I was at a conference this summer where Patrick Harran gave a lecture on his group’s approach toward the synthesis of callyspongiolide. It was special. Harran has charisma that, as best as I can gather, comes from a combination of a near photographic memory, tremendous creativity, deference to the contributions others in his field, and a great sense of humor. His was the most enjoyable presentation in a program full of fascinating stories. That lecture reminded me of a comment I’d made when collecting opinions on well written total synthesis papers. I had observed that, “The Harran diazonamide synthesis communications are also quite good. They struck me as being classy and respectful in their reassignment of the structure. Also, the synthesis really simplifies a complicated molecule for me.” When I put those observations together I was compelled to revisit his diazonamide synthesis as an “active reading” example for my organic synthesis grad class.
Diazonamide was a hot target during an interesting time – the turn of the century. We were still afloat on the dotcom bubble of the roaring 90’s. And as you well know, that bubble burst. On a much smaller scale (a molecular scale!), Harran helped burst the bubble on the errant structural assignment of the diazonamides. That’s the real subject of the communication to be dissected. That’s the funny thing, it doesn’t actually report the synthesis of a natural product. The “nominal” and “proposed” in Harran’s title are harbingers, though by the time the papers were published I assume those in the game already knew the story. The chase for a “wrong” natural product in organic synthesis has plenty of precedents. Nicolaou, a major player in the diazonamide story, even wrote a review about those examples and some of the emotions that surrounded them. On the subject general, KCN observes, “We know, however, what it is like as a synthetic chemist to be in the midst of a total synthesis or at its “end”, only to find out that the molecule we were chasing was never there!” And on the diazonamides in particular, “Although we certainly admired the beautiful synthesis of Harran and his team as well as the logic behind the proposed structural revision, our initial reaction could only be described as intense disappointment and frustration.” You don’t see a lot of that type of honesty in print. I think it’s a valuable reminder to students and professionals alike: science is a contact sport and its practice can leave scars. What follows is a written version of the explication, paragraph by paragraph, of Part 1 of Harran’s back-to-back communications that popped the diazonamide dot-com structure bubble and reassigned it as the true diazonamide structure.
Total Synthesis of Nominal Diazonamides – Part 1: Convergent Preparation of the Structure Proposed for Diazonamide A1
Jing Li, Susan Jeong, Lothar Esser, and Patrick G. Harran
Paragraph 1: Harran could have stolen the Alan Greenspan “irrational exuberance” phrase to use in this opening paragraph. Here he proceeds quickly to the curious conditions that qualify the “purported” structures of the diazonamides. What had to be reconciled were an X-ray crystallographic structure of diazonamide B analog 2 (Scheme 1)2 that contained an acetal rather than the hydroxy hemiacetal motif of parent compound 1b. A C2 amine was invoked (Using a sound rationale because it meant that the amino acid valine was at this position of the molecule.) to make the proposed structure “more consistent with the available data”. More analysis of the spectral and MS data, along with true structure are in the second communication in the back-to-back set. He closes the paragraph with his Alan Greenspan moment, essentially saying, “We should have known. But in our irrational exuberance, we became mesmerized by the beauty and potency of the diazonamides and we (the community) inadvertently overlooked some important details.” The end of the paragraph suggests that those oversights will be explained and will set the stage for his proposal for the true structures of diazonamides A and B.
Paragraph 2: Though this was early in his career, the second paragraph evidences that Harran was already running with the big dogs in pursuit of the diazonamides. Citation five, at the end of the first sentence, gives a list of groups working on this target: Magnus, Vedejs, Wipf, Nicolaou, Stoltz/Wood, Moody, Konopelski, and Pattenden. Those are all household names, at least at my house. The “word efficiency” of the next sentence must be high. Harran writes, “Our own tactics for constructing a diazonamide ring system have likewise evolved steadily.” He’s sending the reader back to three papers, each interesting in its own right, that chart advancements in their approach to the target in incremental stages. The responsible reader who hasn’t already read those papers would do well to stop and follow the evolution of their approach. If I tried to summarize them, I’d go with: 1- first synthesis of the “western” macrolactam; 3 2- validation the pinacol ring contraction; and 3 – lessons from attempts at forming the indole biaryl linkage. Then it’s the big reveal: the old structures for diazonamide A and B were like pets.com – highly touted but ultimately unsustainable. The true structure (fully rationalized in the second communication in this back-to-back sequence) of diazonamide A is that of 3.
