The first thing you'd have to check is to make sure that your analyte isn't interacting with your chip. Have you immobilized BSA to the reference cell or have you just activated and inactivated it? If your analyte is interacting more strongly with the reference cell than the immobilized enzyme, then this will result in a negative curve produced.

It could even be as simple as you may have setup your reference cell as the experimental cell, thereby producing your negative curve.

So the two things you should do are, 1) check to make sure your reference cell is setup properly; 2) check tot see if your analyte is interacting with your reference surface. You may need to immobilize BSA or even periodate-treated BSA in order to make a relevant reference cell.

Sometimes the CM5 and NTA chips are not the best option for binding studies, because the carboxydextrane chains as a sticky hydrogelmatrix sometimes capture the analytes in solution non-specifically. So if the reference surface is less saturated by an immobilized ligand, negative binding curves occur. I also had these cases. So in this cases I started to make my own Biacore chips on plain gold surfaces with n-alkanethiol self-assembling monolayers and I could generate really nice binding curves (for details see this freely available publication: Clin Chem. 2007 Jun;53(6):1137-43). It is not so difficult to prepare such chips and the alkanethiols (you also can choose different lengths and differential functional groups) are cheap.

Hi, as Devon said, i would check first the reference cell. Maybe you have a high signal on it. If it's the case, you should test an other way to make your reference cell.

Mick

A negative signal would imply that the analyte binds more strongly to the reference channel than the ligand channel. This can be the case when the analyte binds non-specifically to the chip surface with a slight difference between the channels.

More commonly it results from a high bulk refractive index change. If your small molecules are dissolved in DMSO or your antibody has glycerol or something else in the buffer then the mismatch with the running buffer will cause a high refractive. A small difference in response between the reference and ligand channel can lead to negative curves.

So check the absolute response in your raw sensograms. If there is a high response in the reference channel then check the buffers of the analytes. If they are ok, is there non-specific binding?

If none of the above is the case and you are working with high concentrations (above millimolar) of sugars, you are changing refractive index by the sugars that hang around longer than wanted in the matrix, after their short-lived specific interaction with the immobilized ligand. Short-lived because of low affinity, which is why you use high concentrations of analyte, but then millimolar affinity is a bit too low to measure using Biacore.

This phenomenom is not matched by the reference cell when there is indeed a normal absolute response in the reference cell, because there there is no specific interaction. Thus the negative curves in the measurement really mean something and, in contrast to what you do when correcting for DMSO negative signals, you could take them for real, especially if another technique outputs affinities in the same range.

Test options are 1) increase the flow rate and thus decrease the refraction due to prolonged presence of analyte in the matrix (also called mass-transfer effect)

2) reflect about the optical properties of your analyte, for a change immobilize the analyte?

I agree with the previous answers regarding the reference channel. A higher response in the reference channel with respect to the ligand channel will result in a negative response. Further the reference channel should compensate for a high bulk refractive index shift and additionally a way to compensate for this effect is by double referencing with an sample injection that contains only the (high refractive index buffer, e.g. DMSO) without analyte. Because of sensitivity differences between the ligand channel and the reference channel a rest response still exists and by subtracting this response from the analyte response the data will be cleaned up using this double referencing procedure.

In theory a reference channel without immobilized ligand is more sensitive to bulk refractive index shifts because the total evanescent field of the reference channel is replaced by the high refractive index buffer, while at the ligand channel only a part of the evanescent field is replaced by the high refractive index buffer and the ligand (with higher RI) remains and is not replaced. Effectively the reference channel overcompensates for a RI shift. A way to reduce this, except double referencing, is to reduce the ligand density, but a drawback is that the low molecular weight compound cannot be measured anymore.

Further to express off-rate is using kd and equilibrium dissocation constants is with capital KD . Kd does not exist in the affinity vocabular. In order to measure high quality affinity constants I refer to the paper: R.B.M. Schasfoort, W. de Lau, A. van der Kooi, H. Clevers, G.H.M. Engbers, (2012) Method for estimating the single molecular affinity. Analytical Biochemistry, 421, (2) 794–796

I agree with the previous answers regarding refractive index changes. In my experience, this is the more common cause of negative curves. However, large analyte concentrations can both cause this and also cause negative curves through nonspecific adsorption as has been mentioned. I would run a dilution series in a fresh set of cells to see if the problem persists.

I must thank you all for your kind answers. I'm going to evaluate our situation taking into account your suggestions, and in case of further doubts I'll come back to ask again.

Thank you again!

