Photoaffinity labeling reagents is a valuable way by which to study the interactions between two molecular entities. Often times this is a ligand and it’s receptor. The strategy is to attach the photolabel to a ligand, incubate with it’s binding protein, and expose to UV light. The photolabel will then be activated and react with the portion of the protein that it is in proximity with to create a covalent bond. After digestion, the peptides can be analyzed and structural information such as the site of binding can be elucidated by seeing where the ligand is attached. Dizairines have been commonly employed as photolabels. Upon exposure to UV light they degrade to a reactive carbene which can then undergo reaction. By attaching a separation tag to the photoaffinity label, the digest can be enriched for only those peptides that are covalently bound to the ligand. Biotin is, not surprisingly, a common tag choice.
While still the gold standard in affinity enrichment, biotin is not without its issues including non-specific binding by other moieties to avidin, incomplete elution from the avidin, and complicating fragmentation during MS/MS analysis. This has led researchers to produce cleavable versions of biotinylated photoaffinity labels. An example is shown below containing a diol linker. After capture using avidin or streptavidin the diol linker is cleaved by periodate to release the photoaffinity linked ligand and peptide. The problem is not only the extra steps, but the darn photoaffinity label is getting pretty big and complicated which may adversely effect ligand binding.
Fluorous enrichment overcomes many of the issues associated with biotin, so it’s not surprising that fluorous methods are beginning to gain increased use in proteomics and metabolomics. Fluorous tags are relatively small, easy to elute from the capture media, and non-fragmentary in MS/MS.
Given all this the development of fluorous photolabels is not surprising and two research groups have previously reported the synthesis of fluorous modified photoaffinity labaling agents. As shown below, these reagents replaced the trifluoromethyl substituent on non-fluorous diazirine 1 with a perfluorinated alkyl. Compound 2 was reported by Zhang while compound 3 was from the Grond group. In both instances the synthesis was reported along with attachment of the photolabel to a ligand and successful separation of the fluorous photolabeled ligand by fluorous solid phase extraction (FSPE). In neither case, however, have actual experiments linking receptors with the ligands been reported yet.
With this backdrop, a new report from the Manetsch group at the University of South Florida describes their efforts at fluorous photolabel design. They prepared two fluorous photolabels, 4 and 5. Diazirine 4 was essentially the same as that described by Zhang, but contained two hydroxyls for ligand attachment instead of one. Diazirine 5 is a bit more interesting because the fluorous chain is not directly attached to the diazirine ring, but rather to the aryl ring.
The researchers than studied the reactivity of each photolabel by exposing them to UV light in the presence of methanol-d4. What they found was very interesting. Compound 5 reacted as expected with the resultant carbene insertion into the methanol-d4 occuring quickly and cleanly. Under identical conditions compound 4 resulted in three products. The expected insertion product and two other new products. One of the new products was not identified, but the other was identified as the olefin from what appears to be elimination of DF after carbene insertion. Presumably, the fluorous group increases the acidity of the benzylic proton relative to the trifluoromethyl group resulting in the elimination. This is, of course, not what one wants to see as it complicates the analysis. Based on the above result it would seem that 5 would be the better reagent. Surprisingly, however, neither Zhang nor Grond mention this type of reaction occurring in their reports. Since none of these three reports have yet to use the fluorous labels in actual crosslinking experiments, it will be interesting to see what the labeling efficiencies and analyses actually result in between these differing designs.