I use a home-made sandwich ELISA (96-well format) to detect the HIV p24 protein in cell lysates and virus-containing supernatants. The setup is as follows:
96-well Immulon 2HB polystyrene microtiter plate, coat at 4C overnight with monoclonal mouse Ab
Wash 4x (PBS/0.2% Tween-20) then block with wash buffer/10% FBS, 60 min (everything from here on is at room temp)
Samples (kept on ice) are diluted in sample diluent (PBS/1% Triton X-100/10% FBS), transferred into a separate 96-well tissue culture plate, and serially diluted twice in the plate using a multichannel pipettor
Wash 4x and load samples using multichannel pipettor from tissue culture plate into ELISA plate, 90 min
Wash 4x, human primary Ab @1:8000 in sample diluent, 45 min
Wash 4x, anti-human-HRP secondary Ab @ 1:8000 in sample diluent, 45 min
Wash 4x, TMB substrate (Thermo), until standard curve has developed well, stop with H2SO4, read at 450nm, blank on sample diluent alone.
I have recently noticed that samples I load on the left side of the plate tend to have higher absorbance than samples loaded on the right side. In order to confirm this, I multichannel-loaded all 96 wells with the same sample out of a trough and looked for drift. I was able to detect a smooth gradient running from left-to-right, with samples at the right edge of the plate being estimated about 1.3 times lower than samples at the left edge (I have seen variability across the plate as high as 2-fold in other cases).
In parallel to that plate, I did the exact same thing on another plate except for loading the plate right-to-left (i.e. the opposite of what I normally do). The gradient in that case followed my direction of loading, and not the plate's inherent orientation. This suggests that the gradient is not due to the plate but due to some aspect of the handling.
I have been trying to troubleshoot this and have thought of the following possibilities:
- Pipetting inconcistencies: I use a multichannel pipettor for all the steps. I change tips and forward-pipette (i.e. aspirate from first stop, expel to second stop) when loading my samples. I use the same set of tips but reverse-pipette (i.e. aspirate from second stop, expel to first stop) out of a trough for coating, washing, blocking, antibody and reagent steps. I am confident that by reverse-pipetting, I should be pipetting the same volume in each well (I have weighed water pipetted onto parafilm and can very consistently get the correct weight when I reverse-pipette).
- Washing: Tiny inconsistencies during pipetting might be amplified due to the high number of washes in the same direction. I therefore now flip the plate over by 180 degrees for every other wash. This should eliminate any directionality from washing as an equal number of washes are done in each direction.
- Time it takes to load the plate: I have timed myself and am able to load a 96 well plate with reagent in 23 seconds. Considering that most of my reagents are incubated for an hour or more, 23 seconds should not contribute at all to any variability. I have even optimized my sample loading step by using the intermediate extra 96-well plate (instead of using individual tubes) so that when it comes to loading samples, it is also very quick (usually about 2 minutes per plate).
- Incomplete mixing of reagents: While this is a possibility, there is no reason I can think of for this to result in a consistent left-to-right gradient. It could contribute to variability in random positions across the plate, but not a consistent directional drift.
- The sample is sticking to the intermediate 96-well plate, or the trough (in the case where I loaded the whole plate with the same sample): Once again, I pipette fast enough that I really doubt that between pipetting the leftmost wells and the rightmost wells, 25% of the protein in the sample has stuck to the trough.
- Bad batch of plates: This should be addressed by the fact that I loaded two plates in parallel but in opposite directions and the gradient followed my loading direction, not the plate direction, so it is probably not the plates.
Sorry for the long post. I am at a loss right now and cannot afford to waste plates and reagent with wild stabs in the dark. I am hoping that someone out there has seen this issue before, or can think of something I haven't thought of. I am hoping that it is something really simple that is just staring me in the face. Thanks.