Yes, I agree that reference cell problems, bulk refractive index changes, and low-affinity or non-specific interactions may cause these negative curves. Also conformational changes in the dextran matrix or the protein may cause negative curves. Most of these problems may be solved by using Label-Enhanced SPR technology, i.e. using dye-labelled analyte analogues together with algorithms that make the mesurement specific with respect to the labelled species. www.episentec.com

Given that you have tested the suggestions above and suspects that you have non specific binding, you could test other vendors having better surface with respect to non specific binding. For instance the LNB surface from Attana.

Another consideration to add to the excellent ideas already described: dissociation of your immobilized enzyme. Since I'm guessing you have a His-tagged enzyme, it is possible that there is a difference in immobilized protein dissociation kinetics between the sample channel and the blank. This can lead to a negative shift. Are you using a control protein or just a blank NTA surface in your reference channel? Is the net negative response immediate (like a bulk RI shift) or gradual? I have seen hexahistidine-tagged proteins dissociate from Ni-NTA enough that dissociation rates are impossible to measure accurately, especially for tight binders. You could try a 10X histidine tag instead on your enzyme if you think that this is happening, or you could try covalent immobilization with better cross-linking chemistry than NTA/His.

I agree with Matthew. Much in SPR biosensorics depends on the choice of the right immobilization method. I also used the NTA chips and it was really frustrating to immobilize 6x His-tagged proteins. I never get a stable baseline and after removal of the bound ligand you have to immobilize your ligand again. If you want to compare the binding curves this is completely inacceptable. So my advice is to covalently couple the ligand to the surface when ever possible. If the ready-made chips do not offer the right functional groups, use alkanethiols. Not everything depends on EDC/NHS, there are a lot of other coupling chemistries available. There are blank gold chips available (AU chips) and also chip supports to externally assemble the chips. Sometimes the plain SAM surfaces composed of alkanethiols are a better choice than the dextran hydrogelmatrix in respect to better accessibility of functional groups for immobilizing the ligand and because of prevention of non-specific binding during ligand binding of the analyte. Another advantage is that these SAM surfaces are very stable and you can clean the surfaces very rigorously (i.e. with NaOH, Guanidine-HCl, SDS).

I agree with previous answers, it just means that you have more signal in the reference cell than with your immobilized ligand. It's sometime the same with ELISA control. When possible, I try to immobilise streptavidin in both Flowcell then inject biotynilated ligand in one flow cell. Usually I do not have much background in the reference coated by streptavidin. Good luck.

I agree with previous answers, it just means that you have more signal in the reference cell than with your ligand. It's sometime the same with ELISA control. When possible, I try to immobilise streptavidin in both Flowcell then inject biotynilated ligand in one flow cell. Usually I do not have much background in the reference coated by streptavidin. Good luck.

The negative signal is usually result of more binding to the reference cell either unspecific or it could also be a bulk effect due to differences in the sample and the running buffer. From my experiences is that antibodies has low unspecific binding to CM5 chips in Buffers with 150mM NaCl and 0,005% p20 detergent.

With small molecules it is common that you dissolve them in DMSO. DMSO has a high refractive index and then you need to do a DMSO calibration curve to get the correct reference subtraction. It can also be of course that you have unspecific of the small molecules to the reference cell. Changing to a different type of chip can solve it. If that does not work I have had success with adding CM-dexstran in the sample and running buffer to eliminate the unspecific binding. BSA in the sample and running buffer has also worked for me to remove unspecific binding.

The negative signal can also be the result of unstable surface, NTA chip are usually not stable with his tag alone to circumvent that I activate the chip first with NHS/EDC the use the histag to concentrate the protein to the surface to immobilize it with amine coupling. This gives a stable surface and the advantage is that you do not need to expose the protein to low pH for concentrating the protein to surface for the normal amine coupling to CM5 chip that could inactivate some proteins.

I have seen unstable surfaces on Amine coupled proteins on CM5 chips. When this has been the case the immobilized protein has been a dimer that has dissociated over time. If that is the case SPR could be problematic and I would start to investigate for alternative methods.

Hi

I have a question Kristoffer. You said that, adding BSA in running buffer as well as in sample reduces the non specific binding on the surface. What is the final concentration of BSA should I use ? If I have to immobilized BSA at the reference surface what should be the target RU?

Hi

I would start with the concentration of 1mg/ml of BSA. If 1mg/ml does not solve the problem you could try to increase the concentration to 5-10 mg/ml. I recommend that you sterile filtrate the BSA containing running buffer since bacteria grows easily in it. BSA in the sample and running buffer does not always work to remove the unspecific binding. For BSA immobilization on the reference, I would start to immobilize around the same RU as your ligand

Thanks Kristoffer

Hi Schott

Another possibility to check, in the case of sensor chip CM5, is the reference cell was blocked with ethanolamine using amine coupling method. I have seen two cases where this procedure was not performed and generated data similar to that reported.

best regards

Hi Scott

I was in similar situation and good negative curve. I was told that you get negative curves when analyte binding is more on the reference surface compared to active surface, which means that analyte is non-specifically binding to the chip surface. In this case I would not trust my data. I hope that would help.

