Insulin resistance is not a single measurement — it is a phenotype (an overall biological state) that you build up across several layers of evidence, so the first decision in any study is choosing the right insulin resistance model. This application note lays out a general experimental workflow, from whole-body physiology down to the signalling proteins inside a single cell, and shows where each assay — and each reagent — fits.
Insulin is the hormone that tells cells to take glucose (sugar) out of the blood. Insulin resistance is when cells stop listening to that signal properly, so the body has to make more and more insulin to get the same effect. A defensible study rarely rests on one assay; it combines physiological, functional, and molecular readouts so that each supports the others.
Insulin action vs insulin secretion: the two arms
Before choosing methods, it helps to fix one distinction that runs through the whole field:
- Insulin action — how well target tissues (skeletal muscle, liver, and fat) respond to insulin. When they respond poorly, that is insulin resistance in the strict sense. This is where most of the signalling and glucose-uptake toolkit lives.
- Insulin secretion — how much insulin the pancreas actually makes. Insulin is produced by β-cells ("beta cells"), specialised cells in the pancreas. Early on, β-cells compensate for resistance by making extra insulin; eventually they can fail. Related to insulin action, but a separate axis with its own methods.
Keeping these two arms distinct prevents the most common credibility problem in this literature: mixing a β-cell (secretion) result into an insulin-action argument as if they answered the same question. The stages below follow the action arm as the backbone, with the secretion arm treated as its own stage.
The insulin resistance workflow at a glance
Once you have set up a model, a complete insulin resistance study answers five questions in sequence — each one building on the last:
| The question | Answered by |
|---|---|
| Does resistance exist at all? | Whole-body physiology (Stage 2) |
| Is the defect real at the cell level? | Functional glucose uptake (Stage 3) |
| Where in the signalling relay does it break? | Molecular signalling (Stage 4) |
| Why is it happening? | Upstream stress drivers (Stage 5) |
| How is the body compensating? | The β-cell secretion arm (Stage 6) |
×
No single stage answers the whole question. The confidence comes from stacking them — which is why the workflow is worth following end to end.
Stage 1 — Choose and establish an insulin resistance model
A "model" here means the experimental system you study — either cells in a dish (in vitro) or a living animal (in vivo). Every later result inherits the strengths and quirks of this choice, so it comes first.
In vitro insulin resistance models (cells in a dish)
Researchers use cell lines — populations of one cell type that grow indefinitely in culture. Three standard ones each stand in for one of the insulin-target tissues: 3T3-L1 (mouse) or SGBS (human) for adipocytes (fat cells); C2C12 / L6 for myotubes (muscle cells); and HepG2 or primary hepatocytes for liver cells.
These cells start out insulin-sensitive, so resistance has to be deliberately induced — commonly with palmitate (a saturated free fatty acid) at ~0.2–0.75 mM for 16–24 h to mimic lipotoxicity (fat-overload damage); with chronic high glucose; with chronic hyperinsulinemia (long, high-dose insulin, which itself blunts the later response); or with an inflammatory hit (TNF-α ± low oxygen). Each method pushes the phenotype toward a different mechanism, so the induction should match the biology you want to study. (refs 1–3)
In vivo insulin resistance models (whole animals)
- Diet-induced obesity (HFD = high-fat diet) — closest to how humans develop the condition; milder and slower.
- Monogenic models (one gene defect) — ob/ob and db/db mice, which disrupt leptin (an appetite/body-weight hormone). Very obese and insulin resistant, but a single strong lesion isn't typical of human disease.
- Polygenic and other models (many genes) — NZO and TALLYHO mice; ZDF and Goto-Kakizaki rats.
Worth stating plainly: genetic and diet-induced models reproduce different slices of human disease and are not interchangeable, even at matched body weight. (refs 4, 5)
- 3T3-L1 cell — mouse preadipocyte; the adipose model
- C2C12 cell — mouse myoblast; the skeletal-muscle model
- HepG2 cell — human hepatocyte-like; the liver model
Stage 2 — Whole-body physiological phenotyping (GTT, ITT, HOMA-IR, clamp)
This stage answers a simple question: is the whole organism actually insulin resistant, and how badly? It sits above any molecular work — and it is the tier that carries over almost unchanged between mice and humans, which makes it the natural bridge from animal research to clinical studies. All three methods below are routine in mouse work as well as in people.
