Published as part of the Cluster Iterative Synthesis
Key words
ring expansion - ring enlargement - rearrangements - macrocycles - medium-sized rings
- iterative synthesis
1
Introduction
Medium-sized rings and macrocycles are important in a wide array of scientific fields
and technologies,[1]
[2] propagating the development of several methods for their synthesis. Most of these
synthetic methods are based on the end-to-end cyclization of a linear precursor, which
is often a challenging process;[3] achieving efficient cyclization whilst avoiding unwanted intermolecular reactions
such as dimerization or polymerization is usually the main problem to overcome. Various
innovative strategies to help favor cyclization have emerged over the years, including
the use of high-dilution conditions,[4] pseudo-high-dilution conditions,[5] templated systems,[6] and the incorporation of internal structural features to help bias conformation
towards cyclization,[7] amongst others.[8]
All these methods are based on improving the efficiency of an often-difficult long-chain
end-to-end cyclization step. Ring-expansion approaches are different as they allow
this difficult step to be avoided completely.[9]
[10] Consequently, several creative ring-expansion processes have been developed and
used to facilitate the synthesis of medium-sized rings and macrocycles, with this
topic reviewed by our group in 2017.[9b] Synthetic methods where ring-expansion reactions have been applied consecutively are far less common. Nonetheless, those that have been reported clearly demonstrate
the high potential of such strategies for the iterative construction of large ring
molecules.
The current state-of-the-art in the application of consecutive ring-expansion strategies
is discussed in this Account. Our intention is to highlight the power of ‘growing’
large ring systems via the sequential or iterative insertion of smaller linear fragments.
We hope that bringing together the various strategies described herein will help to
propagate the use of iterative ring-expansion processes in synthetic chemistry,[11] and in turn will enable functional large ring molecules to be designed with greater
freedom and ambition than is possible using traditional cyclization approaches. The
procedures are grouped into three main categories of ring expansion based on ‘insertion’,
‘pericyclic’, or ‘fragmentation’ reactions, and we have decided to predominantly focus
on methods where the ring size is increased by two or more atoms in each iteration.
As this Account is focused on fully controlled ring-expansion approaches, methods
for the synthesis of cyclic polymers via ring-expansion polymerization have not been
included.[12]
2
Insertion Reactions
The most common class of ring-expansion reaction used as part of a consecutive ring-enlargement
sequence are side-chain insertion reactions. Within this subcategory, transamidation- and transesterification-type processes appear most frequently.
2.1
Transamidation/Transpeptidation
Hesse (unquestionably one of the all-time greats in the ring expansion field) and
co-workers[13] pioneered a series of ring expansions coined ‘zip reactions’ that enable polyamine-based
macrocycles to be prepared via innovative cascade processes. A prototypical example
of this class of reaction is illustrated in Scheme [1]; thus, sodiated laurolactam 1 can be alkylated with acrylonitrile, reduced to form amine 2, and then alkylated and reduced again in the same way to form diamine 3a. The treatment of 3a under strongly basic conditions (using KAPA = potassium 3-aminopropylamide) at reflux
then sets up an equilibrium in which ring expansion (3a → 3b) and ring expansion again (3b → 4) is achieved via two sequential acyl transfer reactions. Even more impressively,
there is no need to stop at just two expansions, provided the requisite starting material
can be prepared; for example, 33-membered polyamine lactam 5 was synthesized in high yield using the same approach (this time via five sequential
ring-expansion reactions).
Scheme 1 ‘Zip’ reactions developed by Hesse et al.
The Hesse group went on to establish a significant body of work in this area,[9a] with projects based on both extending the synthetic methodology and in natural product
target synthesis. Arguably, the most striking example was their synthesis of 54-membered
macrocycle 7 in 38% yield from a 13-membered lactam polyamine derivative 6 (Scheme [2]).[14]
Scheme 2 ‘Zip’ reaction to prepare 54-membered polyamine macrocycle 7
The ‘zip reactions’ described above all start from macrocyclic precursors. Ring expansion
of smaller rings (e.g., 5–7-membered rings) can often be more challenging in comparison, especially when
the reactions are under thermodynamic control and the desired products are medium-sized
rings, as these ring systems often suffer from destabilizing transannular interactions.
