Written and maintained by CASRAI Editorial Board
Last updated
SDS-PAGE (sodium dodecyl sulfate–polyacrylamide gel electrophoresis) is the standard method for separating proteins by approximate molecular weight and is the workhorse first step behind most western blots, protein-purity checks, and mass-spectrometry sample prep. This guide covers the mechanism that makes it work, how to choose an acrylamide percentage for a given target, sample-prep pitfalls that cause avoidable failures, staining tradeoffs, acrylamide handling safety, and a symptom-organized troubleshooting reference. For antibody detection after the gel — transfer, blocking, and imaging — see our western blot protocol walkthrough and no-bands troubleshooting guide; for the DNA/RNA equivalent, see agarose gel electrophoresis basics.
What SDS-PAGE Separates, and Why Some Proteins Don’t Behave
Sodium dodecyl sulfate (SDS) is an anionic detergent that binds along the length of a denatured polypeptide chain in a roughly constant ratio relative to protein mass. That coating does two things at once: it unfolds the protein into an extended, rod-like conformation, and it swamps the protein’s own intrinsic charge with a large, uniform negative charge proportional to chain length. Because the charge-to-mass ratio becomes approximately constant across different proteins, electrophoretic mobility through the polyacrylamide matrix is governed almost entirely by size (or, more precisely, by how much a molecule’s hydrodynamic radius impedes it moving through the gel’s pores) rather than by the protein’s native charge or shape. That is the whole trick of SDS-PAGE: it converts a size-and-shape-and-charge separation problem into a size-only one.
The same mechanism is exactly why some proteins still run at an “apparent” molecular weight that doesn’t match their real mass:
- Glycosylation. Attached carbohydrate adds real mass but does not bind SDS in the same proportion as polypeptide backbone, so heavily glycosylated proteins typically migrate slower than their true molecular weight would predict — they run “big.”
- Extreme isoelectric point (pI). Very acidic or very basic proteins can bind SDS non-uniformly, distorting the assumed constant charge-to-mass ratio the whole method depends on.
- Incomplete SDS binding or renaturation. Some proteins, particularly those with unusual amino acid composition or strong internal structure, don’t fully denature or fully coat with SDS under standard conditions, leaving mobility partly governed by residual native shape.
- Membrane proteins. Highly hydrophobic transmembrane domains can bind SDS in different amounts than soluble domains and often migrate anomalously — a well-known example is that many multi-pass membrane proteins run faster (lower apparent MW) than their calculated mass.
- Intrinsically disordered regions. Extended, unstructured stretches of sequence can behave hydrodynamically as if they were larger than their calculated mass, again pushing apparent MW upward.
None of this means SDS-PAGE is unreliable — it means “apparent molecular weight on a gel” and “calculated molecular weight from sequence” are related but not identical quantities, and a mismatch between the two is diagnostic information, not automatically an error.
The Laemmli Discontinuous System: Why the Stacking Gel Exists
The dominant SDS-PAGE format in use today is the discontinuous buffer system described by Ulrich K. Laemmli in 1970 (Laemmli, U.K., “Cleavage of Structural Proteins during the Assembly of the Head of Bacteriophage T4,” Nature 227, 680–685). Most explanations of SDS-PAGE skip the mechanism behind its most distinctive feature — the low-percentage stacking gel sitting on top of the higher-percentage resolving gel — and simply say it “sharpens bands.” Here is why it actually works:
- The stacking gel runs at a lower pH (commonly around pH 6.8) and lower acrylamide percentage than the resolving gel (commonly around pH 8.8).
- The running buffer supplies two ion species with very different mobilities: chloride ions (from the gel/sample buffer, fast-moving) act as the “leading” ion, and glycine (from the Tris-glycine running buffer) acts as the “trailing” ion. At the stacking gel’s lower pH, glycine is mostly in its zwitterionic, nearly-uncharged form and moves slowly.