- Glucose tolerance test (GTT) and insulin tolerance test (ITT) — "challenge" tests, and among the most common readouts in mouse metabolic studies. You give a dose of glucose (or insulin) and track blood sugar over ~2 hours. In mice the glucose goes in through the abdomen (IPGTT) or by oral tube (OGTT), with blood sampled from the tail; in humans the OGTT uses a standard glucose drink.
- HOMA-IR — a quick estimate from a single fasting blood draw (fasting glucose × fasting insulin). Used in both species, but the standard formula was derived in humans, so mouse studies often report the raw product or a rodent-adjusted version. The fasting insulin is usually measured by ELISA.
- Hyperinsulinemic-euglycemic clamp — the gold standard, first described in humans by DeFronzo and colleagues in 1979, and also performed in mice and rats. Insulin is infused at a fixed high level while glucose is dripped in to hold blood sugar normal ("euglycemia"). The amount of glucose needed to hold blood sugar steady is a direct readout of insulin sensitivity. Precise, but slow, expensive, and surgically demanding in rodents. (refs 6, 7)
- Human/Canine/Porcine Insulin ELISA Kit (PicoKine®) — quantifies insulin in serum/plasma. For mouse or rat fasting insulin, use a species-matched rodent insulin ELISA.
Stage 3 — Cellular glucose uptake assay (the functional readout)
Now zoom to the cell: does insulin still push glucose into it? This is the glucose uptake assay that sits at the heart of any cell-based insulin resistance study.
- Glucose uptake — the core functional test. Cells are given a traceable form of glucose — 2-NBDG (a glucose molecule that glows) or 2-deoxyglucose / 2-DG (measured by radioactivity or an enzyme reaction). The key is to compare basal (no insulin) versus insulin-stimulated uptake, because resistance shows up as a weak response to insulin, not necessarily a low baseline.
- GLUT4 translocation — GLUT4 is the main glucose "door" (transporter protein) in muscle and fat. Normally it sits inside the cell in storage packets and, when insulin arrives, translocates to the cell surface to let glucose in. Whether that movement happens is a location question, so it is studied by imaging (confocal or TIRF microscopy) or by separating the membrane from the cell interior and checking where GLUT4 ended up.
(refs 8, 9, 10)
Stage 4 — Molecular signalling analysis: the insulin signaling pathway
This stage asks where in the chain the breakdown happens. Inside the cell, insulin's message is passed along the insulin signaling pathway like a relay, each protein switching on the next:
Insulin receptor → IRS-1/2 → PI3K → AKT → AS160 → GLUT4 movement
×
Most of this relay works by phosphorylation — tagging a protein with a phosphate group to switch it on or off. So the experiments below mostly measure how much of each protein is phosphorylated (switched on) after insulin.
- Western blot ("WB"), phospho vs total — the quantitative workhorse. It separates proteins by size and detects a specific one. The signature readout is p-AKT (the switched-on form) compared to total AKT after insulin — a weak p-AKT response is the molecular fingerprint of resistance. Upstream, IRS-1 shows less of its "on" tag and more of an inhibitory "off" tag when resistant. Everything is read as phospho:total ratios, basal vs stimulated.
- Immunofluorescence ("IF") — uses glowing antibodies under a microscope to show where a protein is. Good for spatial questions a blot can't answer: is GLUT4 at the membrane? Has FOXO moved into the nucleus?
- Gene expression (qPCR / RNA-seq) — measures which genes are switched on, e.g. liver "glucose-making" (gluconeogenic) genes such as PEPCK and G6Pase.
This is where Western blot and immunofluorescence are complementary, not competing: WB tells you how much signalling or protein there is; IF tells you where the action happens. GLUT4 is the clean example — one target, read two ways, answering two different questions. (refs 10, 11, 12)
Two ways to read these targets — Western blot (antibodies) and, as a plate-based quantitative alternative, ELISA:
- Phospho-AKT (Ser473) ELISA — mouse, human, rat, paired with Total AKT ELISA (human) — this pairing gives the exact phospho:total AKT readout, quantified from lysates without a gel.