In some cases, the problems associated with the instability of medium-sized ring products
can be overcome by performing consecutive ring-expansion reactions in one-pot. An
instructive example concerns the double ring expansion of 6-membered ring barbiturate
8a into 14-membered product 9. In this case, the intermediate 10-membered ring 8b is unstable with respect to ring contraction back to its 6-membered precursor, however,
as a second ring expansion is possible, an equilibrium is established in which 14-membered
ring product 9 is formed, albeit in modest yield (Scheme [3]). Consecutive ring-expansion cascade sequences like this, enable ring-enlarged products
to be formed via unstable intermediates such as medium-sized ring 8b that would be very challenging to access and isolate directly.[9a]
Scheme 3 Reversible consecutive 6- to 10- to 14-membered ring expansion/contraction
Scheme 4 Synthesis of 21-membered lactam 18 including two reversible transamidation ring-expansion reactions
An early example of similar reactivity being used during a natural product synthesis
can be found in the synthesis of desoxo-indandenine,[15] a member of the spermidine family of natural products (Scheme [4]). Starting from 13-membered cyclic ketone 10, α-nitration was achieved via an enol acetate intermediate, and this followed by
a conjugate addition reaction of 11 into acryladehyde. Reductive amination of the resulting aldehyde 12 with partially protected triamine then afforded ketone 13, which was found to exist in equilibrium with ring-expanded amide 14 under the conditions used for its formation. Switching to mildly basic, aqueous conditions
was sufficient to drive the equilibrium towards the formation of the desired ring-expanded
product 14, which was isolated in 55% yield. Next, hydrolysis and a three-step reductive Nef
reaction sequence removed the nitro group to form lactam 16. Finally, electrolysis was used to remove the tosyl groups, and treatment of the
resulting diamine with p-toluenesufonic acid (PTSA) at reflux in xylene produced a 1:1 mixture of 17- and
21-membered lactams 17 and 18. This sequence demonstrates the power of transamidation reactions for the construction
of complex aza-macrocyclic scaffolds. However, it also highlights that the reversibility
of the transamidation rearrangement can be a problem, and this must be considered
when designing reactions of this type; indeed, similar problems with the reversibility
of the ring-expansion steps were also found in Hesse’s work.[9a]
Independently, Takahashi and co-workers[16] published a similar transamidation approach during their synthesis of macrocyclic
tetra-amines structurally related to some polyamine alkaloids, and also showed that
these products are efficient Fe(II) binders in aqueous solutions. Thus, coumarin 19 was reacted with tetra-amine 20 in methanol at reflux for two weeks, and macrocyclic product 21d was produced in 20% yield (Scheme [5]). The reaction is proposed to operate via an initial 1,4-conjugate addition, followed
by an intramolecular amidation and one-atom ring expansion to liberate the tethered
phenol (19 → 21a → 21b). A series of transamidation reactions then takes place (presumably under thermodynamic
control) to relieve the strain associated with the medium-sized ring. Although not
stated in this manuscript, it is possible that some of the smaller ring systems invoked
in this presumed equilibrium were also formed in this reaction, which could account
for some of the mass balance in this highly impressive, but modest yielding cascade
reaction.
Scheme 5 Cascade ring-expansion reactions for the conversion of coumarin 19 into 14-membered lactam 21d
2.2
Transesterification
There is less precedent for transesterification reactions (or translactonizations)
being performed consecutively in macrocycle synthesis, which perhaps reflects the
relative rarity of polylactones in nature compared with polyamides (i.e., cyclic peptides
are relatively common). Nonetheless, Corey and Nicolaou[17] did show that reversible transesterifications processes can be performed, provided
that there is a thermodynamic preference for accessing a particular ring size relative
to the starting materials. Thus, 15-membered lactone 27 was produced via two intramolecular transesterifications as shown in Scheme [6]. In this work, lactonic acid 22 was converted into the corresponding pyridine thioester 23 and was reacted with Grignard reagent to produce keto-lactone 24. Reduction with sodium borohydride in ethanol, followed by desilyation with TBAF
in THF gave dihydroxylactone 26a in 90% yield. Ring expansion was then promoted using catalytic PTSA to produce 15-membered
lactone 27 in an impressive 90% yield, as a 1:1 mixture of diastereoisomers (with the mixture
of diastereoisomers arising during the sodium borohydride reduction step).