- Between a fast leading ion and a slow trailing ion, a steep, self-sharpening voltage gradient forms (the Kohlrausch regulating function). Proteins, whose mobility falls between the two, get swept into an extremely thin, concentrated zone at the moving boundary — regardless of how dilute or how large the original loaded volume was.
- When that stacked zone crosses into the resolving gel, the higher pH there increases glycine’s ionization and mobility enough that it overtakes the proteins, dissolving the stacking effect. From that point on, proteins separate by size in a uniform field through the smaller resolving-gel pores.
The practical payoff: samples loaded as a few-millimeter-thick band of dilute protein start the separation as an extremely thin, concentrated line, which is what makes discontinuous SDS-PAGE produce sharp, tight bands instead of the diffuse ones a single-buffer, single-percentage gel would give.
Choosing an Acrylamide Percentage by Target Molecular Weight
Acrylamide percentage sets the resolving gel’s pore size: higher percentage means smaller pores, which better resolve small proteins but run large proteins poorly (or exclude them). Gel and reagent suppliers (Bio-Rad, Thermo Fisher, and others) publish broadly consistent approximate ranges; treat the cutoffs below as typical starting points, not exact boundaries — the precise range shifts slightly by manufacturer and gel chemistry.
| Acrylamide % | Typical effective separation range | Common use |
|---|---|---|
| 7.5% | ~50–500 kDa | Very large proteins, high-MW complexes |
| 8% | ~25–200 kDa | Large-to-mid-size proteins |
| 10% | ~15–100 kDa | General-purpose, broad mid-range |
| 12% | ~10–70 kDa | Most common default for unknown or typical targets |
| 15% | ~3–45 kDa | Small proteins and peptides |
| 4–20% gradient | ~10–250 kDa | Unknown or very wide MW range on one gel |
Gradient Gels: When They Earn Their Cost
A gradient gel (commonly 4–15% or 4–20%) has pore size that narrows continuously from top to bottom, letting one gel resolve both large and small proteins reasonably well — useful when the target’s molecular weight is unknown, when a lysate needs a broad-range overview, or when co-resolving proteins that differ substantially in size (for example, an intact antibody alongside its cleaved fragments). The tradeoff is cost (gradient gels are more expensive, whether cast in-house or purchased precast) and slightly reduced resolution within any single narrow MW window compared with a fixed-percentage gel specifically chosen for that window. If the target’s approximate size is already known and the goal is maximum resolution around it, a fixed-percentage gel from the table above is usually the better choice.
Reducing vs. Non-Reducing Conditions
Reducing SDS-PAGE includes a reducing agent — commonly dithiothreitol (DTT) or beta-mercaptoethanol (BME) — in the sample buffer to break intramolecular and intermolecular disulfide bonds, denaturing proteins down to their individual polypeptide subunits. This is the default for most western blots and general protein identification, because it gives a consistent, predictable migration pattern independent of the protein’s native disulfide bonding.
Non-reducing SDS-PAGE omits the reducing agent, preserving disulfide-linked subunit associations. This matters when the disulfide-linked quaternary structure is itself the thing being examined — the standard example is antibodies: an intact IgG runs at roughly 150 kDa under non-reducing conditions but separates into ~50 kDa heavy chains and ~25 kDa light chains once reduced. Non-reducing conditions are also used when an antibody’s epitope depends on a native disulfide-bonded conformation that reduction would destroy.
Sample Preparation: The Boiling Trap With Membrane Proteins
Standard sample prep mixes the protein sample with SDS-containing loading (Laemmli) buffer — SDS, glycerol, a tracking dye such as bromophenol blue, Tris at the stacking-gel pH, and (for reducing gels) DTT or BME — then heats the sample to fully denature the protein and ensure even SDS coating before loading.
