Insulin resistance begins in skeletal muscle as a transport defect triggered by intracellular diacylglycerol (DAG) accumulation — not plasma triglycerides — activating novel PKC theta and epsilon, which block IRS-1 phosphorylation upstream of glucose entry.
2
An acute 45-minute bout of moderate-intensity exercise bypasses the entire DAG-PKC block via AMPK-induced GLUT4 translocation independent of PI3-kinase, opening the glucose transport door even in severely insulin-resistant individuals.
3
Muscle insulin resistance drives hepatic fat synthesis through forced diversion of ingested carbohydrate from glycogen to de novo lipogenesis, creating the atherogenic dyslipidemia (high TG, low HDL) and NAFLD that precede overt type 2 diabetes by a decade or more.
4
Metformin lowers gluconeogenesis at clinically relevant concentrations (~50 µM) by inhibiting mitochondrial glycerol-3-phosphate dehydrogenase — not complex I — shifting cytosolic redox to block only lactate- and glycerol-derived gluconeogenesis, which explains its rarity of hypoglycemia.
Protocols
Concrete recipes — what, when, how much, and why
7 items
Moderate-intensity aerobic exercise to acutely bypass muscle insulin resistance
WhatThree 15-minute bouts of stairmaster or equivalent aerobic exercise at approximately 65% VO2max per session; or a single 45-minute continuous bout for acute benefit. The protocol normalizes insulin-stimulated muscle glycogen synthesis via AMPK-independent GLUT4 translocation.
WhenFor anyone with insulin resistance, metabolic syndrome, or high triglycerides/low HDL. Acute benefit begins with the first bout; chronic DAG reduction and improved insulin signaling require weeks.
DoseSingle bout: 45 minutes at moderate intensity. Sustained reversal: three 15-minute sessions/day (~45 min total), six weeks minimum.
For whomInsulin-resistant lean young adults, obese adults, offspring of T2D patients; also effective in T1D for near-insulin-independent glucose control when combined with carbohydrate restriction.
WhyAMPK activated by exercise drives GLUT4 to the plasma membrane independent of the PI3-kinase step that is blocked by DAG-PKCtheta. Chronic exercise also lowers intramyocellular DAG, restoring insulin-dependent signaling.
CaveatsAcute AMPK bypass does not persist beyond hours without ongoing exercise. High-intensity or resistance exercise may provide additional benefit through distinct mechanisms.
Shulman's New England Journal study demonstrated that a single 45-minute bout was sufficient to open up glucose-6-phosphate accumulation and normalize glycogen routing in insulin-resistant offspring of T2D parents. This secondarily reduces hepatic DNL (the liver gets more of the ingested glucose as glycogen, reducing portal hyperinsulinemia and SREBP-1c activation) and acutely reduces plasma triglycerides by reducing the substrate push for VLDL synthesis.
Mechanism
AMPK is activated by a rise in AMP/ATP ratio during exercise. Activated AMPK phosphorylates AS160 (TBC1D4), releasing GLUT4-containing vesicles to fuse with the plasma membrane independently of the PI3K/Akt pathway blocked by DAG-PKCtheta.
Here we can see this was after six weeks of being on a stairmaster three 15-minute bouts at about 65 percent mvo2 max and here we're normalizing insulin stimulated muscle glycogen synthesis and we usually measuring glucose 6-phosphate we've opened up that door of getting glucose into the myocyte
Hypocaloric feeding (~1,200 kcal/day) to reverse hepatic DAG accumulation and restore insulin sensitivity
WhatShort-term caloric restriction to approximately 1,200 kcal/day reverses hepatic fat accumulation, normalizes DAG and acetyl-CoA levels, restores liver insulin sensitivity, and can reverse overt T2D. The mechanism is purely thermodynamic: less total energy relative to the hepatocyte's oxidative capacity.
WhenIn established T2D or significant NAFLD; most effective before advanced fibrosis.
Dose1,200 kcal/day short-term; NMR-measured hepatic normalization demonstrated in Shulman's group and replicated by Roy Taylor's UK primary care trials.
For whomT2D patients with NAFLD; also insulin-resistant patients who cannot exercise. Roy Taylor demonstrated reversal in a UK primary care setting.
WhyHepatic insulin resistance is driven by DAG (activating PKCe, blocking the receptor) and acetyl-CoA (activating pyruvate carboxylase, accelerating gluconeogenesis). Both are downstream of excess lipid flux to the liver. Caloric restriction reduces all upstream inputs.
CaveatsMaintaining the caloric restriction long-term is the primary challenge; weight regain reverses all benefits.