Scheme 6 A double translactonization approach to lactone 27
2.3
Transthioesterification
Inteins are sections of proteins which have the ability to excise themselves from
the peptide sequence in a process called splicing. Intein splicing operates through
a series of acyl shifts which is commonly facilitated by a cysteine residue which
can induce an N,S-acyl shift to break the peptide bond. Tam and co-workers[18] have developed a clever ‘thia-zip reaction’ to produce macrocyclic peptides via
a ring-expansion cascade that bears striking similarity to the biological approach
used by inteins. Thus, starting from a cysteine-rich linear peptide, an initial activation
step using an external thiol produces thiolactone (represented as 28, Scheme [7]). The cysteine thiol residues then promote a series of reversible thiolactone exchange
reactions as drawn, which operate at physiological pH. The equilibrium is driven towards
the ring-expanded product 29d via a final irreversible S- to N-acyl migration (29d → 30) of an N-terminal cysteine residue, thus completing this ingenious multiple ring-expansion
cascade sequence.
Scheme 7 A ‘thia-zip’ transthioesterification cascade process
2.4
Aminyl Radical Cascade
Pattenden and Schulz[19] serendipitously discovered a double radical ring-expansion cascade (Scheme [8]) starting from acetylene oxime 31 using (TMS)3SiH and radical initiator AIBN. It was proposed that this process is initiated by
radical silylation at the terminal alkyne and that the resulting vinyl radical 32a cyclizes to form aminyl radical 32b. Fragmentation of the fused bicycle to form carbon-centered radical 32c is then followed by transannulation to the more stabilized α-silyl radical. Cyclopropanation
via a 3-exo-trig cyclization then gave 32d, which quickly ring-expanded with elimination of its silyl group, furnishing bicyclic
oxime product 33. Hydrolysis of the oxime was then shown to deliver the enone in good yield.
Scheme 8 Radical cascade conversion of 4-membered cyclic oxime 31 into bicyclic cyclic oxime 33
2.5
Iterative Synthesis of Lactones
Seyden-Penne, Rousseau, and Fouque[20] developed an iterative procedure where lactones can be ring-enlarged via iterative
one-carbon ring-expansion reactions through the clever use of carbenes. This is exemplified
by the iterative expansion of caprolactone 34 (Scheme [9]). First, the lactone is treated with LDA and then TMSCl to form silyl enol ether
35. Treating 35 with pre-formed ethyl carbenoid (itself made with dichloroethane and butyllithium)
then yields cyclopropane 36, which upon heating at reflux in toluene undergoes desilylation and ring expansion
(36 → 37). Olefin hydrogenolysis is then required before another ring-enlargement iteration
can be performed (37 → 38). This then gives an 8-membered lactone that itself can be converted into a silyl
enol ether and re-enter this ring-expansion sequence, allowing subsequent iterations
of the same process to be performed (either with the same or different chlorocarbene
reagents), thus enabling 9- and 10-membered lactones 39 and 40 to be synthesized in the same way. Note that while we set out to only include ring-expansion
processes where the ring is enlarged by two or more atoms, we feel that the elegance
and simplicity of this iterative process justified its inclusion in this Account as
a notable exception.