The classic avoidable failure: don’t reflexively boil every sample. For soluble, cytosolic proteins, heating at or near 95–100°C for several minutes is standard and reliable. But for membrane proteins — multi-pass transmembrane proteins, transporters, GPCRs, and similar hydrophobic targets — boiling frequently causes irreversible aggregation of their hydrophobic domains rather than clean denaturation. The result is a smeared high-molecular-weight aggregate, or the target protein missing from its expected monomeric band entirely, even though it was present in the original sample. Standard practice for membrane-protein samples is a lower, gentler heating step (commonly in the 37–70°C range, well short of boiling) for a longer time instead of a short boil — check your specific target’s published handling if aggregation is a known issue for that protein family. When a band that should be present is simply missing at the expected size on a membrane-protein prep, aggregation from over-heating is one of the first things to check before assuming an antibody or expression problem.
Choosing a Molecular Weight Ladder
A prestained ladder is dyed so bands are visible during the run and after transfer, without requiring a stain step — useful for tracking progress and confirming a successful transfer to membrane, and it’s the standard choice for western blotting. An unstained (or “precision”) ladder gives a more accurate molecular-weight estimate on gels that will be directly stained (Coomassie or silver), since the dye conjugated to a prestained ladder can shift its apparent migration slightly relative to its stated molecular weight. Choose a ladder whose range comfortably brackets the target protein’s expected size, with enough marker bands near that size to interpolate confidently.
Running Conditions and Heat Effects
Discontinuous gels are commonly run at a lower, gentler voltage while the sample stacks, then a higher voltage once the dye front enters the resolving gel — exact voltage and current depend on the specific gel system, buffer, and gel thickness, so follow the manufacturer’s recommended range for the specific cassette or casting system in use rather than a single universal number. The consistent principle across systems: too much voltage or current generates excess Joule heating, and heat is a major source of downstream artifacts — distorted “smiling” bands (edge lanes run faster than center lanes because heat dissipates unevenly across the gel), buffer depletion, and reduced resolution from band broadening. Running gels in a cold room, with active cooling, or simply at a lower voltage for a longer time reduces these effects, particularly on longer runs.
Comparing Stains: Sensitivity vs. Mass-Spec Compatibility
Which stain to use depends heavily on what happens to the band afterward — a decision many quick comparisons skip.
| Stain | Relative sensitivity | Mass-spec compatibility | Notes |
|---|---|---|---|
| Coomassie (colloidal / R-250) | Lower (roughly sub-microgram to low-microgram range per band, depending on formulation) | Generally good — the standard default for downstream MS | Simple, inexpensive, widely regarded as the safest default when a band may go for identification by mass spectrometry |
| Silver stain | High (commonly cited as roughly one to two orders of magnitude more sensitive than Coomassie) | Variable — many conventional protocols use glutaraldehyde or formaldehyde fixation that covalently modifies proteins and interferes with trypsin digestion | Use an MS-compatible silver stain kit specifically (these omit or minimize glutaraldehyde) if the band will be excised for MS |
| Fluorescent stains (e.g., ruthenium- or SYPRO-type dyes) | High, comparable to or better than silver, with a wider linear dynamic range | Generally good | Requires a fluorescence imager; often preferred when accurate relative quantitation across a wide abundance range matters, not just detection |
In short: reach for Coomassie by default, especially if mass spec is a possibility; reach for silver or a fluorescent stain only when sensitivity genuinely requires it, and confirm MS compatibility of the specific kit before excising a band for identification.
Acrylamide Safety: A Real EHS Obligation, Not an Afterthought
Unpolymerized acrylamide monomer is a potent neurotoxin and is classified as a probable human carcinogen, and this is one of the more genuinely under-covered angles of SDS-PAGE content aimed only at bench mechanics. Key points for anyone casting gels from powder or liquid stock rather than using precast commercial gels:
- Neurotoxicity is cumulative and dose-related. Repeated low-level exposure to unpolymerized acrylamide is associated with peripheral neuropathy (numbness, weakness, incoordination) that develops with cumulative dose, not necessarily from a single exposure.
- It is readily absorbed through intact skin, not just by inhalation, so skin contact with acrylamide powder or unpolymerized solution is a genuine route of systemic exposure, not merely a local irritant concern.
