ELISA

Why it matters

The enzyme-linked immunosorbent assay (ELISA) measures how much of a specific protein, hormone, antibody or other molecule is in a sample. Research labs use it for cytokines and biomarkers; clinical and public health labs use ELISA-type immunoassays for infections and antibodies; biomanufacturing uses it to measure product and impurities such as host-cell proteins. Running the plate is half the job. The other half is turning optical density readings into concentrations you can defend.

The core idea: a sandwich ELISA

The most common format for measuring a protein is the sandwich ELISA in a 96-well plate:

1. Capture antibody is coated on the bottom of each well.
2. Blocking with a protein solution covers the remaining plastic.
3. Sample or standard is added; the target binds the capture antibody. Washing removes everything else.
4. Detection antibody, which binds a different part of the target, is added. It is linked to an enzyme (often HRP), directly or through a biotin and streptavidin step.
5. Substrate (often TMB) is added. The enzyme turns it blue; more target means more enzyme and more colour.
6. Stop solution (an acid) turns the colour yellow and stops the reaction, and a plate reader measures optical density (OD) at 450 nm.

Other formats exist. A direct or indirect ELISA detects target stuck to the plate; a competitive ELISA, used for small molecules, gives less signal when there is more target.

The standard curve

OD is not a concentration. You convert it using standards: known concentrations of the target, run on the same plate. Typical good practice:

• Run standards, samples and blanks in duplicate or triplicate, and use the mean OD.
• Subtract the blank (zero standard) OD from everything.
• Fit a curve to the standards. Many ELISAs are sigmoidal over their full range and are fitted with a four-parameter logistic (4PL) model in plate-reader software. Over a narrower range the curve is often close to a straight line, which is what this lesson's activity uses.
• Read unknowns only inside the range of the standards. A result above the top standard is reported as "above range" and the sample is diluted and rerun; you never extrapolate.

Worked example. After blank subtraction, a linear fit of the standards gives slope 0.00208 OD per pg/mL and intercept 0.0 (these standards happen to fall on a perfect line). A sample diluted 1 in 2 reads OD 0.440, or 0.390 after subtracting the blank 0.050.

concentration in well = (0.390 - 0.0) / 0.00208 = 187.5 pg/mL
original sample       = 187.5 x 2 (dilution factor) = 375 pg/mL

Forgetting the dilution factor is one of the most common reporting errors.

Quality checks on a plate

• Replicate agreement: the coefficient of variation (CV = SD / mean x 100) of duplicate ODs is often required to be below about 15 to 20 percent. A high CV points to pipetting or washing problems.
• Curve fit: check R-squared, and back-calculate each standard from the curve; each should be close to its nominal value.
• Controls: kit or in-house controls with known values should fall in their acceptable ranges, just like the QC rules from the lab data course.
• Edge effects and timing: uneven temperature or adding substrate slowly across a plate can shift ODs, so follow incubation times exactly.

Clinical and regulated context

Clinical immunoassays run on validated, regulated systems, and results are released only by qualified staff under CLIA in the US. Medical laboratory scientists qualify through a NAACLS-accredited program (https://www.naacls.org, checked on 2026-10-23) and ASCP Board of Certification exams (https://www.ascp.org/content/board-of-certification, checked on 2026-10-23). Plate work itself happens in a lab, under supervision.

Common mistakes

• Forgetting to apply the dilution factor.
• Reporting values above the top standard or below the lowest standard as numbers.
• Not subtracting the blank, or subtracting it twice.
• Ignoring high CV between duplicates.
• Mixing standards from one kit lot with plates from another.

Quick recap

• Sandwich ELISA: capture, block, sample, detection antibody, enzyme, substrate, stop, read OD.
• Convert OD to concentration with a standard curve on the same plate (4PL or, over a linear range, a straight line).
• Blank-correct, stay within the curve, apply dilution factors.
• Check replicate CV, the curve fit and controls before reporting.

Try it

Open the Do tab. A Python script fits the standard curve; you write the line that turns a sample OD into a concentration.