Working Paper · Nutritional Biochemistry

The Aryl Hydrocarbon Receptor & Metabolic Disease

How one transcription factor links gut permeability, tryptophan metabolism, seed-oil oxidation, hyperinsulinaemia and B-vitamin status to the circadian collapse of NAD⁺.

Abstract

Obesity is usually modelled as an energy-balance problem with inflammation as a complication. This paper argues something narrower and more testable: that a single ligand-activated transcription factor, the aryl hydrocarbon receptor (AhR), sits at the point where several independent metabolic insults converge, and that its chronic activation is sufficient to produce the characteristic obese phenotype — impaired fat oxidation, circadian disruption, NAD⁺ depletion, and a gut microbiome that extracts more calories from the same food.

The causal case does not rest on correlation. AhR inhibition prevents Western-diet obesity in mice3, reverses established obesity and steatosis (fatty liver disease)4, and kynurenine-driven AhR signalling alone causes weight gain, liver steatosis and hyperglycaemia5. What makes AhR unusual as a target is that its inputs are dietary: tryptophan catabolism, gut-barrier integrity, membrane fatty-acid composition, insulin pulse pattern and B-vitamin sufficiency each feed it, and each is modifiable at the plate.

The argument in brief

Six steps, one per part

  1. Part IChronically active, AhR produces six outputs that compound rather than add: NAD⁺ depletion via TiPARP, CLOCK displacement, IDO1 induction, methyl-pool drain via PEMT, CYP1B1–HETE signalling, and a Firmicutes-shifted microbiome.
  2. Part IIWhat switches it on falls into three tiers — direct activation, ligand supply, and loss of restraint — and each fails differently, which is why no single intervention covers all of it.
  3. Part IIIThree of those six outputs regenerate their own inputs, which is why the state persists once established rather than resolving with whatever triggered it.
  4. Part IVThat persistence matters because two circuits break as a result: the circadian NAD⁺ supply collapses, and the saturated-fat thermogenic circuit that depends on it stops running.
  5. Part VEach tier has a corresponding dietary lever — bile-acid signalling, flavonol NAD⁺ sparing, B-vitamin repletion, membrane fat composition, direct NADH reduction — that intervenes on mechanism rather than symptom.
  6. Part VIThose levers collapse into a short, ordinary food list; no supplement is load-bearing.

Not that AhR explains obesity — that it is the node where several partial explanations turn out to be the same explanation.

Reader's guide

Parts I–IV build the mechanism: what AhR does when chronically active, what activates it, the loops that make it self-sustaining, and the downstream damage. Parts V–VI are the intervention half. Caveats are collected in Limitations rather than scattered through the text; where a specific claim is weaker than the prose implies, it carries a grade.

Strong Moderate Animal–in vitro Inferred Contested Speculative

PRIMER

What AhR is, and why it is the right place to look

AhR is a ligand-activated transcription factor. At rest it sits in the cytoplasm held by a chaperone complex; when a ligand binds it sheds the chaperones, moves into the nucleus, pairs with a partner protein called ARNT, and switches on a defined set of target genes.

It was discovered as the receptor dioxin (TCDD, the potent AhR agonist used in most of this literature) binds, which is why most of the literature on it is toxicology rather than metabolism. That history is the reason it is under-appreciated as a metabolic node: for decades it was studied as the thing that makes industrial pollutants harmful, not as something the body operates deliberately.

The best way to understand it, though, is as a normal daily rhythm rather than a disease state. AhR is switched on after every meal by insulin signalling and switched off in the late fed state by bile-acid signalling2 — a routine postprandial pulse. The pathology is not that AhR turns on; it is that in obesity it never turns off.

Part of why a single receptor can sit at this many crossroads at once is that it is unusually promiscuous. Most nuclear receptors are locked to one ligand family — the estrogen receptor binds estrogens, PPARγ binds fatty acids. AhR binds bacterial metabolites, tryptophan breakdown products, dietary compounds, oxidation products and xenobiotics alike. Usually described as a quirk; read as a design feature, it is why gut health, diet, pollution exposure and oxidative stress all land on the same switch.

PART I

What AhR does when it is chronically on

Six outputs matter metabolically. They are worth taking as a set, because the claim of this paper is that the set is the obese phenotype.

DIRECT ACTIVATION KynurenineIDO1-derived · dominant ligand 4-HNE / oxidised ω-6electrophilic, covalent Insulin → PKB/Aktligand-independent LIGAND SUPPLY LPS → TLR4 → NF-κB Adipose hypoxia → HIF-1α IFN-γ / inflammation all three converge on IDO1 LOSS OF RESTRAINT B12 / folate deficiency B6 deficiency Blunted FGF19 / SHP AhR + ARNT nuclear, on XREs OUTPUTS TiPARP (PARP7)consumes NAD⁺ → SIRT1/3 starved Displaces CLOCK from BMAL1NAMPT transcription collapses IDO1more kynurenine — feedforward PEMT + GNMTspends SAM → homocysteine ↑ CYP1B1 → mid-chain HETEsNF-κB ↑ · PPARγ → SCD1 → MUFA Firmicutes ↑ / Bacteroidetes ↓more calories from same food IDO1 feedforward — the output becomes the input NAD⁺ loss → NADH backup → ROS → more 4-HNE

How to read it. Left, three tiers of input: things that directly switch AhR on, things that manufacture its dominant ligand, and conditions that remove the brakes. Centre, the activated AhR/ARNT complex on its target genes. Right, the six outputs. The two dashed rust arrows are the reason the system does not settle — two of the outputs loop back and become inputs.