- OSHA’s permissible exposure limit (PEL) for acrylamide is 0.3 mg/m³ (8-hour time-weighted average), per OSHA/NIOSH occupational exposure documentation.
- Powdered acrylamide is the highest-risk form — airborne dust is easily generated when weighing out powder, and should be handled inside a certified chemical fume hood with the sash appropriately positioned, not on an open bench.
- PPE for anyone weighing, mixing, or handling unpolymerized acrylamide should include a buttoned lab coat, chemical-resistant gloves (check your institution’s glove-selection guidance — some general-purpose gloves are not adequately protective for extended acrylamide contact), and eye protection.
- Polymerized (set) acrylamide gel is much lower risk than the monomer, but a freshly poured gel can still contain a small fraction of unreacted monomer, so treat handling of fresh, unset, or partially set gels with the same care as the liquid stock until fully polymerized.
- Waste disposal is a compliance obligation, not a convenience choice. Unpolymerized acrylamide stock, liquid waste, and any acrylamide-contaminated disposables should go into your institution’s designated hazardous chemical waste stream, following your local Environmental Health & Safety (EHS) office’s labeling and pickup procedures — never down the drain.
Most modern labs substantially reduce this exposure risk by using precast commercial gels for routine work, reserving hand-cast gels for cases needing a non-standard percentage, gradient, or buffer system not available precast.
Troubleshooting SDS-PAGE by Symptom
| Symptom | Likely causes | What to try |
|---|---|---|
| “Smiling” bands (curved, edges ahead of center) | Uneven heat dissipation, voltage/current too high, gel too warm | Reduce voltage, run in a cold room or with active cooling, use fresh cold running buffer |
| Streaking or smearing within a lane | Protein overload, incomplete denaturation, sample aggregation (often from over-heating), degraded or partially precipitated sample, old/expired gel | Reduce loading amount, confirm correct heating step for the target protein type, check sample integrity and gel expiry |
| No bands visible at all | Insufficient protein loaded, protein degraded before loading, stain/detection step failed, wrong side of gel loaded, target ran off the gel (percentage too low for a small target, or run too long) | Re-quantify starting material (see our Bradford assay guide), verify loading order, confirm gel percentage suits the target size; for antibody-detection-specific no-bands cases after a good gel, see our western blot no-bands guide |
| Distorted or curved lanes | Air bubbles trapped between plates or under the comb, uneven or incomplete polymerization, warped or dirty comb, gel loaded before fully set | Re-pour carefully avoiding bubbles, confirm the gel has fully set before loading (typically indicated by a visible interface line), inspect combs for damage |
| Poor resolution (bands not clearly separated) | Wrong acrylamide percentage for the target size, insufficient run distance or time, voltage too low, running buffer diluted incorrectly or reused past its useful life | Match acrylamide percentage to target MW using the table above, run further/longer, prepare fresh running buffer at the correct concentration |
| Protein appears at an unexpected molecular weight | Post-translational modification (glycosylation, phosphorylation) shifting apparent MW, proteolytic degradation, incomplete reduction leaving a disulfide-linked species, oligomerization/aggregation, alternative splicing or isoform, or one of the anomalous-migration causes described earlier in this guide | Compare reducing vs. non-reducing runs, add protease inhibitors during lysis, confirm antibody specificity, and check whether the target protein family is known for anomalous SDS-PAGE migration |
| Gel fails to polymerize (stays liquid) | Expired or degraded ammonium persulfate (APS) or TEMED, incorrect APS/TEMED ratio, oxygen inhibition at the gel’s exposed top surface, gel poured too cold | Use fresh APS (make fresh or store properly — APS solution degrades over days to weeks), confirm reagent ratios, overlay the resolving gel with a thin layer of water or isopropanol while it sets to exclude oxygen |
The Research-Integrity Angle: Gel and Blot Image Manipulation
Gel and blot figures — SDS-PAGE and western blot images specifically — are disproportionately represented among figures flagged in published audits of image problems in the biomedical literature, and inappropriate splicing or duplication of gel/blot lanes is a documented, recurring cause of corrections and retractions. The core rule most journal image-integrity policies converge on: it is not acceptable to combine lanes from different gels, or from the same gel exposed or processed differently, and present them as if they were run together, without disclosing it. Visible discontinuities such as vertical lines between lanes are a common tell that lanes have been spliced together.
