What’s your question?
To avoid wasting time and money, it is critical you start with the end in mind, and spend time clearly deciding what question you’re looking for an answer to. Do you want to know the kinetics of a single interaction or against a large panel of antibodies? Do you just want a yes / no screen of a panel of antibodies or targets? Do you want to know the affinity of an interaction but don’t mind if you don’t know the kinetics? Do you want to toggle-switch epitope select antibodies?… The list is endless! Building on the information from ‘know your protein ‘ I’ll show how I would set up an assay from scratch and what pitfalls I try and avoid.
“What do I want to know?”
No matter the assay, I plot them out and it always starts with that very simple question. As a very visually orientated person,I work by drawing the molecules of interest interacting together and then capture that down in a more structured way. In this article, I’ll walk through an typical project, in this case VEGF binding to bevacizumab:
“By the end of this study I want to know the kinetics of bevacizumab to the VEGF isoforms 121, 165 and 189. I am going to use the VEGF isoforms as the ligand and use amine coupling to attach them to the sensor chip surface. As I am looking at a high affinity interaction (antigen binding to a FAb) I want to avoid avidity issues so I will look at keeping Rmax low and will choose a Sensor Chip Series S C1 as my starting point because it contains no dextran chains and this movement of the immobilised VEGF will be minimised. I will start with a standard buffer HBS-EP+.“
Keep it clear. Keep it concise.
“How much ligand do I need?”
The first step above is to keep a low Rmax but how much ligand do I need to use to get started? Using the following equation, the amount of ligand required to achieve an Rmax of ~50 is calculated (although I don’t go into detail here about why a lower Rmax is lower, I will cover that in a future blog post):
Which is more commonly expressed as:
As we’re using a sensor chip C1 and a low level of ligand immobilisation we can assume a 1:1 binding of VEGF to bevacizumab and therefore, an immobilisation level as detailed below is appropriate:
Selecting immobilisation conditions
Now we need to find the optimum conditions to immobilise 13.5 RU of VEGF. Although this seems simplistic, it’s important to know what you’re aiming for in terms of immobilisation as you can’t hit a target if you don’t know what you’re aiming for!
If we revisit the table from the ‘know your protein’ section:
You’ll recall that we can use this data to help us to determine the optimum concentration of each VEGF isoform and which acetate buffer to use. As the sensor chip has a negatively charged surface in order to preconcentrate to the surface the protein must be positively charged so you work at 0.5 – 1.0 pH units below the pI and a starting point of 5 – 20 µg/ml of ligand is a good choice.
For amine coupling the standard buffer of choice is Sodium Acetate, which are commercially available from GE Healthcare at pH 4.0 – 5.5.
As all isoforms of VEGF are positively charged at pH 4.0 – 5.5 the choice of acetate buffer is less obvious so it’s a good idea to screen all the available acetate solutions at concentrations such as 5, 10 and 20 µg/ml in order to determine the ability to preconcentrate to the surface. This is commonly performed using pH scouting.
Above is an example of pH scouting of a protein from 20 – 5 ug/mL assessed at pH 4.0 – 5.5. As you can see in channel 8, at low concertation none of 5 ug/mL protein reaches saturation of the surface but as we move left and increase the protein concentration we start to observe surface saturation (pH 4.0 and 4.5 in channel 6) and as we reach the highest concentration in channel two, all but pH 5.5 has saturated the surface. In situations like this it’s tempting to jump straight in at pH 4.0 but an important consideration here is that the longer a protein sits in acid the more it gets ‘eaten’ so it’s best to choose the weakest acetate solution viable to minimise damage to your ligand during immobilisation or you risk immobilising a lot of inactive protein and having a lower Rmax than you predict.
Once all combinations of pH and protein concentration have been tested the best combination can be chosen. Usually it’s best to go low and slow instead of fast and frantic for amine coupling as if you make a mistake in prep then too much protein is added to the surface (especially if you are using an ‘aim for’ approach). So go for the middle ground in the protein concentration and acetate pH.
It’s worth here talking about immobilisation and capture as that’s your next port of call.
Attaching your protein to the surface
In my assay statement before I said I was going to use amine coupling to attach the VEGF to the surface, this isn’t my preferred method of presenting a ligand during SPR but it’s a great starting point to learn about how to build an assay Although there are many types of chemistry available for covalently attaching proteins to your chip of choice, the most commonly used is amine coupling. Benefits of amine coupling include its stability and that the protein does not have to be modified in any way and that as shown above, scouting for appropriate conditions to immobilise is well established. In addition, the amount of ligand required is low and it’s generally easy to hit the same immobilisation target level again. The main drawback that’s associated with amine coupling is random orientation. Due to most proteins having a large number of accessible primary amines there are numerous orientations the protein can be covalently attached to the sensor chip surface. This can cause obstruction or blocking of binding sites, which can lead to Rmax varying a lot between assays (and most usually the kinetic rate constants and KD too) and the low pH solutions required to drive the protein towards the chip surface can cause protein degradation too.
As mentioned in ‘know your protein ‘ It’s important to remember that chemically biotinylated proteins can be captured in a similar ‘random’ orientation if they contain multiple primary amines but even if the protein contains only one primary amine suitable for labelling then there’s no guarantee that capture via the biotin does not occlude the binding site.
Specific ligand orientation
The most common solution to problems with amine coupling is to use my preferred assay setup that uses specific ligand orientation. In this setup the molecules are all orientated in the same way and this helps prevent any differences in epitope presentation.
Some of the most common ways of capturing the ligand include His tag, Fc tag, biotin-streptavidin and capturing via a domain within the test antibody itself, as discussed in the ‘know your protein’ .
Choosing your analyte range
Now we’ve chosen our sensor chip (C1) and have optimised conditions for immobilisation, the only thing we’re missing is the analyte and what concentration to use? It’s important to get this stage correct as equally spaced data that maximises the interaction with the ligand can make or break an assay. If you don’t use a wide enough concentration series then the sensorgram response can be clustered and it becomes difficult to accurately determine kinetic rate constants and thus affinity.
From the literature, VEGF165 exhibits an affinity to bevacizumab of low nM by SPR so this is where the Biacore 8K 2D kinetics capability displays its power.
A wide-ranging concentration screen can be performed simultaneously so we could use a series of concentration series like this:
From this you can choose your best concentration series. Speed in this area comes with practice and time, but even in this example you can see concentration series 1 is too low and most of the ligand on the sensor chip surface is not being used and concentration series 8 is too high as the last injection adds very little to the results. Here, 5 and 6 look like they are worth investigating more as the steps between concentrations are even and the sensor chip surface does not look saturated at the higher concentrations (refinement and possible reduction of the number of injections would help to optimise the analyte concentration series as the lower concentrations add very little to the overall data here).
In addition to too low or too high, an example of an inappropriate concentration series is shown below, where the final injection delivers over 50% of the response:
Therefore, this concentration series should be reviewed to ensure a more even distribution of data.
The basics of SPR don’t change much once you have a good thought process in place. Although this has been a simple worked example, I hope that it gives you more of a feel for how my brain works and you can see the importance of starting with the end in mind. So take the information above, use it to drive better sensor chip preparation and analyte concentrations and watch your assays bloom.
Getting the correct immobilisation conditions can take a while so I’d love to hear any challenges you have had getting your proteins immobilised.