Just a bit of sodium ethoxide does the trick …
The Knorr pyrazole synthesis, broadly defined, involves the condensation of hydrazines with 1,3-dielectrophiles, e.g., 1,3-diketones, beta-ketoesters, alpha-cyanoketones, beta-alkoxyacrylonitriles, alkoxymethylenemalonates, etc. When a nitrile electrophile is involved, an aminopyrazole typically results, producing compounds that are very useful in the pharmaceutical field. The first two equations in the general scheme below use hydrazine as the dinucleophile, but what happens when an unsymmetrically-substituted hydrazine is involved? Wouldn’t it be nice to be able to produce either the 3- or 5-aminopyrazole regioisomer in a selective fashion?
This problem has been kicking around the literature for a long time. The 5-aminopyrazole is generally the major product, leaving the 3-aminopyrazole as a useful-yet-expensive poor cousin. Much effort has been spent trying to rationalize and predict the outcome of these processes, so a recent paper from Fandrick, et al., from Boehringer Ingelheim is a most welcome arrival.
Heating the nitrile 3 with the alkyl-substituted hydrazine 4 in ethanol produces the 5-aminopyrazole 7, presumably through the adduct 5, a result that is consistent with the prior literature. At first glance, this should be rather surprising, since it is known that alkylhydrazines are more nucleophilic at the most substituted nitrogen, i.e., k1 should be larger than k2, thus producing adduct 6 and the 3-aminopyrazole 8. (Arylhydrazines are different; they are generally more nucleophilic on the unsubstituted nitrogen, though it depends on the nature of the arene.) Fandrick and co-workers proposed that 6 is indeed the kinetically-formed adduct, but its cyclization to 8 (under typical neutral conditions) is slower than isomerization to the more stable adduct 5, producing 7 instead.
In order to obtain the more rare and often desirable 3-aminopyrazole 8, Fandrick simply introduced sodium ethoxide to the mix with the idea that the kinetically favored 6 might be transformed to 8 before it can isomerize to 5. It worked, producing 3-aminopyrazoles with selectivities of up to 99:1 depending on the exact example.
Increasing the size of the alkyl group on the hydrazine should erode the selectivity for 8 in the kinetically-controlled ethoxide reaction, since the internal nitrogen of the hydrazine, while electronically more nucleophilic, is less accessible due to steric hindrance. Indeed, moving to cyclohexylhydrazine and then t-butylhydrazine leads to diminished and even reversed selectivities of 72:28 and 5:95 for 8:7. Under neutral thermodynamic conditions, cyclohexyl- and t-butylhydrazine gave >99:1 ratios of 7:8, as expected.
Arylhydrazines are known to highly favor 5-aminopyrazoles, which was supported by the current work when employing neutral (thermodynamic) conditions. Interestingly, reasonable quantities of 1-aryl-3-aminopyrazoles 8 were formed under kinetically-controlled (ethoxide) conditions. Hence, with phenylhydrazine, a 1:1 ratio of 8:7 was formed, indicating that a good deal of 6 was formed kinetically. Using the more electron-rich p-methoxyphenylhydrazine under kinetically controlled conditions produced even more of the 3-aminopyrazole, giving a 78:22 ratio of 8:7. This presumably reflects and increase in the nucleophilicity of the internal hydrazine nitrogen due to the resonance donating effect of the p-methoxy substituent.
Near-simultaneous reports of an interesting new pyrrole (and dihydropyrrole) synthesis have appeared recently from the labs of Frank Glorius (Westfälische Wilhelms-Universität Münster) and Naohiko Yoshikai (Nanyang Technologial University, Singapore). Optimization of the additives, solvent, and temperature led to essentially the same conditions from each laboratory. A balloon of oxygen is preferred: yields are much lower when air is used. A Heck-type mechanism is shown, but a Wacker-type mechanism involving nucleophilic attack of an enamine onto a Pd-complexed alkene could not be ruled out.
The scope of the reaction is decent, as indicated below. Failures include R3 = CF3 or cyclopropyl and R2 = Ph. N-Homoallylic and N-cinnamyl imines were also unsuccessful.
Specific examples are shown below:
Interested in having a short course on Heterocyclic Chemistry at your company?