You can always try with another technology to see if they provide you with the same results or of the potentially are better for your curent applications. You can for instance test with Fortébio's interferometry technology or Atana's QCM technology.

I agree with Teodor but to check what really is happening in the Biacore, one should measure it with SPR. A ligand density series from very high to undetectable low should be immobilized including referencing using various ligands as BSA, etc. For this purpose one may consider to use the IBIS MX96. In this instrument a researcher can spot 96 different ligands using continuous flow microspotting. In this way many cross checks can be build in the experiment. Google to ibis-spr and you will observe the capacity of this instrument.

Someone recommended using 5-10 mg/ml BSA to block non specific binding. In my experience 0.1-0.2 mg/ml is usually sufficient, and I have found that going above 1 mg/ml BSA can clog an IFC (even when sterile filtered, ultracentriuged, etc., at least in a T100 and Biacore 3000.). This was also the advice of the biacore application specialists. If you do put BSA in the running buffer, then be sure to flush it out of the system and perform a sanitize at the conclusion of your experiment.

Hi everyone! I have read all your comments and I am in almost the same problem. I want to do a binding assay between the protein and its receptor. So, the protein that I will inmobilize to the CM5 chip will be the receptor that is specific for my ligand of interest. The ligand is a recombinant protein that will be a drug, so in all of cases, the pharmaceutics use BSA like a stabilizer, so they add in the excipient. I think that BSA can not bind to the receptor because the receptor is specific to one protein, but BSA can bind to the dextran matrix of the chip, so I thought to inmobilize the receptor on the ligand channel and inmobilize BSA on the reference channel at the same RU, then I was thinking to block with ethanolamine using amine coupling method. Thus, when I will pass the fluid with my protein of interest that also contains BSA, only the protein of interest binds to its receptor and probably BSA wont bind to the dextran matrix because there wont be more free sites.

Do you think my idea is OK? Please let me know your opinions. It will be helpful

Thanks everyone!

Hi Claudia, I agree withTeodor, you can try another technology as well. One of the technologies that can work for you is MicroScale Thermophoresis, form NanoTemper Technologies, that allows the study of protein-protein interaction in solution (not ned to immobilize).

I work for NanoTemper Technologies in Latin America and right now I´m organizing a trip to Mexico, I will be at the Mexican Biochemistry meeting on November and doing demos afterwards. If you are interested to know more about it let me know I will be glad to visit your institute, give a semianr, measure some samples.

Saludos cordiales,

Daniel Maturana

NSB solution is available from Biacore i.e Carboxy Methyl dextran solution) incubation of analyte with this solution can minimize non specific interaction in reference channel. 

 Hi, Claudia

i agree with you. BSA contained buffer will decrease non-specific binding. but BSA can bind to the dextran matrix of the chip. However, i don think your experiment design is OK. because if you immobilize the  receptor on the ligand channel and immobilize BSA on the reference channel at the same RU. the BSA immobilized surface can be take as reference cell.  Because the relative receptor immobilized level will proberlly less than 0. i think analyte pass order should be reference channel-BSA-receptor . 

Hi, 

Suggested answers are really useful and can improve your result. I want to add that try your experiment with different buffer and prepare your sample in that same buffer. I was facing some unusual binding of analyte so I tried with different buffers and got improvement in results.

The new Handbook of Surface Plasmon Resonance 2nd edition is available.

In chapter 7 authored by Robert Karlsson of Biacore the ABA strategy of switching and testing buffers is reveiled. Success !

negative sensorgrams show up when binding to reference channel is higher than the active (with ligand) channel. this happens if the analyte has affinity towards dextran matrix. you can rectify this issue by using other sensor surfaces (like CM3-has shorter dextran; C1 chip- no dextran; PEG chip- PEG matrix). For your assay I will suggest using sensorchip PEG (here only the matrix is changed, ligand coupling chemistry remains same).

Thanks for the inputs...i had a query...i am studying the interactions of BSA with small molecules. In this case if i use BSA in the running buffer, there is a chance of my analyte interacting with BSA (which is in running buffer) and may results in the false positive results...what shall i do..

thanks in advance.

I have a question: Does the sensorgram of the negative response look like the attached figure? I use capture and my ligands are heat-treated. I would to check if it is normal to not see the capture curve on the sensorgram on treated samples. I have used a control (non-treated samples) and the curves look pretty (positive RU).

Thank you