- GLUT4 ELISA — human, rat. These measure total GLUT4 protein ("how much"), not translocation ("where") — for translocation you still need imaging or membrane fractionation.
- Anti-GAPDH antibody (WB/IF) — a housekeeping loading control for normalising Western blots.
Stage 5 — Upstream drivers (why insulin resistance happens)
Insulin resistance is usually the downstream result of one or more stresses on the cell. Pinning these down turns a description into a mechanism.
- Lipotoxicity — fat building up where it shouldn't ("ectopic" fat, e.g. inside muscle or liver) and jamming the signalling relay.
- Chronic low-grade inflammation — fat tissue releasing inflammatory proteins (TNF-α, IL-6, MCP-1) that activate stress pathways (NF-κB and JNK), which add the "off" tag to IRS-1. Measured by ELISA or multiplex panels, plus Western blot for the pathways.
- ER stress — the endoplasmic reticulum (the cell's protein-folding factory) triggers a stress response (the "unfolded protein response," run by sensors PERK, IRE1α, ATF6) when overwhelmed; detected via markers like GRP78.
- Oxidative stress — a buildup of ROS (reactive oxygen species), measured with dedicated ROS assays.
(refs 13, 14, 15)
Stage 6 — The secretion arm: β-cell compensation and GSIS
Kept deliberately separate from Stages 3–5. The question here is how much insulin the β-cell makes, not how tissues respond to it.
- Glucose-stimulated insulin secretion (GSIS) — the core β-cell test. Islets (pancreatic clusters that contain β-cells) are exposed to low then high glucose, and the insulin they release is measured (usually by ELISA). Done in a static dish or by perifusion, where fluid flows continuously past the cells to capture the timing of release.
- Islet histology / immunofluorescence — staining tissue slices to see structure: insulin and glucagon in different cells, β-cell mass, cell division (Ki67), and cell death (TUNEL).
The storyline is compensation (β-cells making extra insulin to cover for resistance) sliding into failure — the point where insulin resistance tips over into full type 2 diabetes. (refs 16, 17, 18)
- Human/Canine/Porcine Insulin ELISA Kit (PicoKine®) — measures the insulin released into the medium during a GSIS run. For mouse or rat islet GSIS, use a species-matched rodent insulin ELISA.
Working across species: mouse and human insulin resistance models
The same workflow spans preclinical and clinical research, but the stages don't all translate equally:
| Stage 2 — physiology | Portable. GTT, ITT, HOMA-IR, and the clamp are done in both mice and humans with the same logic — the bridge between the two. |
|---|---|
| Stage 1 — models | In humans you can't ethically induce resistance, so you recruit cohorts (lean vs obese, healthy vs type 2 diabetic). |
| Stages 3–4 — cellular & molecular | Need biopsies (muscle or fat), often taken before and during a clamp, then analysed ex vivo with the same methods. Whole-body/organ glucose disposal can also be traced with labeled-glucose tracers, and tissue uptake imaged by PET. |
A practical upside for reagent selection: many key ELISA targets and antibodies are cross-reactive or available in species-matched versions, so one mechanistic panel can serve both mouse discovery and human translational samples. On BioHippo, both phospho-AKT (Ser473) and TNF-α are stocked in mouse, rat, and human versions, and GLUT4, total AKT, and insulin ELISAs are available for human (and, for GLUT4, rat) samples.
Convergence and interpretation
The five questions from the start only become a conclusion when their answers line up. A mechanistic claim is strongest when the functional and molecular layers agree — for example, weak insulin-stimulated glucose uptake and low p-AKT and GLUT4 failing to reach the membrane, all in the same model.
- Keep insulin action (target tissues) separate from insulin secretion (β-cells).
- Always report basal vs insulin-stimulated, and phospho:total ratios — never a lone stimulated value.
- Match species and tissue across a claim; don't stretch a fat-cell result to muscle without evidence.
- Don't equate a quick estimate (HOMA-IR) with a direct measurement (clamp).
Frequently asked questions
How do you induce insulin resistance in vitro?
