediction that selecting for tameness drags white markings along.

What would falsify it: The marking is random across litters with no link to belly lightness or temperament, which would make it a cosmetic one-off rather than a window into domestication.

Sources: Trut L 1999, American Scientist 87:160^[11].

[G6] Why did the melanistic (MC1R E94K) coat become common in commensal black rats: relaxed predation plus founder drift indoors, soot camouflage, plague tolerance, or a commensal-fitness syndrome (bolder, lower-stress around people, thriftier metabolism) carried with the black allele?

Our rats are uniquely suited to answer this.

Where it stands: The genetics are largely settled (pending peer review for the European samples): black is the dominant MC1R E94K gain-of-function allele, of a single recent origin, that hitchhiked on the commensal lineage out of Asia, and it is an indoor-commensal trait, not a European one (agouti dominates the wild native range and feral island populations, while the black morph tracks buildings and attics). The open part is the SELECTION. In the wild, dark coats are purged by predators, which is why natural populations stay agouti; indoors that penalty is relaxed, soot-blackened attics may even make a dark rat cryptic (the pre-industrial peppered-moth idea), and founder effects and drift in isolated city and ship populations push the allele up. A preliminary genomic analysis reportedly finds no strong selection signature on the coat-colour gene itself, consistent with relaxed selection plus drift rather than a hard sweep. On top of that, the melanocortin system MC1R belongs to is pleiotropic, linked across species to boldness, stress resistance, and energy balance, so if the black allele also carries (or co-segregates with) a commensal-fitness advantage, calmer and bolder around people and thriftier on patchy human food, that would favour black in the human niche even with predators gone, and a behaviour-linked advantage need not leave a signature on the pigment gene. Whether such an advantage is caused by the MC1R change or merely linked to it is open. The plague could add an episodic reinforcing push where rats were the vector (itself debated for medieval northern Europe). Brian notes black is the dominant coat in our own colony.

The test: In our colony, where black is dominant and we already select for calmness around people, raise black and agouti littermates together in a common predator-free indoor environment and compare boldness and handling tolerance, stress physiology (cortisol, adrenal size), and weight gain and feed efficiency on a standard ration; PCR-confirm the E94K allele and test whether it tracks any behavioural or metabolic difference. Historically, map melanistic frequency against plague geography and indoor-soot exposure, and replicate the genomic selection test in peer review.

What you would see if it holds: Black littermates are consistently bolder, lower-stress, or thriftier than agouti littermates in a predator-free indoor setting, showing a commensal-fitness advantage of the allele (or of variation linked to it) that selects for the dark coat independent of camouflage and plague.

What would falsify it: Black and agouti littermates are indistinguishable in temperament, stress, and growth, leaving relaxed predation, soot camouflage, founder drift, and the debated plague leg as the only candidates; or the genomic neutrality result holds with no niche advantage found, making drift the main story.

Sources: Li D, Taylor AW 2008, Journal of Leukocyte Biology 84:191-198^[12]; Kambe Y et al. 2011, Zoological Science 28:560-567^[4]; Kambe Y et al. 2012, Genes & Genetic Systems 87:29-38^[6]; Tollenaere C, Rahalison L, Ranjalahy M, et al. 2008, Infection, Genetics and Evolution 8:891-897^[13]; Letts JB 1999, English Heritage / University of Reading^[14]; Cook LM, Grant BS, Saccheri IJ, Mallet J 2012, Biology Letters 8:609-612^[15]; Pagès M, Corbet G, Orth A, Volobouev V, et al. 2011, Journal of Mammalogy 92:659-670^[16]; Yiğit N, Çolak E, Kandemir İ, Kankılıç T 2008, Zoology in the Middle East 45:19-28^[17]; Gales RP 1982, New Zealand Journal of Zoology 9:463-466^[18]; Hufthammer AK, Walløe L 2013, Journal of Archaeological Science 40:1752-1759^[19]; Ducrest AL, Keller L, Roulin A 2008, Trends in Ecology & Evolution 23:502-510^[10].

[G7] The black MC1R allele is incompletely dominant, giving three coat classes (agouti, heterozygous black, homozygous black), and heterozygotes rust toward agouti with age. Do MC1R’s non-pigment (pleiotropic) effects follow the same gene-dose gradient across the three classes, and do they shift with age as the heterozygotes rust?

Our rats are uniquely suited to answer this.

Where it stands: Two features of this colony make it an unusually clean pleiotropy testbed. First, incomplete dominance turns the allele into a three-level dose series rather than a black-versus-agouti binary, and a graded series lets us test for an ordered trend, which is far harder to explain away than a two-group difference. MC1R shows a clean gene-dose effect on pigment itself (wild-type homozygote greater than heterozygote greater than variant homozygote), and in wild vertebrates the expression level of the melanocortin system, not genotype alone, tracks the associated traits, so graded signalling plausibly yields graded outcomes. The honest caveat is that heterozygotes are not guaranteed to be simple intermediates: some MC1R variants are dominant-negative, so a pleiotropic dose-response could be non-additive. Which traits are the best bets matters, and here a caveat from G3 and the Ducrest framework applies: the broad colour-and-behaviour syndrome in animals usually runs through shared melanocortin ligands acting on MC3R and MC4R centrally, not through the MC1R receptor itself, so a peripheral MC1R coding change should not be expected to produce a generic darker-equals-bolder effect. The defensible dose-responsive pleiotropy for our allele lies in MC1R’s own roles: its anti-inflammatory and Nrf2-protective signalling (G3), the biased constitutive signalling of the gain-of-function receptor, and the melanin-with-body-mass-and-immunity covariance seen in other species, consistent with Brian’s firsthand note that black rats often seem larger. Second, heterozygotes rust toward agouti with age (a heterozygous black female in the colony started solid black and is rusting), which adds a within-animal time axis: if MC1R output effectively wanes, an age-linked drift in a pleiotropic trait should track the rusting in heterozygotes while homozygotes hold steadier. This age arm is the more speculative one, because coat rusting is driven by melanocyte stem-cell depletion and differentiation, and MC1R is in fact dispensable for acute stress-induced graying (ACTH and MC2R dominate there), so the rusting may reflect follicle stem-cell dynamics rather than a system-wide change in MC1R activity. A practical consequence: once heterozygotes rust, coat colour stops being a reliable genotype readout, so the three classes must be set by PCR (the E94K genotyping of G2), not by eye, to keep the dose series honest. A final caveat: in some black rats melanism is driven by ASIP rather than MC1R, so genotype must be confirmed before any dose claim.

The test: PCR-genotype colony rats into the three MC1R classes (agouti, heterozygous, homozygous black) rather than scoring coat by eye, which fails once heterozygotes rust. Raise the genotypes as common-garden littermates to hold rearing constant, then measure candidate pleiotropic traits: response to a standard inflammatory or oxidative challenge plus Nrf2 tone, stress physiology (cortisol and adrenal size), and body weight with feed efficiency, and test for an ordered dose trend across the three classes rather than a black-versus-agouti contrast. Add a longitudinal arm: track the same traits and the degree of coat rusting in heterozygotes versus homozygotes across age, asking whether the pleiotropic measures drift in parallel with the rusting.

What you would see if it holds: A monotonic gradient across the three genotype classes on one or more non-pigment traits, most plausibly inflammatory or oxidative tone or body weight, together with an age-linked drift in heterozygotes that tracks the coat rusting, would show that MC1R dosage and its age-dependent expression drive pleiotropy in a live mammal: the gain-of-function, dose-resolved complement to the human MC1R literature, which is studied almost entirely from the loss-of-function redhead end.

What would falsify it: No ordered trend (heterozygotes indistinguishable from one homozygote, or black-versus-agouti differences that do not grade by dose), or no age-linked drift in heterozygotes, would mean the pigment dose-effect does not extend to the measured traits, or that coat rusting is a follicle-melanocyte phenomenon decoupled from MC1R’s systemic roles. ASIP rather than MC1R driving the melanism, or trait differences tracking linked genetic background rather than the allele itself, would likewise undercut a dose interpretation.

Sources: Naysmith L, Waterston K, Ha T, et al. 2004, Journal of Investigative Dermatology 122:423-428^[20]; Emaresi G, Ducrest AL, Bize P, et al. 2013, Molecular Ecology 22:4915-4930^[21]; Beaumont KA, Shekar SN, Newton RA, et al. 2007, Human Molecular Genetics 16:2249-2260^[22]; Benned-Jensen T, Mokrosinski J, Rosenkilde MM 2011, PLoS ONE 6:e24644^[5]; Li D, Taylor AW 2008, Journal of Leukocyte Biology 84:191-198^[12]; Kim SY, Fargallo JA, Vergara P, et al. 2013, Heredity 111:139-146^[23]; Ducrest AL, Keller L, Roulin A 2008, Trends in Ecology & Evolution 23:502-510^[10]; Sarin KY, Artandi SE 2007, Stem Cell Reviews 3:212-217^[24]; Inomata K, Aoto T, Binh NT, et al. 2009, Cell 137:1088-1099^[25]; Zhang B, He M, Rachmin I, et al. 2021, Experimental Dermatology 30:572-577^[26]; Sasamori S, Yoshida MA, Suzuki H, et al. 2017, Zoological Science 34:513-522^[7]; Kambe Y et al. 2011, Zoological Science 28:560-567^[4].