I.1NAD⁺ depletion through TiPARP

TiPARP (PARP7) is a canonical AhR target gene7. PARP enzymes consume NAD⁺ stoichiometrically while ADP-ribosylating their substrates — each reaction burns one NAD⁺ to attach an ADP-ribose tag to the target protein — so inducing TiPARP is inducing an NAD⁺ drain. Dioxin measurably lowers NAD⁺ in liver and thymus and deactivates SIRT3 downstream, the mitochondrial enzyme that keeps proteins like SDH deacetylated and active6. Strong

I.2Circadian displacement

CLOCK:BMAL1 is the transcription-factor pair that runs the body's ~24-hour rhythm, setting the timing of sleep, hormone release and metabolic activity. AhR interacts with BMAL1 and displaces CLOCK from the heterodimer; separately it interferes with CLOCK:BMAL1 already bound at E-boxes; and separately again it displaces the coactivators the complex needs8. Three mechanisms, one consequence: dampened Per1 and a flattened clock. Because there is no single point of failure, partial AhR activation produces disproportionate disruption. Strong

CLOCK:BMAL1 disruption is not a metabolism-only concern — the same clock genes are implicated well outside this paper's scope, including in mood-disorder pathophysiology30, which is consistent with one transcription-factor pair sitting upstream of more than one downstream system.

I.3IDO1 induction — the output that is also an input

AhR transcribes IDO1, the rate-limiting enzyme that makes kynurenine — a breakdown product of the amino acid tryptophan, not an amino acid itself — which is AhR's own dominant endogenous ligand. This is the tightest of the loops in Part III and the reason the system is hard to exit once running. Strong

I.4PEMT and methyl-group consumption

In the fed state AhR induces PEMT and GNMT2. PEMT spends three SAM molecules per phosphatidylcholine synthesised — SAM is the cell's universal methyl-group donor, so each PC made costs three separate methyl transfers — each transfer yielding S-adenosylhomocysteine and then homocysteine, whose rise is itself a marker of cardiometabolic risk. GNMT (glycine N-methyltransferase) is a second SAM-consuming methyltransferase induced alongside it. A transient postprandial pulse of this is unremarkable. Sustained, it is a continuous drain on the methyl pool that B12 and folate exist to replenish. PEMT, AhR and phosphatidylcholine are all elevated in simple-steatosis patients2. Moderate

I.5CYP1B1 and the mid-chain HETEs

AhR induces CYP1B1, which converts arachidonic acid into mid-chain HETEs (5-, 12-, 15-HETE). Those HETEs activate NF-κB and MAPK and feed back to induce more CYP1B124. One side branch matters here specifically: 12-HETE activates PPARγ, raising SCD1 and monounsaturated-fat synthesis. That is AhR actively pushing membrane composition away from the saturated profile — and SCD1 overexpression lowers NAD⁺ and AMPK activation while SCD1 ablation raises both — Part IV.2 walks through why16. Animal–in vitro

I.6Microbiome shift

AhR activation by dietary and endogenous ligands raises Firmicutes and lowers Bacteroidetes — two of the dominant bacterial phyla in the human gut. Firmicutes are the better energy harvesters, extracting more calories from identical intake, so the same meal leaves a larger surplus of absorbed energy to be stored as fat. The phylum-level ratio is contested as an obesity marker across human cohorts, though the same Firmicutes shift tracks fat accumulation unambiguously in hibernating animals29, and is reported here as a directional AhR effect, not a diagnostic. Contested

PART II

What turns it on

The inputs are not parallel switches. They fall into three categories that fail in different ways, and the distinction matters because it determines what kind of intervention works on each.

Tier 1 — Direct activation

Insulin via PKB/Akt. Insulin drives AhR into the nucleus through PI3K/PKB with no ligand involved2. Physiological and self-limiting in health; its pathological form is Loop 2 in Part III. Strong

Kynurenine. IDO1 in inflamed adipose tissue converts tryptophan to kynurenine, which binds the AhR ligand-binding domain. Adipocyte-specific IDO1 depletion abolishes kynurenine accumulation and protects mice against obesity; adipocyte AhR ablation abolishes the effect of kynurenine entirely5. The dominant endogenous route. Strong

4-HNE and oxidised ω-6 phospholipids. Polyunsaturated membrane fatty acids — linoleic and arachidonic acid — carry the double bonds that hydroxyl radicals attack. Peroxidation yields 4-hydroxynonenal, an α,β-unsaturated aldehyde that modifies AhR cysteine residues covalently. Oxidised LDL activates AhR and downregulates circadian transcripts28. Saturated membrane fatty acids have no double bonds and cannot initiate the chain. Inferred

Tier 2 — Ligand supply

Three upstream conditions never touch AhR directly. They all raise IDO1, and therefore all act through kynurenine: bacterial LPS (lipopolysaccharide, a fragment of the outer wall of gram-negative gut bacteria) crossing a permeable gut barrier and signalling through TLR2/4 to NF-κB3; hypoxia (oxygen starvation) in adipose tissue that has expanded faster than its blood supply can keep up, signalling via HIF-1α; and inflammatory IFN-γ, released by immune cells infiltrating that same inflamed, oxygen-starved fat tissue, signalling via JAK/STAT1.

Calling these separate AhR routes would overstate their independence. What they are is three separate reasons the kynurenine tap runs — which is why gut-barrier repair and anti-inflammatory measures register as AhR interventions at all. Strong

Tier 3 — Loss of restraint

These are not activators. They are the removal of brakes that would otherwise hold a given level of activation in check — which is why two people with identical kynurenine can have different AhR activity.