Shulman described reversal of all hepatic abnormalities (DAG normalization, acetyl-CoA reduction, insulin-stimulated glycogen synthesis restoration, normalization of gluconeogenesis flux) in T2D patients after short-term 1,200 kcal/day feeding. Roy Taylor replicated this in primary care with a low-calorie formula diet and demonstrated remission of T2D at 1 year in approximately 50% of participants. Bariatric surgery achieves the same reversal when matched for weight loss, suggesting weight reduction is the operative variable regardless of composition.
We've taken these type 2 diabetics and short-term hypocaloric feeding 1200 calories a day we basically can reverse all these abnormalities through reduction in the topic lipid which we can get into molecular mechanisms and reverse their diabetes and this has now been shown by many many other investigators and most recently roy taylor my colleague who trained with us is now doing this in the primary care clinic back in the uk
Triglyceride/HDL ratio as early insulin resistance screen — flag TG >100 mg/dL and TG/HDL >2
WhatUse fasting triglyceride divided by HDL cholesterol as an early screening signal for muscle insulin resistance and hepatic fat accumulation. Attia's threshold: anything over 100 mg/dL TG is abnormal; TG more than 2x HDL is a strong red flag, not the typical lab reference range of 150-200 mg/dL.
WhenAt any routine metabolic panel. Most relevant in asymptomatic, euglycemic patients who would otherwise be reassured by normal fasting glucose.
DoseOne-time fasting blood draw; repeat 3-6 months after dietary or exercise intervention to track response.
For whomAll adults, especially lean-appearing individuals in their 20s-40s who may have no clinical signs of metabolic disease. One in four lean young Americans is insulin resistant.
WhyMuscle IR forces ingested carbohydrate into hepatic de novo lipogenesis, elevating VLDL-TG export and reciprocally lowering HDL. This atherogenic dyslipidemia precedes fasting hyperglycemia by years. Kit Peterson's 2007 PNAS data showed euglycemic insulin-resistant young adults already had ~100% higher TG than insulin-sensitive peers.
CaveatsThe TG/HDL ratio is a marker, not a direct measure of insulin resistance; HOMA-IR or an oral glucose tolerance test with insulin sampling provides more direct quantification.
Shulman emphasized resetting the normal range: the insulin-sensitive cohort should define normal. The ~80% difference in TG between insulin-sensitive and insulin-resistant lean young adults at euglycemia makes the standard TG cutoff of 150 mg/dL a lagging indicator that only catches patients who are already a decade into the disease process. Attia's clinical practice uses TG >100 and TG/HDL >2 as the threshold for lifestyle intervention.
We view anything over a hundred as abnormal that's a red flag and if your trigs are more than 2x your hdl cholesterol that's a very big red flag although most people would accept triglycerides of three or four if not five times above hdl cholesterol before the sirens would go off
Oral glucose tolerance test (OGTT) with insulin sampling — the clinical screen that fasting glucose misses
What75-100 g oral glucose load with insulin and glucose sampled at 30-minute intervals. A peak glucose of ~200 mg/dL with peak insulin of ~70 microU/mL at 30-60 minutes in an otherwise euglycemic fasting individual (fasting glucose 90, fasting insulin 5) is a positive screen for muscle insulin resistance, potentially a decade before diabetes diagnosis.
WhenAny patient with TG/HDL >2, family history of T2D, unexplained weight gain, or NAFLD on imaging — even with completely normal fasting glucose.
For whomAsymptomatic lean individuals at metabolic risk; offspring of T2D parents (Joslin study: insulin resistance in this group was the single best predictor of future T2D).
WhyFasting glucose remains normal for years while beta cells compensate with 2-3x normal insulin secretion. The OGTT unmasks impaired postprandial glucose disposal that NMR studies confirm is caused by a 50% reduction in muscle glycogen synthesis flux.
CaveatsA crude test — the glucose and insulin AUC reflect the net result of multiple processes. NMR-based glycogen synthesis measurement remains the gold standard but is research-only.
Shulman screens approximately 1,000 volunteers to identify the bottom quartile of insulin sensitivity. Even within lean, non-smoking, sedentary young adults (BMI 22-23), the bottom quartile shows muscle glycogen synthesis down 50% under matched insulin and glucose concentrations, with both glucose-6-phosphate and intracellular glucose reduced by NMR — confirming the block is at transport, not downstream. Joslin Center research confirmed that insulin resistance (not family history alone) predicted T2D progression: two T2D parents plus insulin resistance equaled the highest risk group.