[BH1] Do roof rats live by territory rather than by rank, and is the dominance hierarchy everyone assumes even the wild Norway rat’s pattern?

Where it stands: This is already live on the site as Brian’s firsthand model: roof rats have no alpha, they are more cats than dogs, and they drive out rather than rank. It matters because some hobby sources import fancy-rat assumptions (keep in same-sex groups, introduce like Norway rats) that conflict with what the colony shows. A correction, made 2026-08-18: this entry used to name Barnett’s classic 1958 analysis as the Norway-rat status model being contrasted, and that was a misreading of our own source. Barnett does sort male Norway rats into alphas, betas and omegas, so a status vocabulary is genuinely in the paper, but the paper denies the hierarchy and puts the fighting somewhere else. His own summary states that there is no reason to think dominance hierarchies ever develop in wild rat colonies, and that the fighting of wild rats is essentially territorial, not for any specific object. He locates aggression where Brian locates it, most readily evoked in males established in a familiar area faced with a strange adult male of the same species, and he reports that males, but not females, added to established colonies were attacked by the resident males. That is the drive-out pattern, in the source we had filed as its opposite. The same study also carries rare roof-rat data: a small number of experiments with R. rattus showed that this species possesses all the components of amicable and aggressive behaviour observed in R. norvegicus, but that it is less fierce and more agile. So the hypothesis is now stated more strongly than it was, and Brian made that call after being shown the argument against it. The stronger claim is this: the drive-out model is not a roof-rat heresy against an established finding. It is what the foundational study of wild rat social behaviour reported, for Norway rats, in 1958. The stable-hierarchy picture that hobby advice rests on does not come from that work. Two limits travel with the claim and are not decoration. Barnett studied WILD rats, and the advice this argues against is about animals in cages, so a captive Norway-rat hierarchy would not contradict him. And his roof-rat observations were explicitly a small number of experiments, so they are a signpost and not a result. We still have no citable source for the strict Norway-rat dominance hierarchy, and we are not naming a substitute, because naming a paper we have not read is how the first error happened. If you know the right source for the status-hierarchy model, the box at the bottom of this page is where to tell us, and it would sharpen the question rather than settle it.

The test: Run a systematic study of roof-rat social interaction in a tame colony and compare it head to head with Norway rats: do they form a stable dominance hierarchy, or do they tolerate cagemates and simply drive out an intruder?

What you would see if it holds: Roof rats show no stable alpha or rank order and fight to drive out an intruder rather than to climb a hierarchy. Barnett reported the same for WILD Norway rats in 1958, so on this reading the hierarchy picture is an artefact of the pet-keeping literature rather than a species difference at all, and the colony’s model is the older and better supported one.

What would falsify it: Roof rats settle into a stable dominance hierarchy, which would mean the colony’s drive-out model is wrong or captivity-specific. It would equally weaken the stronger claim if wild Norway rats turn out to form stable hierarchies after all, since that would restore the species contrast this hypothesis was first written against.

Sources: BARNETT SA 1958, Proceedings of the Zoological Society of London 130:107-152^[27].

[BH2] Do arboreal roof rats build 3D cognitive maps and beat Norway rats on vertical tasks while doing worse on flat ones?

Where it stands: The entire rodent spatial-cognition canon (cognitive maps, place cells, mazes) was built on a burrowing species in essentially flat apparatus. An arboreal rat is an obvious, clean, untouched test of whether 3D living builds 3D maps. Nothing in the literature addresses it.

The test: Test roof rats and Norway rats on a true 3D climbing maze and on a glass-topped flat 2D maze (glass-topped so the climbers cannot leave the plane), and compare learning and accuracy.

What you would see if it holds: Roof rats outperform Norway rats in the vertical, volumetric maze and underperform on the flat maze, showing spatial cognition tuned for a tree-dwelling life.

What would falsify it: Roof rats do no better in 3D than Norway rats, which would mean arboreality does not reshape their spatial cognition.

[BH3] Is the colony’s tameness retained neoteny, an extended childlike window for accepting new humans and new rat friends?

Where it stands: Brian’s protocol selects for it on purpose: no handling until past a month, then keep only the animals still calmly accepting handling at two to three months, around sexual maturity. The idea is that this keeps childlike openness to new people, and he argues to new rat friendships, into adulthood. It connects directly to the Belyaev farm-fox domestication work. There is also outside support for the timing of his protocol. In the Norway-rat selection experiments the standard tameness test is run at 45 to 60 days, deliberately after adolescence, because adolescent rodents are transiently more exploratory and less anxious about novelty, so a friendly score taken then reflects exploratory drive rather than temperament. Brian’s rule of judging at two to three months lands in the same window, which means the protocol was independently arrived at rather than borrowed, and the colony tool’s nine-week maturity line sits just past the top of it. The mechanism worth borrowing here is developmental TIMING rather than the domestication syndrome as a whole. In dogs and wolves the senses mature on the same schedule and only the socialization window moves, which is neoteny expressed as a shifted timer rather than as a retained trait. Whether the colony shows the same shift is BH13.

The test: Measure the stress response (cortisol) and adrenal-gland size of the tame line against wild-type roof rats, since domesticated Norway rats have smaller adrenals and a blunted stress response.

What you would see if it holds: The tame line carries smaller adrenals and a lower stress response, consistent with the domestication syndrome and with a friend-acceptance window that stays open into adulthood.

What would falsify it: No difference in stress physiology, which would mean the tameness is learned handling rather than a heritable shift.

Sources: Trut L 1999, American Scientist 87:160^[11]; Albert FW, Shchepina O, Winter C, et al. 2008, Hormones and Behavior 53:413-421^[28]; Albert FW, Carlborg O, Plyusnina I, et al. 2009, Genetics 182:541-554^[29]; Lord K 2013, Ethology 119:110-120^[30]; Lu Y, Shi C, Jin X, He J, Yin Z 2022, Frontiers in Endocrinology 13^[31].

[BH4] Did earlier attempts to domesticate this species stall because selecting for tameness accidentally selected for behaviorally infertile males?

Where it stands: Brian saw exactly this firsthand: his especially friendly early males were behaviorally infertile until he found males that were both friendly and fertile. It is a real, novel domestication hypothesis with a clear practical lesson for any successor breeder.

The test: Across the breeding program, score the friendliest males for an appropriate mating response to a clearly receptive female, and track whether tameness and breeding competence can be held together over generations.

What you would see if it holds: The very friendliest males tend to be poor breeders at first (no appropriate response to a receptive female), so a naive tameness-only selection would lose fertility, and only deliberately keeping the friendly-and-fertile males moves the line forward.

What would falsify it: Tameness and male breeding competence are unrelated, which would remove this as a domestication bottleneck.

[BH5] Does selecting for tameness drag along white markings and a lighter belly, the way it does in foxes and Norway rats?

Where it stands: The colony already shows the predicted bundle: a white tail tip, lighter bellies, and tameness, with lighter-bellied animals tending calmer. The domestication-syndrome link is real but contested in the fox data, so this is framed as a hypothesis the colony is unusually well placed to test. Since then the closest thing to a direct test has come in, and it cuts against the link: in Norway rats bred for tameness over 60-plus generations, genome mapping put white coat spotting on a separate locus from tameness with no significant effect on tame behaviour, and individual rats with white spots were not tamer than those without. So the markings did appear alongside tameness in that line, but they did not travel with the tameness genes. The roof-rat observation that marked animals stay tame while solid-coloured littermates stay skittish comes only from breeder reports and has never been mapped, which is precisely what makes this colony worth measuring rather than a settled case.

The test: Over successive generations of the tameness-selected line, track the frequency of the white tail tip and lighter bellies and test whether they rise together with tameness.

What you would see if it holds: White markings and belly lightness become more common as the line gets tamer, matching the Belyaev pattern (silver foxes went from under one percent to about twelve percent white; Trut’s tame Norway rats became largely piebald in about thirty generations) through the shared neural-crest mechanism.

What would falsify it: Markings stay flat while tameness rises, which would break the domestication-syndrome link in this species.

Sources: Trut L 1999, American Scientist 87:160^[11]; Albert FW, Carlborg O, Plyusnina I, et al. 2009, Genetics 182:541-554^[29]; Albert FW, Shchepina O, Winter C, et al. 2008, Hormones and Behavior 53:413-421^[28].