B12 and folate are direct competitive antagonists at the AhR ligand-binding domain, blocking nuclear localisation, XRE binding and target-gene induction. Deficiency in mice induces AhR transcriptional activity on its own1. Strong

B6 never touches the receptor. As pyridoxal-5-phosphate it is the cofactor for kynureninase, the enzyme that clears kynurenine downstream18. Without it the ligand accumulates — see II.1.

Weak bile-acid signalling removes the physiological off-switch for the postprandial insulin pulse. Bile acids activate hepatic FXR, which induces SHP, and SHP is what actually binds ARNT and blocks AhR/ARNT9. When bile-acid signalling is weak, that shutoff doesn't fire: SHP occupancy at AhR target promoters is reduced in NAFLD patients and obese mice while AhR occupancy rises8. Strong

II.1The B6 branch point — the same molecule as fuel or as poison

Kynurenine is not intrinsically harmful. It is an intermediate, and what it becomes depends on whether the enzyme that processes it has its cofactor.

Tryptophan dietary IDO1 Kynurenine the fork Kynureninase active PLP (vitamin B6) present → 3-HAA → quinolinic acid NAD⁺ de novo synthesis Kynureninase idle PLP deficient kynurenine accumulates AhR activation → TiPARP → NAD⁺ ↓ same pathway, opposite outcome — and the two ends move NAD⁺ in opposite directions

How to read it. Roughly 60 mg of tryptophan yields 1 mg of nicotinamide by the upper route, so an intact kynurenine pathway is a genuine NAD⁺ source. The lower route is the same molecule with the exit blocked. The irony worth noticing: the pathway that makes NAD⁺ when B6 is sufficient destroys it when B6 is not, because accumulated kynurenine activates AhR and AhR induces an NAD⁺-consuming enzyme.

Supplementing 80 mg pyridoxine for eight weeks in overweight and obese women reduced fat mass, visceral adiposity, fasting insulin, HOMA-IR, triglycerides and leptin, and raised adiponectin against control17. The trial measured neither kynurenine nor AhR activity, so the mechanism here is inferred from the enzymology rather than demonstrated by the trial. Inferred

PART III

Why it does not settle

A transcription factor with a working off-switch does not cause chronic disease. What makes AhR different in the obese state is that three of its outputs regenerate its own inputs.

AhR shared by all three LOOP 1 · LIGAND FEEDFORWARD IDO1 inducedAhR target gene Kynurenine ↑AhR's own ligand LOOP 2 · METHYL DRAIN PEMT ↑ → SAM ↓3 SAM per PC B12 / folatedrawn down, cannotantagonise receptor LOOP 3 · REDOX TiPARP → NAD⁺ ↓SIRT3 starved NADH backs upSDH acetylated, RET lost Complex I superoxide→ hydroxyl radical ω-6 PUFA → 4-HNEelectrophilic AhR ligand Loop 3 is the one dietary fat composition breaks — saturated membranes have no double bonds to peroxidise

How to read it. All three loops pass through the same central node, which is why they compound rather than add. Loop 1 regenerates the ligand. Loop 2 removes the antagonist. Loop 3 manufactures a second, chemically different ligand out of the redox damage the first two cause. Every individual arrow is documented; none of the three rings has been demonstrated end-to-end in a single experiment.

III.1Loop 1 — the ligand regenerates itself

AhR induces IDO1; IDO1 makes kynurenine; kynurenine activates AhR. The tightest of the three, and the reason kynurenine does not simply decay once the initiating inflammation resolves. Strong

III.2Loop 2 — the antagonist is consumed

AhR induces PEMT; PEMT spends SAM; regenerating SAM draws on B12 and folate; B12 and folate are the receptor's competitive antagonists. The receptor consumes the molecules that would restrain it. In health this never runs long enough to matter, because FGF19/SHP terminates the pulse. In obesity, with the shutoff blunted and insulin chronically elevated, it runs continuously. Inferred

III.3Loop 3 — redox damage makes a second ligand

AhR induces TiPARP; TiPARP depletes NAD⁺; NAD⁺-starved SIRT3 leaves SDH hyperacetylated14; NADH backs up at Complex I and produces superoxide in the forward direction25; the hydroxyl radical that follows attacks membrane polyunsaturated fatty acids; 4-HNE results and activates AhR covalently.

That redox damage has a second exit besides the ligand it makes directly, and this one closes back through insulin rather than through AhR's ligand-binding domain. The same NADH backup is a mitochondrial-overload signature — incomplete fatty-acid oxidation and depleted TCA intermediates, which Koves et al. tie directly to skeletal-muscle insulin resistance31. And the superoxide is not just a stepping-stone to 4-HNE: reactive oxygen species are independently sufficient to cause insulin resistance across several experimental models, an effect reversed by antioxidant treatment32. Either route lands in the same place the ligand does — insulin resistance drives the compensatory hyperinsulinaemia that feeds Tier 1's PKB/Akt route back into AhR, the mechanism in the flag below. So Loop 3 has two closures sharing one upstream cause, not one. Inferred

This loop has a dietary lever the other two do not. Its substrate is membrane ω-6 content, and a saturated-fat-rich membrane cannot sustain the peroxidation chain at all, because there are no double bonds for the radical to attack. Inferred

Where insulin resistance makes this worse rather than better

Hepatic insulin resistance is selective: glucose disposal fails while lipogenic signalling stays intact or amplifies. If the PKB→AhR arm sits on the preserved side — and the fed-state mechanism suggests it does — then AhR activation increases in proportion to how insulin-resistant someone becomes, driven by the compensatory hyperinsulinaemia the resistance itself produces. Speculative The selective-resistance principle is well established; this particular branch has not been quantified in insulin-resistant human liver. What is no longer speculative is that resistance gets there in the first place: Loop 3 supplies a documented route into it — mitochondrial overload and superoxide excess, independent of whatever else is causing resistance elsewhere in the body.