You challenge them with 75 to 100 grams of glucose but say 30 or 60 minutes later their fasting glucose is 200 their insulin is 70 we call that insulin resistance and we impute from that that something has broken down in the pathway that prevents their muscle from taking in glucose
Reduce dietary fructose and sucrose to blunt hepatic de novo lipogenesis
WhatRestrict sucrose, high-fructose corn syrup, and fruit-juice fructose specifically. Fructose is funneled almost entirely into the liver's DNL pathway (unlike glucose, which can be distributed to muscle glycogen), disproportionately upregulating fat synthesis.
WhenIn insulin-resistant patients with elevated TG, NAFLD on imaging, or elevated hepatic fat on MRI. Also as primary prevention in euglycemic individuals.
For whomAnyone with NAFLD, high TG, or family history of T2D. Specifically relevant to individuals consuming high-sugar processed foods or sweetened beverages.
WhyFructose bypasses phosphofructokinase regulation and enters hepatic metabolism unregulated, driving acetyl-CoA and DNL with no brake. Combined with muscle insulin resistance (which prevents glucose from entering muscle glycogen), fructose creates a liver-specific substrate overload.
CaveatsFructose restriction alone will not reverse established NAFLD without a caloric deficit; it reduces one major driver but does not address adipose tissue lipolysis or overall caloric balance.
DNL is minimal in the fasting state or with moderate carbohydrate intake in insulin-sensitive individuals — consistent with Hellerstein's 1994 studies. In insulin-resistant individuals on high-fructose diets, DNL becomes a significant contributor to hepatic TG accumulation. Fructose specifically lacks the glycogen-storage buffer that glucose has — when muscle glycogen synthesis is blocked by insulin resistance, ingested fructose has nowhere to go but fat synthesis.
Fructose basically gets funneled into the liver into the dnl pathway it's ubiquitous you can push dnl to be significant and it is a significant contributor to metabolic fatty liver disease and it's upregulated with peripheral sensitivity
GLP-1 agonist therapy for NAFLD reversal via reduced energy intake and weight loss
WhatGLP-1 receptor agonists (semaglutide, tirzepatide class) reduce appetite centrally, producing 10-15% weight loss, partial hepatic fat reduction, and improvement in TG/HDL dyslipidemia — though not full normalization of hepatic insulin sensitivity.
WhenFor T2D or obese insulin-resistant patients who cannot achieve adequate weight loss through diet and exercise alone.
DosePer standard prescribing: semaglutide 0.5-2.4 mg weekly; tirzepatide 2.5-15 mg weekly. Benefit persists only with continued use.
For whomInsulin-resistant or T2D patients with BMI >=27 and at least one weight-related comorbidity. Not indicated for lean insulin-resistant individuals.
WhyGLP-1 agonists work via central nausea/appetite suppression at therapeutic doses. Weight loss reduces hepatic fat flux and partially reverses the thermodynamic imbalance driving NAFLD.
CaveatsDoes not fully normalize hepatic insulin sensitivity at the molecular level; TG/fat reduction is incomplete. Benefits reverse on discontinuation.
Shulman explicitly frames GLP-1 agonists as working through energy intake reduction, not through any direct insulin-sensitizing mechanism. Unlike a liver-targeted mitochondrial uncoupler, GLP-1 agonists do not address the thermodynamic imbalance inside the hepatocyte — they reduce the input side. SGLT2 inhibitors add approximately 400 kcal/day urinary glucose loss but plateau after several weeks with minimal hepatic fat reduction.
Glp-1 agonists i believe are having its major effect is weight loss and they are what they are they do accomplish reversal fatty liver to some degree they don't normalize but it does come down in the right direction
Intralipid + heparin FFA elevation challenge (research model for lipid-induced IR; clinical corollary: avoid prolonged elevated plasma FFA)
WhatRaising plasma free fatty acids to ~1.5 mM for 3-4 hours via triglyceride emulsion (Intralipid) plus low-dose heparin (to activate LPL) produces insulin resistance equal to obesity or T2D within hours. The clinical corollary: any state that chronically elevates plasma FFA (obesity, physical inactivity, high-fat hypercaloric feeding) drives DAG-PKCtheta activation and insulin resistance.
WhenResearch model; clinical insight: avoid prolonged fasting FFA elevation. Interventions that lower plasma FFA (exercise, caloric restriction, GLP-1 agonists via weight loss) will reduce DAG accumulation.
DoseResearch protocol: 3-4 hours of FFA elevation to produce maximal insulin resistance. Chronic exposure produces persistent DAG accumulation.
For whomResearch tool. Clinical implication: patients with any condition elevating circulating FFA chronically are at risk for progressive muscle insulin resistance via DAG accumulation.