[BH6] Are roof rats already partly self-domesticated, so taming them need not change their appearance much?

Where it stands: Roof rats are commensal, living alongside people for a very long time, and Brian’s view is that they are already largely self-domesticated. That predicts a faster, lower-cost domestication than the classic fox experiment.

The test: Compare how quickly tameness genes spread and fix in a roof-rat selection program against the fox and Norway-rat timelines, and watch whether appearance changes as little as the self-domestication idea predicts.

What you would see if it holds: Many individuals already carry tameness, so selection moves fast and, unlike the foxes, does not need to reshape the body or coat much.

What would falsify it: Tameness is rare and hard to fix and only appears alongside large appearance changes, which would argue against meaningful self-domestication.

[BH7] If the blonde (Rab38-null) coat dilutes eye pigment and weakens vision, how does that change climbing and jumping?

Where it stands: This links the genetic and behavioral question sets: the Rab38 hypothesis predicts a pigment and possibly vision effect, and an arboreal animal lives or falls by its depth perception. Whether blonde vision is actually impaired is itself unproven.

The test: Compare climbing accuracy, gap-crossing, and jumping in blonde and pigmented roof rats, alongside a direct check of visual acuity and light sensitivity.

What you would see if it holds: Blonde rats show measurably weaker vision and lean more on whiskers and care when climbing or jumping, a behavioral readout of the Rab38 pigment defect.

What would falsify it: Blonde and pigmented rats climb and see equally well, which would mean the dilute coat does not carry a vision or behavior cost.

[BH8] Norway rats emit 50-kHz laughter ultrasonic vocalizations and seek out heterospecific tickling as a reward. Do roof rats, which are arboreal, semi-solitary and lack the Norway rat’s rough-and-tumble social play, respond the same way, and does tickle-responsiveness track our tameness-selected line (and the face-rub bruxing and eye-boggle display we already see)?

Our rats are uniquely suited to answer this.

Where it stands: In the Norway rat the tickling and 50-kHz response is a well-validated, dopamine-driven positive-affect system: manual tickling reliably elicits 50-kHz calls and reward-seeking, the call rate is a stable trait-like individual marker (and inversely linked to anxiety), it is heritable enough to select high and low lines within a few generations, and a systematic review supports its use as a welfare measure. Every bit of that work is in Rattus norvegicus, with no published data in R. rattus. This is genuinely open rather than a foregone conclusion because of the very thing that makes roof rats different (BH1): they are semi-solitary and territorial with no stable hierarchy, and the rough-and-tumble juvenile play that tickling taps into is far more central to the social Norway rat. So the play-and-laughter system may be conserved, reduced, or qualitatively different in a less-social commensal, and a muted or absent response would be as informative as a positive one about how the system evolves with sociality. There is a suggestive opening on our side: Brian’s tame rats already show a candidate positive-affect display, a head-tilt with closed eyes and bruxing during face rubs, and whether that maps onto the Norway rat’s 50-kHz tickle response is exactly what a collaborator with ultrasonic recording and a tickling background could test. The honest cautions: tickling can be aversive if an animal dislikes it (so it needs an escape-proof setup, since roof rats solve problems by fleeing toward the highest point), and rearing history and individual differences are strong moderators, so any tame-versus-wild comparison has to separate innate response from how the animals were raised.

The test: With an ultrasonic microphone, apply a standardised tickling protocol in an escape-proof arena to roof rats from the tame line (and, where available, wild-derived rats), recording 50-kHz call rate and structure plus reward-seeking (approach and hand-following), and compare against published Norway-rat baselines. Test whether tickle-responsiveness correlates with the tameness phenotype and with the spontaneous face-rub bruxing and eye-boggle display, and follow the refinement protocols that separate innate predisposition from early-handling effects. Score blind.

What you would see if it holds: Roof rats emit 50-kHz vocalizations and seek the tickling, especially the tame line, showing the play and positive-affect system is conserved in a semi-solitary arboreal commensal and giving the colony a new objective welfare and affect readout, potentially one that tracks tameness selection.

What would falsify it: Little or no 50-kHz response and no reward-seeking, or active aversion, would indicate the rough-and-tumble-play and laughter system is reduced or rerouted in this less-social species, an interesting divergence from the Norway rat rather than a failure of method.

Sources: Burgdorf J, Panksepp J 2001, Physiology & behavior 72:167-73^[32]; Burgdorf J, Panksepp J, Brudzynski SM, et al. 2005, Behavior Genetics 35:67-72^[33]; Engelhardt K, Schwarting RKW, Wöhr M 2018, Psychopharmacology 235:83-98^[34]; LaFollette MR, O’Haire ME, Cloutier S, et al. 2017, PLoS One 12:e0175320^[35]; Bombail V, Brown SM, Martin JE, et al. 2022, F1000Research 11:1053^[36].

[BH9] When a roof rat is suddenly startled (a bang, a knocked-over object) or after an accidental fall, it FREEZES and then gives a distinct, audible, repetitive staccato call (eh eh eh / uh uh uh), heard firsthand in the colony. Is this a species-typical R. rattus alarm or warning call that has simply gone undocumented?

Our rats are uniquely suited to answer this.

Where it stands: This call is undocumented. A deep literature search (scientific, pest and ecology, laboratory, and fancier sources) found no description of a startle- or fall-triggered audible staccato alarm in R. rattus, or in this exact form in any rat. That is an argument from absence, not a disproof, and there are reasons the gap looks real rather than meaningful. The famous rat alarm call is the ULTRASONIC 22 kHz signal, inaudible to people, and a sudden startle is exactly what triggers that ultrasonic call in lab rats, so the documented startle-to-call link is to a signal nobody can hear without equipment. The audible calls that ARE described in Norway rats are pain or distress squeals and defensive threat calls that escalate as a threat gets closer, none of which is a freeze-then-call startle alarm. R. rattus’s own audible repertoire rests on essentially two old studies (Kaltwasser 1990, a single colony; Watts 1980), neither of which names a startle alarm, and no field bioacoustic study of free-living black-rat alarm behaviour was found. So the call may simply never have been recorded in the one setting where it is obvious: a tame, close-up, indoor colony, a context that barely exists in science despite humans living alongside R. rattus for millennia. The freeze-then-call sequence is behaviourally coherent as an antipredator response (freeze, assess, signal) and presents as a structured, distinct signal rather than a graded squeal, so the breeder’s read that it is a real, species-typical roof-rat call is the leading possibility; a shared-call explanation (the homology argument, where R. rattus and R. norvegicus differ mainly in call intensity) is the null to rule out, not the favoured answer. The arboreal hypothesis (that roof rats alarm-call like monkeys or squirrels because they live up high) is not supported by comparative data: across rodents DIURNALITY, not arboreality or sociality, predicts alarm calling, both Rattus species are nocturnal and are mapped as not reported to alarm call, and the audible-alarm-calling squirrels are themselves diurnal, confounding arboreality with diurnality. A complementary AI deep-research pass (Gemini) overstated the case, concluding the call is well documented; its own sources do not support that (mostly pest-control blogs, fancier sites and forums, with Kaltwasser cataloguing the repertoire but no startle alarm). It did surface one real primary precedent: Barnett and Stewart (1975) described audible signals during intolerant (agonistic) behaviour in the Australian bush rat R. fuscipes, which shows Rattus use audible social signals but is agonistic, not startle-specific. Footage now exists: an independent online video shows the same call in another keeper’s roof rat, and the colony’s own clip (a rat scolding after a fall) has had a first rough acoustic look from a single phone recording, a pulse train of about ten elements per second, broadband with energy up to roughly 6 to 7 kHz, harsh and pulsed rather than tonal, and clearly distinct from the human voice on the same recording. This argues against a single-colony idiosyncrasy, but it is a first characterization from one phone clip, not a controlled measurement, so a clean quiet-room recording remains the real test.

The test: Record the colony in the SONIC range (not only ultrasonic) at the moment of startle and post-fall, and capture a spectrogram: dominant frequency, the staccato rate and number of elements, intensity, and whether the freeze, a startle-jump, or hind-foot drumming accompanies it. Compare its structure against the published Norway-rat audible calls and against Kaltwasser’s and Watts’s R. rattus descriptions to test distinct-signal versus louder-shared-call. Where possible, check whether wild-derived roof rats give the same call under the same trigger, which separates a species-typical signal from a colony idiosyncrasy. Brian’s description is of a voiced eh/uh sound rather than clicking, so the recording should also confirm it is a true voiced vocalization (a forced-exhalation huff or grunt, or a distinct call) and not mechanical tooth-chatter (bruxism), since those differ in mechanism and meaning.