Scheme 9 Iterative one-carbon ring expansion of lactones using carbenes
2.6
Successive Ring Expansion of β-Ketoesters and Lactams
In our laboratory, we have developed a system based on the successive ring expansion
of cyclic β-ketoesters that allows 3- or 4-atom amino acid or hydroxy acid fragments
to be inserted into ring-enlarged products iteratively.[21] This process is exemplified by an iterative triple ring-expansion reaction sequence,
starting from 12-membered β-ketoester 41 (Scheme [10, a]). Thus, the starting β-ketoester 41 was treated with amino acid chloride 42 in the presence of magnesium(II)chloride and pyridine to promote C-acylation and form tri-carbonyl species 43. Cleavage of the Fmoc protecting group (43 → 44a) then initiates spontaneous cyclization (44a → 44b) and ring expansion (44b → 45) in high yield over the acylation/deprotection/rearrangement sequence. As the 16-membered
product 45 is another cyclic β-ketoester, it is itself a suitable substrate for further ring
expansion and can be subjected to the same sequence to produce 20-membered and 24-membered
macrocyclic products 46 and 47 via two further iterations. There is considerable freedom to vary the ring size of
the starting material and use both α- and β-amino acids. Indeed, by varying the linear
fragment, it is possible to prepare ring-expanded β-ketoester by the insertion of
different functional groups in sequence (e.g., macrocycles 48 and 49, Scheme [10, b]). In a later study, the same chemistry was used to create a library of leadlike
medium-sized ring scaffolds for inclusion in a high-throughput-screening compound
collection.[22]
Scheme 10 Successive ring-expansion (SURE) reactions of cyclic β-ketoesters with amino acid
derivatives
The same idea can also be applied to hydroxy acid derived linear fragments, allowing
ring-expanded lactones to be prepared in a similar way; in these reactions, the protected
amine is replaced by a benzyl-protected alcohol that is revealed by hydrogenolysis
to initiate ring expansion. Indeed, both methods can be readily combined to allow
mixed lactam/lactone products (e.g., 53–55, Scheme [11]) to be prepared.
We more recently extended this methodology to change the β-ketoester moiety to a simple
lactam functionality.[23] Using 13-membered amide 56, N-acylation was achieved using modified thermodynamic conditions with amino acid chloride
57. The amine moiety of imide 58 was then revealed using DBU, and under these conditions the amine that was formed
spontaneously rearranged (59a → 59b → 60) to furnish 17-membered ring-expanded product 60 in excellent yield. This product can then undergo another two iterations using the
same chemistry to afford both 21- and 25-membered macrocycles (61 and 62, respectively, Scheme [12]). A broad array of amino acids and peptoid monomers are compatible with this method,
which also works well successively, enabling the synthesis of a range of tri-peptide
mimetics. Note that a conceptually related ring-expansion method to this one was independently
developed and published around the same time by Yudin and co-workers.[10c]
Scheme 11 Successive ring-expansion (SURE) reactions of cyclic β-ketoesters with hydroxy acid
derivatives
Scheme 12 Successive ring-expansion (SuRE) reactions of lactams with amino acid derivatives
The most recent extension to this chemistry involves the incorporation of hydroxy
acids into lactam starting materials, allowing ring-expanded lactones to be made via
a similar strategy.[24] For example, N-acylation of 8-membered lactam 63 with hydroxy acid chloride 64 can be achieved using the same acylation conditions as before, and hydrogenolysis
can then be used to cleave the benzyl protecting group. Unlike the analogous amino
acid chemistry, spontaneous ring expansion did not ensue in this system, instead an
equilibrium was established between imide 66a, fused bicyclic intermediate 66b, and ring-expanded product 67, but pleasingly, a solvent switch to chloroform and the addition of triethylamine
was successful in driving the equilibrium towards the desired ring-expanded lactone
67, which was isolated in high yield (Scheme [13, a]). A range of both α- and β-hydroxy acids are well tolerated and crucially, this
method and our earlier amino acid ring-expansion reactions are fully compatible, meaning
that mixed lactam/lactone macrocycles can be prepared via the iterative insertion
of either functionality; selected examples of scaffolds prepared in this way are highlighted
in Scheme [13] (b). The iterations can also be telescoped into an ‘assembly line’ type process,
in which chromatography is not performed until after the final iteration; for example,
8-membered lactam 63 was taken through three iterations of the typical hydroxy acid ring-expansion procedure
to form macrocycle 73 in 48% yield over the entire sequence (Scheme [13, c]).