In late 2011, I put together a new two-day short course entitled “Heterocyclic Chemistry – A Drug-Oriented Approach” to present at companies. In a nutshell, it’s an intensive, preparatively-oriented course on the synthesis of the types of molecules encountered in the pharma and agricultural endeavors. There is a heavy focus on practical, proven methodology that people actually use. I keep it updated with lots of current chemistry.
Beyond the two days of instruction at your company, participants are left with a great resource: A book of over 500 slides of information, nicely organized, well-referenced, and containing many specifics on the best ways to make heterocycles.
I’m booking courses for 2013 and early 2014. Learn more by checking out the course web site, where you’ll find more on what the course covers. You can also download a two-page summary of the course to pass around to your colleagues. If you know of other chemists who might be interested in the course, I’d appreciate passing this along. You can reach me at will at pearsonchemsolutions dot com.
Here’s a process that I think needs continued attention from the synthetic community: Making saturated nitrogen heterocycles from simple N-alkylamines by intramolecular CH amination reactions. There’s a lot of great chemistry out there for related process where there is an electron-withdrawing group attached to the nitrogen within the tether (vide infra), but let’s focus on N-alkyl groups. With all the activity on CH functionalization chemistry in general, I hope this reaction will become routine at some point. Let’s take a look at some recent work in this area.
The Hoffmann-Löffler-Freytag reaction – An medicinal chemistry application
To exemplify the need for such a reaction, consider the compounds shown below, appearing in a recent J. Med. Chem. paper by McClure and coworkers at Pfizer. The diazatricyclodecane (or diazaadamantane) heterocycles in the dotted boxes were proposed as conformationally restricted piperidines that might make good agonists of G-protein-coupled receptor 119.
The Pfizer group settled on a Hoffmann-Löffler-Freytag (HLF) reaction to form the heterocycle. In their initial work, they were unable to reproduce Rassat’s route to such diazatricyclodecanes (JACS 1974), which involved heating the N-bromoamine in acid. Switching to the N-chloroamine led to only 14% of the desired compound accompanied by 40% of an elimination product involving the N-benzyl group.
Ultimately, forgoing the protecting group was fruitful. N-Chlorination of the primary amine shown below was followed by photolysis with a 450 W mercury lamp to provide multigram quantities of the crude cyclization product. Acylation followed by demethylation of the other amino group provided the key diazatricyclodecane for their studies.
One curious bit is the chlorination reaction: The authors do not state how many equivalents of t-butylhypochlorite are used. This seems rather important, since primary amines are well-known to form dichloroamines. I’ve contacted to authors, so hopefully we’ll know soon whether they were dealing with the monochloroamine or the dichloroamine. [Update: Dr. McClure responded that 1.2 equivalents of t-BuOCl were used.]
The Pfizer compounds did not pan out, so we’ll never know if their process research wizards would be able to employ the HLF route in a scaleup setting, but I imagine it would be an uphill battle. If photochemistry is ruled out, I imagine the “acid and heat” HLF would have to be sorted out somehow.
Now it’s easy to see why a more modern CH functionalization reaction with a catalytic transition metal would be useful, right?
Toward a practical intramolecular CH functionalization reaction
This seems like an excellent strategy, and I look forward to seeing where this research will go. How about metal-free, light-free, halogen-free versions? There’s a worthy goal!
Finally, I’d be remiss if I didn’t mention the extensive work in the literature on intramolecular aminations using nitrenoids that are substituted by strong electron-withdrawing groups (DuBois, Sanford, Davies, White, Lebel, Panek, and others, leading reference here). It’s a bit different from what we’re talking about here, since the electron-withdrawing group ends up in the tether, making cyclic sulfamates, carbamates, and the like.
By the way, if you’re interested in this topic, you might also look at some nice recent work by Tom Driver at UIC, who is using aryl azides as the nitrogen source for metal-catalyzed intramolecular CH aminations. His paper is also a good entry to the literature of CH aminations in general.
A recent paper by Xingwei Li and co-workers in Angewandte Chemie International Edition has some nice chemistry in it, but take a look at the title, “Rhodium(III)-catalyzed oxidative C-H functionalization of azomethine ylides.” One might expect to see some azomethine ylides, right? No, they are azomethine imines:
I suppose one could make the case that azomethine imines are a subset of azomethine ylides, but I’ve never seen it done. What am I missing here?
