Wednesday, March 22, 2017

Thermodynamic Peptide Macrocyclization

Frequenters of the Open Flask may have noticed the increased occurrence of peptide substrates in some of our recent methods reports. Nearly two years ago, we were joined by an exceptionally talented post-doctoral researcher, Dr. Lara Malins; her expertise in the field of peptide synthesis has proven a boon for the group, and her contributions can be found scattered throughout many of our recent papers. “Peptide Macrocyclization Inspired by Non-Ribosomal Imine Natural Products” (J.Am. Chem. Soc. 2017, ASAP) however, represents the Baran Lab’s first true foray into the field of peptide chemistry.


While the project has been rewarding, it was not without its practical challenges. As an historically total synthesis-based group, we are not equipped with the standard supplies and instrumentation found in even the smallest of peptide synthesis laboratories; instead we have relied on both the generosity of other TSRI groups (shout out to the Ghadiri lab!) and our own creativity to accomplish this study. Reactions that required heating—specifically, the loading of amino aldehydes onto our manually prepared Rink TG resin—posed a unique problem, as our orbital shakers are limited to room temperature agitation. As such, we were forced to use rubber bands and rotavaps, sans vacuum, to simultaneously stir and heat the reaction vessels. While this was the best method available, it wasn’t great… We fished soaking syringes out of the water bath on more occasions than we care to remember. We were also on the receiving end of judgement from labmates about (a) what on earth we thought we were doing and (b) whether this was really the best use of a rotavap…

“I think your peptide is dry, Justine…” – fellow labmate after rotovap resin-loading for over 5 h…
It is also worth mentioning that the majority of substrates were made without the aid of an automated peptide synthesizer, relying instead on our own two (four) hands. Scale-up for initial reaction screening meant running multiple 300–400 μmol scale reactions in parallel; tennis elbow has nothing on peptide elbow!

When Phil asks for a gram-scale reaction, you find a way…
After all of the work (and swollen elbows), we were thrilled to submit a JACS communication in the fall of 2016, wherein we described a novel approach to thermodynamic peptide macrocyclization. This unconventional method allows for late-stage diversification of high-value substrates by virtue of an inherently reactive imine intermediate, a process inspired by a non-ribosomal peptide synthetase/reductive release mechanism. Incorporation of isotope labels and bioorthogonal functional handles proved facile, as did further manipulation of the installed moieties. Additionally, four distinct natural products and associated analogues were synthesized, each in a time frame of about 2 days (start to finish). As per Baran Lab dogma, the reactions require no protecting groups and are tolerant of all proteinogenic functionality.

While much of the initial feedback was positive, we received a few valid critiques to address prior to resubmission. The first point raised concerns over whether conformational predisposition of the chosen amino acid sequences contributed to the robust protocol. Given that many of our sequences were inspired by natural products, we held similar concerns; in the full paper (out today), we opted to use an online random sequence generator to minimize the possibility of practitioner bias in sequence selection. We were pleased to discover no notable difference in reaction rate or yield between our original substrates and those picked at random. Our observations were further corroborated by variable temperature NMR studies done by our friends at Bristol-Myers Squibb (thanks, Kevin and Paul!). Unsurprisingly, Kevin was able to identify hydrogen-bonding networks in the four natural products synthesized, which likely contribute to the facility of cyclization in these cases. In contrast, the other substrates investigated showed few, if any, of these interactions. While we cannot claim that conformational bias plays no role, we have determined that it is not a strict prerequisite.

A second, but related, matter raised by the referees was the origin of selectivity in sequences that possess an internal lysine residue (i.e., cyclization at the terminal amine versus the lysine ε-NH2). In our original report, we had included only one example with an internal lysine, which—as pointed out by one of the reviewers—was an consideration that should be further explored. We agreed, and have now expanded our study to include six examples, one of which was generated at random, and are confident that reaction conditions, regardless of protocol, can be successfully tuned to favor terminal amine cyclization. Once again, VT-NMR studies suggest no internal bonding networks in the lysine-containing peptides analyzed, indicating that conformational bias is not the sole contributing factor.  

We are immensely grateful for the insightful criticism received from our first submission—the suggestions provided helped guide our revision process, and served as a prime example of peer-review at its finest. We now present a thorough, and much expanded, study on our approach to thermodynamic peptide macrocyclization. Enjoy!

Justine

4 comments:

  1. By the way, you can take TentaGel-OH (PS-PEG-OH) and attach FmocNHCH2CH(OMe)2 by acetal exchnge with py.TsOH under azeotropic conditions, or use BrCH2CH(OEt)2, make the acetal resin and do halogen displacement with a primary amine. Aldehyde is revealed upon cleaving with formic acid, to close rings to provide polycyclic peptidomimetics. We did this already in mid 90s, and there was extensive parallel synthesis/combichem work done on this at Aventis since then [J. Org. Chem. 63(10) 3162-3 (1998) + lots of patents from Patek et al]

    ReplyDelete
  2. Thanks for the comment, milkshake! Pátek has certainly made some important contributions to the field of solid-phase heterocycle synthesis, much of which factored into our initial literature survey. Given the structure of our targets (and those of BMS), however, the employment of a hot acyliminium intermediate wasn’t a practical option in our head-to-tail peptide macrocyclizations (though we continue to brainstorm possible methods for "traceless" imine activation, as this would allow us to explore additional trapping agents!).

    Likewise, the acetal exchange is a clever approach to the formation of aldehydes on resin cleavage! As mentioned in the post, when we began the project, we were fairly limited in resources and had only Rink amide resin on hand; as such, our early investigations began with the attachment of aminoacetaldehyde dimethyl acetal through a Glu side-chain, providing access to the glycinal series. Installing the masked aldehyde at the end of the synthesis also provided a nice point of divergence (the thioester S19 we used in our NCL reaction―see scheme 3B―was made by installing a preformed glycine thioester onto the same Glu side-chain acid!) When we expanded the method to more elaborate substrates, we needed a way to access peptides with C-terminal α-chirality―this led us to use preformed Fmoc-amino aldehydes in combination with TG-functionalized resin to form an oxazolidine linkage. Upon resin cleavage and global deprotection, peptide aldehydes could be directly employed in a variety of transformations, including the synthesis of a few non-ribosomal imine natural products. Thanks again for taking the time to read and respond!

    ReplyDelete
  3. Some aa-CHO pre-loaded resins are commercially available using the same oxazolidine linkage (from Anaspec if I am correct). Is there any particular reason you did not use these?

    ReplyDelete
  4. Thanks for taking the time to read and post, MN! We did consider using pre-loaded AA–CHO resins initially, but opted instead to build our own variant for a couple of reasons. First, the use of a standard Rink linkage allowed for immediate divergence: we were able to use the same resin to build both the glycinal and the α-amino aldehyde series, and had the option to install any AA–CHO at the C-terminal, including those with non-natural α-stereocenters. Next, while some suppliers (e.g., Anaspec) do have a wide range of pre-loaded amino acids available for purchase, a quick comparison revealed that the cost of buying each individual pre-loaded resin would far surpass that of a large-ish (25 g) bottle of the standard Rink. Having said that, however, there is no reason to believe that the pre-loaded would not work just as well when applied to our chemistry! Thanks again!

    ReplyDelete