On November 10, 2026, we ran one HRP-labeled chicken anti-Protein A antibody at three dilutions in a single experiment. The zero-calibrator wells read an OD450 of 0.636 at 1:4,000, 0.328 at 1:8,000 and 0.129 at 1:16,000. The antibody, the capture layer and the substrate were the same. Background still moved almost fivefold.
The HRP conjugate dilution is the factor by which the labeled detection antibody is diluted before it goes onto the plate in the detection step of a sandwich ELISA. In a Protein A assay the plate holds an anti-Protein A capture antibody, the sample supplies Protein A, and the conjugate turns every captured molecule into color. Too much conjugate raises the baseline and flattens the top of the curve. Too little starves the signal.
The run below is our worked example. The dilution factors belong to our conjugate. The method carries over to yours.
Step 1: Fix the sandwich and vary only the conjugate
A dilution test only means something when everything else stays put. Capture antibody, plate, sample series, incubation times, washing and substrate time were held constant, and the conjugate dilution was the one variable.
The format was a standard non-competitive sandwich:
- Capture: a mouse anti-Protein A antibody on a 96-well plate.
- Analyte: Sekbio Protein A, diluted into a ten-fold series.
- Detection: a chicken anti-Protein A antibody labeled with HRP.
The detection antibody is chicken IgY on purpose. IgY is not bound by Protein A, so the labeled reagent is not captured through its Fc region. A published multi-product residual Protein A ELISA uses chicken IgY for the same reason. Specifications for the unlabeled reagent are on the chicken anti-Protein A antibody (IgY) page.
Plate coating, blocking and drying were finished beforehand. Those parameters are Sekbio proprietary and are not listed here. The steps on the day were:
- Add 100 µL/well of Protein A standard and incubate for 1 h at 37 °C.
- Wash three times with 400 µL/well of wash solution and tap the plate dry.
- Add 100 µL/well of HRP conjugate at the dilution under test and incubate for 1 h at 37 °C.
- Wash three times again.
- Add 50 µL/well of chromogenic Solution A, then 50 µL/well of Solution B, and incubate for 15 min at 37 °C.
- Add 50 µL/well of stop solution and read OD450.
Step 2: Run each dilution against the full Protein A range
One mid-range point cannot show you the background or the ceiling, so every dilution met the whole range. Protein A went in at eight levels, 0, 0.002, 0.02, 0.2, 2, 20, 200 and 2,000 ng/mL, six orders of magnitude in ten-fold steps, each in duplicate wells. The conjugate was tested at 1:4,000, 1:8,000 and 1:16,000.
This is a one-axis cousin of a checkerboard titration, the usual way to optimize an ELISA. Boster's checkerboard titration guide says to choose the condition with the best signal-to-noise ratio and to avoid concentrations that saturate the assay. Table 1 holds the raw means.
| Protein A (ng/mL) | 1:4,000 | 1:8,000 | 1:16,000 |
|---|---|---|---|
| 0 | 0.636 | 0.328 | 0.129 |
| 0.002 | 0.889 | 0.476 | 0.202 |
| 0.02 | 2.647 | 1.401 | 0.397 |
| 0.2 | 3.675 | 1.923 | 0.803 |
| 2 | 3.901 | 2.580 | 1.522 |
| 20 | 3.921 | 3.308 | 1.920 |
| 200 | 3.859 | 3.515 | 2.419 |
| 2,000 | 3.863 | 3.737 | 3.209 |
Table 1. Mean OD450 of duplicate wells for each HRP conjugate dilution.
Step 3: Read the zero wells before anything else
The zero wells carry conjugate but no Protein A, so everything they read is background. In our run it fell from 0.636 at 1:4,000 to 0.328 at 1:8,000 and 0.129 at 1:16,000, a drop of about 4.9 times across a fourfold change in dilution.
A high zero OD costs you twice.
- It uses up reading range. At 1:4,000 the baseline alone takes about one sixth of an OD scale that tops out near 3.9.
- It inflates the denominator of every S/N, so a good-looking raw OD can hide a poor ratio.
Excess conjugate sticking to the plate or the capture layer is a common cause of an elevated zero. We did not run a no-conjugate control, so we cannot say how much of the 0.129 floor at 1:16,000 comes from substrate and plate rather than conjugate. If you can spare a few wells, add some that receive diluent in place of conjugate. They split the two sources.