WhyDirect proof that intracellular FFA metabolites (DAG, acyl-CoA) — not triglycerides themselves — cause the transport block. Glucose-6-phosphate and intracellular glucose both fall (not rise, as Randle predicted), confirming the defect is at GLUT4 translocation, not downstream glycolysis.
Mechanism
Elevated FFA leads to acyl-CoA accumulation in the myocyte exceeding mitochondrial oxidation capacity, building sn1,2-DAG in the plasma membrane, recruiting and activating PKCtheta, causing serine phosphorylation of IRS-1 (blocking tyrosine phosphorylation), reducing PI3K activation, and impairing GLUT4 translocation.
We gave them an intra-lipid infusion just raised plasma fatty acids for three to four hours and found that after three to four hours we can make them as insulin resistant as anyone with type 2 diabetes... it's due to this block in glycogen synthesis and it's the same block it's that block and transport
What's new
Personal practice updates, fresh positions, predictions
7 items
DAG-PKCe threonine-1150 phosphorylation: the molecular pin of hepatic insulin resistance
~1 h 20 min
Using untargeted phosphoproteomics, Shulman's lab identified that activated PKCe phosphorylates threonine-1150 on the insulin receptor kinase activation loop — one amino acid from the tyrosines required for catalytic activation — directly inhibiting receptor kinase activity. This threonine is conserved from humans to fruit flies, suggesting it is a protective starvation mechanism co-opted by obesity.
Why this matters: First molecular proof that PKCe acts at the insulin receptor itself, not downstream in the cascade — and the evolutionary framing explains why a mechanism designed to preserve glucose for the brain during starvation becomes the engine of type 2 diabetes under chronic overnutrition.
Background
Prior models placed lipid-induced insulin resistance somewhere between the receptor and PI3K without pinpointing the exact residue. Randall's earlier citrate/PFK model predicted glucose-6-phosphate should rise with fat-induced IR; Shulman's NMR data showed it fell, ruling out that mechanism.
Jesse Reinhardt and Max Peterson at Yale performed untargeted phosphoproteomics using purified insulin receptor and activated PKCe. They found phosphorylation at threonine-1150 (mouse homolog of human T1160), one residue from tyrosines Y1158 and Y1162 that must be phosphorylated to open the activation-loop door for IRS-1 docking. Knock-in mice where T1150 was replaced with alanine (non-phosphorylatable) were completely protected from high-fat-diet-induced hepatic insulin resistance despite identical liver fat and DAG accumulation — proving the threonine is both necessary and sufficient. The same mutation has been conserved from Homo sapiens to Drosophila; a Mexican cave fish that lives in a near-permanent starvation state developed a spontaneous insulin receptor mutation producing constitutive hepatic insulin resistance — presumably adaptive.
Jesse reinhardt who is our collaborator in mass spec maven identified using purified receptor purified pkc epsilon that when you add activated epsilon to the receptor you phosphorylate this threonine and that got us very excited because kali that's one amino acid away from these two tyrosines that are required for activation
Also said
“Mice are perfectly normal normal chow normal insulin sensitivity nothing near normal size normal growth but when max fed these mice a high fat diet the wild type mice get profound hepatic insulin resistance... when you simply mutate that three nutrient alone now you have perfectly normal hepatic insulin sensitivity as reflected by insulin's ability to suppress hepatic glucose production and this is despite the same amount of liver fat same amount of liver dags in the liver”— The knockout experiment proving T1150 is necessary: same fat, same DAGs, but protected from IR when threonine cannot be phosphorylated.
sn1,2-DAG in the plasma membrane — not total DAG — is the specific IR-inducing lipid species
~1 h 15 min
A Cell Metabolism paper from Shulman's group showed that measuring total intracellular DAG misses the active species. It is specifically the sn1,2 stereoisomer located in the plasma membrane — not in the cytosol, lipid droplet, ER, or mitochondria — that translocates PKCe and drives insulin resistance. Other isoforms and compartments are irrelevant or inert.
Why this matters: Resolves why some studies found DAG uncorrelated with insulin resistance: they measured the wrong compartment or stereoisomer. Triglycerides track DAG but are inert — the apparent paradox of TG correlating with IR is fully explained by TG being a marker for the plasma-membrane sn1,2-DAG pool.
Background
For decades researchers debated whether ceramide, total DAG, or triglycerides were the causative lipid intermediates. Shulman's group had long argued DAG but could not fully explain why TG sometimes dissociated from IR.
