In previous work, including the paper previously discussed, the authors used mirror image phage display to identify D-amino acid sequences that bound the MDM2-p53 binding pocket with greater affinity than p53 (~low nanomolar) [27, 28]. In this work, they aimed to create a D-peptide that bound even better than the ones that they had identified earlier.
Figure 1: Structure of the D-PMI-gamma D-peptide (purple) inhibitor bound to MDM2 (gray surface). The MDM2 residues which directly interact with peptide side chains are highlighted in lime. D-Phe7, a key residue in the interaction, is marked. It can be seen that the phenyl ring does not extend all the way into the cleft (PDB: 3IWY) [27].
The availability of crystal structures for many of these D-peptide MDM2 antagonists enabled first a detailed structural analysis to determine particular residues in the interactions that destabilized the binding between the peptide and receptor. The structures of the D-PMI family of inhibitors revealed D-PMI-β and D-PMI-γ (Figure 1) possessed three crucial hydrophobic residues that directly interfaced with a hydrophobic cleft on MDM2: D-Trp3, D-Phe7, and D-Leu11 [28]. In particular, D-Phe7’s phenyl ring did not fully extend into the binding pocket, and this was identified as a target for further optimization of binding affinity.
The D-PMI-β (Kd = 37.8 nM) peptide was selected as the starting point for optimization because it had the best binding affinity of the D-PMI family members that had thus far been synthesized. The first three derivatives synthesized were chlorinated at three different substituent sites on the phenyl ring to determine which position was best poised to increase the peptide’s potency (Figure 2).
Figure 2: Chlorination of D-phenylalanine at the para (left), ortho (middle), and meta (right) positions.
The o-chlorinated peptide had a moderate increase in dissociation constant (149 nM), while the m-chlorinated’s dissociation constant increased well over 10-fold (3370 nM). The p-chlorinated peptide’s dissociation constant was reduced by about 50%. Due to this result, a number of D-PMI-β analogs with different functional groups in the p position were synthesized and screened for greater binding affinity (Figure 3). The best binder was the one with a trifluoromethyl group (Kd = 0.45 nM), named p-CF3-Phe7-D-PMI-β. It has been found that fluorination and trifluoromethylation of aromatic compounds can increase their hydrophobicity [30], so the good performance of the trifluoromethylated peptide is not entirely unexpected.
Figure 3: p-trifluoromethyl-Phe7-D-PMI-beta (pink) in complex with MDM2 (gray surface). The same MDM2 residues are highlighted in lime. In this case, the trifluoromethyl group in the para position of the phenyl group extends further into the binding cleft. It is thought that this interaction is responsible for the significant decrease in dissociation constant compared to the non-trifluoromethylated peptide (PDB: 3TPX) [29].
Crystallization and structure determination of the complex between this inhibitor and MDM2 revealed that indeed the p-CF3D-Phe7 residue makes a deeper contact into the binding cleft (Figure 3). In a separate series of experiments, fluorination of D-Trp3 at carbon 6 was found to also increase binding affinity. In all cases, binding affinity was assessed using a surface plasmon resonance (assay). In this type of experiment, free MDM2 (30 nM) is introduced to a concentration of a specific inhibitor. The MDM2 which does not bind to inhibitor is then bound to the natural binding partner, p53, which is immobilized on a surface. The amount of free MDM2 is determined based on a calibration curve.
Figure 4: Binding curves for D-PMI and some optimized derivatives. The lines represent the fits, and the points are the experimentally observed concentrations of free MDM2. The dissociation constant can be found for each derivative where [MDM2] = 15 nM. Reproduced with permission from the publisher from reference 29. Copyright © 2012 American Chemical Society.
The binding curves for selected inhibitors are shown in Figure 4. Since the experiments all were started with 30 nM of MDM2, the Kd is the concentration of ligand at which 15 nM of free MDM2 was observed. After it was determine that modifications to residues 3 and 7 individually could improve efficacy, a final analog incorporating both of these modifications was synthesized. This analog, termed D-PMI-δ, performed better than the peptides with only a single modification, with Kd = 0.22 nM [29]. Overall, the optimization effort resulted in an over ten-fold reduction of dissociation constant.
This work of optimization of D-peptides that target this interaction ultimately resulted in the identification and synthesis of a peptide with a binding affinity more than twice as strong as that of those previously identified by mirror-image phage display. While this work still does not completely imply that such peptides can be used as therapeutics just yet (previous experiments [27] still suggest critical drug delivery problems), it is demonstrative of a general scheme of sequence identification using mirror-image phage display and structural optimization that could be used to create novel therapeutics that not only target the p53-MDM2 interface, but other important protein-protein interactions.
There is a lot of data presented in this section that makes it hard to flow through the texts. Instead of telling numbers, it might help to state what was deduced from these numbers (i.e. data collected from this experiment showed that functionalization at the para position produced the best binding [hyperlink to paper] this was further used to synthesize p-CF3-Phe7-D-PMI-β…). With this edit, there is less complex text filled with numbers that could bog the reader down. It would also be helpful to include a picture of the structure of p-CF3-Phe7-D-PMI-β. You include its structure in the context of binding the MDM2 protein but it could also help to include its standalone chemical structure so that readers can visualize what parts are interacting with the protein. I am curious about what make the trifluoromethyl group such an attractive functional group for MDM2. Does the paper address what the functional group mimics on p53 maybe?
I completely agree with Suraj that figures would make this section a lot easier to read. On the pathway page, you extensively discussed the structure of p53. If you could bring back some of that discussion here and combine it with the inhibitors you talk about here, it would be really interesting and would answer Suraj’s questions.
I don’t have much to add to this that Katherine and Suraj haven’t said. I just want to emphasize what Suraj said about adding the picture of the structure. It would really help break up the blocks of text, and give the audience something to follow along with as they read through.
To the tufts.edu webmaster, Thanks for the informative post!