PART IV

The damage: two circuits AhR takes down

IV.1The circadian NAD⁺ supply

NAD⁺ is not a static pool that happens to be low in obesity. It is manufactured rhythmically. CLOCK:BMAL1 transcribes NAMPT, the rate-limiting enzyme of the salvage pathway; NAMPT makes NMN; NMNAT converts NMN to NAD⁺; NAD⁺ fuels SIRT1; and SIRT1 is recruited back to the NAMPT promoter to sustain synthesis of its own coenzyme1213. Intracellular NAD⁺ oscillates on a 24-hour cycle as a result.

This is what makes AhR's displacement of CLOCK more serious than a timing problem. It shuts down NAMPT transcription, which is the rate-limiting step of NAD⁺ production. The NAD⁺ deficit in obesity is therefore not only a consumption problem — TiPARP, PARP1, CD38 — but a production failure, and the production failure is downstream of AhR.

The human anchor is a monozygotic-twin study: in BMI-discordant identical twins, the heavier co-twin had lower NAD⁺/SIRT pathway expression in adipose tissue, tracking with inflammation and insulin resistance20. Because the twins are genetically identical, the deficit is acquired. The companion study found weight loss raised NAD⁺/SIRT1 and lowered PARP activity21. Moderate

One implication worth stating plainly. If the deficit is a broken oscillator rather than a shortage of raw material, then NMN and NR supplementation floods a system that has lost its rhythmic regulation instead of repairing it — a candidate explanation for why precursor trials read so much better in rodents than in humans. Speculative

Sleep is the other input here, and no dietary intervention substitutes for it. Delaying sleep onset progressively — without reducing total sleep — collapses circadian transcript expression including CLOCK and BMAL1 severalfold in humans. citation to pin

IV.2The FADH₂ / reverse-electron-transport circuit

The second casualty is a thermogenic circuit that runs on saturated fat and only works when NAD⁺ is replete.

Saturated fat18 carbons, no double bonds High FADH₂ : NADH9 FAD cycles via ETF Q pool reducedSDH feeds it too Reverse electron transportsuperoxide at Complex I flavin SOD → H₂O₂the useful signal, not the damage Glutathione peroxidasespends NADPH, makes GSSG NNTNADH + NADP⁺ → NAD⁺ + NADPH NAD⁺ regenerated, heat producedenergy dissipated without ATP WHERE OBESITY BREAKS IT 1 · SDH hyperacetylated SIRT3 has no NAD⁺ — Q pool not driven, no RET 2 · NNT suppressed palmitate lowers it; NADPH not regenerated 3 · Membranes are ω-6 SCD1/PUFA give the radical something to hit

How to read it. Follow the top row left to right, then down the right side and back along the bottom — it is a cycle, and the point of the cycle is that it burns energy without making ATP. The superoxide here is a signal, not damage: small, controlled, immediately buffered. Note that it emerges at Complex I running backwards, not at Complex II. SDH and β-oxidation FADH₂ create the conditions; Complex I produces the radical.

Saturated fatty acids reduce FAD at every carbon pair during β-oxidation. Polyunsaturated fatty acids cannot — their double bonds are already oxidised at those positions — so they yield proportionally less FADH₂ and a less reduced Q pool. The in-vivo evidence is the SCD1 line: mice that cannot convert saturated to monounsaturated fat show higher AMP/ATP, higher AMPK activation and higher NAD⁺ with greater SIRT1 activity, while SCD1 overexpression produces the reverse16. Animal–in vitro

The same NAD⁺/sirtuin gate decides whether the pulse is useful. In an NAD⁺-replete cell SIRT3 keeps SDH functional and the burst is a discrete, buffered H₂O₂ signal. In an NAD⁺-depleted one the circuit is broken at three points at once — SDH acetylated, NNT suppressed23, membranes peroxidisable — and what is produced instead is unbuffered forward-direction superoxide from backed-up NADH. Same organelle, opposite meaning.

Why antioxidant supplements are the wrong tool here

Reductive stress is an excess of reducing equivalents, and chronic antioxidant supplementation adds to it. Over-reducing the glutathione couple impairs the redox-sensitive protein chemistry that depends on it, and a reductive-stress → oxidative-stress → reductive-stress feedback has been described27.

The corrective is to lower the NADH:NAD⁺ ratio, not to add more reducing capacity on top of it. Alpha-lipoic acid does exactly that — reduced by lipoamide dehydrogenase in an NADH-dependent reaction, it lowered cellular NADH by about 30% with a smaller and slower effect on NADPH26. It is an electron acceptor, not a scavenger. Moderate

PART V

The interventions

Each lever below acts on a different tier of Part II. The coverage matrix at V.6 is the single place where the mapping is laid out; these sections explain the mechanisms rather than repeat it.

V.1Theabrownin — the bile-acid lever

Theabrownin is the dominant bioactive polymer in fermented Pu-erh tea, and its mechanism in humans is unusual: it acts on FXR in two anatomical places with opposite local effects.

In the gut it suppresses bile-salt-hydrolase-active microbes, so conjugated bile acids accumulate in the ileum. These are weak FXR agonists, so intestinal FXR signalling falls and serum FGF19 drops — which sounds like the wrong direction until you follow it: less FGF19 means CYP7A1 stays active, so bile-acid synthesis and turnover stay high. Meanwhile the bile acids reaching the liver through the portal circulation activate hepatic FXR, which induces SHP10. SHP binds ARNT and blocks AhR9.