The paper measured three DAG stereoisomers (sn1,2; sn1,3; sn2,3) across five intracellular compartments (plasma membrane, cytosol, lipid droplet, ER, mitochondria). Only plasma-membrane sn1,2-DAG correlated with PKCe translocation and insulin resistance. The hydroxyl group on the third carbon of sn1,2-DAG sits in the cytoplasm after the two fatty acids anchor into the bilayer, creating the cytoplasmic hook that recruits PKCe. This was confirmed in the fat cell as well: same pathway, same isoform selectivity, generalizing across all three major insulin-responsive tissues.
Just to summarize this paper that just came out in cell metabolism we were able to show by measuring the three different stereoisomers of diazoglycerols it's really the sn12 isoform and measuring these different isoforms in five different intracellular compartments the plasma membrane the cytosol lipid droplet er and the mitochondria it's really specifically the sn12 isoform in the plasma membrane that's important
Muscle IR precedes and causes hepatic fat — humans and rodents progress in opposite directions
~1 h 05 min
In lean 20-year-old Yale undergraduates who are insulin resistant (bottom quartile of glucose tolerance), NMR shows complete muscle glycogen synthesis block with zero liver abnormality. Only once fatty liver develops do liver abnormalities appear. Rodent models go in reverse — liver fat and hepatic IR develop first. This means the majority of mechanistic rodent IR literature applies to the wrong tissue in the wrong order.
Why this matters: Overturns the assumption that NAFLD drives peripheral IR in humans. In the human progression, hyperinsulinemia from muscle IR redirects ingested carbohydrate to hepatic de novo lipogenesis first — the liver becomes a victim, not an initiator.
Background
Kit Peterson's 2007 PNAS paper showed that lean, euglycemic insulin-resistant young adults already had 2.3-fold higher liver TG and greater than twofold higher de novo lipogenesis after two high-carbohydrate meals — all without any detectable hyperglycemia.
The progression Shulman traces: (1) muscle insulin resistance from DAG-PKCtheta blockade of GLUT4 translocation; (2) compensatory hyperinsulinemia (2-3x normal peripheral; 5-6x normal in portal vein); (3) SREBP-1c activation by portal hyperinsulinemia; (4) upregulation of DNL enzymes; (5) hepatic TG accumulation and NAFLD; (6) hepatic DAG-PKCe insulin resistance; (7) loss of insulin suppression of gluconeogenesis via acetyl-CoA rise; (8) fasting hyperglycemia and overt T2D. An estimated 25% of lean, asymptomatic US young adults are already at stage 1-2.
We don't see liver abnormalities in these young 20 year olds it's all muscle and maybe a little bit of the fat cell which we'll come to at the end but it's the muscle there's no alterations in the liver until they get fatty liver once they get fatty liver then we see both insulin resistance in liver and insulin resistance and muscle a very important distinction between humans and rodents we've studied both models quite extensively rodents develop insulin resistance in the reverse direction they get liver fat first liver insulin resistance and then muscle
Acetyl-CoA — not FOXO/Akt/PEPCK transcription — is the dominant acute regulator of hepatic gluconeogenesis
~1 h 40 min
Insulin suppresses gluconeogenesis within five minutes — far faster than transcriptional mechanisms can operate. Shulman's data show that insulin's primary antigluconeogenic action is indirect: by suppressing peripheral lipolysis, it reduces fatty acid flux to the liver, lowering hepatic acetyl-CoA, the allosteric activator of pyruvate carboxylase. Liver from poorly controlled T2D patients showed no upregulation of PEPCK or glucose-6-phosphatase protein expression despite doubled gluconeogenesis rates.
Why this matters: Challenges the canonical biochemistry textbook view that insulin turns off gluconeogenesis via Akt/FoxO1 and PEPCK transcriptional repression. FOXO and Akt knockouts can still suppress gluconeogenesis acutely — this model explains why.
Background
The direct hepatic model dominated for decades: insulin binds liver receptor, Akt phosphorylates FoxO1 causing nuclear exclusion, PEPCK and G6Pase transcription is repressed. But knockout experiments failed to abrogate acute gluconeogenic suppression.
Shulman's model: insulin in the periphery (not liver) acts on adipocytes to suppress lipolysis, reducing glycerol and fatty acids to liver, reducing beta-oxidation and acetyl-CoA, reducing pyruvate carboxylase allosteric activation, lowering gluconeogenesis. In T2D, peripheral lipolysis is accelerated (insulin resistance in fat cell) — fatty acid flux to liver doubles, acetyl-CoA doubles, pyruvate carboxylase flux doubles, fasting hyperglycemia results. The glycerol substrate push from lipolysis adds another 10-15% to gluconeogenesis. Transcriptional mechanisms may play a role over days-weeks but do not explain 5-minute glucose suppression.