What you would see if it holds: A consistent, structurally distinct audible startle alarm, especially one absent or much rarer in Norway rats under the same trigger, would be a first description of a species-typical R. rattus alarm call: a real piece of natural history hiding in plain sight, and exactly the kind of thing only a tame, observable colony can document.

What would falsify it: If the recorded call matches a known Norway-rat audible call in structure and differs only in loudness or how easily it is triggered, it is a louder elicitation of a shared call rather than a unique roof-rat alarm. If only this colony gives it and wild-derived roof rats do not, it may be a line or rearing idiosyncrasy rather than species-typical.

Sources: Kaltwasser MT 1990, Journal of Comparative Psychology 104:227-232^[37]; Watts CHS 1980, Journal of Zoology 191:531-555^[38]; Litvin Y, Blanchard DC, Blanchard RJ 2007, Behavioural Brain Research 182:166-172^[39]; Shelley EL, Blumstein DT 2005, Behavioral Ecology 16:169-177^[40]; McRae TR, Green SM 2017, Ethology Ecology & Evolution 29:54-63^[41]; Barnett SA, Stewart AP 1975, Australian Journal of Zoology 23:103-112^[42].

[BH10] Almost everything known about rat empathy comes from Norway rats, the species behind laboratory and fancy pet rats. A recent framework (Newen and colleagues, 2026) reframes animal empathy as a profile across five dimensions rather than a yes-or-no trait, but every rat score in it was measured in Norway rats. Where would tame roof rats (R. rattus), a more solitary and territorial rat that still bonds intensely with people, fall on those five dimensions, and would they pass the cage-freeing helping test that has been run many times in Norway rats and, as far as we know, never in roof rats?

Our rats are uniquely suited to answer this.

Where it stands: When people talk about rat empathy, they are almost always talking about Norway rats. The multidimensional framework scores an animal across five dimensions: noticing another’s feelings, understanding another’s situation, taking another’s perspective, adjusting behaviour instead of running on autopilot, and taking others into account by synchronizing, cooperating, comforting or helping. Rats do well on some of these and poorly on others, but every one of those rat scores was measured in Norway rats. Roof rats are a different animal. They are more solitary and territorial, they do not form the same stable social groups, and males who were close can fall out as they mature. And yet, raised tame, they bond deeply with people. Whether that mix pulls a roof rat’s profile lower on the social dimensions, or whether the strength of their human bonds shows up somewhere you would not expect, is genuinely unknown. No one has scored roof rats on the five dimensions, so both the honest answer and the interesting one are the same: we do not yet know.

The test: Score tame colony roof rats across the five dimensions using the framework’s own criteria, and run the cage-freeing helping test: a free rat learns to open a restrainer holding a trapped cagemate, with chocolate available as a competing reward, and one records whether it frees the cagemate and whether it shares the food. That paradigm is well established in Norway rats (Ben-Ami Bartal and colleagues, 2011), which gives a direct point of comparison. Where possible, compare tame and wild-derived roof rats under the same test to separate a species trait from an effect of tameness or rearing. Careful records across the colony over time would let the profile firm up rather than rest on a single pairing.

What you would see if it holds: If roof rats show a distinct empathy profile, for example freeing a trapped cagemate less readily than Norway rats do, or in a different pattern across the five dimensions, that would be the first comparative empathy data for the species and would make tame roof rats a useful contrast for how social structure shapes prosocial behaviour. Because roof rats are almost never kept and closely observed, a tame colony with careful records is one of the only settings where this can be measured at all.

What would falsify it: If tame roof rats score within the Norway-rat range on all five dimensions and free a trapped cagemate just as readily, there is no distinct roof-rat empathy profile to report and the species mainly confirms that the pattern generalizes. If any difference turns out to be driven by tameness or rearing rather than the species, wild-derived controls would show it, which is why they matter to the test.

Sources: Newen A, Griem M, Huber L, Bugnyar T, Blaisdell A, Pika S 2026, Biological Reviews^[43]; Ben-Ami Bartal I, Decety J, Mason P 2011, Science 334:1427-1430^[44].

[BH11] Do black roof rats, assuming their black coat is the constitutively active MC1R E94K allele (question [G2]), differ in temperament from their agouti siblings, bolder and quicker to escalate when provoked? And if they do, is it the coat receptor itself acting outside the skin, or a neighbouring gene dragged along with it on the chromosome?

Our rats are uniquely suited to answer this.

Where it stands: Across many vertebrates darker, more eumelanic animals tend to be bolder, more aggressive and more stress-resistant, and the usual explanation is that pigment and temperament are co-regulated by the shared melanocortin (POMC) system, in which the same peptides (alpha-MSH, ACTH, and the antagonist ASIP) that set coat colour also tune behaviour and the stress axis across the five melanocortin receptors. But that covariation is ligand-driven, and our case is different and harder, because our black is a coding change in ONE receptor, MC1R, and the researchers who built the pleiotropy idea (Ducrest, Keller and Roulin) treat MC1R as the exception to it, largely melanocyte-restricted. So for a receptor-only change the disciplined prior is NO temperament difference, and a real difference would be the surprising, high-bar finding rather than the expected one; the wild darker-equals-bolder pattern is itself only influential, not settled, since a 2024 meta-analysis found colour-behaviour associations are real overall but not specific to melanin. On the adrenal question the answer is clean: the coat allele cannot reach the adrenal directly, because the adrenal cortex runs on a DIFFERENT melanocortin receptor, MC2R, which responds only to ACTH and needs the accessory protein MRAP even to reach the cell surface, while MC1R does not drive steroid output (if anything the arrow runs the other way, since chronically high ACTH can weakly darken pigment through MC1R as in Addison’s disease, but pigment does not drive the adrenal). What keeps a receptor-level effect on the table is that MC1R is not pigment-only: it also sits on immune cells (macrophages, microglia) and in specific brain regions such as the periaqueductal gray, giving a constitutively active E94K two receptor-intrinsic routes. First, a central one, since an MC1R change already alters central sensory and opioid gating (the loss-of-function redhead has altered pain and analgesia), so an always-on receptor could shift reactivity; second, an immune one, since MC1R activation is anti-inflammatory (it suppresses NF-kB and lowers cytokines such as IL-1beta and IL-6) and those same cytokines are the brakes that normally induce caution and behavioural inhibition, so a permanently anti-inflammatory MC1R could lift them (our question [G3] carries this immune angle, and is also why the effect cannot simply be pinned on the receptor). The E94K allele is a constitutively activating gain-of-function, the same class as the mouse sombre and tobacco E92K, which signals (biased cAMP and arrestin recruitment) without its ligand, so if any of these routes matter this is the genotype that would show it. The shape Brian sees fits a known ethology: in coping-style terms boldness and provoked aggression are two faces of one ‘proactive’ type, and the step from bold to aggressive is conditional rather than fixed, since in sticklebacks the boldness-aggression correlation appeared only after the fish had met a predator, so a black rat that reads as merely bold at rest and escalates mainly when provoked or already in a bad mood is exactly that predicted shape. The honest brake is the strongest competing explanation, and it is not the receptor: in a small closed colony the black allele rides in a haplotype, and genes physically next to Mc1r are dragged to high frequency with the coat when it is selected (genetic hitchhiking), while the Mc1r neighbourhood is dense with neuro and immune genes (the neurodevelopmental gene TUBB3, Class III beta-tubulin, sits only a few kilobases downstream of Mc1r and is even co-transcribed with it). A littermate design does NOT break this, because siblings that differ at the coat also differ across the whole linked block, and the murine tell points the same way, since the dark sombre and tobacco E92K mouse is not reported as notably aggressive, which argues against the receptor causing the temperament by itself. So a linked neighbour, not the receptor, is the disciplined default until breeding or a knock-in can separate them. What is genuinely open is whether black and agouti animals actually differ here, which is untested, and this colony is unusually placed to test it, with a single, PCR-confirmable allele (MC1R E94K, pending confirmation in [G2]) segregating in tame, handleable siblings; Brian tracks it in the colony’s black-versus-agouti study, and a second breeder in the collaboration independently reports the same direction, a blonde rat carrying the black MC1R allele being more aggressive than a blonde without it.

The test: Score tameness, boldness and exploration (open field, elevated-plus-maze), and provoked aggression (a resident-intruder or tube-dominance test) in black versus agouti animals, with the scorer BLINDED to coat colour (coat-masked video, or a rater who cannot see the animal), because an unblinded ‘the blacks are meaner’ impression is expectancy-laden and would not survive review. Treat litter and dam as the statistical unit (a mixed model with litter as a random effect), since pups in one litter are not independent. Then separate the receptor from its neighbours: do not rely on littermates alone, because siblings that differ at the coat also differ across the whole linked chromosome block, but breed black by agouti across several generations so recombination can uncouple the coat from adjacent genes, then test whether temperament tracks the MC1R allele or the coat block. The decisive versions are a cross-population test (does an independently arisen E94K line show the same temperament?) or a CRISPR knock-in of E94K on a clean, uniform background. Reading the stress axis (baseline and post-handling corticosterone) can place any difference on the proactive-reactive map, but a stress-axis difference is downstream, not proof of the mechanism.

