Insulin Resistance and Glucolipid Metabolic Models

Desensitisation and substrate overload produce different states from the human disease.

Three different things are called the same thing

A cultured cell treated for 24 hours is not a person with insulin resistance. Three distinct perturbations are used interchangeably in the literature, and they produce different molecular states with different reversibility.

ModelProtocolMolecular stateReversible
Chronic hyperinsulinaemia100 nM insulin, 16–24 hReceptor and IRS-1 downregulation; reduced signal per cellYes, within 24–48 h of insulin-free medium
Lipid overloadPalmitate 100–400 µM, 16–24 hDAG–PKCθ activation, ceramide, ER stress; kinase-mediated serine phosphorylation of IRS-1Slowly, and incompletely
InflammatoryTNF-α 10 ng/mL or macrophage-conditioned medium, 24 hSOCS-mediated IRS degradation, phosphatase activationPartly, after cytokine removal

The first row is desensitisation: fewer receptors, same machinery. The second is signal interference: normal receptor number, inhibited transduction. A study that reports "reduced p-AKT" without distinguishing them has reported a phenotype, not a mechanism.

Important distinction. Desensitisation is homologous and adaptive — it is what any cell does when exposed to a supraphysiological hormone concentration. It is a poor model of the human disease, in which insulin is elevated for years but signalling is impaired at normal receptor density. If the claim concerns human insulin resistance, the lipid and inflammatory models are the better-matched perturbations, and the hyperinsulinaemia model should be described as desensitisation.

The readout decides the conclusion

The single most common error in this literature is comparing phospho-AKT at one insulin concentration. That measurement conflates two independent parameters.

ParameterWhat changesHow it appears on a dose–responseInterpretation
Sensitivity (EC₅₀)Affinity or coupling of the receptor-proximal stepsCurve shifts right, same maximumTrue resistance to insulin
Responsiveness (Vmax)Total signalling capacity or GLUT4 contentCurve flattens, EC₅₀ unchangedReduced capacity, not reduced sensitivity
BothCommon in chronic modelsShift plus flatteningReport both

A right-shifted curve with an unchanged maximum says the cell needs more insulin for the same effect. A flattened curve says it cannot reach the same effect at any concentration. These have different implications for a drug: a sensitiser should shift the curve, a GLUT4-inducing agent will raise the maximum.

Readout note. A full dose–response requires 5–6 concentrations spanning at least three log units, a fixed stimulation time (typically 5–10 minutes for p-AKT), and serum starvation before stimulation. Without starvation, basal phosphorylation is high and the fold-change is compressed.

Which phosphorylation site

ReadoutDepends onConfounded by
p-AKT S473mTORC2mTOR inhibition, amino acid shortage, growth factor withdrawal
p-AKT T308PDK1Largely insulin-specific in this context
p-IRS-1 (Ser307/312)Stress kinasesAny stress, including handling
Total IRS-1, total IRProtein stabilityTranscriptional change over days
GLUT4 at the membraneTranslocationTotal GLUT4 may be unchanged, so a lysate blot shows nothing
2-NBDG uptakeTransporter-mediated influxProbe saturation; not linear with transport rate

Caveat on S473. Because S473 phosphorylation requires mTORC2, any treatment that inhibits mTOR or depletes amino acids lowers it without touching insulin signalling. A "palmitate-induced insulin resistance" phenotype measured only at S473 can therefore be an mTOR effect, particularly at concentrations that also reduce proliferation.

Caveat on total GLUT4. Translocation is the insulin-regulated step, and total cellular GLUT4 often does not change. A whole-cell lysate blot is the wrong measurement; membrane fractionation, surface biotinylation or a pH-sensitive surface reporter is the right one.

Tissue matters, and the readouts do not transfer

Cell typeInsulin-responsive processAppropriate readout
Skeletal myotube (C2C12, primary)Glucose uptake, glycogen synthesis2-DG uptake; glycogen mass
Adipocyte (3T3-L1, human ASC)Glucose uptake, lipolysis suppression2-DG uptake; glycerol release ± isoproterenol
Hepatocyte (primary, HepG2)Gluconeogenesis suppression, lipogenesisGlucose output; de novo lipogenesis
EndothelialCapillary recruitmentNot routinely modelled

A myotube does not perform gluconeogenesis, and a hepatocyte does not express GLUT4 as its principal insulin-responsive transporter. Applying an adipocyte readout to a hepatocyte model is not a simplification; it is a different experiment.

Important distinction. In the liver, insulin resistance is selective: glucose output becomes resistant while lipogenesis remains sensitive. A hepatocyte model therefore has to measure both arms, because a treatment that reduces p-AKT uniformly will look like "insulin resistance" while producing the opposite of the human hepatic phenotype.

Why "glucolipid" and not two separate models

High glucose and high insulin together activate SREBP-1c and increase de novo lipogenesis; palmitate in turn generates DAG and ceramide that interfere with insulin signalling. The two arms form a loop.