Paragraph 3a: Paragraph three is a behemoth. I’ve chosen to split it in half to make sense of it. The first half does three things. First it gives you another homework assignment in terms of retrosynthesis. Harran assumes you know the general approach because you’ve read the earlier “evolution” papers; if not, you have go do it. Embedded in this reminder is that the synthetic design elements had to include consideration for some of the stereochemical features (especially the atropisomerism of the poly-aryl bit on the eastern macrocycle). There will be more on that shortly. Second, and relatedly, Harran identifies the starting materials for the synthesis: two commercial products (These are not explicitly stated, again it’s put on you to reason through it; they are tryptamine and valine.), and compounds 4–6 in Scheme 2. References 6a and 7 give you details on their synthesis. 4 (ref 7) is readily prepared by protecting iodo-L-tyrosine. 5 (ref 6a) is from condensation of Boc-protected valine and aminomalonitrile using EDC. 6 (ref 7) comes from benzyl protection of 2-hydroxyacetophenone followed by dehydrative chlorination. The third, and main, thing this part of paragraph three does is report on the synthesis of macrolactam 12, a key intermediate in the synthesis. Here styrenyl zirconocene 7 (derived from 6) is cross coupled with bromo-oxazole 5 and then acylated with tyrosine derivative 4 to give 10 (Scheme 2). Intermediate 10 is the acyclic precursor for a Heck macrocyclization they’d established earlier. An observation noted in this section is that the modified Takahashi variant of a Negishi coupling “neither requires nor benefits from a Zr-to-Zn transmetallation.” My interpretation is that he’s referring to the stoichiometric zirconocene that is required for the reaction is less than ideal. On this scale, though, the efficiency of the coupling makes me think, “So what?”. The Heck conditions were largely figured out in that earlier work, so conversion of 10 to 12 via pallado-intermediate 11 (Scheme 3) proceeds as expected. The import of 12 to Harran is that it has the content of the diazonamide core and only needs to be “oxidatively restructured” en route to the target. To me, Scheme 2 would have been more effective if it included structures 11 and 12. Further, this choice could have dovetailed with a paragraph three that ended here too. Just a preference.
Paragraph 3b: The second half of the behemoth takes macrolactam 12 all the way to bicyclic lactone 18. Central to this part of the synthesis is the dihydroxylation of 124, converting it to syn-β-diol 14 in Scheme 3 which is followed by the pinacol rearrangement to deliver 16. There’s a lot going on in those steps. First the dihydroxylation: in reference 12, Harran observes that the inherent diastereoselectivity for dihydroxylation of 12 is to give the syn-α-diol in a 6:1 dr. This is an interesting example of macrocyclic diastereoselection – something our group has explored in the context of some other macrocycles. The macrocycle presumably populates multiple conformers where one face or the other of the olefin is accessible to reagents; remarkably, osmium complex 13 facilities the mismatched dihydroxylation. That’s pretty cool. Next the pinacol; here Harran signals euphemistically, with the flourish of a total synthesis chemist, that the pinacol, while low yielding, is valuable because the “stereochemical communication is near perfect.” Agreed. It’s a good trade. Stereochemical fidelity, they argue, is mediated by phenonium ion 15. Then he drops this pearl, “The axial asymmetries of the diazonamide polycycle can now be made an artifact of their assembly.” I like that sentence. My translation is that they’re going to leverage the asymmetry of the macrocycle in 16 to selectively set the atropisomers of the biaryl linkages that are about to follow. Conversion of 16 to 18 involves protecting group manipulations, reduction of the aldehyde that came about in the pinacol, bromination of the E-ring phenol, and lactone formation. One could wonder about the specifics of the sequence, but Harran doesn’t belabor them.
Press pause for a second and assess what has to happen for Harran to get home from 18. He’s amongst those lofty heights of an advanced intermediate here. Now they have to form the hemiacetal to establish the dihydrobenzofuran and also grow the “ter”-biaryl, “eastern” macrocycle onto the macrolactam of 16.
Paragraph 4: Harran calls 18, now at the beginning of Scheme 4, the right platform to launch attempts on the polycycle. You can feel the blood sweat and tears of his co-workers in the sentence, “While this goal was elusive for some time, nonproductive forays finally gave way to success.” Flash-forward to maoecrystal V, the compound whose synthesis launched 1000 reactions. Success took the form of amidation via aluminum amide opening of the lactone, re-oxidation of the alcohol to an aldehyde, and photolysis of an o-nitrobenzyl protecting group on the C16 phenol that had been installed en route to 18. This sequence gave a hemiacetal that was trapped to give one diastereomer, acetate 21. Oxidation at the benzylic position of the tryptamine followed by dehydration “parlays” this part of the molecule into the (bis)-oxazole-indole of product 22. In this short paragraph, Harran has nearly built the core skeleton of diazonamide. He just has to link the indole to the dihydrobenzofuran to close-down the eastern macrocycle.