Scheme 13 Successive ring-expansion (SuRE) reactions of lactams with hydroxy acid derivatives
The SuRE methods described above are believed to be under thermodynamic control, and
ring size has been shown to play a key role in their outcomes. For example, for the
hydroxy acid chemistry described in Scheme [13], only lactams of 8-members or above undergo ring expansion using α-hydroxy acids,
whereas for β-hydroxy acids, lactams of 6-members or above work well. A relatively
simple density functional theory (DFT) study has been performed on these reactions
in which it was found that when considering the three isomeric species (e.g., 66a, 66b, and 67) presumed to be in equilibrium in these reactions, there is strong correlation between
the reaction outcome and their DFT-calculated ground-state energies. For example,
for the 5-membered system (which did not ring-expand) imide 74a was calculated to be significantly lower in energy than either isomer 74b or 74c, whereas for the successful 8-membered example, the ring-expanded product 75c was calculated to be the lowest in energy and was indeed isolated in good yield in
the synthetic reaction (Scheme [14]). This method used a B3LYP/6-31G* DFT method; a more rigorous and detailed DFT study
on these and related reactions is currently ongoing and will be reported in due course.
Scheme 14 DFT study of SuRE reactions
3
Pericyclic Reactions
There has been a significant amount of work done on sequential sigmatropic rearrangements.
These rearrangements typically require a strong thermodynamic driving force to avoid
the formation of multiple products in equilibrium and have often been used in natural
product synthesis.
3.1
Sulfur-Mediated Rearrangements
Sulfur-mediated ring-expansion reactions are often based on ylide reactivity, which
can be advantageous as charged pericyclic rearrangements often proceed faster than
their neutral counterparts. This is exemplified by some impressive work from Vedejs
and Reid,[25] who used a 2,3-sigmatropic rearrangement as part of their synthesis of the carbocyclic
cytochalasin natural products, to construct a key 11-membered ring (Scheme [15]). Thus, iodide 76 was heated in acetonitrile/K2CO3 to form sulfonium ylide 77, which spontaneously underwent a 2,3-sigmatropic rearrangement to form 78. Then, a second ring-expansion reaction was performed via methylation of sulfide
78 using Meerwein salt followed by treatment with Zn/acetic acid to cleave one of the
C–S bonds and furnish ring-expanded product 80, which was converted into the final target molecule 81 via three additional steps.
Scheme 15 Consecutive sulfur-mediated ring-expansion reactions in the total synthesis of carbocyclic
cytochalasin natural product 81
Vedejs and co-workers were the first group to develop the idea of using consecutive
2,3-sigmatropic rearrangements (Scheme [16]) in an iterative sense,[26] exemplified by a sequence starting from α-vinyl cyclic sulfide 82. This simple 5-membered ring starting material was first converted into sulfide salt
83 upon treatment with a diazocarbonyl reagent. Then, upon addition of base, a stabilized
ylide 84 was formed, which was found to undergo spontaneous 2,3-sigmatropic rearrangement
in situ to afford ring-expanded cyclic olefin 85. A Wittig reaction was then performed to convert the ketone side chain of 85 into an olefin, thus furnishing new α-vinyl cyclic sulfide 86 primed to enter the same ring-expansion sequence. A different diazo carbonyl species
and reaction conditions were used for the second iteration, but the chemistry proceeds
in broadly the same way; thus, α-vinyl cyclic sulfide 86 was shown to react with diazomalonate in the presence of copper at 100 °C, to form
11-membered ring product 88 via the same sequence of S-alkylation, ylide formation, and 2,3-sigmatropic rearrangement.
Scheme 16 Iterative 2,3-sigmatropic rearrangements of cyclic vinyl sulfides with diazocarbonyl
reagents
Using a similar strategy, Schmid and Schmid[27] developed an allylation-based iterative cyclic sulfide expansion (Scheme [17]). Starting from the same α-vinyl cyclic sulfide 82 used in the Vedejs study, it was shown that S-allylation can be performed by reacting with allyl bromide under acidic conditions
and that subsequent treatment with aqueous potassium hydroxide promoted ylide formation
(89 → 90) and 2,3-sigmatropic rearrangement (90 → 91) to deliver the ring-expanded product 91. An advantage of this method compared with the previous Vedejs work is that no further
manipulation of the product is required to complete additional iterations, and thus
it was shown that 8-membered cyclic sulfide 91 could be further expanded into 11- and 14-membered derivatives 94 and 95 using the same method.