We can turn off gluconeogenesis within five minutes and that's much faster than you'd expect from transcriptional translational mechanisms... it's really insulin putting the break on peripheral lipolysis less fatty acid delivery to liver less generation of acetylcholine we've shown this the more fatty acids that flux the liver tract almost perfectly with acetyl-coa content less pyruvate carboxylase activity
Also said
“We got liver from patients with poorly controlled diabetes so when patients go in for a roux and y or bariatric surgery the surgeon you can take a little piece of liver... i expected pepsi k and 6 phosphatase and fructose 1 6 by phosphatase all to be up regulated two to three fold in the poorly controlled diabetic... and we found no relationship between protein expression of these enzymes gluconeogenic enzymes and at least fasting glucose”— Direct human tissue data ruling out PEPCK/G6Pase transcriptional upregulation as the driver of fasting hyperglycemia in T2D.
Liver-targeted mitochondrial uncoupling as a NAFLD/NASH therapeutic — 100-fold safer window than systemic DNP
~2 h 30 min
Shulman's lab is developing a liver-targeted mitochondrial uncoupler that forces hepatocytes to burn more fat to generate the same ATP, melting intrahepatic lipid and reversing NAFLD/NASH/fibrosis and T2D in rodent and non-human primate preclinical models — without the systemic hyperthermia that killed people on DNP in the 1930s. An IND has been filed.
Why this matters: DNP is the concept that works (confirmed weight loss in millions in the 1930s) but kills at too-high doses through whole-body hyperthermia. Liver-targeting confines uncoupling to the organ where the pathological fat accumulates, expanding the therapeutic window ~100-fold.
Background
DNP was used as a weight-loss drug over-the-counter in the 1930s until the newly created FDA pulled it after deaths from hyperthermia. The concept was abandoned for 80 years.
The liver-targeted compound reverses hepatic steatosis, NASH, liver fibrosis, insulin resistance, and diabetes in Zucker diabetic fatty rat models, as well as safety/efficacy in non-human primates. Mechanistically: less hepatic fat reduces DAG, which reduces PKCe activation, restoring insulin receptor function, while lower acetyl-CoA reduces pyruvate carboxylase flux and gluconeogenesis. Secondary effect: less VLDL-TG export lowers plasma TG, raises HDL, lowers LDL — cardioprotective vs. the lipid-raising profile of many competing NASH drugs. Unlike GLP-1 agonists (energy intake) or SGLT2 inhibitors (urinary glucose loss), this targets the thermodynamic imbalance directly inside the hepatocyte.
If we could just melt the fat away within a liver specific manner maybe we can have that beneficial effect without the toxicity and so in a series of studies we were able to show proof of concept that by simply uncoupling the liver you could avoid hyperthermia and all the toxicities that have typically been associated with the parent compound dnp and increase the therapeutic window every drug has a therapeutic window even aspirin and tylenol by a hundred fold
AMPK-induced GLUT4 translocation bypasses the entire DAG-PKC-PI3K block with a single exercise bout
~1 h 55 min
A single 45-minute bout of 65% VO2max exercise in insulin-resistant young adults normalized muscle glycogen synthesis measured by 13C NMR — opening the glucose-6-phosphate block completely. AMPK activation causes GLUT4 translocation independent of PI3-kinase, short-circuiting the insulin resistance at the exact point where DAG-PKCtheta causes the blockade.
Why this matters: Exercise is not merely good for metabolism — it mechanistically bypasses the molecular defect causing insulin resistance in muscle and acutely redirects ingested carbohydrate back to glycogen, reducing hepatic DNL and plasma TG within a single session.
Background
The 6-week stairmaster study (three 15-minute bouts/day, 65% VO2max) in insulin-resistant offspring of T2D parents showed normalization of insulin-stimulated muscle glycogen synthesis. The follow-up New England Journal study showed a single bout was sufficient.
AMPK activated by energy-sensing during exercise causes GLUT4 vesicle translocation to the plasma membrane via a pathway that does not depend on PI3-kinase — routing around the IRS-1 block. Chronically, exercise also reduces intramyocellular DAG accumulation (melting the lipid that triggers PKCtheta), so long-term exercise provides dual benefit: improved insulin-dependent signaling and preserved AMPK-independent uptake. The T1D phenotype — near-insulin-independent glucose control with brisk walking — shows the maximal extent of the bypass when both arms are maximized.
