March 15, 2009 (Vol. 29, No. 6)
Eric J. Sundberg
SPR Facilitates the Understanding of Relationships that Can Lead to More Accurate Predictions
Nearly all cellular processes depend on protein-protein interactions. Therefore, understanding protein function and protein-protein interactions is key to gaining insights into the biochemical mechanisms that underlie disease and developing new drugs. Our current understanding of protein molecular recognition, however, remains far from comprehensive and impedes the drug development process.
Label-free biosensor-based surface plasmon resonance (SPR) systems are now commonly used in the study of diverse biomolecular interactions, and have become, perhaps, the most widely accepted technology for quantifying their kinetic parameters, as these instruments allow real-time measurements. SPR analysis can be used, not only to provide fundamental insights into molecular recognition but throughout the protein therapeutic development process.
Using SPR systems, one interaction partner is immobilized on a sensor chip surface, while the other flows over the surface via a microfluidic flow path. Binding is monitored in real time via changes in the refractive index that are proportional to alterations in mass concentration at the surface. The technique is noninvasive, enabling direct analysis in buffer, as well as in complex media such as serum or cellular extracts. The resulting plot of binding response over time is known as a sensorgram; it provides comprehensive quantitative information on the entire interaction process, including the kinetic parameters of the interaction. Additionally, when SPR analysis is applied at numerous temperatures to a particular molecular interaction, thermodynamic parameters can be determined.
Our lab has used SPR technology extensively (Biacore™ 3000 from GE Healthcare), to investigate the molecular basis of disease caused by a family of bacterial toxins known as superantigens (SAGs) and to develop therapeutics against them. SAGs are proteins secreted predominantly by Staphylococcus aureus and Streptococcus pyogenes that bind simultaneously to class II major histocompatibility complex (MHC) and T-cell receptor (TCR) proteins on the surfaces of antigen-presenting cells and T lymphocytes, respectively (Figure 1a).
SAG engagement of these cell-surface receptors leads to the hyper-proliferation of T cells and a systemic release of inflammatory cytokines. This results in a condition known as toxic shock syndrome that is characterized by high fever, erythematous rash, hypotension and, eventually, multiorgan failure and death.
Currently, no drug or vaccine exists to specifically inhibit SAG-mediated disease. Thus, as a multilaboratory collaborative team, we have embarked on a strategy in the last several years to engineer protein therapeutics that abrogate the interactions between SAGs and their TCR-binding partners, thereby inhibiting SAG-mediated T-cell activation and disease (Figure 1b). Although SAGs bind to both MHC and TCR molecules in numerous ways, they nearly always bind to a single immunoglobulin domain of the TCR called the Vb domain (e.g., the variable domain of the TCR b chain). We have used yeast display directed evolution methods to create variants of TCR Vb domain fragments that bind specifically to individual SAGs, but with vastly increased affinities than the wild-type TCR molecule from which they originate.
Yeast display is a directed evolution technique similar to phage display in that iterative rounds of random mutagenesis and the selection are carried out to mature the affinity of the protein of interest. Yeast display is distinguished by the fact that the protein that is of interest to evolve is displayed on the surface of yeast cells and the selection of improved binding variants is carried out by cell sorting in a flow cytometer.
Our first SAG target was staphylococcal enterotoxin C3 (SEC3), and numerous rounds of yeast display-directed evolution applied to one of its TCR ligands, the mouse TCR Vb8.2 domain (mVb8.2), resulted in a highly mutated and affinity-matured variant TCR molecule. To determine precisely the degree of improvement between the affinity of the variant versus the wild-type mVb8.2 molecules, we used SPR analysis to measure the affinities and kinetics of binding to their common SAG target, SEC3.