Scheme 17 Iterative 2,3-sigmatropic rearrangements of cyclic vinyl sulfides via S-allylation
Vedejs and co-workers then went on to extend their previous work,[28] and during these studies an unwanted competing 2,3-sigmatropic rearrangement was
observed (shown in Scheme [18] to form 98) originating from the formation of endocyclic ylide 97.[28a] It is stated that the undesired endocyclic ylide forms only when making medium-sized
rings, which is likely to be a consequence of the additional thermodynamic difficulties
associated with making medium-sized rings (e.g., transannular strain in the products and reaction transition states). This highlights
the importance of achieving the correct chemo- and regioselectivity when planning
iterative ring-expansion processes and also the major impact that the size of the
rings involved can have on the outcomes of ring-expansion reactions.
Scheme 18 Desired ring expansion and unwanted allyl transfer pathways
3.2
Nitrogen-Mediated Rearrangements
An interesting double 3,3-sigmatropic rearrangement approach was reported by Back
and co-workers[29] as part of a natural product synthesis (Scheme [19]). The sequence started from 2-vinylpyrrolidine 99, which was reacted with acetylenic sulfone to form a zwitterionic species 100, which then underwent 3,3-aza-cope rearrangement in situ to produce ring-expanded enamine product 101. The authors then realized that this aza-Cope rearrangement could be performed for
a second time if a similarly placed vinyl group could be introduced into this product.
To do this, olefin hydrogenation was followed by treatment with triflic acid and the
addition of a vinyl Grignard reagent to afford 103. Removal of the tosyl group was not required to perform the next ring-expansion iteration,
but it was needed for their motuporamine natural product target and was performed
using a sodium-mercury amalgam to form 104. This system was then set up to undergo the same conjugate addition/aza-Cope rearrangement
sequence that had previously been performed on pyrrolidine 99, and it was successful in enlarging the ring further to furnish 13-membered amine
106. An additional five steps were then needed to complete the total synthesis of motuporamine
A.
Scheme 19 Consecutive 3,3-aza-Cope rearrangement ring-expansion reactions
Suh and co-workers utilized an intriguing aza-Claisen rearrangement (ACR) during their
total syntheses of fluvirucinines A1and A2,[30]
[31] in which two ring-expansion rearrangements were performed with excellent control
of the stereogenic centers formed in construction of the 14-membered skeleton (Scheme
[20]). The sequence started from piperidinone 108, which was synthesized by an asymmetric Evans alkylation and stereoselective vinylation
as described previously by the same group. Then, treatment with LiHMDS formed enolate
109 which rearranged into lactam 110 as a single stereoisomer, via a 3,3-sigmotropic rearrangement. The fact that a single
product was formed suggests that the reaction proceeds via Z-enolate 109 as shown. Olefin hydrogenation, amine protection, and partial lactam reduction/silylation
then yielded N,O-acetal 113. Then, treatment with Lewis acid and an allyl tin reagent promoted a highly stereoselective
amidoalkylation and concomitant deprotection to furnish 114. Re-protection of the amine was followed by oxidative cleavage of the olefin with
osmium tetraoxide and sodium periodate, and then silylation of the resulting aldehyde
produced the required (E)-enol ether 116 for the second ACR. Finally, N-acylation with activated trans-pentenoic acid delivered compound 117 in which all the functionality needed to perform the second ACR was in place. LiHMDS
was again used to form amide-enolate and initiate the second iterative ACR, which
was successful in forming 14-membered ring 119 with >10:1 diastereoselectivity. Further manipulation of 119 was performed to produce the final target molecules fluvirucinine A1 and A2 (not shown).[30]
[31]
Scheme 20 Consecutive ACR reactions used in the total synthesis of fluvirucinine natural products
Later on, Suh and co-workers streamlined their synthesis of fluvirucinine A2.[31] These second-generation ACR exhibited significant kinetic improvements, and they
also demonstrated the importance of the geometry of the enol ether in the ACR precursor
(Scheme [21]). A chair transition state is postulated to account for the large degree of stereocontrol.
