Experimental methods

Protein production and purification

TCR and HLA-DR4 were produced in vitro from inclusion bodies found in E. coli. Affinity maturation of CD4 was undertaken using yeast and 12 residues in the D1 domain were mutated using degenerate primers to generate a mutant library. This CD4 library was then displayed on yeast cells and was probed by flow cytometry. This allowed mutants with increased affinity to be found. One specific mutant which contained 2 mutations produced a full length protein and this protein was then fused to various signals and tags, including a C terminal FLAG tag. The protein was purified using affinity chromatography. Other purification methods were then employed to further purify the protein.

Surface plasmon resonance

 

This was used to analyse the interaction between the affinity matured CD4 and HLA-DR4. The HLA was tagged and attached to a chip and different concentrations of CD4 were then injected onto these chips or onto buffer as a control. The data was fitted with a 1:1 binding model and the dissociation constant was then obtained by computational analysis. An example of a sensor chip is shown below, in Fig. 8, and in our example the bound ligand is HLA-DR4 and the analyte is CD4.

Figure 8: a sensor chip showing covalently or non-covalently bound ligand

 

Crystallisation and data collection

TCR, HLA-DR4 and CD4 D1-D4 were mixed in equimolar concentrations and then the solution was concentrated until the ternary complex began to crystallise spontaneously. Crystals up to a size of 0.2x0.1x0.1mm were transferred to a cryoprotectant solution and were then flash-cooled in liquid nitrogen. In silico analysis was then performed and one ternary complex was found per asymmetric unit.

Structure determination and refinement

The TCR-MHC-CD4 complex was solved by molecular replacement using MS2-3C8-MBP-DR4 and the D1-D3 domains of CD4 as search models. Various computer programs were then used to build models and refine the structure. The final Rwork and Rfree values compared well with those for many other similar resolution structures deposited in the protein data bank in the past 3 years. Rwork and Rfree values show how similar the experimental x-ray crystallographic data is to the data produced during the refinement process. 

Rwork = 23.8%
Rfree = 30.5% 

These values were low however the actual resolution of the structure was 4.0 angstroms which is considerably lower than the best resolution structures currently being produced. This shows that although the researchers had been able to produce a structure for the complex there are many improvements which can still be made.