So theabrownin restores the physiological AhR shutoff — the same SHP-dependent brake that terminates the postprandial insulin pulse in health. It also inhibits α-glucosidase and α-amylase, blunting the glucose rise and therefore the insulin signal that opens the pulse in the first place. One compound acting on both the trigger and the brake of Tier 1's insulin route.

Three further arms, all downstream of Pu-erh's fermentation microbes and polyphenol content: it promotes gut Lactobacillus enrichment, which diverts tryptophan to indoles, competing with IDO1 for substrate11; it also promotes Akkermansia muciniphila enrichment, which tightens the gut barrier and reduces the LPS flux driving Tier 2; and cinnabarinic acid, present in Pu-erh itself, occupies the AhR ligand-binding domain as a partial agonist, competitively displacing kynurenine.

The cholesterol arm is separate and addresses reductive stress rather than AhR. Sustained bile-acid excretion depletes hepatic cholesterol, SREBP-2 responds, and cholesterol synthesis consumes roughly 12–16 NADPH per molecule across the mevalonate pathway — a genuine NADPH sink, shown to lower NAD(P)H and improve function in a complex-I-deficient model15. Moderate for the human FXR/SHP arm; Animal–in vitro for the NADPH sink.

V.2Quercetin and myricetin — the NAD⁺ lever

Both flavonols competitively inhibit PARP1 at its NAD⁺ pocket, sparing NAD⁺ that would otherwise be consumed. This complements rather than duplicates the theabrownin arm: theabrownin reduces TiPARP by suppressing its transcription, flavonols reduce PARP1 by inhibiting the enzyme. Two different PARP family members, two different mechanisms.

What the restored NAD⁺ buys is SIRT3 activity, and SIRT3's two relevant substrates are exactly the two failure points in the Part IV circuit: SDHA, which it deacetylates at thirteen identified sites14, and NNT. Restoring both reinstates the RET circuit and — by preventing the NADH backup that generates hydroxyl radicals — reduces the 4-HNE production feeding Loop 3.

Secondary arms: both flavonols activate NRF2, the transcription factor that senses oxidative and electrophilic stress and switches on the cell's antioxidant enzymes in response — glutathione synthesis, thioredoxin reductase (which keeps the thioredoxin system reduced and able to neutralise peroxides) and NQO1 (which detoxifies reactive quinones before they can do damage) — cutting the ROS burden upstream of both PARP activation and lipid peroxidation. Quercetin also inhibits PI3K, reducing PKB (protein kinase B, i.e. Akt) directly — the same insulin-driven arm that opens Tier 1's nuclear-translocation route. SIRT1 deacetylates LKB1, activating AMPK, which improves insulin sensitivity and so lowers the chronic insulin driving Tier 1. Quercetin also inhibits mTORC1 — as does AMPK, via TSC2 and Raptor — promoting mitophagy, and inhibits COMT, slowing catecholamine breakdown and supporting lipolysis. Animal–in vitro for most arms; see Limitations on bioavailability.

V.3Whole-food dairy — the antagonist lever

B12 and folate are the only interventions here that act at the receptor itself, and their delivery vehicle matters more than usual. Milk B12 is bound to haptocorrin and related binding proteins that survive gut transit and support cubilin-mediated ileal absorption, and its bioavailability exceeds synthetic cyanocobalamin at equivalent dose22b. Milk folate behaves similarly, and milk lowers homocysteine more effectively than matched supplement doses22. Low serum B12 is inversely associated with BMI in a nationally representative US sample after adjustment19.

Whey contributes separately. Dietary tryptophan has two competing fates: the IDO1/kynurenine route this whole paper is about, and a smaller route through tryptophan hydroxylase (TPH) into serotonin and then melatonin. Both enzymes draw on the same circulating tryptophan pool, so when IDO1 is running hot — as this paper's model has it doing throughout obesity — more of that pool is pulled toward kynurenine and less reaches TPH. Whey is the densest dietary tryptophan source per gram of protein, which raises the size of the pool itself rather than trying to out-compete IDO1 for a fixed one; more tryptophan reaching TPH means more serotonin and melatonin, and melatonin is itself a circadian amplitude signal that the Part IV mechanism erodes. Pairing tryptophan-rich protein with B6 compounds this twice over: B6 is the cofactor that converts 5-HTP to serotonin, and — per II.1 — the same cofactor that clears any kynurenine that does form before it accumulates and activates AhR. Whey also preserves lean mass during fat loss and blunts postprandial triglycerides. Asparagus and orange juice cover folate outside dairy. Moderate

V.4Dietary saturated fat — the membrane lever

Saturated fat does two things nothing else here does. It supplies the high FADH₂:NADH ratio that drives the Part IV thermogenic circuit, and it builds membranes that cannot sustain a peroxidation chain — closing Loop 3 at its substrate rather than downstream of it. Both follow from the same structural fact: no double bonds.

This is where this paper and the linoleic-acid argument meet from opposite directions. That paper's case is that excess ω-6 supplies the oxidisable substrate; this one's is that AhR supplies the oxidant and then consumes the product as a ligand. The 12-HETE → PPARγ → SCD1 branch of I.5 means AhR is also pushing membrane composition toward monounsaturated and away from saturated — the direction that keeps the circuit off. Inferred

V.5Alpha-lipoic acid — the redox lever

Covered in the box at the end of Part IV. Its distinguishing property is that it targets the NADH excess directly rather than scavenging the radicals that excess produces, which makes it the one supplement in this framework that does not carry the reductive-stress objection. Moderate