Where splicing genuinely is necessary and appropriate (for example, removing irrelevant lanes from the same original gel image), current best practice — reflected in policies from major publishers including Nature Portfolio and society journals such as the American Society for Microbiology’s journals — is to mark the join clearly (a visible dividing line in the figure) and disclose it in the legend, and to retain the full, unprocessed original image so it can be produced if requested. Many life-science journals now require submission of uncropped, unprocessed source images of gels and blots as supplementary material at the time of submission, specifically so reviewers and readers can verify what was actually run. For the broader mechanics of what counts as acceptable image processing versus manipulation, see our guide on detecting image manipulation in figures and on what must be documented in a figure legend; for what happens when manipulation is confirmed, see what an ORI misconduct finding actually means.
The practical takeaway for anyone running SDS-PAGE as part of a study intended for publication: photograph or scan the full, unedited gel or blot before any cropping, keep that original file, and if a figure needs to show non-adjacent lanes from the same gel, mark the splice rather than presenting it as continuous.
Frequently Asked Questions
What does SDS-PAGE stand for?
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis: SDS is the detergent that denatures proteins and gives them a uniform charge-to-mass ratio, and polyacrylamide is the gel matrix the proteins migrate through.
What’s the difference between SDS-PAGE and native PAGE?
SDS-PAGE denatures proteins and separates them almost entirely by size. Native (non-denaturing) PAGE runs proteins without SDS or heat, preserving their native folded structure, charge, and any complexes they form, so migration depends on size, shape, and native charge together — useful for studying protein-protein interactions or enzymatic activity that requires the native conformation, but not for a clean molecular-weight estimate.
Can SDS-PAGE be used to separate DNA or RNA?
No — SDS-PAGE is a protein technique. Nucleic acids are separated by agarose gel electrophoresis (for larger fragments) or non-denaturing/denaturing polyacrylamide gels without SDS (for very small fragments), which rely on nucleic acids’ own uniform negative charge rather than an SDS coating. See our agarose gel electrophoresis guide for the DNA/RNA equivalent.
What percentage gel should I use for a ~50 kDa protein?
A 10% or 12% gel is the typical choice for a target in the 40–60 kDa range — see the acrylamide percentage table above for the fuller range-by-target breakdown, and use a gradient gel instead if the sample also contains proteins well outside that range that need to be resolved on the same gel.
Why do I need a stacking gel at all — can I just run a single-percentage gel?
A single-percentage gel without a stacking layer will still separate proteins, but bands will generally be more diffuse because the sample enters the resolving gel at whatever concentration and volume it was loaded at, rather than first being concentrated into a thin zone. For most applications where band sharpness and resolution matter, the discontinuous (stacking + resolving) Laemmli system is worth the extra pour step.
Why boil (or not boil) my sample before loading?
Boiling ensures full denaturation and even SDS coating for most soluble proteins. For membrane proteins specifically, boiling often causes aggregation instead of clean denaturation — use a gentler heating step for those targets. See the sample-preparation section above.
How long does an SDS-PAGE run take?
Run time depends on gel percentage, thickness, voltage, and how far the target needs to migrate to achieve adequate separation from neighboring bands — anywhere from under an hour to a few hours is typical for a mini-gel format. Follow the specific gel manufacturer’s recommended run time and voltage rather than a fixed universal duration.
Further reading: figure legends — A figure legend (also called a figure caption in most journal style guides — the two terms are used interchangeably at outlets such as Nature, Cell Press, and PLOS) is the descriptive text accompanying a figure that expl.
