’Twas brillig, and the spiroketals
Did gyre and gimble in the flasks…
A while back, we had some behind-the-scenes talks about narrating some of our research projects here on the blog. Ken got us started with his delightful tale of his recent publication. I’ll go next and tell you about one of my grad school projects. My story will not be as intriguing as Ken’s because a) the project ultimately failed to achieve its objective and b) we didn’t publish the results. But I’ll tell you about it anyway, as the project made up the bulk of my dissertation.
I will have to leave out a few details, though, because my PI may want to eventually revisit the project, and I may sit down here soon and churn out a short comm manuscript and submit it for publication at some point.
The project centers around the synthesis of spiroketals in a Diversity-Oriented Synthesis project. DOS is a strategy for making molecular libraries similar to combichem, but perhaps with a bit more purpose and a bit less reliance on random chance/luck. In our project, we attempted to synthesize a series of 6,6-spiroketals with orthogonally differentiable functional groups in various positions around the spiroketal core.
A quick primer on spiroketals – spiroketals are spirocyclic tetrahydropyran rings where the rings are fused through a ketal carbon atom. Spiroketals were chosen because the two rings are historically very rigid – the 3-dimensional orientation is governed by the anomeric effect – a topic I’ve blogged about before.
Additionally, as functional groups are rotated to different positions about the spiroketal framework, the vector relationship between the two functional groups changes. This was the purpose of synthesizing a library of spiroketals. We wanted to probe the ability of the spiroketal to act as a scaffold upon which we could position a number of functional groups at unique and specific relative orientation.
Back to DOS. We wanted to synthesize spiroketals through a convergent approach. We would position simple functional groups in the various positions through this convergent approach to make a small library of purposely designed spiroketals. These simple functional groups would be orthogonally differentiable, like an aryl bromide and a terminal alkene. This would allow us to differentiate each spiroketal at each position using reactions that are orthogonal to each other (that is, the Pd-catalyzed cross coupling reaction would likely not interfere with the terminal alkene and the cross metathesis reaction would likely not interfere with the aryl bromide)
Using this approach, we could prepare a library of spiroketals in short order. Subsequently, each spiroketal could be used as the starting point for a second library by functionalizing the aryl bromide and the terminal alkene. The same secondary functionality could be introduced in each secondary library, but each secondary library would be different because of the unique vector relationship between the two functional groups.
All mimsy were the aldols,
And the phosphonates outgrabe…
As shown in the following retrosynthesis, we split the spiroketal precursor (the dihydroxyketone) in half through a Horner-Wadsworth-Emmons olefination to lead back to an aldehyde and a β-keto phosphonate. The chirality in both fragments arises from an enantioselective aldol addition mediated by a thiazolidinethione chiral auxiliary.
The enantioselectivity issue had been worked out in advance and guided our decision to use the thiazolidinethione-mediated aldol addition. Additionally, the thiazolidinethione is preferred over the more traditional oxazolidinone because the reduction of the chiral auxiliary can be stopped directly at the aldehyde oxidation state – shortening our synthesis by one step. Another cool feature of the thiazolidine-mediated aldol addition is that three of the four possible aldol diastereomers can be accessed starting with the same thiazolidinethione starting material simply by changing the reaction conditions (click for larger).
The next interesting reaction is the 1,4-conjugate reduction of the α,β-unsaturated enoate in the presence of the aryl halide. Because of the aryl halide, typical transition metal hydrogenation is an unfavorable reaction. We accomplished this reduction by treating the enoate with tosylhydrazine and aqueous sodium acetate in refluxing dimethoxyethane. The aqueous base reacts with tosylhydrazine to form diimide. Diimide acts as a reducing agent by engaging in a [4 + 2] reaction with the alkene, delivering the elements of hydrogen across the double bond and releasing elemental nitrogen as the byproduct.
A modified Claisen condensation reaction using the ester and lithiated dimethyl methylphosphonate prepared the β-ketophosphonate in high yield (but only if the internal temperature of the reaction is held steady at -78 °C. The reaction is completed essentially instantaneously, but if the internal temperature is any warmer than -78 °C, the reaction suffers from dramatically lower yields and very messy reaction mixtures. To ensure the dropwise addition of reagents without warming the internal temperature, I got to use one of my new favorite pieces of glassware – the jacketed addition funnel (product # UI-4980)). Another aldol/reduction sequence provided the aldehyde necessary for the Horner-Wadsworth-Emmons olefination.
To carry out the Horner-Wadsworth-Emmons reaction, we utilized barium hydroxide as the base. This allowed us to deprotonate the β-ketophosphonate under relatively mild conditions. Unfortunately, without vigorous stirring, the reaction mixture turns into a gel. It then stops stirring and the reaction suffers from disappointingly low yields. As long as vigorous stirring is maintained, I obtained consistent yields in the 70-88% range.
Again, a 1,4-conjugate reduction was needed, this time of an α,β-unsaturated ketone in the presence of both the aryl halide and a terminal alkene. A very interesting reaction was utilized which allowed for consistent yields without over reduction. A catalytic amount of copper(I) iodide is dissolved in THF and an equal amount of methyl lithium is added. To the mixture we add hexamethylphosphoric triamide and diisobutylaluminum hydride. The mixture is kept at -50 °C for a while, then the enone is added. Presumably, some sort of copper hydride species is formed and facilitates the 1,4-addition of hydride to the enone olefin, without interacting with the terminal olefin.