V.6Coverage matrix

InputTierWhat addresses it, and how
KynurenineDirect Theabrownin — SHP blocks AhR/ARNT; Lactobacillus diverts tryptophan; cinnabarinic acid competes at the site. B12/folate — competitive antagonism. B6 — clears the ligand downstream.
4-HNE /
oxidised ω-6
Direct Saturated fat — removes the peroxidisable substrate (primary). Flavonols — NRF2 and restored SIRT3/NNT cut radical production. ALA — lowers the NADH driving it. B12/folate antagonism is not established against covalent modification.
Insulin / PKBDirect Theabrownin — α-glucosidase inhibition lowers the trigger, FXR/SHP restores the brake. Quercetin — PI3K inhibition. Flavonols — AMPK improves sensitivity, lowering chronic insulin.
LPS /
gut barrier
Supply Theabrownin — Akkermansia enrichment and tight-junction integrity.
Adipose hypoxia,
IFN-γ
Supply Addressed only indirectly, through fat loss and reduced inflammatory tone. The weakest coverage in the framework.
B12 / folate
status
Restraint Milk (superior bioavailability), asparagus, orange juice.
B6 statusRestraint Animal protein, whey.
FGF19 / SHP
shutoff
Restraint Theabrownin — hepatic FXR activation restores SHP induction.
PART VI

What to eat

The framework reduces to a short list. None of it requires supplements, and the food-matrix forms are in several cases better absorbed than the isolated compounds.

  1. Pu-erh tea, daily. The bile-acid and microbiome arms respond to consistent low exposure rather than large doses — the gut microbiome is entrained by regularity, and the human trial ran four weeks. A cup a day is the realistic form of this.
  2. Blueberries. Quercetin and myricetin in glycoside form, delivered with anthocyanins that carry their own NRF2 and NF-κB effects, in a fibre matrix that extends gut residence.
  3. Whole milk and whey. The B12 and folate delivery vehicle, plus the tryptophan density that keeps the serotonin and melatonin routes supplied while IDO1 competes for the same pool.
  4. Pastured animal protein. B6 for the kynurenine exit; saturated fat for the membrane and the FADH₂ ratio.
  5. Asparagus and orange juice. Folate outside dairy.
  6. Less ω-6. Not a food to add but a substrate to withdraw — the one lever that closes Loop 3 rather than mitigating it.
  7. Consistent sleep timing. Not a food, and not optional. The circadian arm has no dietary substitute; a flattened CLOCK:BMAL1 oscillation means low NAMPT transcription regardless of what is on the plate.

VI.1How much, and from where

The restraint-lever nutrients from V.3 and the coverage matrix (V.6), quantified. Amounts are USDA-typical per common serving; %DV uses current adult label values (B12 2.4 μg, folate 400 μg DFE, B6 1.7 mg). Whey's tryptophan density is unusual — most protein sources run closer to 1–1.3% of protein by weight.

NutrientRestraint onFood, servingAmount
Vitamin B12KynurenineWhole milk, 1 cup (240 ml)~1.2 μg (~50% DV) — haptocorrin-bound, higher bioavailability than synthetic cyanocobalamin
FolateKynurenineAsparagus, 1 cup cooked~260 μg (~65% DV) — the richest non-dairy source here
FolateKynurenineOrange juice, 1 cup~75 μg (~18% DV) — carries a fructose load; time relative to meals
FolateKynurenineWhole milk, 1 cup~12 μg (~3% DV) — modest per serving, but the vehicle drinkers hit daily
Vitamin B6KynureninePastured chicken or beef, 3 oz (85 g)~0.4–0.5 mg (~25–30% DV)
TryptophanSerotonin / melatoninWhey protein, 1 scoop (~25 g protein)~500 mg — densest common source per gram of protein
TryptophanSerotonin / melatoninPastured chicken or turkey, 3 oz (85 g)~300 mg

Moderatefor the amounts — standard USDA per-serving values, but brand, cut, and fortification shift them; this quantifies V.3’s food list, it does not newly establish the antagonism mechanism itself (that's ref 1).

LIMITS

What would weaken this argument

  • No loop has been closed end-to-end. All three rings in Part III are assembled from individually documented arrows. No single experiment traverses any of them, and most constituent work is cell and rodent. This is the central structural weakness and applies to the framework as a whole rather than to particular claims within it.
  • 4-HNE as a direct AhR activator is the weakest documented link. Electrophilic cysteine modification of AhR by reactive aldehydes is chemically plausible, supported by the oxidised-LDL/AhR/circadian work28 and by structurally comparable quinones binding AhR directly. A direct 4-HNE binding assay at the ligand-binding domain has not been published. If that link fails, Loop 3 still functions through its inflammatory arm — 4-HNE → TLR4 → NF-κB → IDO1 → kynurenine — but becomes a Tier 2 supply route rather than a direct activation route.
  • B12/folate antagonism has not been tested against covalent modification. Their competitive antagonism is established against reversible ligand binding1. 4-HNE modifies cysteines covalently, which is different chemistry. The matrix reflects this: route 3 is addressed by preventing 4-HNE formation, not by receptor competition after the fact.
  • Flavonol bioavailability is a real problem. Unformulated oral quercetin produces peak plasma concentrations one to two orders of magnitude below the in-vitro IC₅₀ values for PI3K and PARP inhibition. Phytosome and cyclodextrin formulations improve this roughly tenfold to twenty-sevenfold, reaching the low micromolar range at best. Myricetin is worse — under 10% oral bioavailability in rats, with no comparable human pharmacokinetic data. The defence offered here is that pathway-level signalling may operate below enzyme-inhibition thresholds measured in isolated systems, but that is an argument, not evidence.
  • The theabrownin human dose is not achievable from brewed tea. The four-week human trial used 50 mg/kg/day of Pu-erh preparation — several grams of extract for a 75 kg adult. Ordinary tea drinking delivers substantially less. The microbiome and bile-acid arms plausibly respond at lower chronic exposure, but the human FGF19 and bile-acid data were generated at the higher dose.
  • The insulin-resistance branch is the most speculative claim in the paper. That the PKB→AhR arm sits on the preserved side of selective insulin resistance follows from the fed-state mechanism but has not been measured in insulin-resistant human liver. If it sits on the desensitised side instead, that route weakens considerably in exactly the population the paper is about.
  • The Firmicutes/Bacteroidetes claim is contested. The original mouse and human finding has replicated inconsistently; large human datasets find variation between studies exceeding variation between lean and obese individuals within them. It is included as a directional AhR effect, not an obesity marker. The Akkermansia-specific mechanism is on firmer ground than the phylum ratio.
  • Human quantification is thin throughout. The causal AhR work345 is murine. The strongest human anchors are the monozygotic-twin NAD⁺ data20, the theabrownin trial10, the B6 trial17 and the NAFLD promoter-occupancy finding8. None tests the framework as a whole; each tests one arm.
  • Citation gaps to close. The SHP–ARNT interaction paper needs its primary reference pinned rather than described. The human forced-desynchrony sleep data in IV.1 needs a specific citation. The 12-HETE → PPARγ → SCD1 branch is carried over from the linoleic-acid paper's chain and needs its own reference here.

