Late last year Prof. Qisheng Zhang at the University of North Carolina published a paper describing his group’s efforts at developing fluorous diazirines as photoaffinity labeling reagents. For more details on Prof. Zhang’s work, you can view the presentation he gave at ISoFT ‘09. Such reagents appended to appropriate substrates could then be used to investigate various intermolecular interactions such as those that occur between a protein and a small molecule. The strategy is that photoactivation of the substrate in the presence of the protein will result in a covalent bind being formed between the two molecules. Digestion, enrichment by FSPE, and analysis by MS can then provide information regarding the active site of the protein. Previous methods for identifying the cross-linked molecules include radioisotope tagging and biotin based enrichment, each of which has some disadvantages. Hence the application of fluorous techniques.
Prof. Zhang isn’t the only person thinking along these lines as researchers from die Universität Tübingen have now also reported using fluorous diazirines as photoaffinity labels. The approaches were essentially the same in term of designing and synthesizing the fluorous diazirines. The researchers, led by Prof. Stefanie Grond, then attached the fluorous photoaffinity tag to two V-ATPase inhibitors, 21-deoxyconcanolide A and bafilomycin A. The modified inhibitors were then purified by FSPE.
With the fluorous photoaffinity lavels in place they then compared the inhibitory activity of the compounds with unlabeled concanolide A and bafilomycin A in a V-ATPase holoenzyme assay. In the case of concanolide A they found a slight decrease in IC50 with the fluorous photoaffinity tag compared to unmodified concanolide, but the labeled compounds maintain enough inhibitory potential to be useful probes. This was not the case with bafilomycin where the addition of the photoaffinity label substantially changes the inhibitory activity of the substrate to the point where the labeled balfilomycin is no longer a useful compound.
The authors conclude by stating that the first studies using fluorous photoaffinity labeled concanolide A to separate and identify labeled protein fragments using FSPE and MS are underway.
An Org. Letters ASAP paper from the Merck Research Labs in Boston reports the formation of ureas and carbamates using CO2. The researchers reacted amines and anilines with CO2 catalyzed by DBU to form a carbamic acid intermediate. This intermediate was then reacted with amines or alcohols under Mitsunobu conditions, i.e. PBu3 and DBAD, to provide carbamates and ureas. They then used a modified Bohdan Miniblock to run reactions in parallel. Using this set up they were able to conduct 24 reactions at a time. Each of these were then purified using automated flash chromatography on a Biotage.
The chemistry itself is interesting from a mechanistic standpoint. When starting from a primary amine, the carbamic acid is dehydrated under the Mitsunobu conditions to form the isocyanate. Support for this mechanism is found in that secondary amines do not form any urea product. When forming ureas, however, the reaction seem to take place through standard Mitsunobu mechanism rather than dehydration. This is supported by the fact that secondary amines, incapable of dehydration to the isocyanate, are good substrates for the reaction and when using a chiral alcohol inversion at the stereocenter was observed. A nice one-pot methodology which avoids using phosgene or other active acylating agents.
From a fluorous perspective, this is the type of reaction that could easily be renedered fluorous using fluorous tagged phosphine and fluorous DIAD. The reactions could then be run in parallel with parallel purification. This could have several advantages. First, if this reaction were in the middle of synthesis it would allow for quick intermediate purification without having to use chromatography. Even automated, 24-48 purifications takes time, particularly instrument time that others could use. Second, for the final step this would allow for a quick pre-purification prior to final HPLC resulting in greater recoveries and purities.
At FTI we’ve successfully implemented fluorous methods into our library syntheses and production. Through a NIH Pilot-Scale Libraries grant the value of using fluorous techniques has been demonstrated by the number and quality of the libraries we have submitted to the Molecular Libraries Small Molecule Repository (MLSMR).
In June of last year, the Yu group from the University of Maryland published a report in Angewandte Chemie describing the synthesis of a fluorous dendrimer as a 19F MRI imaging agent. The molecule, dubbed FIT, is pictured below and you can read more about the preparation and initial MRI studies in this F-Blog post.
As part of their continuing studies in this area, they endeavored to prepare other fluorous tagged dendrimers as potential 19F MRI agents. In preparing FIT they used FSPE as primary mode of purification. To make analogs of FIT, however, they chose to use fluorous mixture sythesis (FMS) instead and have now published their results in J. Org. Chem. Using this approach they were able to synthesize 4 fluorous dendrimers in a series of single pot reactions. These compounds were differentiated by the number of dendrimer arms each contained. The compounds were demixed at the end of the synthesis using HPLC.