Further study of these enol ether precursors was completed and was well described
with more details of such aza-rearrangements in a focus review.[32]
Scheme 21 Stereochemical outcomes in ACR reactions
4
Fragmentation Reactions
Dowd and Zhang[33] reported a double ring-expansion process which involves both side-chain insertion
and fragmentation (Scheme [22]). In this study, cyclic silyl enol ether 123 and ketene 124 were reacted via a [2+2] cycloaddition to generate cyclobutanone 125, with this reaction proceeding with excellent endo/exo selectivity when cooled to –20 °C. Classical AIBN and Bu3SnH conditions were then used to form the tributyltin radical which abstracted the
pendant bromide group to generate primary radical 126a, which cyclized and fragmented (126a → 126b → 127) to complete the first ring-expansion reaction. Release of ring strain in the cyclobutyl
ring is presumably an important driving force in this reaction. In the same pot, radical
de-chlorination (promoted by the tributyltin radical) also took place to furnish reduced
compound 129. Next, lithium aluminum hydride reduction of the ketone, followed by mesylation and
desilylation afforded compound 131 and set up the second ring-expansion reaction, through a base-mediated Grob fragmentation
to form 11-membered ketone 132 with cis-olefin geometry in excellent yield. A similar strategy was also used to generate
homologous 12-membered ketone 134 as a mixture of geometrical isomers, with a base-mediated Grob fragmentation again
being a key step (see box in Scheme [22]).
Scheme 22 A consecutive radical and Grob ring-expansion sequence
Thommen and co-workers[34] accessed 15-membered ketones via consecutive Grob fragmentations (Scheme [23]), starting from tricyclic diol 135, which was itself prepared via a trimolecular aldol-type process. First, diastereoselective
reduction of the ketone group of 135 to cis-triol 136 was achieved selectively using Red-Al®. This was followed by tosylation using n-butyllithium, and the first Grob fragmentation was performed upon treatment with
potassium tert-butoxide to furnish 138 as the major product. Lithium aluminum hydride reduction of 138 followed by tosylation of the secondary alcohol to form 140 then set up a second Grob fragmentation, which again was promoted by potassium tert-butoxide, to afforded 15-membered ring-expanded dienone 141.
Scheme 23 A double Grob fragmentation sequence
Another impressive example of the use of sequential ring-expansion reactions involving
Grob-type fragmentation was reported by Ikeda and co-workers.[35] In this work, a traditional Grob-type fragmentation (144 → 145) was followed by an oxidative fragmentation (145 → 146), allowing an advanced precursor to the natural product 147, (±)-phoracantholide M to be formed. The key cyclobutane intermediate 143 was itself formed via an elegant [2+2] cycloaddition strategy (Scheme [24]).
Scheme 24 A sequential Grob/oxidative fragmentation sequence
Maio and co-workers[36] have reported that sequential ring expansions can be used to form medium-sized lactones,
including the synthesis of natural product (–)-phoracantholide J in six linear steps
and an overall 26% yield (Scheme [25]). Thus, silyl enol ether 148 was treated with methyllithium to form the corresponding enolate and then reacted
with epoxide in the presence of boron trifluoride to form hemiketal 149. Then, oxidative fragmentation was achieved using diacetoxyiodobenzene and iodine,
to form geometrically pure cis-olefinic 8-membered lactone 150. Treatment with TBAF revealed primary alcohol 151 and this set up the second ring-expansion reaction, which occurred via a spontaneous
translactonization to form hydroxy lactone 152; the thermodynamically favorable change in ring size (8- to 10-membered) likely facilitated
this facile ring expansion. Radical deoxygenation was then used to complete the synthesis
of (–)-phoracantholide J (not shown). The synthesis was performed on gram scale, demonstrating
the scalability of this impressive reaction sequence.
