A negative response with SPR.

Sometimes a negative binding response is seen in a SPR experiment. In general, negative responses are visible after reference subtraction meaning that the reference surface binds more analyte than the active channel. However, more and more experiments show that in some situations real interaction can give negative SPR responses (1).

This document is divided in two parts:

• Negative responses due to experimental conditions
• Negative responses due to real interactions

Negative responses due to experimental conditions

Negative SPR responses due to experimental conditions are mainly by buffer mismatch and non-specific binding of the analyte to the reference surface.

Buffer mismatch

Some negative binding responses originate from buffer mismatches. In general, a low ionic analyte solution will give a negative jump compared to the flow buffer. A buffer mismatch of 1 mM NaCl will give a jump of 20 RU in a Biacore type of instrument on a carboxylated dextran sensor chip. In addition, all components in the analyte solution contribute to the buffer mismatch. Component such as glycerol, DMSO and high salt content give large positive response but sometimes an analyte is provided in low salt solutions. Therefore, the dilution of the analyte in the flow buffer may alter the total salt content of the sample to be injected. The buffer mismatch can be eliminated by dialysing the analyte.
However, if the problem persists and is proportional to the analyte concentration it is most likely originating from the analyte itself (see below).

Immobilisation

The most used immobilisation on a carboxylated dextran matrix is the amine-coupling. This procedure uses a low ionic strength solution with a lower pH than the flow buffer. During injection of the ligand the baseline drops sharp and when pre-concentration occurs will go up again. When the ligand injection finishes the baseline returns with an added response when the ligand is covalently bound. In general, non-covalently bound ligand is washed away by the flow buffer. This type of negative response is normal.

Volume exclusion

Some negative binding responses originate from different behaviour of the reference and ligand channel to the injected analyte solution (buffer composition, additives, pH, analyte concentration). This is explained by the difference in ligand density on the sensor chip. When immobilizing, the actual amount of the ligand differs between the sensor surfaces. This difference is not only in response units or in molarity but also in the volume the ligand is occupying in the matrix. When the flow buffer is replaced by the analyte (buffer), the matrix can swell or shrink depending on the difference. Because of the different ligand volume in the matrix, the volume exclusion of the buffer is different, hence the reaction on buffer changes (2),(3).

In situations with for instance DMSO or glycerol in the analyte buffer, differences can be relatively large and result in negative responses in the original curves, after reference subtraction or double referencing. In these cases, it can be beneficial to make a calibration plot to compensate for the volume exclusion (2),(4).

Non-specific matrix interaction

Some negative binding responses originate from the interaction of the analyte and sensor chip matrix. Especially the dextran-based sensor chip can have high affinity for small compounds. Adding CM-dextran (0.1 - 1 mg/ml) to the flow buffer can lower the non-specific interaction with the dextran matrix. Other solutions are to use sensor chips without dextran or other types of linkers, like alginate (Bio-Rad) or dendritic polyglycerol (Xantec). It is even possible to make your own sensor chip surface. In this way you can add a SAM to the sensor gold layer that will suit you best (5).

Non-specific reference interaction

Some negative binding responses originate from non-specific binding of the analyte to the ligand on the reference channel. This should be obvious by overlaying the raw data of the reference and ligand channel. From the sensorgram you can sometimes deduce if this is a strong or weak non-specific interaction (observe the dissociation). If the interaction seems to be weak, adding extra detergent (e.g. 0.02%), extra salt (e.g. 250 mM NaCl) or changing the pH can suppress this interaction. Adding BSA (0.1 – 1 mg/ml) or CM-dextran (0.1 - 1 mg/ml) to the flow buffer can lower the non-specific interaction. In addition, reversing the interaction system can be a answer to this problem.

The reference surface is in all situations crucial. The easiest surface is one which is native and thus unmodified. Taking the carboxylated dextran based sensor chips, the carboxyl groups (-COOH) become negatively charged at pH below ~7. Activating with (NHS/EDC) and deactivating with ethanolamine replaces the carboxyl groups with a hydroxyl group (-OH) which are less negatively charged at a physiological pH.
A better reference channel contains a non-related ligand (protein, oligo)but finding one is not always straight forward. Often BSA is chosen because it is readily available and pure. However, BSA can bind a lot of molecules and should not be the first choice. In addition, albumin (when ionized in water at pH 7.4, as found in the body) is negatively charged.

Other molecules that can be used are for instance non-related antibodies (IgG). Most laboratories have IgG for detection of proteins of interest and one can choose an antibody that is not binding to the analyte of interest and use as a reference surface. The selection of a proper reference channel can be a challenge and will cost some time.

The amount of ligand to be immobilized as a reference surface is the next difficult part. Start with immobilizing an equal amount of response units compared to the ligand of interest. This will give a similar buffer volume exclusion compared with ligand of interest. Only testing and comparing will show if you have to immobilize less or more.

Solving negative curves

Steps you can take to test the suitability of your reference channel (you can make a dedicated chip for this):

• Inject the analyte at the highest concentration over a native surface
• Inject the analyte at the highest concentration over a deactivated
• Inject the analyte at the highest concentration over a protein surface

Steps you can take to reduce non-specific binding and lower difference between reference and ligand channel

• Add extra detergent or salt to the flow buffer
• Add BSA or CM-dextran to the flow buffer
• Add blank injections for double referencing
• Add calibration injections and make a calibration plot to correct for volume exclusion effects

In addition, consider reversing the interaction system and different sensor chip matrices.

Negative responses due to real interactions

There are more and more reports that some real analyte-ligand interactions give negative SPR responses (1). And although in the past measurement of conformational changes with SPR was not accepted it seems that some interactions can only be explained by changes of the refractive index at the sensing interface.

In the experiments of Crauste, et al. the mycobacterial transcriptional repressor EthR and a mutant EthRG_106W were immobilized on a sensor surface and several analytes where tested. Several know binders produced a dose-dependent negative interaction curves while control analytes gave no significant response. In addition, only the EthR coated surface gave this behaviour and the EthRG_106W mutant not. In control experiments it was shown that more immobilized ligand gave a deeper negative response. The negative curves were multiplied by -1 to obtain positive curves and analysed as done normally. All kinetic result of the different surface densities and equilibrium fittings reported equivalent values.

It is proposed that the structural changes upon analyte binding are the driving force of the negative dose-dependent responses (1).

Two more publications which may be of interrest are:

Note: Although these publications show the possibility of actual measuring negative binding responses or possible detection of conformational changes at the sensor surface, it is imperative to show by other measurement techniques that the interaction/conformational change is real.