[Edit: Correspondence with Professor Li reveals his point of view: “the structure can be called both azomethine ylides and azomethine imines,” (sic) and there “is not much difference,” (I strongly disagree on both counts) and “we simply want to emphasize the ylidic character of our substrate.” Okay, I can understand wanting to emphasize that something is an ylide, but there’s no need to choose an incorrect name to do so. Learning organic nomenclature is hard enough. Enough said. I’m now going to retire my Professor Pearson hat on this issue.]
[Edit: Here’s another paper from the same lab, still referring to azomethine imines as azomethine ylides. Apparently the referees at Advanced Synthesis and Catalysis aren’t minding the store either.]
Let’s take a moment to appreciate the challenges of synthetic organic chemistry. Not exactly stamping out widgets, is it? Witness the recent retraction of an approach to the lundurine alkaloids from Steve Martin’s group (original Organic Letters paper here, retraction here). Since the early days of RCM, Martin’s group has recognized its potential for the construction of alkaloids, but the presence of nitrogen atoms in RCM precursors can lead to problems. Such may be the case here.
The Kopsia lapidilecta alkaloids
The Kopsia lapidilecta species of Asian flowering plants produces numerous alkaloids that bear the novel 5,6,12,13-tetrahydro-11a,13a-ethano-3H-pyrrolo[1′,2′:1,8]azocino[5,4-b]indole ring system. Representative examples include lapidilectine B and lundurine B.
Back at the University of Michigan, we became interested in making these alkaloids using our 2-azaallyl anion cycloaddition chemistry. It was a war, but postdoc Ill Young Lee and Ph.D. student Patrick Stoy were up to the task. The key step was the cycloaddition of the 2-azaallylanion shown with phenyl vinyl sulfide. The resultant pyrrolidine cycloadduct was then converted on to lapidilectine B, completing the first total synthesis of any of the Kopsia lapidilecta alkaloids (JACS, JOC).
Martin’s group at Texas (link) and Sarpong’s group at Berkeley (link) have published approaches to the lundurines and lapidilectines, respectively, the former using RCM to make the central azocine core and the latter using an intramolecular electrophilic aromatic substitution approach.
Trouble with the RCM approach
Martin and co-workers reported a potentially simple way to access these alkaloids. They carried out an RCM reaction on 24, which itself was made by an RCM assembly of the pyrroline ring. The closure of the eight-membered ring was reported in 26% yield, producing 1.4 mg of 25. A larger amount (12.9 mg) of 25 was then hydrogenated selectively to produce 26. Their plan was to use this approach to make lundurine B. (Edit: Corrected yield of RCM.)
In a recent retraction, Martin has now withdrawn this work “on the basis that the RCM of 24 to give 25… is not reproducible; thus, the reduction of 25 to give 26… is also not reproducible.” The wording of this sentence is curious to me, since the reproducibility of the reduction should not depend on the reproducibility of the RCM (the RCM product was used after purification), but nonetheless, this is a setback for this potentially powerful approach.
I hope the situation can be remedied. Our synthesis of lapidilectine B was linear and involved a lot of steps, and I’d like to see the Martin or Sarpong approaches succeed; they have the potential to be considerably shorter. I imagine there are other groups working on these alkaloids; they’re too beautiful to resist. In my opinion.
You might recall my fascination with cocaine (no, not for that reason), which culminated in a total synthesis of its enantiomer, (+)-cocaine, using our 2-azaallyl anion chemistry (Mans and Pearson, 2004, blog post here). Davies and co-workers at Oxford have now published a short and elegant synthesis of (+)-pseudococaine, a diastereomer of natural (-)-cocaine.
The key step is a highly diastereoselective transannular iodoamination with concomitant N-debenzylation by iodide ion to produce the tropane skeleton:
It’s worth pointing out that intramolecular iodoamination reactions of simple primary and secondary amines can be problematic due to N-iodoamine formation and subsequent shenanigans, including potential post-cyclization aziridinium ion formation. Davies’ reaction is well-behaved, since transannular iodoaminations are especially favorable, the amine is tertiary, and the product is not susceptible to aziridinium ion formation.
The cyclization precursor is made by a very nice sequence involving conjugate addition of a chiral lithium amide to an enoate with subsequent in situ trapping of the enolate by a diasteroselective aldol condensation (7% of the other aldol diastereomer is not shown).