Step 4: Look for a plateau at the top of the curve
At 1:4,000 the curve runs out of room early. The mean OD was 3.675 at 0.2 ng/mL, then 3.901, 3.921, 3.859 and 3.863 at 2, 20, 200 and 2,000 ng/mL. A thousand-fold change in Protein A moved the signal by less than 2%, and S/N stopped at about 6.1. An OD near 3.9 is also close to the reading ceiling of many plate readers, so check your reader's specification before you blame the chemistry.
At 1:8,000 the curve bends over late. The OD climbed from 2.580 at 2 ng/mL to 3.308 at 20 ng/mL, then to 3.515 and 3.737 at 200 and 2,000 ng/mL. The last two ten-fold steps each added about 6% to the S/N (10.1, 10.7, 11.4).
At 1:16,000 the curve is still climbing at the highest level: 1.522, 1.920, 2.419 and 3.209 over the same four steps, with S/N reaching 24.9. That is about four times the S/N of 1:4,000 at 2,000 ng/mL. No plateau appeared within the range we tested.
Step 5: Check the low end with CV as well as S/N
The low end is where the three dilutions looked most alike, and where they differed most in precision.
S/N at 0.002 ng/mL was 1.4, 1.5 and 1.6 for 1:4,000, 1:8,000 and 1:16,000. The duplicate wells sat above the zero wells, but the ratio is below the S/N of 3 often used as a screening threshold. At 0.02 ng/mL it was 4.2, 4.3 and 3.1, so every dilution cleared S/N 3 at that level. Changing the dilution moved the background and the ceiling. It did not move the level where the response began.
| Protein A (ng/mL) | S/N 1:4,000 | CV 1:4,000 | S/N 1:8,000 | CV 1:8,000 | S/N 1:16,000 | CV 1:16,000 |
|---|---|---|---|---|---|---|
| 0 | 1.0 | 8.1% | 1.0 | 5.6% | 1.0 | 5.5% |
| 0.002 | 1.4 | 4.0% | 1.5 | 6.5% | 1.6 | 23.1% |
| 0.02 | 4.2 | 5.6% | 4.3 | 5.7% | 3.1 | 30.3% |
| 0.2 | 5.8 | 0.2% | 5.9 | 4.4% | 6.2 | 7.3% |
| 2 | 6.1 | 0.4% | 7.9 | 1.9% | 11.8 | 2.3% |
| 20 | 6.2 | 0.3% | 10.1 | 1.4% | 14.9 | 3.2% |
| 200 | 6.1 | 1.9% | 10.7 | 1.6% | 18.8 | 2.0% |
| 2,000 | 6.1 | 1.6% | 11.4 | 1.8% | 24.9 | 2.4% |
Table 2. S/N (mean OD of the level divided by mean OD of the 0 ng/mL wells) and CV between the two duplicate wells.
Precision tells a different story. At 0.002 and 0.02 ng/mL the duplicate CV was 4.0% and 5.6% at 1:4,000, 6.5% and 5.7% at 1:8,000, but 23.1% and 30.3% at 1:16,000. Duplicate wells differed by 0.04 to 0.21 OD at these two levels at every dilution, so the absolute scatter was similar. At 1:16,000 it sat on a signal of only 0.20 to 0.40 OD, and that is what pushes the CV above 20%.
Two wells per level is a thin basis for a CV. Read these figures as a pointer and add replicates before you lock a dilution.
Step 6: Pick the dilution that matches your working range
No dilution won every column. The right one depends on the range you need to read.
| Dilution | Zero-well OD | Top of the curve | CV at 0.002 and 0.02 ng/mL | Where it fits |
|---|---|---|---|---|
| 1:4,000 | 0.636 | Flat at OD 3.86 to 3.92 from 2 ng/mL | 4.0%, 5.6% | Little reason to pick it here: 1:8,000 covers the same low end with about half the background |
| 1:8,000 | 0.328 | Still rising, but flattening above 20 ng/mL | 6.5%, 5.7% | 0.02 to 200 ng/mL |
| 1:16,000 | 0.129 | Still rising at 2,000 ng/mL | 23.1%, 30.3% | 0.2 to 2,000 ng/mL, when the lowest levels are not the target |
Table 3. How the three dilutions compare in our run.