ModelWhat it testsWhat it cannot show
High glucose aloneHexosamine flux, osmotic and methylglyoxal stressLipogenic drive, which requires insulin
High insulin aloneDesensitisationSubstrate-driven lipotoxicity
Palmitate aloneLipid-mediated signal interferenceHyperinsulinaemic drive
Glucose + insulin + palmitateThe loop as it exists in vivoWhich arm initiated it

The fourth row is the one closest to the disease and the one least able to attribute causation. That is the trade-off, and it should be stated rather than resolved.

What a cell model cannot claim

  • • Systemic compensation. Hyperinsulinaemia in vivo maintains normoglycaemia for years; a dish has no β cell, no incretin axis and no counter-regulation.
  • • The chronic set point. Two weeks in culture is a long experiment; the human disease develops over a decade.
  • • Tissue crosstalk. Adipose-derived fatty acids reaching liver and muscle are the mechanism in vivo and are absent in a single-cell-type model.
  • • A therapeutic effect. Ameliorating palmitate-induced signalling in a line is not evidence that a compound improves insulin sensitivity.

Failures that recur

  • • Supraphysiological insulin as "activation". 100 nM is roughly ten times fasting insulin; at that concentration, differences in sensitivity are saturated away.
  • • One concentration, one timepoint. Reported as "reduced insulin signalling".
  • • No osmotic control for high-glucose conditions (mannitol or L-glucose).
  • • No vehicle control for the BSA used to conjugate palmitate; BSA alone changes free fatty acid availability.
  • • Confluence ignored. Glucose uptake per cell falls sharply at high density, which mimics resistance.
  • • Passage number ignored. HepG2 and C2C12 drift in metabolic phenotype with passage.

Where this connects

  • • Glucose pillar supplies the flux rules: uptake and lactate release are different measurements.
  • • Lipid pillar supplies the species rules: esterification is protective, and total triglyceride is uninformative.
  • • Disease pillar supplies the tissue-and-timescale frame: which of the five insulin-responsive processes is being modelled.
  • • Downstream, the operational protocols (concentrations, timings, controls) are in the companion article on inducing insulin resistance in vitro.

Scientific interpretation and experimental design

  • • Desensitisation is not resistance. Chronic insulin downregulates receptors; the human disease does not.
  • • Report EC₅₀ and maximum. A single concentration cannot separate them.
  • • S473 is mTORC2-dependent. Use T308 for receptor-proximal signalling.
  • • GLUT4 translocation needs a surface measurement. Total protein is usually unchanged.
  • • Hepatic insulin resistance is selective. Measure output and lipogenesis separately.
  • • Match the readout to the tissue. Myotube, adipocyte and hepatocyte answer different questions.
  • • Name the loop, do not claim the initiator. Combined models cannot attribute causation.

Selected references for further reading

  • • Petersen MC, Shulman GI. Mechanisms of insulin action and insulin resistance. Physiological Reviews. 2018;98:2133–2223.
  • • Czech MP. Insulin action and resistance in obesity and type 2 diabetes. Nature Medicine. 2017;23:804–814.
  • • Samuel VT, Shulman GI. Mechanisms for insulin resistance: common threads and missing links. Cell. 2012;148:852–871.
  • • Manning BD, Toker A. AKT/PKB signaling: navigating the network. Cell. 2017;169:381–405.

Related Products

Product name Cat.No. Species Gene ID
PBRM1 Knockout A-549 Cell Line EDJ0001-K01 Human 55193 Details Get a Quote
YTHDC1 Knockout A-549 Cell Line EDC07652 Human 91746 Details Get a Quote
TP53 Knockout HCT 116 Cell Line EDC07854 Human 7157 Details Get a Quote
Sirt2 Knockout MH-S Cell Line EDJ-KQ78173 Mouse 64383 Details Get a Quote
TFE3 Knockout KGN Cell Line EDJ-KQ48 Human 7030 Details Get a Quote
Stub1 Knockout MB49 Cell Line EDJ-KQ53 Mouse 56424 Details Get a Quote
ARID1A Knockout SNK-6 Cell Line EDJ-KQ64 Human 8289 Details Get a Quote
Tet1 Knockout Kupffer Cell Line EDJ-KQ88 Mouse 52463 Details Get a Quote
NR1H3 Knockout HEK293T Cell Line EDJ-KQ109 Human 10062 Details Get a Quote
NR1H2 Knockout HEK293T Cell Line EDJ-KQ110 Human 7376 Details Get a Quote
RPS6KB2 Knockout HEK293 Cell Line EDJ-KQ125 Human 6199 Details Get a Quote
SMAD6 Knockout HEK293 Cell Line EDJ-KQ126 Human 4091 Details Get a Quote
JUN Knockout HEK293 Cell Line EDJ-KQ176 Human 3725 Details Get a Quote
JUN Knockout HEK293T Cell Line EDJ-KQ184 Human 3725 Details Get a Quote
ALKBH5 Knockout HEK293 Cell Line EDJ-KQ186 Human 54890 Details Get a Quote
Contact Us
*
*
*
How did you hear about us: