I have a problem with negative signals too. I cover a HPA chip with a lipid layer, run gangliosides on top, which bind to about 200-300 RU and then pass antiganglioside mAbs. Sometimes I get beautiful sensograms some times the signal shoots up and immediately starts going down forming a weird negative profile of a binding curve (no subtraction). After the end of the injection all comes down to baseline. I wish I could demonstrate the curve instead of describing it.
Q: Is it possible this is due to the mAb binding the ganglioside with higher affinity than the ganglioside has for the lipid layer and removing it from the chip?
Thank you, Arnoud, for the useful tips.
I have to make it clear that this effect is specific for the channel coated with the ganglioside, for which the mAb is specific but not on other channels coated with other gangliosides. BSA in the same buffer displays regular sensograms. Most important - different mAb to the same ganglioside displays a regular sensogram. And - yes, the shape of the sensogram is more or less the same any time we run it. Lower concentration of the antibody smooths the effect but as a rule there is a peak followed by a steady decline until the end of the injection.
My other observation is that this mAb has a high dissociation rate compared to the mAb with regular sensogram, which makes me think the explanation could be related more to crowding effects.
Your description gives me more the idea that you have a solution with two antibodies/proteins. One is fast and has weak binding and the second is slow and has an strong binding. The second is replacing the first.
The problem with this theory is, that replacing one antibody with the second should not give the downward curvature because they are approx. the same size.
Some more observations:
- the profile changes from run to run, yielding increasing binding after regeneration and more pronounced negative curvature;
- increasing the flow rate (50 mcl/min) does not remove the effect;
- another antibody (MAB 2) with lower on_rate is much less sensitive and fits well with bivalent binding + mass transport;
- the best fits of both antibodies have beautiful patterns of oscillating residuals (couple of periods per assoc./dissoc. phase), which correspond in their extreme amplitudes to the weird variations in the sensograms for MAB 1;
- increasing the concentration of MAb 1 DECREASES the effect by raising the frequency of the "oscillations" and lowering their amplitude relative to the main trend;
- raising the temperature from 25oC to 37oC changed apparently the kinetic constants for MAb 1 (to be expected) but not the profiles of the sensograms;
- the only way for this effect to disappear seems to be very low level of "coupling" of the ganglioside to the lipid layer (bellow 30 RU), which actually may be the good level since gangliosides have MW less than a 1000 Da.
I don't think heterogenous analyte explains it although the problem you mention of constant size may be solved if the solution contains active fragments of the antibody. Could it be that binding and dissociation rearranges the lipids and the surface has different properties for the next run? Still - whence the dip?