Paragraph 5: /Harran, to himself,/ “Of all the configurations of all the atropdiastereomers in all the precursors of diazonamide, Witkop delivered mine.” A Witkop-type reaction now converts a dilute solution of precursor 22 to 25. As with the pinacol, Harran takes an aside to walk the reader through some details of how the reaction is works, complete with structures to illustrate his comments. Intramolecular electron transfer from the indole to bromo-arene, initiated by the irradiation, creates an intermediate akin to 23. This next sentence puts it all in a nutshell (parentheticals are my explanations). “Mesolytic elimination of bromide from the resultant radical ion pair 23 (bromide leaves now creating a neutral radical of the erstwhile bromo-arene), biradical collapse (the radical just formed from the bromide departure plus the indoyl radical that comes about via deprotonation of that moiety; these couple to make the new, sought after bond), and prototropy in 4H-indole in 24 would give 25 (tautomerization – proton)”. His vocabulary enables the smooth synergy between the text and the scheme; it’s enviable. A recent review gives greater insight into Witkop reactions in synthesis. Harran then lists results from other experiments (Li+ ion effects – there are two eq. of LiOAc in the reaction; inability phenol to inhibit the reaction, etc.) that are consistent with the proposed mechanism. Harran shows some of the physical organic underpinnings that must have been inspired by his time in Yale’s Sterling Chemistry Lab. Ultimately, it’s closure, and we’re beginning to feel – gratification dot com!
Paragraph 6: End game. Or is it mind game? With the skeleton set, it’s time for some clean-up. Uneventful chlorination of 25 gave 26 (Scheme 5) was followed by hydrogenolyis of the Cbz group on the C2 amine (and the benzyl group on the F-ring phenol) followed by amidationwith a Cbz-protected valine residue. The α-amino group of 27 of the valine is a key player in the structural reassignment. Hydrolyis of the acetate that protects the hemiacetal and another hydrogenolysis to remove the Cbz on the valine residue gives “nominal” diazonamide A, 1a. Think about it. Harran’s group was the first to achieve the synthesis of “the structure originally proposed for (-)-diazonamide A.” Imagine being Jing, Susan, or Lothar – whichever of them it was to take that sample to the NMR. Were they triumphant? Were they already fearful? Had the storm clouds already been seen on the horizon?
Paragraph 7: “[Peewee], There’s no basement at the Alamo!” Harran realizes that the spectra of the stuff they’ve synthesized does not match the sample of natural diazonamide A procured from Fenical; moreover, he finds that 1a is not particularly stable. Look at footnote 205: “Synthetic 1 a appears (1H NMR) as an ~4:1 mixture of C11 epimers. However, unpredictable degradation prevents detailed characterization of these materials. Protected derivative 27 is serviceable with respect to handling and analysis.” The main degradation pathways are deformylation at C10 and diketopiperazine formation where the free amine on valine attacks the C1 amide carbonyl of the macrolactam. Those pathways are giving hints about the true structure. Wait, that’s part of the traditional role of synthetic organic chemistry – reactivity as one parameter (in addition to spectra) of the target to reveal its structure and function. To gather more information from his intermediates, he synthesizes the “nominal” diazonamide B derivative that contains a p-bromobenzamide as the pendant to the macrolactam in place of valine (Scheme 5); this material, too, is “subtly different” than a sample derived from diazonamide B isolated from nature. Last, 26 was converted to acetal 28 (Figure 1); an X-ray structure of 28 confirms that the skeletons of 1a and 2, are as presented and synthesized. Harran has made it to the end of the Alamo tour and there is no basement. However, he has set the stage for the reassignment of the structures for natural diazonamides which is the subject of the proceeding “Part 2” communication.
The correction of the market – the bursting of the dotcom bubble – was toward companies that had a solid foundation. To belabor the analogy to the diazonamides, the bubble was the errant structures. It was not the compounds themselves, Harran, or synthetic organic chemistry as a discipline. In fact, each of these showed their true value. The details of the diazonamide structures were clarified. Harran leveraged his strategy toward the revelation of the true diazonamide structures. Organic synthesis proved its inherent value in an advanced age of modern spectroscopic techniques. In particular, Harran’s characterization of the degradation pathways of the late intermediates localized the NH-O transposition that led to the true structures. So it was win, win, win for diazonamide dot-com.
1 It’s safe to say that Jing, Susan, and Lothar did most if not all the experiments here. I don’t have a clue about who did the writing, but certainly Harran did the editing at a minimum. Nonetheless I refer to the team throughout as “Harran” mostly for simplicity. We all know the students and post-docs do the labor and the team, together, formulates and refines the strategies and ideas. Thanks to Patrick for sharing the original chemdraws (wow, that’s organization) of the figures and schemes from the manuscript and for doing a read-through of an advanced draft of this post.
2 The “nominal” structure of diazonamide A can be generally described as follows: an oxidized, highly dessicated and consequently crosslinked, polycyclic peptide characterized by two fused, 12-membered ring macrocycles. The “western” macrocycle is a macrolactam that also contains a valine residue pendant attached to the macrocycle via an amide bond. The “eastern” macrocycle has four contiguous aromatic rings linked to each other via single bonds (biaryl linkages); two of the aryl rings in the array – one oxazole and an indole – are chlorinated. The structure is further rigidified by a hemiacetal unit at the interface between the two macrocycles.
3 This was my personal favorite of the three. An eloquent and old-fashioned synthesis paper.
4 actually the C16 bromoethyl ether of 12
5 References end at ref 16 in the pdf version of this paper but all the refs are with the online version. FYI.