As shown in Figure 2a, the mVb8.2 variant incorporated mutations at nine amino acid positions and bound SEC3 with an approximately 1,500-fold increased affinity relative to the wild type, largely through a reduction in the off-rate of the interaction. Once we observed such a large difference in affinity between the wild-type and variant mVb8.2 molecules, we then went on to show that the affinity-matured variant could act as a competitive inhibitor of SEC3-mediated T-cell activation in vitro.
This mVb8.2 variant was an ineffective inhibitor of SEC3-mediated T-cell activation and disease in vivo, however. We suspected that the still relatively low affinity of the variant (with a KD in the nanomolar range) was impeding its efficacy, and thus, we set out to further improve the affinity.
While most directed evolution applications maintain sequence length while altering the sequence itself, we had strong evidence from some of our previous x-ray crystallographic and SPR analyses of a related SAG-TCR complex, that of streptococcal pyrogenic exotoxin C (SpeC) binding to the human TCR Vb2.1 domain (hVb2.1), which provided us the impetus to rationally alter our yeast display methods to “grow” a specific region of mVb8.2 toward its SAG-binding partner.
We did this by selecting variants from clones that encoded longer, but still randomized, sequences of a particular loop. This time we targeted the SAG staphylococcal enterotoxin B (SEB), which is highly homologous to SEC3 and also binds to mVb8.2. Variants of mVb8.2 that exhibited longer loops consistently bound SEB with relatively higher affinity, as determined by SPR analysis (Figure 2b). We found that one of these variants bound SEB with an affinity of 50 picomolar, again primarily due to reduction in the off-rate, and was effective in protecting animals from lethal injected doses of SEB (Figure 2c).
The use of SPR analysis of SAG-TCR interactions has been essential for pushing our work into the development of novel therapeutics, but has also helped bring us from applied back to basic research, allowing us to address some fundamental questions in protein-protein interactions.
Protein-protein interfaces are structural and energetic mosaics. Within these interfaces, a subset of amino acid residues contributes to a higher proportion of the binding energy, while many other residues are energetically silent. These residues are termed hot spots and contribute disproportionately to the affinity of the complex.
The binding energetics of protein-protein interactions are networked, in that some residues communicate, resulting in protein-protein binding energies that are not necessarily a sum of the binding energies from the individual residues within the interface. Furthermore, energetic networks within protein-protein interfaces display a degree of fine structure with numerous sub-networks existing within a single interface.
Several groups have shown that hot-spot residues tend to cluster together into hot regions rather than being distributed evenly across the protein-protein interface. The binding energies between residues within a hot region are thought to be cooperative, while energetics between residues in distinct hot regions are proposed to be strictly additive.
By using combinatorial mutagenesis and SPR binding analysis of yet another yeast display evolved TCR molecule in complex with its SAG target, in this case affinity-matured variant of hVb2.1 binding to toxic shock syndrome toxin-1 (TSST-1), we were able to dissect energetic additivity and cooperativity in this protein-protein interaction. The general experimental design is shown in Figure 3a. We showed that residues within distinct hot regions, and at distances >20 Å apart, contribute in a significantly cooperative manner to the binding energetics of the complex (Figure 3b).
Our results suggest that a broader consideration of cooperativity within protein-protein interactions may ultimately lead to more accurate predictions of these interactions. The presence of hot regions within these relatively planar interfaces offers prospective binding sites for small molecule inhibitors of protein-protein interactions, which have previously proven difficult to develop. If, as our results suggest, some distinct hot regions are linked energetically, the potency of a small molecule inhibitor that targets a cooperative hot region may be amplified relative to a small molecule region targeting one that is strictly additive.
In our research, SPR analysis of protein-protein interactions has been critical to our basic understanding of the molecular basis of SAG-mediated disease and has enabled us to capitalize on this understanding to develop novel protein therapeutics. Coming full circle, SPR analysis has provided an avenue by which to take the end products of our drug development process to address some outstanding fundamental questions in protein molecular recognition.
Eric J. Sundberg ([email protected]) is a scientist at Boston Biomedical Research Institute.