What you would see if it holds: Black animals are reliably bolder and quicker to escalate under blinded scoring, AND the temperament still tracks the MC1R allele after recombination has broken the linkage (or in a knock-in). That would be a rare demonstration of receptor-level melanocortin pleiotropy from a single, known gain-of-function allele, the always-on mirror of the loss-of-function redhead, and would most plausibly run through MC1R’s jobs in the brain and immune cells rather than the adrenal.

What would falsify it: Two ways. If blinded scores show no black-versus-agouti difference, the coat allele carries no temperament corollary here, consistent with the receptor-restricted view in question [G3]. If a difference exists but does NOT track the MC1R allele once recombination breaks the linkage block (or fails to reappear in a knock-in), then a hitchhiking neighbour, not the receptor, was driving it.

Sources: Ducrest AL, Keller L, Roulin A 2008, Trends in Ecology & Evolution 23:502-510^[10]; Roulin A, Ducrest AL 2011, European Journal of Pharmacology 660:226-233^[45]; Ruckman SN, Humphrey EA, Muzzey L, Prantalou I, Pleasants M, Hughes KA 2024, Ecology and Evolution 14:e70655^[46]; Mafli A, Wakamatsu K, Roulin A 2011, Animal Behaviour 81:859-863^[47]; Almasi B, Jenni L, Jenni-Eiermann S, Roulin A 2010, Journal of Evolutionary Biology 23:987-996^[48]; Kambe Y et al. 2011, Zoological Science 28:560-567^[4]; Benned-Jensen T, Mokrosinski J, Rosenkilde MM 2011, PLoS ONE 6:e24644^[5]; Mogil JS, Wilson SG, Chesler EJ, et al. 2003, Proceedings of the National Academy of Sciences 100:4867-4872^[9]; Chen X, Chen H, Cai W, et al. 2017, Annals of Neurology 81:395-406^[8]; Koolhaas JM, Korte SM, De Boer SF, Van Der Vegt BJ, Van Reenen CG, Hopster H, De Jong IC, Ruis MA, Blokhuis HJ 1999, Neuroscience & Biobehavioral Reviews 23:925-935^[49]; Bell AM, Sih A 2007, Ecology Letters 10:828-834^[50]; Maynard Smith J, Haigh J 1974, Genetical Research 23:23-35^[51]; Dalziel M, Kolesnichenko M, das Neves RP, Iborra F, Goding C, Furger A 2011, Nucleic Acids Research 39:2378-2392^[52].

[BH12] When a female becomes less tame while nursing her first litter, does she come back to her old self after weaning, or is some of it permanent?

Where it stands: Brian’s observation from his own colony: many females become less tame while nursing a first litter, and it is unclear whether that is permanent or hormonal and reversible. In Norway rats the equivalent behaviour is well mapped and is clearly temporary. Maternal aggression appears after birth, peaks in early lactation and fades toward weaning; it is gated by the pups themselves, so five hours of separation markedly reduces it and ten minutes of reunion restores it, through a posterior-amygdala to hypothalamus circuit whose drive falls as oxytocin falls. But the physical recovery is slower than the behaviour: the stress axis is still not back to normal 72 hours after weaning, and is only partially recovered at 14 days, so a score taken days after weaning reads a still-recovering animal as a permanent drop. None of this has ever been measured in Rattus rattus, and the species differ in the direction that matters, since wild roof-rat females are reported as more aggressive than males at baseline, the reverse of the Norway pattern. This is one of the few questions here that needs no laboratory: it needs one tameness score at each end of one litter, in the same female, which is exactly what a breeder can collect and what published work almost never has.

The test: Score the same female twice with the same handler: once before she is bred, and again at least two weeks after her litter is weaned. Compare the pair. Repeat across her second litter, which separates a first-litter effect from simply getting older. Scores taken while she is still nursing are deliberately excluded, because during lactation the score measures a hormonal state rather than her temperament.

What you would see if it holds: Her after-weaning score returns to her pre-breeding level, meaning the nursing drop is a temporary state and not a change in the animal. That is what happens in Norway rats, where the behaviour is switched on by the pups and switches off without them.

What would falsify it: Her score stays down two weeks after weaning and again after the next litter. That would be a lasting change in temperament with first motherhood, which has never been described in Norway rats and would be a genuine species difference.

Sources: Yamaguchi T, Yan R, Khan M, Kuno S, Tewatia K, Osakada T, et al. 2026, Nature 654:1012-1022^[53]; Yamaguchi T, Lin D 2026, Current Opinion in Neurobiology 98:103194^[54]; Bosch OJ 2013, Philosophical Transactions of the Royal Society B 368:20130085^[55]; Windle RJ, Wood S, Shanks N, Perks P, Conde GL, da Costa AP, Ingram CD, Lightman SL 1997, Journal of Neuroendocrinology 9:407-414^[56]; Windle RJ, Wood SA, Kershaw YM, Lightman SL, Ingram CD 2013, Endocrinology 154:749-761^[57].

[BH13] Is selecting for tameness moving the colony’s socialization window later, so that pups can meet people for the first time after their eyes open instead of before?

Where it stands: In dogs and wolves the senses come online on the SAME schedule, about two weeks for smell, four for hearing and six for sight. What domestication changed is when the socialization window opens: two weeks in wolves, four in dogs. So a wolf pup explores while still blind and deaf and a dog pup explores with everything working, and that difference in experience, not a difference in the senses themselves, is what makes one tameable and the other not. Roof rats show a suggestive echo: breeders selecting for tameness reported that early generations had to be hand-fed before their eyes opened, while later generations could stay with the mother until near weaning and still be tame. Reading one species against the other is reasonable here because the systems involved are deeply conserved — the stress axis that tameness acts on is described as “typically conserved across vertebrates”, holding from fish to mammals. Two cautions. The roof-rat observation is a breeder report, not a study. And the mechanism behind the wider domestication syndrome is genuinely contested: the neural-crest explanation is influential and has been argued to be no unified explanation at all, so the conserved part to lean on is the stress axis and the developmental timing, not the full syndrome.

The test: For each litter, record the age at which pups first tolerate being handled, and whether that first friendly contact happened before or after their eyes opened. Track it across successive generations of the selected line. The comparison is within the colony, over time, so no wild-type animals are needed.

What you would see if it holds: Later generations socialize successfully when their first human contact comes AFTER eye-opening, while earlier generations needed contact before it. That would mean selection moved a developmental timer rather than changing behaviour directly, which is exactly what separates a dog from a wolf.

What would falsify it: Every generation needs the same pre-eye-opening contact, or tameness turns out to track handling effort rather than generation. Then the window is not moving and the resemblance to the canid case is only an analogy.

Sources: Lord K 2013, Ethology 119:110-120^[30]; Lu Y, Shi C, Jin X, He J, Yin Z 2022, Frontiers in Endocrinology 13^[31]; Wilkins AS, Wrangham RW, Fitch WT 2014, Genetics 197:795-808^[58]; Johnsson M, Henriksen R, Wright D 2021, Genetics 219^[59].

[P1] Are roof rats spared the tumor burden of fancy rats because they were never bred for fecundity and stay lean?

Where it stands: Fancy rats were bred hard for fecundity and fast growth, which raises the lifetime estrogen and prolactin load that drives rat mammary tumors (ovariectomy cut incidence from about 74 percent to 5 percent in one model), and pet rats trend obese, and obesity raises the risk of at least 13 cancers. Roof rats stayed lean and were never bred for fecundity. This is a distinct question from the ergothioneine cancer pharmacology one (H6); do not conflate them.

The test: Compare lifetime tumor incidence and body composition in roof rats and fancy rats matched for age and diet, separating the effect of reproductive-hormone load from the effect of body fat.

What you would see if it holds: Roof rats develop far fewer mammary and other tumors than fancy rats, and the gap tracks with their leaner bodies and their lack of selection for early, rapid, heavy breeding, making them a natural lean, never-bred-for-fecundity control for fancy-rat tumor biology.

What would falsify it: Roof rats get tumors at the same rate as fancy rats once age and diet are matched, which would remove the fecundity-and-obesity explanation.

Sources: Planas-Silva MD, Rutherford TM, Stone MC 2008, Cancer Detection and Prevention 32:65-71^[60]; Lauby-Secretan B, Scoccianti C, Loomis D, et al. (IARC Handbook Working Group) 2016, New England Journal of Medicine 375:794-798^[61]; Moore SC, Lee IM, Weiderpass E, et al. 2016, JAMA Internal Medicine 176:816-825^[62].