What would break the argument outright is a demonstration that AhR inhibition improves metabolic outcomes only through one of the six outputs rather than the set — that would make AhR a proxy for something narrower rather than the integrator this paper claims it is.

REFS

References

AhR causation & core biology
  1. Kim SY, et al. Vitamin B12 and folic acid alleviate symptoms of nutritional deficiency by antagonizing aryl hydrocarbon receptor. PNAS 2020;117(27):15837–15845. PMID 32571957. Direct competitive antagonism at the AhR ligand-binding domain; deficiency alone induces AhR activity.
  2. Bhattacharya S, et al. Aryl hydrocarbon receptor regulates the fed-state one-carbon cycle and hepatic phosphatidylcholine synthesis. Nat Commun 2018;9:540. Insulin/PKB drives AhR nuclear translocation; FGF15/SHP terminates it in the late fed state; PEMT, AhR and PC elevated in steatosis patients.
  3. Moyer BJ, Rojas IY, Kerley-Hamilton JS, et al. Inhibition of the aryl hydrocarbon receptor prevents Western diet-induced obesity. Toxicol Appl Pharmacol 2016;300:13–24. Oxidised-LDL → TLR2/4 → IDO1 → kynurenine → AhR model.
  4. Rojas IY, Moyer BJ, Ringelberg CS, Tomlinson CR. Reversal of obesity and liver steatosis in mice via inhibition of the aryl hydrocarbon receptor. Int J Obes 2020;44(4):948–963. Reversal of established disease, not prevention only.
  5. Rojas IY, et al. Kynurenine-induced aryl hydrocarbon receptor signaling in mice causes body mass gain, liver steatosis and hyperglycemia. Obesity 2021;29(2):337–349. Kynurenine alone is sufficient; adipocyte AhR ablation abolishes the effect.
  6. Diani-Moore S, et al. NAD⁺ loss, a new player in AhR biology: prevention of thymus atrophy and hepatosteatosis by NAD⁺ repletion. Sci Rep 2017;7:2268. AhR activation lowers NAD⁺ and deactivates SIRT3.
  7. MacPherson L, et al. TiPARP (ARTD14) is a mono-ADP-ribosyltransferase and repressor of AhR transactivation. Nucleic Acids Res 2013;41(3):1604–1621. PARP7 as a canonical AhR target gene.
  8. Kwon E-K, et al. FXR agonist reduces hepatic miR-802 via SHP-mediated repression of AHR. Diabetes 2021;70(2). PMID 33328206. SHP occupancy reduced and AHR occupancy raised at the miR-802 promoter in NAFLD patients and obese mice; obeticholic acid improves insulin resistance and steatosis.
  9. Small heterodimer partner inhibits AhR/ARNT transcriptional activity; ARNT interacts directly with SHP. 2001. primary citation to pin
Bile acids, theabrownin & the microbiome
  1. Huang F, et al. Theabrownin from Pu-erh tea attenuates hypercholesterolemia via modulation of gut microbiota and bile acid metabolism. Nat Commun 2019;10:4971. Human arm: intestinal FXR–FGF15/19 inhibited while hepatic FXR–SHP is activated; serum FGF19 falls; conjugated bile acids accumulate in the ileum.
  2. Pu-erh tea modulates tryptophan metabolism via the gut–liver–brain axis under circadian disruption. J Agric Food Chem 2022. Indole and 5-HT routes boosted; serum kynurenine, xanthurenic acid and 3-hydroxykynurenine reduced. Cinnabarinic acid identified as a Pu-erh constituent affecting food intake and weight gain in circadian-disrupted mice.
  3. Degregori S, Johnson GC, Barber PH, Blumstein DT. Firmicutes and Bacteroidetes contribute to mass gain variation in female obligate hibernators. J Mammal 2024;105(1):2–12. In wild yellow-bellied marmots, higher Firmicutes tracks greater mass gain and higher Bacteroidetes tracks less, across active-season fattening ahead of hibernation.
Circadian NAD⁺, sirtuins & the RET circuit
  1. Nakahata Y, Sahar S, Astarita G, Kaluzova M, Sassone-Corsi P. Circadian control of the NAD⁺ salvage pathway by CLOCK-SIRT1. Science 2009;324(5927):654–657.
  2. Ramsey KM, et al. Circadian clock feedback cycle through NAMPT-mediated NAD⁺ biosynthesis. Science 2009;324(5927):651–654.
  3. Finley LWS, et al. Succinate dehydrogenase is a direct target of sirtuin 3 deacetylase activity. PLoS One 2011;6(8):e23295. PMID 21858060. Thirteen SDHA acetylation sites; SDHA hyperacetylated and SDH activity reduced in SIRT3-null mice.
  4. Schirris TJJ, et al. Stimulation of cholesterol biosynthesis in mitochondrial complex I deficiency lowers reductive stress and improves motor function and survival in mice. Biochim Biophys Acta Mol Basis Dis 2021;1867(2):166062. PMID 33385517.