The twist they’ve put on FMS is they’ve used the same fluorous tag on all 4 species. Previous FMS preparations have used fluorous tags of differing length on analogs of small molecules. The longer the tag the longer the molecule is retained on the fluorous HPLC column. In this paper, however, the non-fluorous dendrimer portion of the molecules within the mixture is sufficiently different that even when using the same fluorous tag, they will have different retention times on the fluorous media due to large differences in the % fluorine content. The difference in %F content between the molecules is on the order of 10% of total molecular weight. How does this compare to small molecules with a C4F9, C6F13 and a C8F17 tag. Assuming that the non-fluorous part has a MW ~ 350, the %F for the three molecules would be 38.5%, 47.6%, and 54.5%. About a 8% difference. Not that far off from what Prof. Yu has in his mixture. So while we have generally thought of FMS as separating by fluorous tag length, we might want to amend that to thinking about it in terms %F content.
As seen in Fig. 2 from their paper, the researchers also compared the demixing of these compounds using FluoroFlash® HPLC column, a normal phase column, and a reverse phase column. The fluorous and the normal phase column both gave sharp peaks with good separation, while the reverse phase column did not have good resolution. Also note that the fluorous column and the normal phase resulted in opposite elution orders.
An interesting paper using FMS which makes one think about other applications of fluorous HPLC. Regardless of how one might want to think about it, in practical terms mixtures of fluorous tagged compounds can be separated by fluorous HPLC in a predictable elution order based on fluorine content.
Fluorous compounds generally have a steep temperature dependent solubility curve. This has been exploited by several researchers in the design of catalysts and reagents which are in solution at higher reaction temperatures, but then precipitate out of solution upon cooling. These catalysts are known as thermomorphic catalysts.
Just available online is a report of a thermomorphic metal scavenger from the Baker group at the University of Dublin. They prepared a fluorous ketone using the route pictured below. Starting from the 1H,1H,2H,2H-perfluoroiodide, they prepared the Grignard which was quenched with CO2 to provide the carboxylic acid. Conversion to the acid chloride and reaction with the zincate formed from the same perfluoroiodide gave the symmetrical ketone.
Yields were modest and they also reported that they could not make the C8F17 analog using the Grignard approach. Other methods for forming the anion also were deemed unsuitable in their hands. (They could have just purchased the acids from FTI! Probably would have saved them a lot of work. That’s why we’re here.) The zincate formation and coupling to the acid chloride seemed to work OK. They made no mention of trying to use phosgene with 2 equiv. of zincate.
With the ketone in hand they then tested it’s ability to complex various metals in water. The procedure described calls for stirring the ketone in the aqueous solution for 5h at 80ºC, cooling to room temperature, then filtering the ketone complexes. The Table below shows some of their results. Since they call this thermomorphic scavenging, I assume that the ketone is in solution at 80ºC, although this is not explicitly stated. How much fluorous ketone may have remained in the aqueous solution was also not provided, but would clearly be important if such a process were to be utilized. So some work remains to be done, but this does provide a proof of concept at least for the use of thermomorphic fluorous metal scavengers.
One of the post-ISoFT ‘09 F-Blog posts discussed some of the work presented by Prof. Kenichi Hatanaka describing his group’s work in the in vivo glycosylation of fluorous tagged saccharide primers. You can read that post here. Late last year, they published their results in the Bull. Chem. Soc. Jpn. In this report they made a lactoside primer and a GlcNAc primer and tagged each with various fluorous tags. The tags were 10-12 carbons in length with varying composition of CH2 and CF2 moieties. They then introduced these primers in the cell cultures to study what effect the different fluorous tags had on glycosylation. The process requires several steps including a) uptake by the cells, b) introduction into the Golgi, the site of glycosylation, c) recognition of the primers by the cellular enzymes, and d) eventual excretion of the glycosylated products back to the culture medium. The fluorous tagged products could then be isolated using fluorous separation methods. The researchers found that most of the tagged primers worked to varying degrees. I won’t go into further detail here, but would like to address one specific point of the paper.
Unlike some of the earlier publications from Prof. Hatanaka where they used fluorous liquid-liquid extraction (FLLE) to isolate the products, in this work they used fluorous solid phase extraction (FSPE) instead using FluoroFlash® cartridges. What’s interesting about this work is that instead of using the FSPE in a binary mode as is usually done, meaning a fluorophobic wash to elute the non-fluorous compounds then a fluorophilic wash to elute the fluorous compounds, they used the cartridges in more of a chromatography mode. In one instance they used a water wash to remove non-fluorous cellular components then MeOH/H2O step gradient to separate the fluorous molecules. The step gradient went from 0-100% MeOH in 10% increments. This pseudo-chromatography was necessitated due to the presence of unreacted primer which was also fluorous tagged.
Usually we recommend that reactions are conducted so that only one fluorous compound is present, avoiding just such a situation, but in this case that was not possible. One of the nice parts about FSPE is it’s flexibility. It’s essentially a partitioning-based separation like other chromatographic methods, but it’s selectivity, specificity, and partitioning for fluorous chains is high enough that it can function like an affinity-based separation, which is how most people use it. The major selector in FSPE is fluorous content, but there is a minor selector for polarity. It’s this minor component that the researchers exploited in this case. There are other ways they could have achieved the separation using fluorous methods, but we’ll leave those up to your imagination.
As seen in Fig. 2 from their paper, the researchers also compared the demixing of these compounds using 