There are two main unfortunate circumstances surrounding this reaction, though. I have to use HMPA, and the reduced product has the same TLC Rf value as the enone starting material. Can’t do anything about using HMPA, just gotta be real careful distilling it and syringing it and disposing of it (double glove and wash everything a lot with a lot of bleach).
To work around the TLC issue, we monitor the reaction by NMR. Nothing fancy involved – an hour into the reduction a few dozen microliters are taken from the reaction and quenched. The solvent is removed and the residual oil is analyzed by NMR. The HMPA signal (which is not removed by the quick mini-extraction) is huge and typically drowns out all the other signals. Fortunately, I’m really only interested in the 6.5-7.0 ppm range. By blowing that range up I can see the presence or (hopefully) absence of the characteristic enone proton signals. If they’re gone, the enone has been reduce; if they’re still there, the reaction’s not complete.
“Beware the Jabberwock, my son!
The jaws that bite, the claws that catch!
Beware the diastereomers, and shun
The frumious steric clash!”
Following 1,4-reduction, all that remains is removal of the protecting groups and acidic spiroketal formation. When triethylsilyl protecting groups are used, we can accomplish these transformations concurrently by (carefully!) using 48% HF(aq). The spiroketal we’ve been discussing has the substituents in the ‘naturally occuring’ 2- and 8-positions about the spiroketal ring. This is a useful proof-of-concept spiroketal, but doesn’t actually locate the substituents anywhere they haven’t already been.
So spiroketal #2 was made, now moving the terminal alkene to the 7-position. The synthesis of the linear protected dihydroxyketone was more or less uneventful, but one aspect is worthy of note. We desired to make a highly modular synthetic route to these spiroketals. Since the aryl halide fragment is the same, we didn’t have to remake the β-ketophosphonate fragment. All I had to do was make a new aldehyde in three steps and we were ready for HWE coupling.
We then proceeded to the cyclization. First, I deprotected the silyl ethers using TBAF to give the unprotected dihydroxyketone. Treatment of the dihydroxyketone with catalytic p-toluenesulfonic acid yielded an inseparable mixture of two spiroketals in a 3:1 ratio. Interestingly, treatment of the bis-protected dihydroxyketone with HF resulted in the same inseparable mixture of spiroketals, but with the selectivity reversed 1:13.
Whaaat? If the doubly anomeric spiroketal should be thermodynamically stable, why would I see two different results by cyclizing under two different conditions? And how am I going to tell which is which? We used 2-dimensional NMR (NOESY and COESY were the most helpful, but we also got HMBC, HMQC, 1D proton, 1D carbon, DEPT, and we also asked the NMR tube really, really nicely what the 3-D conformation was).
In the 1:13 sample, we noticed an nOe correlation between protons labeled Hc and the two methyl groups, but not between Ha and Hb (a correlation we would expect to see in the desired spiroketal). This meant the product we could produce the most of was ultimately the singly anomeric spiroketal – the wrong spiroketal diastereomer. A positive nOe correlation was noticed between Ha and Hb in the 3:1 sample… meaning we are forming the doubly anomeric spiroketal – the right spiroketal, but not in synthetically useful selectivity.
It’s worth pointing out that the undesired spiroketal is not undesired because the spiroketal isn’t doubly anomeric, but because the vector relationship between the substituents in the undesired spiroketal is now the same as in the ‘naturally occurring’ spiroketals. This defeats the purpose of putting the functional groups in different positions about the rings.
We thought we could bias the equilibrium toward the desired spiroketal by increasing the bulk of the methyl group. So we repeated the synthesis with an isopropyl group in that position to make spiroketal #3. Again, the modular synthesis only necessitated the synthesis of the aldehyde fragment, and we were ready for HWE coupling and cyclization.
We again performed the cyclization both ways to see what happened. Again, two different spiroketals were formed, but this time as single compounds, not mixtures. Again, 2-D NMR experiments were crucial in helping determine the 3-D configuration. Unfortunately, in neither sample was an nOe correlation noted between protons Ha and Hb, meaning neither spiroketal is in the desired conformation. In the sample where HF was used for cyclization, extensive analysis of the 2-D data led us to believe we did form the right spiroketal diastereomer, but the steric hindrance of the axial allyl group caused one of the 6-membered rings to be oriented in a boat conformation, not a chair conformation. We still don’t know the absolute configuration of the other sample, but it ultimately is irrelevant, because the two substituents are not in the desired vector relationship.
So while we proved a modular synthesis of spiroketals, the major goals of the project were not met, in that we could not predictable control the vector relationship between the two substituents. So we ultimately decided to revamp the project and take our modular synthesis and apply it to the total synthesis of a spiroketal-containing natural product. But perhaps I’ll save that story for another post…
*My Life and Hard Times is the name of James Thurber‘s autobiography. In high school, I played James Thurber in a play called Jabberwock based on his autobiography. It chronicles the hapless Thurber’s teens/early adult life and his mishaps and tribulations in a dysfunctional family. In the middle of the play, when he feels no one gets him and he gets overwhelmed with his comedy-of-errors life, he recites the Lewis Carrol poem Jabberwocky to the girl of his affection. She doesn’t get it. This is how I felt during grad school, so that’s why I framed the post this way.