Here we can see this was after six weeks of being on a stairmaster three 15-minute bouts at about 65 percent mvo2 max and here we're normalizing insulin stimulated muscle glycogen synthesis... i think molecular explanation for this is this protein called ampk which we can talk about gets activated with exercise and that has been shown to cause bohr translocation independent of piatria kinase and so we're kind of short-circuiting that block with exercise
Metformin works via glycerol-3-phosphate dehydrogenase (not complex I) at clinically relevant concentrations
~2 h 45 min
All prior mechanistic studies on metformin used millimolar concentrations to inhibit mitochondrial complex I. Clinical plasma levels are 30-50 microM — 10-20x lower. Shulman's lab showed metformin at 50-100 microM inhibits mitochondrial glycerol-3-phosphate dehydrogenase, raising cytosolic NADH/NAD+, which selectively blocks gluconeogenesis only from lactate and glycerol — not from alanine or DHAP.
Why this matters: Directly challenges the consensus mechanism printed in every pharmacology textbook. The substrate-selectivity of the inhibition is a testable prediction that distinguishes the two models and explains metformin's clinical safety profile.
Background
Guanidine compounds inhibit complex I at high concentrations. Phenformin-associated lactic acidosis was explained by complex I inhibition; metformin's much lower risk was assumed to reflect potency differences, not a mechanistic difference.
Shulman's in vitro and in vivo studies at 50-100 microM showed metformin specifically inhibited gluconeogenesis from glycerol and lactate (both NADH-dependent conversions) but not from alanine or DHAP. This explains: lactic acidosis (lactate-to-pyruvate is the blocked step); rarity of hypoglycemia (alanine-derived gluconeogenesis remains intact); poor efficacy in healthy euglycemic individuals (gluconeogenesis not driven by excess glycerol/lactate flux). Shulman remains undecided on metformin for longevity in healthy insulin-sensitive exercisers: the cytosolic redox shift may or may not be beneficial.
The effects that i do think are clinically relevant that we have observed at 50 and 100 micromolar of metformin are really on the enzyme glycerol 3-phosphate dehydrogenase the mitochondrial isoform that is required to move the protons from outside to inside the mitochondria and when you inhibit this enzyme nadh goes up nad goes down when you have this increase in the cytosolic redox you can't get lactate to pyruvate and you can't get glycerol to dhap
Also said
“Metformin at these clinically relevant doses and concentrations only inhibit gluconeogenesis from glycerol and lactate it doesn't inhibit it from alanine or dhap or anything that does not depend on the cytosolic redox state this also explains why we rarely see clinically hypoglycemia on patients treated with metformin because there's these alternative gluconeogenic substrates that can come in alanine can keep coming out”— The in vivo substrate-selectivity prediction distinguishing G3PDH from complex I inhibition, and explaining metformin's safety profile.
Recommendations
Products, supplements, and tools mentioned in the episode
3 items
Weight loss by any sustainable method to reverse NAFLD and T2D
Practice
Shulman's clinical bottom line: the specific diet (low-carb, low-fat, Mediterranean, intermittent fasting) matters less than the caloric deficit and the patient's ability to adhere long-term.
Shulman explicitly rejects dietary dogma: whatever works to everyone is so different. Bariatric surgery achieves the same metabolic reversal as 1,200 kcal/day feeding when matched for weight lost — confirming that composition is secondary to caloric deficit in established metabolic disease. The challenge is not knowing what to do but finding an adherence strategy the individual can sustain. Attia adds that low-carbohydrate diets improve postprandial hyperinsulinemia and TG more rapidly, and intermittent fasting reduces total caloric intake for many adherents.
vs alternatives
Low-carbohydrate diets (Attia's clinical preference) improve postprandial hyperinsulinemia and TG more rapidly; intermittent fasting reduces total caloric intake for many adherents; GLP-1 agonists provide pharmacological appetite suppression. All converge on the same thermodynamic solution: less hepatic lipid.
I say you find something whatever works for you stick with it that's the challenge because we're very successful in the short term getting patients to lose weight the unfortunate part is they're able to get the weight off and then three months later six months later they come back to the office and they're right back where they started
Fructose and added-sugar restriction as first dietary intervention for NAFLD
Practice
Among dietary modifications, fructose/sucrose restriction is the most mechanistically justified first step for NAFLD: fructose bypasses hepatic regulatory checkpoints and feeds directly into DNL without a glycogen buffer.
Shulman's NMR studies showed de novo lipogenesis increased greater than twofold in insulin-resistant euglycemic young adults after two high-carbohydrate milkshakes. Fructose is distinctively problematic because it enters hepatic metabolism past the PFK regulatory point. Sucrose (50% fructose) and HFCS are ubiquitous in processed foods and sweetened beverages. Even without overall caloric restriction, reducing fructose intake reduces DNL substrate, lowering VLDL-TG export and slowing the lipid accumulation that drives DAG-PKCe insulin resistance.