[P2] Do roof rats have a leaner metabolic set-point than Norway rats, and what are their actual nutrient requirements?

Where it stands: There are no published nutrient requirements for R. rattus; the major feed makers acknowledge they do not have them, so the colony is fed an 18 percent protein diet based on firsthand experience. Roof rats stay lean and are more active, unlike the obesity-prone Norway rat, so the usual senior-rat advice to cut protein and watch obesity does not obviously fit them. Which way an individual goes with age is part of what is open here: our own record of aging roof rats says a well fed older one may put weight on and become less active.

The test: Run a controlled feeding study in Rattus rattus against Rattus norvegicus on matched diets, measuring body composition, the age-to-weight trajectory, and activity, and use it to establish R. rattus nutrient requirements.

What you would see if it holds: Roof rats stay leaner and stay more active than Norway rats on the same food, and their age-to-weight curve is their own rather than a borrowed Norway-rat one, which would make them a useful lean-aging control and would give the species its own requirements instead of borrowed Norway-rat numbers.

What would falsify it: Roof rats and Norway rats converge on the same body composition and aging curve, which would mean there is no distinct roof-rat metabolic type to model.

Sources: Xu X, Cai G, Bu R, et al. 2015, PLOS ONE 10:e0144442^[63]; Moore SC, Lee IM, Weiderpass E, et al. 2016, JAMA Internal Medicine 176:816-825^[62].

[H1] Are blonde (Rab38) rats functionally deficient in ergothioneine?

Our rats are uniquely suited to answer this.

The test: In blonde (Rab38-mutant, fawn-hooded) rats and agouti littermates raised on the same ergothioneine intake, measure how much OCTN1 transporter sits at the proximal-tubule apical surface (surface biotinylation or immunostaining) and measure ergothioneine in kidney and blood.

What you would see if it holds: The blonde rats place less OCTN1 at the membrane and carry less ergothioneine on the same diet, and supplementing ergothioneine narrows a measurable kidney gap between the two coats.

What would falsify it: Equal OCTN1 placement and equal ergothioneine levels, which would mean Rab38 is not the gatekeeper for this transporter.

Sources: Oiso N et al. 2004, Mammalian Genome 15:307-314^[1]; Rangel-Filho A et al. 2013, Journal of the American Society of Nephrology 24:283-292^[64]; Nielsen R, Christensen EI, Birn H 2016, Kidney International 89:58-67^[65].

[H2] Do metformin or L-carnitine competitively reduce ergothioneine transport by OCTN1?

Where it stands: a human genetic signal already exists. In the SHIP-TREND cohort, people homozygous for the favorable OCTN1/OCTN2 haplotype (the OCTN1 c.1507T / L503F allele plus OCTN2 -207C) had about 30% lower all-cause mortality from age 60 to 70 onward. That is a genetic association, not proof of cause, and the study authors say it should be treated as potentially chance until it is replicated, but it is the kind of result you would expect if how much ergothioneine a body holds, set partly by this transporter, really matters for health.

The test: Measure blood and, where possible, kidney ergothioneine in long-term metformin users versus matched controls; in rats, dose metformin or carnitine and measure renal ergothioneine uptake directly. A sharper human design that needs only data the UK Biobank already holds: test an OCTN1 L503F by metformin-dose interaction on microvascular endpoints, since L503F carriers move metformin through OCTN1 more efficiently, so an ergothioneine shortfall should hit them harder at the same dose. And replicate the SHIP-TREND mortality association in a cohort with longer follow-up (SHIP-START or UK Biobank).

What you would see if it holds: Chronic metformin users carry lower ergothioneine than matched controls, with the biggest shortfall in the kidney where the transporter is densest; the L503F-by-metformin microvascular interaction is real; and the mortality association replicates.

What would falsify it: No difference, which would fit the view that the transporter is highly selective and not really a shared drug carrier.

Sources: Chen M, Yi Y, et al. 2024, Toxicology^[66]; Kato Y, Kubo Y, Iwata D, et al. 2010, Pharmaceutical Research 27:832-840^[67]; Tschirka J, Kreisor M, Betz J, Gründemann D 2018, Drug Metabolism and Disposition 46:779-785^[68]; Urban TJ, Yang C, Lagpacan LL, et al. 2007, Pharmacogenetics and Genomics 17:773-782^[69]; Reiter CP^[70].

[H3] Can a wild diet’s OCTN1-shared relatives substitute for scarce ergothioneine, and is scarcity already happening in some roof-rat populations?

The test: In ergothioneine-restricted rats, supplement the relatives that share the same doorway (stachydrine from citrus and alfalfa, selenoneine from fish) and track illness, tissue antioxidant status and lifespan against deficient controls; and survey ergothioneine exposure across wild roof-rat niches (fruit-heavy versus soil and fungal-rich) as a natural experiment.

What you would see if it holds: Rats getting the relatives stay healthier and live longer than deficient controls even without ergothioneine itself, showing real functional overlap; and fruit-niche populations show lower ergothioneine status than soil-foraging relatives.

What would falsify it: The relatives make no difference, meaning ergothioneine cannot be substituted and must be supplied on its own.

Sources: Gründemann D et al. 2005, Proc Natl Acad Sci U S A 102:5256-61^[71]; Yamashita Y, Yamashita M 2010, Journal of Biological Chemistry 285:18134-18138^[72].

[H4] Does ergothioneine raise a cell’s NADPH by making more of the enzyme G6PD, or only by pulling harder on the G6PD already there?

Where it stands: It is established that ergothioneine activates Nrf2 and raises the glutathione-building enzymes; what has never been measured in any mammal is its effect on G6PD or pentose-phosphate flux. Because G6PD is braked by NADPH itself, the pull-harder-on-existing-enzyme route is the more defensible one.

The test: Give cells a physiological dose of ergothioneine, trace labeled glucose through the pentose-phosphate pathway, and repeat with G6PD knocked down or Nrf2 removed.

What you would see if it holds: Flux rises without G6PD levels changing, tracking glutathione-recycling demand rather than enzyme amount.

What would falsify it: Ergothioneine raises G6PD expression itself, supporting the make-more-enzyme story.

Sources: Hseu YC, Lo HW, Korivi M, et al. 2015, Free Radical Biology and Medicine 86:102-117^[73]; Jenny KA, Mose G, Haupt DJ, et al. 2022, Antioxidants (Basel) 11:185^[74].

[H5] Does the 2025 ergothioneine-to-NAD+ signaling cascade work in rats, does it fade with age, and can it be protected?

Where it stands: the cascade fades with age at more than one point, and the two best-documented are the enzyme and the mark. CSE, the enzyme that turns ergothioneine into its hydrogen-sulfide signal, declines with age, and it is vitamin-B6 (PLP) dependent, so the age- and inflammation-driven fall in B6 status throttles it further. Separately, protein persulfidation, the chemical mark the signal writes, falls with age and is erased faster in an oxidized cell, and the methylglyoxal behind sugar-glycation competes for the very same cysteines. So the cascade rusts twice over: less signal is written, and the mark is wiped faster. Its output, NAD+, is also drained in parallel by the age-related rise in the enzyme CD38.

The test: Map CSE activity (with and without added PLP) and protein persulfidation in young versus aged rats, then test ergothioneine against ergothioneine plus a node-specific partner, ranked by how directly each hits the cascade. The smartest single companion is pyridoxamine, because it does two jobs at once: it supplies the B6/PLP cofactor CSE needs, and it traps the methylglyoxal that corrodes the enzyme and the persulfidation sites. Benfotiamine, the obvious anti-glycation pick, is weaker here: it only lowers the glycation precursor indirectly, hits no other node, and failed to lower glycation end-products in a human trial. Two more pairings worth running: ergothioneine plus a CD38 inhibitor (apigenin or quercetin) to stop the parallel NAD+ leak, and ergothioneine plus a slow-release hydrogen-sulfide donor to bypass a failing CSE entirely. The single cleanest readout across all of these is the persulfidation tag-switch assay, alongside tissue NAD+.

What you would see if it holds: Ergothioneine raises NAD+ and persulfidation only where CSE activity remains, and pairing it with B6/PLP or pyridoxamine restores the response in aged tissue that ergothioneine alone cannot; bypassing CSE with an H2S donor rescues it even when the enzyme is gone.

What would falsify it: The cascade is age-independent in rats, or no cofactor pairing changes the aged-tissue response.

Sources: Petrovic D et al. 2025, Cell Metabolism 37:542-556.e14^[75]; Zivanovic J et al. 2019, Cell Metab 30:1152-1170^[76]; Petrovic D et al. 2021, Front Aging Neurosci 13:674135^[77]; Williams ME et al. 2007, Am J Nephrol 27:605-614^[78]; Alkhalaf A et al. 2012, PLoS One 7:e40427^[79]; Camacho-Pereira J et al. 2016, Cell Metab 23:1127-1139^[80].