  5. Dobrzyn P, et al. Stearoyl-CoA desaturase 1 deficiency increases NAD⁺ levels and AMPK activation in skeletal muscle. 2019. PMID 31241768. SCD1 ablation raises AMP/ATP, AMPK phosphorylation, NAD⁺ and SIRT1 activity; overexpression reverses all four.
  6. Nohl H. Generation of superoxide radicals as byproduct of cellular respiration. 1990. PMID 2251677. Foundational: Complex I as a major site, maximal when the NAD pool is highly reduced.
  7. Alpha-lipoic acid decreases cellular NADH with a smaller and slower effect on NADPH. 2002. PMID 11985536. Roughly 30% NADH reduction; reduced by lipoamide dehydrogenase in an NADH-dependent reaction.
  8. Reductive stress in redox biology, and the antioxidant paradox. 2012. PMID 22456698; PMC3401201. Chronic reductive stress induces oxidative stress; chronic antioxidant supplementation can be pro-oxidant.
  9. Tonon AC, Nexha A, Mendonça da Silva M, Gomes FA, Hidalgo MP, Frey BN. Sleep and circadian disruption in bipolar disorders: from psychopathology to digital phenotyping in clinical practice. Psychiatry Clin Neurosci 2024;78(11):654–666. PMID 39210713. Review of circadian clock-gene involvement in mood-disorder pathophysiology — cited here only to show CLOCK:BMAL1 disruption has consequences studied well outside metabolism.
  10. Koves TR, Ussher JR, Noland RC, et al. Mitochondrial overload and incomplete fatty acid oxidation contribute to skeletal muscle insulin resistance. Cell Metab 2008;7(1):45–56. PMID 18177724. Excessive β-oxidation with depleted TCA-cycle intermediates — the mitochondrial-overload signature of NADH backup — drives insulin resistance independent of lipid accumulation.
  11. Houstis N, Rosen ED, Lander ES. Reactive oxygen species have a causal role in multiple forms of insulin resistance. Nature 2006;440(7086):944–948. PMID 16612386. Six independent ROS-lowering interventions each improved insulin resistance in cell models; one confirmed in obese, insulin-resistant mice.
B-vitamins, dairy & human anchors
  1. Effects of pyridoxine supplementation on anthropometric indices, glycaemic control and metabolic markers in overweight and obese women: a randomised controlled trial. 2022. PMID 35715443. 80 mg/day, 8 weeks; fat mass, visceral adiposity index, fasting insulin, HOMA-IR, triglycerides and leptin down; adiponectin up.
  2. Mathematical modelling of vitamin B6 and the kynurenine pathway. 2018. PMID 29401505. PLP as cofactor for kynureninase and kynurenine aminotransferase; PLP restriction limits NAD⁺ production from tryptophan and raises kynurenine, 3-hydroxykynurenine and xanthurenic acid.
  3. Serum vitamin B12 and obesity in US adults. 2019. PMID 31316466. Inverse association after adjustment for age, sex, ethnicity, socioeconomic status, diet and lifestyle.
  4. Jukarainen S, et al. Obesity is associated with low NAD⁺/SIRT pathway expression in adipose tissue of BMI-discordant monozygotic twins. J Clin Endocrinol Metab 2016;101(1):275–283. PMID 26574954. Monozygotic design means the deficit is acquired rather than inherited.
  5. Rappou E, et al. Weight loss is associated with increased NAD⁺/SIRT1 expression but reduced PARP activity in white adipose tissue. J Clin Endocrinol Metab 2016;101(3):1263–1273.
  6. Effect of milk consumption on plasma folate, vitamin B12 and homocysteine. 2000. PMID 10952952. Superior absorption and homocysteine reduction versus matched supplement dosing.
  7. Vitamin B12 bioavailability from milk exceeds synthetic supplement forms, regardless of thermal or filtration processing. Animal Frontiers 2014.
  8. NNT expression in visceral adipose tissue predicts BMI, waist circumference and body-fat percentage; palmitate lowers NNT, NADPH and glutathione in immune cells, and PBMCs from obese subjects show reduced NNT. split into two citations
Oxidised lipids & the CYP1B1 arm
  1. Maayah ZH, El-Kadi AOS. 5-, 12- and 15-hydroxyeicosatetraenoic acids induce cellular hypertrophy in the human ventricular cardiomyocyte RL-14 cell line through MAPK- and NF-κB-dependent mechanism. Arch Toxicol 2015;89(8):1329–1346; with El-Kadi lab, CYP1B1 → mid-chain HETEs, Mol Cell Biochem 2017;429:151.
  2. Oxidised LDL activates AhR and interferes with circadian rhythm-related transcripts. PMC7700619.
AhRKynurenineIDO1TiPARPCLOCK:BMAL1NAMPTSIRT3SDHNNTReverse electron transport4-HNECYP1B1PEMTFXR/SHPTheabrowninQuercetinMyricetinVitamin B6Vitamin B12FolateAlpha-lipoic acidSCD1