Especially if you track it chronically in patients who are continuously high carb feeding especially high sucrose high fructose corn syrup we want to get into that but fructose basically gets funneled into the liver into the dnl pathway it's ubiquitous you can push dnl to be significant
OGTT with insulin sampling as metabolic screening — not just fasting glucose
Service
Attia's clinical protocol: a 75 g OGTT with 30-minute insulin and glucose sampling to catch insulin resistance in euglycemic patients who would otherwise be told their labs are normal.
Standard clinical labs catch insulin resistance only when it has progressed to impaired fasting glucose (>100 mg/dL) or HbA1c elevation. Shulman's NMR data show that insulin-resistant individuals have already developed atherogenic dyslipidemia and doubled hepatic DNL while fasting glucose remains at 90 mg/dL. The OGTT with insulin sampling — even without NMR — reveals the beta-cell hypercompensation (peak insulin 2-3x normal) that is the physiological signature of insulin resistance a decade before clinical diabetes.
You challenge them with 75 to 100 grams of glucose but say 30 or 60 minutes later their fasting glucose is 200 their insulin is 70 we call that insulin resistance and we impute from that that something has broken down in the pathway that prevents their muscle from taking in glucose
Carbon-13 / Phosphorus-31 NMR spectroscopy for in vivo metabolic flux measurement
Tool Sponsored · disclosed
The only non-invasive method to measure intracellular glycogen synthesis flux, glucose-6-phosphate, and intracellular free glucose in living humans without ionizing radiation. Enabled every mechanistic finding in this episode.
DisclosureShulman's lab at Yale pioneered this technique and directly benefits from its adoption.
13C NMR with C1-labeled glucose lets Shulman track the label as it enters the myocyte or hepatocyte and is incorporated into glycogen — measuring flux, not just concentration. 31P NMR measures glucose-6-phosphate from natural-abundance phosphorus to localize the transport block vs. downstream enzymes. Proton NMR (same physics as clinical MRI) quantifies intramyocellular lipid vs. extracellular fat. In combination, the three modalities answered every major mechanistic question about insulin resistance in humans that was previously only tractable in animal tissue.
Using this approach we found fat inside the muscle was the best predictor for this block and transport in all of our volunteers young people old people children sedentary individuals fat inside the cells the best predictor for instant resistance
Lines worth pulling out — contrarian, specific, or perfectly phrased
6 items
Insulin resistance is probably one quarter of our population and one half of our population has it perfectly asymptomatic you don't know you have it we can test for it using sophisticated tools that we can talk about but it's a very common phenomenon
Sets the scale: 25-50% of the US population is insulin resistant right now, most undiagnosed — making it the single most prevalent undetected pathological state in Western medicine.
It's really the sn12 isoform in the plasma membrane that's important — if you just measure total dags you may easily miss this
The key methodological reason decades of insulin resistance research produced conflicting findings on DAG: the wrong compartment and wrong stereoisomer were measured.
We can turn off gluconeogenesis within five minutes and that's much faster than you'd expect from transcriptional translational mechanisms
The single most concise argument overturning the PEPCK/FoxO textbook model of insulin action in the liver.
The reason i think it exists is it's protective for us during starvation when you starve this is true pretty much in all mammals mice rats and humans when we starve we get fatty liver... by promoting hepatic insulin resistance we're promoting glucose in circulation for basically the cns to operate and so that to me is why that threonine is preserved all the way from humans to fruit flies
The evolutionary framing of insulin resistance: not a disease but a starvation-survival mechanism hijacked by overnutrition — from molecular biology (conserved threonine) to evolutionary biology (Mexican cave fish).
I would still predict if you do careful studies of measuring rates of lipolysis by definition they will have insulin resistance in the fat cell and that's because the reason they're holding on to fat they're not happy about it is because it's at hyperinsulinemia so their insulin concentrations are two to three fold
Clarifies the paradox that obese people retain fat despite high insulin: they ARE insulin resistant in adipocytes, but compensatory hyperinsulinemia overcomes the resistance — masking the dysfunction.
If i had to pick one organ to target it's the liver as important as muscle insulin resistance is at the very beginning if we actually want to reverse the disease and make the biggest impact if i had to pick one or it's the liver
Shulman's therapeutic priority hierarchy: the liver is the reversal target even though muscle is the initiating organ.
Sign in to share feedback
Tell us if this brief hit the mark or missed it — feedback feeds back into the next iteration of the prompt.
Reading is free for everyone. A free account adds the personal layer: save protocols, follow experts, and see how the other experts weigh in on this same topic.
Educational summary of the cited expert source — not medical advice. Open the source recording linked above and consult a qualified physician before acting on any protocol.