[H6] In a supplemented rat, does ergothioneine feed an existing tumor or help prevent and fight one?

Where it stands: Two-sided. Tumors can build extra antioxidant-uptake machinery and could hoard ergothioneine; but ergothioneine also prevents the oxidative damage that starts cancers and has killed colorectal-cancer cells outright by triggering programmed cell death.

The test: In a rat mammary-tumor model, randomize ergothioneine supplementation and track tumor incidence and growth alongside transporter expression in tumor versus normal tissue.

What you would see if it holds: If hijack dominates, faster growth with high transporter expression; if protection dominates, lower incidence or slower growth.

What would falsify it: No effect either way.

Sources: D’Onofrio N, Martino E, Balestrieri A, et al. 2022, FEBS Letters 596:1313-1329^[81].

[H7] Are the diets of captive and pet rats functionally short of ergothioneine, the way cat food once was short of taurine?

Where it stands: Ergothioneine comes only from the diet, the body holds it for weeks, and it has been called a longevity vitamin missing from modern diets. Lab blocks and kibble contain almost no mushrooms, the main source. The taurine precedent is exact: cat food left taurine out and cats went blind and died of heart failure until it was added back.

The test: Measure red-cell ergothioneine in colony rats on standard chow versus chow plus dietary ergothioneine, and track kidney, oxidative-stress and lifespan outcomes.

What you would see if it holds: Standard-chow rats run low on ergothioneine and show worse age-related outcomes that supplementation improves.

What would falsify it: Chow rats already carry ample ergothioneine, for instance from gut microbes (see H12), with no outcome gap.

Sources: Beelman RB et al. 2020, Journal of Nutritional Science 9:e52^[82]; Pion PD et al. 1987, Science 237:764-768^[83].

[H8] Does ergothioneine protect the kidney’s filtering barrier (the podocytes and their glycocalyx), not just the tubule, through Nrf2?

Where it stands: The transporter sits in podocytes and glomerular lining cells, and ergothioneine switches on Nrf2 and its protective genes; in rats it activated Nrf2 and protected the kidney in cisplatin and diabetic-kidney models. Whether it specifically repairs the glomerular barrier in a living rat is the open part.

The test: In a rat proteinuria model, give ergothioneine and measure albumin leak alongside podocyte foot-process markers and glycocalyx integrity, checking the Nrf2 targets.

What you would see if it holds: Ergothioneine lowers albumin leak with restored podocyte markers and a preserved glycocalyx, in step with Nrf2 activation.

What would falsify it: No change in barrier markers despite Nrf2 activation.

Sources: Salama SA, Abd-Allah GM, Mohamadin AM 2021, Life Sciences 278:119572^[84]; Dare A, Channa ML, Nadar A 2021, Biomedicine & Pharmacotherapy 141:111921^[85]; Nielsen R, Christensen EI, Birn H 2016, Kidney International 89:58-67^[65].

[H9] Do blonde (Rab38) rats have a lung-surfactant vulnerability that worsens respiratory disease, and does ergothioneine’s antioxidant action matter there?

Our rats are uniquely suited to answer this.

Where it stands: Rab38 runs the lamellar bodies that store lung surfactant, and Rab38-null rats have disrupted surfactant that is rescued by restoring Rab38. Whether this leaves blonde rats more prone to respiratory disease, and whether ergothioneine helps, is untested. There may already be a clue in the fancy-rat record: dilute-coated versus normal-coated Norway rats could differ in respiratory illness.

The test: Compare surfactant measures and respiratory-challenge outcomes in blonde versus agouti rats with and without ergothioneine; and mine existing Norway-rat and mouse colony data for any dilute versus non-dilute difference in respiratory morbidity and mortality.

What you would see if it holds: Blonde rats show a surfactant or respiratory disadvantage that antioxidant support narrows, mirrored by a dilute-coat signal in the fancy-rat record.

What would falsify it: No surfactant or respiratory difference by coat.

Sources: Zhang L et al. 2011, American Journal of Physiology. Lung Cellular and Molecular Physiology 301:L461-L477^[86]; Osanai K et al. 2017, Respiratory Research 18:70^[87].

[H10] Does ergothioneine extend lifespan in rats, and is the small dietary dose enough or does it take the high experimental dose?

Where it stands: A low daily dose raised median lifespan about 16% in male mice, the first hard mammalian lifespan result. The rat work so far is shorter-term healthspan (endurance, muscle) and kidney protection, not lifespan, and the human longevity link is only a correlation.

The test: A lifelong rat study comparing a dietary-equivalent dose (roughly 1 to 3 mg/kg per day) against the higher experimental tier (20 mg/kg per day and up).

What you would see if it holds: A measurable lifespan extension, likely larger at the higher dose.

What would falsify it: No lifespan effect at either dose in rats.

Sources: Katsube M et al. 2024, GeroScience 46:3889-3909^[88]; Petrovic D et al. 2025, Cell Metabolism 37:542-556.e14^[75]; Smith E et al. 2020, Heart 106:691-697^[89].

[H11] Does the roof rat (Rattus rattus) handle ergothioneine the same way as the lab rat (Rattus norvegicus) and humans?

Our rats are uniquely suited to answer this.

Where it stands: Almost all ergothioneine pharmacology is in humans, mice and the lab rat, whose transporter does carry ergothioneine. The roof rat is a different species and how it handles ergothioneine has never been measured, so applying the literature to it is a reasonable assumption, not a fact. It is likely highly conserved among related rodents, but we do not know.

The test: Characterize roof-rat OCTN1 ergothioneine transport kinetics and tissue retention against the lab rat.

What you would see if it holds: Equivalent affinity and retention, validating the transfer.

What would falsify it: Materially different kinetics, meaning the roof rat needs its own numbers.

Sources: Nakamura T et al. 2008, Biological & Pharmaceutical Bulletin 31:1580-1584^[90].

[H12] Does a rat’s gut microbiome supply absorbable ergothioneine (and its relatives), buffering a low-mushroom diet?

Where it stands: Soil and gut bacteria carry the genes to make ergothioneine, and rats eat their own droppings, so some endogenous supply is plausible but uncharacterized. The same goes for ergothioneine’s transporter-sharing relatives (stachydrine, selenoneine), which different wild and captive rat populations encounter in very different amounts. This is the key confound for the pellets-lack-ergothioneine and fruit-rat-scarcity ideas.

The test: Compare red-cell ergothioneine in germ-free versus conventional rats on an ergothioneine-free diet, and survey ergothioneine and relative-compound exposure across rat diets.

What you would see if it holds: Conventional rats retain measurable ergothioneine on an ergothioneine-free diet while germ-free rats do not.

What would falsify it: No microbial contribution, so dietary intake alone sets ergothioneine status.

Sources: Borodina I, Kenny LC, McCarthy CM, et al. 2020, Nutrition Research Reviews 33:190-217^[91]; Gründemann D et al. 2005, Proc Natl Acad Sci U S A 102:5256-61^[71].

[H13] Do blonde (Rab38) rats bleed dangerously more when blood-thinning supplements like fish oil and high-dose vitamin E are stacked with ergothioneine?

Our rats are uniquely suited to answer this.

Where it stands: Rab38 defects cause a platelet storage-pool deficiency and a mild bleeding tendency, and fish oil, high-dose vitamin E (especially the gamma form) and CBD each thin the blood; stacking several is the real risk. A caution on transfer: much of the bleeding data is in Norway rats and mice, and the well-known fawn-hooded rat is not a pure Rab38-null model (it carries other defects too), so how cleanly this maps to a blonde roof rat is uncertain.

The test: Bleeding-time and platelet-function tests in blonde versus agouti rats, with and without these additives.

What you would see if it holds: Blonde rats show prolonged bleeding times, worsened by stacked blood-thinners.

What would falsify it: No coat-linked bleeding difference.

Sources: Ninkovic I et al. 2008, Journal of Thrombosis and Haemostasis 6:2143-2151^[2]; Qureshi AA, Karpen CW, Qureshi N, et al. 2011, Lipids in health and disease 10:58^[92]; Formukong EA, Evans AT, Evans FJ 1989, The Journal of pharmacy and pharmacology 41:705-9^[93].

[A1] Does ambroxol actually help a rat with mycoplasma respiratory disease? It has never been tested in a rodent mycoplasma infection.

Where it stands: The whole rat respiratory rationale is extrapolated: human children with mycoplasma pneumonia did better on ambroxol plus azithromycin than azithromycin alone, and ambroxol’s mucus and anti-inflammatory actions fit the disease, but no rodent mycoplasma study has ever included an ambroxol arm. It is plausible, and we think promising, but unproven in the target disease. Play Inner Space: The Ambroxol Voyage to see how the molecule is proposed to work, and Justice League vs Mycoplasma to see the infection it would have to fight.