Scheme 25 Sequential ring expansions during the total synthesis of (–)-phoracantholide J
Murai and co-workers[37] have reported an interesting successive ring-expansion reaction using bromo-diepoxides,
promoted by AgOTf. As a note, this chemistry is different to the other examples described
in this Account, in that the consecutive ring-expansion reactions do not take place
on the same ring (meaning that it is not suitable for ‘growing’ medium-sized rings
and macrocycles), but we decided to include it nonetheless as an example of a powerful
ring-expansion cascade system. Thus, in a prototypical example of this method, epoxide
153 is activated by Ag(I) and undergoes 5-exo-tet cyclization. The resulting strained bicyclic epoxonium intermediate 154 then undergoes intramolecular cyclization from the second epoxide to form tetrahydropyran
155. An intermolecular nucleophilic attack then takes place, which leads to relief of
ring strain and the formation of fused tetrahydropyran structure 156 (Scheme [26]). Extensive studies have continued utilizing polyepoxide precursors and ring-expansion
cascades to form multiple fused rings in this way, with this topic covered in a detailed
review article.[38]
Scheme 26 Successive ring-expansion cascade of bromo-diepoxide 153
Finally, Clayden and co-workers recently reported a method to access medium-sized
rings via a three-atom ring expansion of metalated ureas via an insertion-type ring
expansion, which was followed by a second ring expansion via an acid-catalyzed fragmentation
reaction.[39] Thus, treatment of indoline 157 with triphosgene furnished urea 158, which was then lithiated with LDA and DMPU (1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone), promoting a migratory ring expansion to form ring-expanded urea 160.[39a] The reaction is driven by an increase in anion stability in migrating the negative
charge from a benzylic carbanion position to a deprotonated urea group. DMPU is critical
to the success of this process as it suppresses a competing 1,2-acyl shift reaction.
The transformation is stereospecific and proceeds with retention of configuration
via a conformationally stable organolithium intermediate; a concerted associative
mechanism is proposed, rather than a stepwise SNAr-type process. The same group later went on to show that cyclic ureas of the form
160 undergo facile ring-contraction reactions to form 1-aryl tetrahydroisoquinolines
and tetrahydrobenzazepines (not shown) and in a single case, also showed that urea
160 could be ring-expanded for a second time following treatment with PTSA, to form 11-membered
162 via the fragmentation mechanism shown in Scheme [27].[39b]
Scheme 27 Insertion/fragmentation double ring expansion of meso-urea 158
Conclusions and Future Outlook
5
Conclusions and Future Outlook
The importance of large ring molecules in a wide range of important applications means
that the continued development of practical and scalable methods for their synthesis
will always be of high value.[1]
[2] Ring-expansion reactions have already proven themselves to be very useful in this
regard, especially as they typically do not require the impractical reaction conditions,
such as high dilution, that are often needed in typical end-to-end cyclization reactions.
Although considerably less well developed than single ring-expansion processes, this
review highlights the synthetic benefits of applying ring-expansion reactions consecutively,
with several of the examples being used either to generate natural products or scaffolds
structurally related to bioactive macrocycles, for example, cyclic peptide mimics.
One of the major trends highlighted in this Account is that expanding a ‘normal’-sized
ring (5–7-membered) into a medium-sized ring (8–11-membered) is often more challenging
than other ring-expansion processes in view of the relative instability of medium-sized
rings. Conversely, the expansion of medium-sized to macrocycles is often much easier,
as in this scenario moving away from the medium-sized scaffold to a more flexible,
less strained macrocycle can bring thermodynamic advantages. In the context of consecutive
ring expansions, especially starting from normal (5–7-membered) ring sizes, this can
mean that while the first ‘normal to medium’ transformation can be challenging, once
this barrier has been surpassed, other ring-expansion iterations will often become
easier, which is useful to consider when designing consecutive sequences.
Arguably, the strategies that can be performed iteratively and with little or no extra
transformations between the ring-expansion steps are the most useful and most likely
to be widely adopted, especially if these methods allow for the versatile introduction
of different functional groups in sequence. This Account also highlights several synthetic
strategies in which two different classes of ring expansion have been performed in sequence to good effect. An interesting
avenue that may be explored in the future is to design starting materials containing
functionality compatible with two (or more) different ring-expansion methods, especially
if each method can be performed orthogonally to the other in two directions.