Our reading: for a working range of 0.02 to 200 ng/mL we would start at 1:8,000. Its CV stayed at 6.5% or lower at all eight levels, and the OD still rose at every step up to 2,000 ng/mL. If your samples can land above 200 ng/mL and you would rather not re-dilute them, 1:16,000 buys the range at the price of a noisy bottom end. We did not test anything between the two. A run at 1:12,000 is the obvious next titration.
Re-Titrate Whenever the Conjugate Lot Changes
A dilution factor is a ratio against one stock. Labeling yield and antibody recovery shift from lot to lot, so 1:8,000 of one preparation is not 1:8,000 of the next. Rerun the eight-level series on every new lot and compare the zero-well OD and the top of the curve with the old one. Our list of HRP conjugation mistakes covers common reasons lots differ.
What this run does not show
- One run, duplicate wells. We have no between-plate, between-day or between-lot data for this conjugate yet.
- No no-conjugate control. The zero-well signal is not split between conjugate, substrate and plate.
- S/N is not a detection limit. A formal LoB and LoD needs replicate blanks and low samples. Our residual Protein A CLIA data shows how we do that in a CLIA, with 20 blank wells under CLSI EP17.
- Protein A in buffer, not in antibody. Real residual Protein A samples contain product IgG, and Protein A bound to IgG can hide from the assay. The published method above dissociates the complex by heating with detergents and chelators before the ELISA. That pretreatment was not part of this run.
- Three dilutions only. Nothing between 1:4,000 and 1:16,000 was tested beyond the three points shown.
For the plate-level basics behind this layout, see our guide to sandwich ELISA development.
Frequently Asked Questions
What HRP conjugate dilution should I start with in a sandwich ELISA?
There is no universal number, because the factor depends on the stock concentration of your conjugate. Check the supplier's recommended range, then run three or four dilutions that bracket it against a full analyte series that includes zero wells. For our Protein A conjugate we tested 1:4,000, 1:8,000 and 1:16,000, and we would start at 1:8,000 for a 0.02 to 200 ng/mL range.
How do I know my HRP conjugate is too concentrated?
Look for a high zero-well OD and a flat top. At 1:4,000 our zero wells read 0.636, and the OD stayed between 3.86 and 3.92 from 2 to 2,000 ng/mL. If the zero wells use up a large share of your reader's range and the top levels stop separating, dilute the conjugate further.
Does a higher signal-to-noise ratio mean a lower detection limit?
No. In our run the 1:16,000 conjugate had the highest S/N at 2,000 ng/mL (24.9, against 6.1 at 1:4,000) and the lowest at 0.02 ng/mL (3.1, against 4.2). A detection limit comes from replicate blank and low-sample wells, not from the ratio of two means.
Why did a higher dilution give a better S/N at the top of the curve?
The zero wells fell faster than the high-level wells. From 1:4,000 to 1:16,000 the zero-well OD dropped about 4.9 times (0.636 to 0.129), while the 2,000 ng/mL reading dropped only 1.2 times (3.863 to 3.209), and the 1:4,000 reading was capped near the top of the reader's range.
Why use a chicken anti-Protein A antibody for detection?
Chicken IgY is not bound by Protein A, so the labeled antibody is not picked up through its Fc region. A published multi-product residual Protein A ELISA uses chicken IgY for this reason. The chicken anti-Protein A antibody (IgY) page lists specifications for the unlabeled reagent.
Do I need to re-titrate when I change conjugate lot or substrate time?
Yes. A dilution factor is tied to one conjugate preparation, and the OD you reach also depends on substrate time and reader range. Rerun the eight-level series on each new lot and compare the zero-well OD and the top of the curve with the previous lot before you release it.
Can Sekbio supply reagents or help set up a Protein A ELISA?
Yes. We supply Sekbio Protein A and a chicken anti-Protein A antibody (IgY), and we develop antibody and immunoassay reagents. See our immunoassay development platforms, or request a quote and tell us your target range so we can plan the titration with you.
Bottom Line: Three Dilutions, Three Trade-Offs
In our run, 1:4,000 gave a zero-well OD of 0.636 and a flat top, 1:16,000 gave the widest range with a CV above 20% at the two lowest levels, and 1:8,000 sat between them with a CV of 6.5% or lower throughout. Which one is right depends on the range your samples need.
Your next step is the same titration on your own capture layer and substrate: eight levels, zero wells, a few no-conjugate wells, and one extra dilution between your two best candidates. If you want Sekbio Protein A or a chicken anti-Protein A antibody for it, browse the IVD antibody and antigen catalog or contact the team.