The test: A controlled Mycoplasma pulmonis challenge in rats, antibiotic alone versus antibiotic plus ambroxol (about 10 mg/kg per day), scoring lung histology, airway inflammation, mucus clearance, weight and breathing effort.

What you would see if it holds: The ambroxol arm resolves faster with lower lung-lesion scores and less airway inflammation, without changing bacterial load directly.

What would falsify it: No difference between the arms.

Sources: Gupta PR 2010, Lung India 27:46-48^[94]; Su X et al. 2004, Intensive Care Medicine 30:133-140^[95].

[A2] At the realistic ~10 mg/kg oral dose, which of ambroxol’s effects does a rat actually get?

Where it stands: The actions sort by dose. Below about 10 mg/kg oral a rat most plausibly gets the anti-inflammatory and a mild mucus-clearing (ciliary) effect; direct mucus dissolving, surfactant-gene induction (about 75 mg/kg), nerve-pain blockade, and the lysosomal enzyme boost need far higher concentrations than an oral dose reaches. So the popular shorthand that it thins the mucus is, at this dose, more a signaling effect on the airway lining than physical dissolving.

The test: A single rat dose-ranging study (1, 10, 30, 75 mg/kg per day) measuring all of these endpoints at once, alongside the measured drug level in the airway lining fluid.

What you would see if it holds: A clean threshold ladder: anti-inflammatory and ciliary effects low, surfactant and lysosomal effects only high.

What would falsify it: Surfactant or lysosomal effects appear even at the low oral dose.

Sources: Seifart C et al. 2005, Toxicology and Applied Pharmacology 203:27-35^[96]; Gillissen A et al. 1997, Research in Experimental Medicine 196:389-398^[97]; Gaida W et al. 2005, Neuropharmacology 49:1220-1227^[98].

[A3] Could ambroxol help a host cell clear an intracellular pathogen like mycoplasma by boosting autophagy, and is that even reachable at the rat oral dose?

Where it stands: Ambroxol induces autophagy at human-approved doses and has been proposed against intracellular pathogens, which is biologically reasonable. But no one has shown it clears any mycoplasma from cells, and the lysosomal and autophagy effects need concentrations an oral rat dose probably does not reach in the lung.

The test: Mycoplasma-infected rat lung-cell cultures with and without ambroxol across a dose range, measuring autophagic flux, lysosomal enzyme activity and intracellular bacterial load, then the in-vivo challenge model.

What you would see if it holds: A dose-dependent rise in autophagy and fall in intracellular mycoplasma, but only at concentrations the lung can actually accumulate.

What would falsify it: No effect on intracellular load at reachable concentrations.

Sources: Deretic V, Timmins GS 2019, Expert Opin Drug Metab Toxicol 15:213-218^[99]; Magalhães J et al. 2018, Scientific Reports 8:1385^[100]; McNeill A, Magalhães J, Shen C, et al. 2014, Brain 137:1481-1495^[101].

[A4] Do blonde (Rab38) rats benefit disproportionately from ambroxol, making them a natural model for ambroxol in trafficking and lysosomal disease?

Our rats are uniquely suited to answer this.

Where it stands: Rab38 runs the lamellar bodies and other lysosome-related organelles, and ambroxol’s published job is exactly boosting lysosomal biogenesis, chaperoning lysosomal enzymes and driving these organelles to secrete. So a cell hampered by a Rab38 defect is, on paper, almost tailor-made to respond to ambroxol. The compensation step (ambroxol rescuing a Rab38 rat) is untested. This is why Rab38-null rats are such a clean test bed for ambroxol across many trafficking and lysosomal-deficiency diseases, not just the lung.

The test: Compare ambroxol’s surfactant and lysosomal effects, and disease-model outcomes, in Rab38-null (blonde / fawn-hooded) rats versus wild-type.

What you would see if it holds: A larger relative response in the Rab38-null rats, confirming a compensation effect.

What would falsify it: Equal responses, meaning ambroxol does not preferentially help the Rab38-deficient cell.

Sources: Zhang L et al. 2011, American Journal of Physiology. Lung Cellular and Molecular Physiology 301:L461-L477^[86]; Osanai K et al. 2017, Respiratory Research 18:70^[87]; Bendikov-Bar I et al. 2013, Blood Cells, Molecules & Diseases 50:141-145^[102].

[A5] For a rat’s airway, is nebulized ambroxol much better than oral or the drinking-water route?

Where it stands: In rats, inhaled ambroxol reaches the airway-lining fluid far better than systemic dosing, and the drinking-water route is the weakest of all (low absorption times unpredictable intake by a sick rat). The same nebulized logic may apply to other drugs: NAC, gentamicin (likely safer nebulized than injected) and hypertonic saline all have an inhaled rationale. A dedicated nebulizer-treatment guide is worth adding to the site eventually.

The test: A rat respiratory model comparing oral versus nebulized ambroxol with matched airway-lining drug levels and clinical and histology endpoints.

What you would see if it holds: The nebulized arm reaches much higher airway concentrations and better airway outcomes at a lower total dose.

What would falsify it: No airway-outcome advantage for the nebulized route.

Sources: Ren YC, Wang L, He HB, Tang X 2009, J Pharm Sci 98:1797-1803^[103].

[A6] Does ambroxol actually improve how well a rat’s front-line antibiotics (doxycycline, enrofloxacin) reach the lung?

Where it stands: The clean review found ambroxol raises lung levels of macrolides (so azithromycin), beta-lactams, rifamycins and a few others, but it reports no such increase for tetracyclines or fluoroquinolones, which is exactly doxycycline and enrofloxacin. Important caveat: that is absence of reports, not proof of no effect, and the review itself flags possible non-reporting. And the question is wider than systemic penetration: thinning mucus or disrupting biofilm could still help locally, and nebulized fluoroquinolones may improve the safety margin in young or pregnant rats. This is an open research area, not a settled no.

The test: Measure doxycycline and enrofloxacin lung-tissue and lining-fluid levels in rats with and without ambroxol, infected and uninfected, plus an azithromycin positive-control arm and a nebulized-fluoroquinolone arm.

What you would see if it holds: Azithromycin lung levels rise with ambroxol; doxycycline and enrofloxacin may not via systemic penetration, but mucus or biofilm effects still improve local delivery.

What would falsify it: No change in any antibiotic’s lung delivery with ambroxol.

Sources: Deretic V, Timmins GS 2019, Expert Opin Drug Metab Toxicol 15:213-218^[99].

[A7] Is ambroxol absent from the US market purely because it is off-patent with no sponsor, not because of any safety failure?

Where it stands: This is the advocacy thesis and it is well reasoned, consistent with ambroxol’s roughly 50-year global over-the-counter record and its total absence from the US. But as stated it is an inference about regulatory economics, not a documented FDA decision; there is no public FDA review to point to. Worth verifying against the FDA’s own records before leaning on it hard.

The test: Search the FDA Orange Book and historical new-drug and investigational records and any citizen petitions for ambroxol, to document the actual regulatory history rather than infer it.

What you would see if it holds: No completed or withdrawn US application on safety grounds, confirming the absence is economic.

What would falsify it: A real US safety rejection turns up.

[A8] Are bromhexine and ambroxol different enough to matter in practice for a rat keeper?

Where it stands: Bromhexine is only partly converted to ambroxol (a fraction, not one-to-one) and has some activity of its own, so they are not strictly interchangeable. The open practical question is whether that difference is big enough to change dosing, or small enough that it is not worth fussing over when a keeper can only get one of them.

The test: A rat pharmacokinetic study measuring the ambroxol produced from a bromhexine dose versus an equivalent direct ambroxol dose.

What you would see if it holds: Bromhexine yields meaningfully lower or more variable ambroxol exposure, so the two should not be dosed one-to-one.

What would falsify it: Comparable effective exposure, so for practical purposes they are interchangeable.

Sources: Gupta PR 2010, Lung India 27:46-48^[94].

[H14] Could heavy L-carnitine supplementation lower antioxidant reserves indirectly, even though it barely competes with ergothioneine at OCTN1?

Where it stands: Carnitine is mainly carried by the sibling transporter OCTN2, so the doorway-competition idea is weak. But carnitine drives fatty-acid burning, and heavy fatty-acid oxidation can raise mitochondrial free radicals that consume glutathione, an indirect drain on the same antioxidant system ergothioneine supports. This is the honest version of the carnitine angle.

The test: Dose carnitine in rats and measure glutathione status, lipid-peroxidation markers and red-cell ergothioneine.

What you would see if it holds: Higher oxidative markers and lower glutathione without necessarily lowering ergothioneine uptake.

What would falsify it: No change in oxidative status.

Sources: Kato Y, Kubo Y, Iwata D, et al. 2010, Pharmaceutical Research 27:832-840^[67].