Standard cell lines (3T3-L1 adipocytes, C2C12 myotubes, HepG2 hepatocytes) start insulin-sensitive, so resistance is deliberately induced — most often with palmitate (~0.2–0.75 mM, 16–24 h) to mimic lipotoxicity, chronic high glucose, chronic hyperinsulinemia, or an inflammatory hit such as TNF-α. Match the induction method to the mechanism you want to study.
What is the gold standard for measuring insulin resistance?
The hyperinsulinemic-euglycemic clamp, first described by DeFronzo and colleagues in 1979. Insulin is infused at a fixed level while glucose is dripped in to hold blood sugar normal; the glucose infusion rate needed is a direct readout of insulin sensitivity. Estimates like HOMA-IR approximate it but are not equivalent.
What is the difference between insulin action and insulin secretion?
Insulin action is how well target tissues (muscle, liver, fat) respond to insulin — poor response is insulin resistance in the strict sense. Insulin secretion is how much insulin the pancreatic β-cells make. They are related but separate axes, and mixing a secretion result into an action argument is a common credibility error.
Which cell lines are used to study insulin resistance?
Three lines each model one insulin-target tissue: 3T3-L1 (mouse adipocytes), C2C12 (mouse myotubes), and HepG2 (human liver). SGBS (human adipocytes), L6 (muscle), and primary hepatocytes are common alternatives.
How do you measure insulin resistance at the molecular level?
Probe the insulin signaling pathway (insulin receptor → IRS-1/2 → PI3K → AKT → AS160 → GLUT4) by phosphorylation. The signature readout is phospho-AKT (Ser473) relative to total AKT after insulin, read by Western blot or a p-AKT ELISA; a weak p-AKT response is the molecular fingerprint of resistance. Always report basal vs insulin-stimulated.
References
- In Vitro Insulin Resistance Model: A Recent Update. PMC9876677
- Odeniyi IA, et al. An improved in vitro 3T3-L1 adipocyte model of inflammation and insulin resistance. Adipocyte, 2024. Full text
- In vitro characterization of the effects of chronic insulin stimulation in mouse 3T3-L1 and human SGBS adipocytes. PMC7469436
- Animal models for type 1 and type 2 diabetes: advantages and limitations. Front Endocrinol, 2024. Full text
- db/db Mice Exhibit Features of Human Type 2 Diabetes That Are Not Present in Weight-Matched C57BL/6J Mice Fed a Western Diet. PMC5606106
- Measuring and estimating insulin resistance in clinical and research settings. PMC9542105
- Defining Insulin Resistance From Hyperinsulinemic-Euglycemic Clamps. Diabetes Care, 2012;35(7):1605. Article
- MicroRNA-506 modulates insulin resistance in human adipocytes… IRS1/PI3K/AKT. (2-deoxyglucose uptake assay example.) PMC8595185
- Insulin signalling and GLUT4 trafficking in insulin resistance. Biochem Soc Trans, 2023;51(3):1057. (PMID 37248992.) Article
- Liu M, et al. Trilobatin ameliorates insulin resistance through IRS-AKT-GLUT4 signaling pathway in C2C12 myotubes and ob/ob mice. Chin Med, 2020. DOI: 10.1186/s13020-020-00390-2
- See ref 9 (Portland Press) — insulin signalling / GLUT4 cascade.
- Impairment of the IR-IRS-PI3K-AKT signaling pathway in insulin resistance. (Review; preprint.) Preprint
- Chronic Adipose Tissue Inflammation Linking Obesity to Insulin Resistance and Type 2 Diabetes. Front Physiol, 2019. Full text
- Adipose tissue and insulin resistance in obese. (FFA / TLR4 / NF-κB / ER-stress / adipokine mechanisms.) Article
- Age-related inflammation and insulin resistance: a review of their intricate interdependency. PMC4246128
- Nutrient Regulation of Pancreatic Islet β-Cell Secretory Capacity and Insulin Production. PMC8869698
- Mechanisms of β-cell functional adaptation to changes in workload. (Ex vivo GSIS / glucose-conditioning assay.) PMC5021190
- Insulin enhances glucose-stimulated insulin secretion in healthy humans. PNAS, 2010. Article
